Datasets:

ccdv
/

arxiv-summarization

Dataset card Data Studio Files Files and versions Community

Subset (2)

section · 216k rows

Split (3)

train · 203k rows

article stringlengths 0 745k	abstract stringlengths 0 92.7k
given a partition @xmath0 of @xmath1 , let @xmath2 denote the associated irreducible complex representation of the symmetric group @xmath3 . the important * kronecker coefficients * @xmath4 are the tensor product multiplicities : @xmath5 one can attempt to understand these coefficients by studying their limiting behavior , in various senses . an important result in this direction is murnaghan s observation ( conjectured by murnaghan in @xcite and proved by littlewood in @xcite ) : @xmath6 is constant for @xmath7 . in @xcite , stembridge proposes a vast generalization of this result , centered on the following concept : a triple @xmath8 of partitions with @xmath9 and @xmath10 is * stable * if , for any other triple of partitions @xmath11 with @xmath12 , the kronecker coefficient @xmath13 is constant for @xmath7 . with this language , murnaghan s theorem is the statement that the triple @xmath14 is stable . stembridge s proposed generalization of murnaghan s theorem is the following conjecture ( ( * ? ? ? * conj 4.3 ) ) : [ mainconj ] a triple @xmath8 is stable if and only if @xmath15 for @xmath16 . murnaghan s theorem is an easy corollary of this conjecture since @xmath17 for all @xmath18 : @xmath19 is the trivial representation of @xmath20 , so @xmath21 . stembridge proves the `` only if '' direction of conjecture [ mainconj ] . this paper proves the reverse direction , thus establishing the conjecture in full . in fact , we also reprove the `` only if '' direction , see [ sec : alg ] . it may be worth emphasizing that our proof only uses schur weyl duality and the borel weil theorem and does not rely on any existing work on kronecker coefficients . for partitions @xmath22 of @xmath1 , let @xmath23 be the @xmath24-multiplicity space in @xmath25 , i.e. , @xmath26 this is a vector space whose dimension is the kronecker coefficient @xmath27 . we will prove conjecture [ mainconj ] by constructing some extra structure on these multiplicity spaces . let @xmath28 , @xmath29 , @xmath30 , @xmath0 , @xmath31 , @xmath32 be partitions such that @xmath9 and @xmath12 . define @xmath33 our main result is : [ mainthm ] the space @xmath34 has the structure of a finitely generated graded integral domain , and @xmath35 has the structure of a finitely generated torsion - free graded @xmath34-module . [ maincor ] conjecture [ mainconj ] is true . suppose @xmath36 for all @xmath16 . then @xmath37 is isomorphic to @xmath38 $ ] , where @xmath39 has degree one . it now follows from the structure theorem for finitely generated @xmath38$]-modules that @xmath40 is isomorphic to @xmath41 $ ] for some @xmath42 , where @xmath43 $ ] denotes a shift in grading . we thus see that @xmath44 for @xmath45 . suppose @xmath46 , and let @xmath47 be the krull dimension of @xmath37 . then @xmath48 for a constant @xmath49 ( depending on @xmath50 ) . more precisely , @xmath51 is a quasi - polynomial of degree @xmath52 with constant leading term for @xmath7 . let @xmath53 . the hilbert series @xmath54 has a pole of order @xmath47 at @xmath55 , by standard properties of krull dimension . let @xmath56 be a non - zero degree one element of @xmath34 . since @xmath57 is torsion free , @xmath58 . as @xmath59 has krull dimension @xmath52 , all poles of @xmath60 have order @xmath61 . thus @xmath54 has a pole of order @xmath47 at @xmath55 and all other poles have order @xmath61 ( and are at roots of unity ) . the corollary now follows from ( * ? ? ? * theorem 4.1.1(iii ) ) , and the fact that the @xmath18th coefficient of @xmath54 is @xmath51 . we will see in [ sec : alg ] that the ring @xmath34 is normal and has rational singularities . we omitted this from the main result for simplicity and because it is not strictly needed for the application to conjecture [ mainconj ] . however , normality can be used to prove the `` only if '' direction of conjecture [ mainconj ] , as we will explain . a general result of meinrenken and sjamaar ( * ? ? ? * corollary 2.12 ) implies that the function @xmath62 is a quasi - polynomial for all @xmath63 . however , the same strengthening can fail for the general functions @xmath64 . vallejo introduces a notion of _ additive stability _ in @xcite and proves that it implies stability in stembridge s sense . additive stability is provided by the existence of a certain additive matrix , and hence is easier to apply , but it is less general ( * ? ? ? * example 6.3 ) . pak and panova show that for any @xmath65 , the triple @xmath66 is stable ( * ? ? ? * theorem 1.1 ) . this is also a special case of vallejo s work just mentioned and stembridge s result that @xmath67 is stable for any partition @xmath28 of @xmath68 ( * ? ? ? * example 6.3 ) . manivel uses geometric techniques in @xcite to produce many more examples of stable triples and to study the cone of stable triples . finally , see @xcite for some general information about the degrees and leading coefficients of the hilbert functions @xmath64 . in [ sec : bg ] we recall some background . the proof of theorem [ mainthm ] is given in [ sec : proof ] . finally , [ sec : rmk ] contains remarks and related results , including a littlewood richardson version and a plethysm version of theorem [ mainthm ] . we thank greta panova , mateusz michaek , john stembridge , and ernesto vallejo for helpful comments and references . we refer to ( * ? ? ? * part 1 ) for further explanation and references for this section . for a partition @xmath0 of @xmath1 define the schur functor @xmath69 by @xmath70 where @xmath71 is a complex vector space . we recall the well - known connection between schur functors and kronecker coefficients : [ prop : sw ] we have a natural identification @xmath72 we have decompositions @xmath73 tensoring these together , and using the decomposition , we find @xmath74 taking the @xmath2 isotypic component yields the stated result . recall that @xmath75 is a nonzero irreducible representation of @xmath76 if @xmath77 , and @xmath78 otherwise . thus , assuming @xmath77 , we have a natural isomorphism @xmath79 . proposition [ prop : sw ] therefore gives a natural isomorphism @xmath80 assuming @xmath81 and @xmath82 . we now recast so that the right side reflects the symmetry of the left , at least superficially . let @xmath83 , @xmath71 , and @xmath84 be finite dimensional vector spaces and let @xmath85 be a trilinear form . assume that @xmath86 is * non - degenerate * in the sense that it induces an isomorphism @xmath87 . note that this implies that @xmath88 . we let @xmath89 be the stabilizer of @xmath86 ; this projects isomorphically to @xmath90 . we can restate as : [ schur - kron ] let @xmath85 be a non - degenerate trilinear form and assume that @xmath91 , @xmath81 , and @xmath82 . then we have a natural isomorphism @xmath92 [ thm : git ] let @xmath93 be a complex reductive group acting on a finitely generated @xmath94-algebra @xmath49 and also compatibly on a finitely generated @xmath49-module @xmath95 , i.e. , the multiplication map @xmath96 is @xmath93-invariant . 1 . the ring of invariants @xmath97 is finitely generated . 2 . @xmath98 is a finitely generated @xmath97-module . \(a ) see ( * ? ? ? * theorem 3.6 ) . ( b ) see ( * ? ? ? * theorem 3.25 ) . let @xmath71 and @xmath84 be graded vector spaces . we define the * segre product * of @xmath71 and @xmath84 by @xmath99 this has the following interpretation in terms of invariant theory . the gradings on @xmath71 and @xmath84 are equivalent to algebraic @xmath100 actions . thus @xmath101 is naturally a representation of @xmath102 , and @xmath103 is the invariants under the diagonal subgroup @xmath104 . from this , we get the following corollary . [ cor : segre ] 1 . if @xmath49 and @xmath105 are graded @xmath94-algebras then @xmath106 is naturally a graded @xmath94-algebra . if @xmath49 and @xmath105 are finitely generated then so is @xmath106 . if in addition @xmath49 and @xmath105 are domains , then so is @xmath106 . if @xmath95 and @xmath57 are graded @xmath49- and @xmath105-modules then @xmath107 is naturally an @xmath106 module . if @xmath95 is finitely generated over @xmath49 and @xmath57 is finitely generated over @xmath105 then @xmath107 is finitely generated over @xmath106 . in addition , if @xmath49 and @xmath105 are domains and @xmath95 and @xmath57 are torsion - free modules , then @xmath107 is a torsion - free @xmath106 module . for partitions @xmath28 and @xmath0 , define @xmath108 [ prop : construct ] let @xmath83 be a vector space with @xmath109 . then @xmath110 naturally has the structure of a finitely generated graded integral domain over @xmath94 , and @xmath111 naturally has the structure of a finitely generated torsion - free graded @xmath110-module . let @xmath112 be the flag variety of @xmath113 . for every partition @xmath28 with @xmath114 , there is a @xmath93-equivariant line bundle @xmath115 on @xmath112 whose sections are @xmath116 ( this is the borel weil theorem , see @xcite ) and they satisfy @xmath117 . let @xmath118 be the total space of the vector bundle @xmath119 , and let @xmath120 be the ring of global functions on @xmath118 . this is an integral domain , since @xmath118 is an irreducible variety . it also has a bigrading given by @xmath121 . since each @xmath122 is a ( non - zero ) irreducible representation of @xmath76 ( by the borel weil theorem ) , and @xmath123 is an integral domain , it follows that @xmath123 is generated as a @xmath94-algebra by @xmath124 and @xmath125 . in particular , @xmath123 is finitely generated as a @xmath94-algebra . the bigrading on @xmath123 can be regarded as a @xmath102 action . then @xmath126 is the ring of invariants under the second @xmath100 and hence is a finitely generated domain ; and @xmath127 is the degree @xmath128 piece under the second @xmath100 action ( this can be interpreted as invariants of a twist by the @xmath129 character ) and hence is a finitely generated torsion - free module over @xmath126 . here we use theorem [ thm : git ] in both cases . in fact , the above proposition holds without the restriction on @xmath130 . let @xmath83 , @xmath71 , and @xmath84 be sufficiently large vector spaces satisfying @xmath88 , and let @xmath85 be a non - degenerate trilinear form . then @xmath131 by proposition [ schur - kron ] . since @xmath110 , @xmath132 , and @xmath133 are finitely generated graded integral domains ( proposition [ prop : construct ] ) , so is their segre product ( corollary [ cor : segre ] ) . since @xmath134 is a reductive group , the above invariant ring is also finitely generated ( theorem [ thm : git ] ) . this shows that @xmath34 is a finitely generated graded integral domain . similarly , we have @xmath135 since @xmath111 , @xmath136 , and @xmath137 are finitely generated torsion - free graded modules over @xmath110 , @xmath132 , and @xmath133 ( proposition [ prop : construct ] ) , their segre product is a finitely generated torsion - free graded module over the segre product of the @xmath49 s ( corollary [ cor : segre ] ) . the invariant module is finitely generated over the invariant ring ( theorem [ thm : git ] ) , and obviously torsion - free . so @xmath40 is a finitely generated torsion - free graded @xmath34-module . we could avoid discussing the segre product by directly constructing the ring @xmath37 and module @xmath138 using a triple product of flag varieties . the ring @xmath139 is the homogeneous coordinate ring of a homogeneous space for @xmath140 and hence has rational singularities @xcite . this property is preserved by taking invariants under a reductive group ( * ? ? ? * corollaire ) . we now give a proof of the `` only if '' direction of conjecture [ mainconj ] . assume that @xmath8 is stable . in particular , @xmath141 is constant for @xmath7 . furthermore , @xmath10 by definition . thus @xmath34 is a finitely generated graded normal domain whose hilbert polynomial has degree 0 and whose first graded piece is non - zero . it follows that @xmath142 $ ] , with @xmath39 of degree one , and so @xmath36 for all @xmath16 . in the above argument , we used that @xmath10 is part of the definition for @xmath8 to be stable . if we dropped that assumption , we could have a case where @xmath143 for @xmath7 , and hence for all @xmath16 since @xmath37 is a domain . this means that @xmath144 and hence that @xmath145 for @xmath7 for all @xmath11 . we can not improve this to @xmath146 : @xmath143 for all @xmath16 if @xmath147 and @xmath148 , but @xmath149 for @xmath150 . from what we have shown , @xmath8 is stable if and only if the krull dimension of @xmath37 is @xmath128 . since @xmath37 is a ring of invariants , this property can be determined algorithmically . although it is probably impractical , there did not seem to be an algorithm previously for determining stability . define @xmath151 to be the dimension of the vector space @xmath152 . ( note the swap of superscripts and subscripts . ) if @xmath8 is stable then @xmath151 is equal to the limiting value of @xmath153 . in particular , @xmath154 is the usual stable kronecker coefficient @xmath155 . it would be interesting if one could say anything about the numbers @xmath151 in general . for example , there is a well - known relationship between stable kronecker coefficients and littlewood richardson coefficients . does this generalize in any way to the @xmath151 ? since the function @xmath62 is given by a quasi - polynomial @xmath156 for all @xmath63 ( see [ ss : related - work ] ) , one can ask if the rings @xmath34 have toric degenerations . this would imply that there is a rational polytope @xmath157 such that the @xmath156 is the number of integer points in the @xmath18th dilate of @xmath158 for all @xmath63 . ehrhart reciprocity ( * ? ? ? * theorem 4.6.9 ) implies that @xmath159 is the number of integer points in the interior of the @xmath18th dilate of @xmath158 . so in particular , @xmath160 . however , from ( * ? ? ? * theorem 2.4 ) , this fails for @xmath161 , @xmath162 , @xmath163 where @xmath164 in particular , @xmath165 and @xmath166 . so no such polytopes exist in general . we thank mateusz michaek for bringing this example to our attention . if we replace @xmath101 in with @xmath167 , then we get a version of theorem [ mainthm ] with kronecker coefficients replaced by littlewood richardson coefficients @xmath168 ( see ( * ? ? ? * ( 3.11 ) ) ) . in this case , a different construction for the analog of the ring @xmath105 is given in @xcite and the rational singularities property was used to deduce that the function @xmath169 is a polynomial for @xmath63 ( * ? ? ? * corollary 3 ) . this gives the following concrete statement ( which can be deduced from previous work , see remark [ rmk : lr ] ) : 1 . the analogs of ( a ) and ( b ) for kronecker coefficients are not equivalent : @xmath179 while @xmath180 . the equivalence of ( b ) and ( c ) can also be deduced from the fact that littlewood richardson coefficients count the number of integer points in rational polytopes @xcite and using ( * ? ? ? * proposition 4.4 ) . the technique used in this paper can also be used to prove stability properties of plethysm coefficients . the ideas are similar , so we omit the details . let @xmath181 be the multiplicity of @xmath182 in @xmath183 ( the composition of schur functors ) . we emphasize that this is not symmetric in @xmath0 and @xmath31 . let @xmath71 be a vector space of large dimension . we can build a finitely generated integral domain @xmath187 and a finitely generated torsion - free @xmath188-module @xmath189 note that @xmath190 and @xmath191 . 1 . take @xmath192 and @xmath193 . this is a result of brion @xcite . 2 . take @xmath194 and @xmath195 . to see that @xmath196 for all @xmath63 , note that we can take @xmath197 . then the ring of @xmath198-invariants on @xmath199 is a polynomial ring on one generator of degree @xmath200 ( the discriminant ) . murnaghan s stability theorem was reinterpreted in @xcite as the fact that the segre product of finitely generated fi - modules is finitely generated . this is a useful reformulation since it turns murnaghan s numerical result into a structural result . the rings @xmath201 are examples of twisted commutative algebras ( see @xcite for an introduction to these objects ) , and modules over the tca @xmath202 are equivalent to fi - modules ( see @xcite ) . one might therefore hope that the stability results in this paper could be reformulated as structural results for @xmath201-modules . we believe that there is such a reformulation , though it is more complicated than the case of murnaghan s theorem . nonetheless , this point of view led to the proof in this paper . we plan to pursue the connection to tca s in future work . emmanuel briand , rosa orellana , mercedes rosas , reduced kronecker coefficients and counter - examples to mulmuley s strong saturation conjecture sh , with an appendix by ketan mulmuley , _ comput . complexity _ * 18 * ( 2009 ) , no . 4 , 577600 , . allen knutson , terence tao , christopher woodward , the honeycomb model of @xmath204 tensor products . ii : puzzles determine facets of the littlewood - richardson cone , _ * 17 * ( 2004 ) , no . 1 , 1948 ,	we prove a conjecture of stembridge concerning stability of kronecker coefficients that vastly generalizes murnaghan s theorem . the main idea is to identify the sequences of kronecker coefficients in question with hilbert functions of modules over finitely generated algebras . the proof only uses schur weyl duality and the borel weil theorem and does not rely on any existing work on kronecker coefficients .
we study recursive networks in which each new vertex is attached to @xmath0 existing ones chosen with probability proportional to a linear function of vertex degree , @xmath22 . this rule generates scale - free correlated networks with a degree distribution exponent @xmath23 . in the infinite network limit , pearson s coefficient @xmath24 , since the third moment of the degree distribution [ see the denominator of pearson s coefficient definition ( [ e10 ] ) ] diverges . here we define @xmath25 , where @xmath26 in the number of vertices in a network . for @xmath20 , we find @xmath27}{(1{+}m)[2(1{+}m ) + a(5{+}7 m ) + a^2(1{+}7 m ) ] } , \label{e30}\ ] ] which shows that for @xmath1 ( i.e. , for random recursive trees ) , pearson s coefficient @xmath28 for any value of @xmath3 . one can also see that @xmath29 for any @xmath0 . in particular , for uniformly random attachment , i.e. , @xmath30 , we have : @xmath31 figure [ f1 ] , obtained by simulation of the networks of @xmath32 , @xmath33 , @xmath34 , and @xmath35 vertices ( the number of runs for each point was between 50 and 500 for @xmath36 and between @xmath37 and @xmath33 for @xmath38 ) , demonstrates the size dependence of pearson s coefficient in these networks . this figure shows that even for large networks , the deviations from the limiting infinite network values are significant . we find the following asymptotic size dependences @xmath39 : @xmath40 \left . \begin{array}{l } \!\!- % % ( 4-\gamma ) t^{-1/(\gamma-1 ) } , \ \ \ \ \ \ m=1 , \\[5pt ] \!\!+ % % \displaystyle\frac{4-\gamma}{(\gamma-3)^2 } t^{-(4-\gamma)/(\gamma-1 ) } , \ m>1 \end{array } \right\ } \ 3<\gamma<4 , \\[21pt ] \left . \begin{array}{l } \!\!-t^{-(\gamma-3)/(\gamma-1 ) } , \ m=1 , \\[5pt ] \!\!+t^{-(\gamma-4)/(\gamma-1 ) } , \ m>1 \end{array } \right\ } \ \gamma>4 . \end{array } \right . \label{e50}\ ] ] here we show the signs of the asymptotes but ignore their factors . the positive sign of the asymptotes at @xmath41 , @xmath42 and @xmath43 means that @xmath44 approaches the infinite size limit @xmath45 from above . in this situation ( @xmath41 , @xmath46 ) , pearson s coefficient varies with size non - monotonously : first increases and then diminishes to @xmath45 ( see the second panel of fig . [ f1 ] ) . introducing exponent @xmath47 : @xmath48 for large network sizes @xmath26 , we arrive at the dependences @xmath49 shown in fig . note that @xmath50 , so the infinite network limit is approached slowly . the relaxation to the infinite network values , @xmath45 , is especially slow if exponent @xmath3 is close to @xmath51 ( at any @xmath0 ) or to @xmath4 ( only if @xmath2 ) . in the specific case of @xmath2 , @xmath52 , we obtain the logarithmic relaxation : @xmath53 we measured the dependence of exponent @xmath47 on @xmath3 in the simulated networks . the result , shown in fig . [ f3 ] , demonstrates an agreement with above analytical predictions . let us derive the pearson coefficient ( [ e50 ] ) of the infinite recursive networks . the derivation is based on rate equations for the average number @xmath54 of vertices of degree @xmath9 in a recursive network at time @xmath26 and the average number @xmath55 of edges connecting vertices of degrees @xmath8 and @xmath9 . we have @xmath56 for vertices , and so @xmath57 where @xmath58 is a degree distribution of the network of size @xmath26 . for edges , we have : @xmath59 and so @xmath60 where @xmath19 is the degree degree distribution for edges . using the standard rate or master equation approaches @xcite to this kind of networks ( specifically , to the recursive networks growing due the preferential attachment mechanism ) , we write the following rate equations for @xmath54 and @xmath55 : @xmath61 & & \!\!\!\!\!\ ! \!\!\!\!\!\ ! m}(t+1 ) = n_k(t ) \nonumber \\[5pt ] & & \!\!\!\!\!\ ! \!\!\!\!\!\ ! + m \frac{k-1+ma}{m(2+a)t}n_{k-1}(t ) - m \frac{k+ma}{m(2+a)t}n_k(t ) % % . \label{e110}\end{aligned}\ ] ] ( note that the mean vertex degree in these networks is @xmath62 ) and @xmath63 & & - m\left[\frac{m+ma}{m(2+a)t}e_{mk}(t ) + \frac{k+ma}{m(2+a)t}e_{mk}(t ) \right ] , \nonumber \\[5pt ] & & e_{jk > m}(t+1 ) = e_{jk}(t ) \nonumber \\[5pt ] & & + m \left[\frac{k-1+ma}{m(2+a)t}e_{j , k-1}(t ) + \frac{j-1+ma}{m(2+a)t}e_{j-1,k}(t)\right ] \nonumber \\[5pt ] & & - m \left[\frac{j+ma}{m(2+a)t}e_{jk}(t ) + \frac{k+ma}{m(2+a)t}e_{jk}(t)\right ] . \label{e120}\end{aligned}\ ] ] we separately write out equations for the case of @xmath64 . here we used the fact that the probability to attach a new vertex to a vertex @xmath65 of degree @xmath66 in this model is @xmath67 assuming a stationary regime at the limit @xmath68 [ @xmath69 , @xmath70 , we reduce these equations to @xmath71 & & \!\!\!\ ! p_{k > m } = m \frac{k-1+ma}{m(2+a)}p_{k-1 } - m \frac{k+ma}{m(2+a)}p_k \label{e130}\end{aligned}\ ] ] and @xmath72 & & \!\!\!\ ! \!\!\!\ ! m(j+k+2 + 2ma+a)e_{jk > m } \nonumber \\[5pt ] & & \!\!\!\ ! \!\!\!\ ! = m(k-1+ma)e_{j , k-1 } - m(j-1+ma)e_{j-1,k } . \label{e140}\end{aligned}\ ] ] to obtain the pearson coefficient , see definition ( [ e10 ] ) , we must find @xmath73 , @xmath74 , and @xmath7 , where @xmath75 multiplying both the sides of the second equation of the system ( [ e130 ] ) by @xmath76 and @xmath77 and summing over @xmath9 , and taking into account @xmath78 which follows from the first equation of the system ( [ e130 ] ) , we obtain @xmath79 and @xmath80 & & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! \times[6(1{+}m ) { + } a(5{+}21m{+}8m^2 ) { + } a^2(1{+}9m{+}16m^2 ) ] . \label{e180}\end{aligned}\ ] ] respectively . to get @xmath7 , we multiply both the sides of the second equation of the system ( [ e140 ] ) by @xmath81 and sum them over @xmath8 and @xmath9 . we also take into account the following general equality : @xmath82 which gives @xmath83 as a result we find @xmath84 % % \nonumber % % \\[5pt ] % % & & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! % % \langle jk \rangle_{\text{e } } = 2m\mu + m\frac{2 + 2 m + 3a + 5ma + 2ma^2 + a^2}{a^2 } , \label{e200}\ ] ] and so at @xmath85 , i.e. , @xmath38 , the numerator of definition ( [ e10 ] ) is @xmath86 & & \!\!\!\!\!\!\!\!\!\!\!\!\ ! = \frac{m-1}{4a^2 } [ 4(1+m ) + 4a(1 + 2 m ) + a^2(1 + 3 m ) ] . \label{e205}\end{aligned}\ ] ] importantly , this numerator is zero if @xmath1 . so we see that @xmath87 at least if @xmath38 . on the other hand , if @xmath2 , the numerator is finite in the range @xmath38 . this shows that @xmath88 must be zero at @xmath89 , since the denominator diverges if @xmath90 . finally , at @xmath91 , @xmath28 at any @xmath0 [ see the next section , eq . ( [ e300 ] ) ] . substituting relations ( [ e170 ] ) , ( [ e180 ] ) , and ( [ e205 ] ) into definitions ( [ e10 ] ) and ( [ e20 ] ) readily gives the resulting expression ( [ e30 ] ) for the stationary value @xmath45 of pearson s coefficient . equations ( [ e110 ] ) and ( [ e120 ] ) allow one to derive the full size dependence of the pearson coefficient . instead of these cumbersome straightforward calculations , we obtain the asymptotic behavior of @xmath44 in an easier way , using known results for the asymptotics of degree degree correlations in these networks @xcite . the derivation is based on the following expression of the pearson coefficient in terms of the average vertex degree @xmath92 of the nearest neighbors of the vertices of degree @xmath9 : @xmath93}{\langle k \rangle\langle k^3 \rangle - \langle k^2 \rangle^2 } % % \nonumber \label{e210 } \\[5pt ] & & = \frac{\langle k \rangle \sum_k k^2 p_k % % p(k ) \overline{k}_{\text{nn}}(k ) - \langle k^2 \rangle^2}{\langle k \rangle\langle k^3 \rangle - \langle k^2 \rangle^2 } . \label{e220}\end{aligned}\ ] ] this expression directly follows from definition ( [ e10 ] ) . the leading asymptotics of @xmath92 can be obtained by using the known exact asymptotics of @xmath94 for these recursive networks @xcite . here @xmath94 is the probability that a descendant vertex of degree @xmath9 is connected to an ascendant vertex of degree @xmath95 . this quantity satisfies the following relations : @xmath96 & & \sum_l ( n_{kl } + n_{lk } ) = kp_k , \nonumber \\[5pt ] & & \sum_l l(n_{kl } + n_{lk } ) = kp_k \overline{k}_{\text{nn}}(k ) . \label{e230}\end{aligned}\ ] ] at @xmath1 , according to ref . @xcite , @xmath97 & & n_{kl } \sim l^{1-\gamma}k^{-2 } \ \ \ \ \text{if } \ \ 1 \ll k \ll l . \label{e240}\end{aligned}\ ] ] for @xmath38 and any @xmath0 , @xmath98 approaches a stationary limit as the network grows . figure [ f4 ] demonstrates this relaxation for two values of the degree distribution exponent @xmath3 , namely , for @xmath99 and @xmath52 in the case of @xmath1 . we can approximate the deviation from the infinite network limit by @xmath100kp_k \sim \int_{k_{\text{cut}}}^\infty dl\ , l ( n_{kl } + n_{lk } ) , \label{e250}\ ] ] where @xmath101 is the time dependent cutoff of the degree distribution . in these networks , @xmath102 . substituting eq . ( [ e240 ] ) into eq . ( [ e250 ] ) results in @xmath103 & & \sim c_1 k^{\gamma-3 } t^{(3-\gamma)/(\gamma-1 ) } + c _ 2 k^{2\gamma-3 } t^{-(2\gamma-3)/(\gamma-1 ) } % % , \label{e260}\end{aligned}\ ] ] at large degrees @xmath104 for the networks with @xmath38 . here @xmath105 and @xmath106 are constants depending on @xmath3 , which we do not calculate . similarly , for @xmath107 , where @xmath98 diverges as @xmath108 at any @xmath0 , we can write an asymptotic estimate @xmath109 substituting eq . ( [ e240 ] ) into eq . ( [ e270 ] ) we obtain @xmath110 at large degrees @xmath104 for the networks with @xmath107 . here @xmath111 and @xmath112 are some constants @xcite . we substitute asymptotics ( [ e260 ] ) and ( [ e280 ] ) into expression ( [ e220 ] ) for the pearson coefficient and take into account the leading terms in the numerator and the denominator . the combinations of these leading terms are different in different areas of @xmath3 and @xmath0 , since the quantities @xmath98 , @xmath73 , and @xmath74 in expression ( [ e220 ] ) change their asymptotic behavior at two special points , namely , @xmath113 and @xmath52 . in particular , at these points ( @xmath113 and @xmath4 ) the second and , respectively , the third moments of the degree distribution become divergent . for example , if @xmath107 at any @xmath0 , relation ( [ e220 ] ) with substituted asymptotics ( [ e260 ] ) takes the form : @xmath114 & & \!\!\!\!\ ! + \ ! \left.c'_2 t^{(3 - 2\gamma)/(\gamma-1)}\!\!\int^{t^{1/(\gamma-1 ) } } \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!dk\,k^2 k^{2\gamma-3 } k^{-\gamma}\!\right ) % % \nonumber % % \\[5pt ] % % & & \!-\ ! c'_3\!\!\left.\left(\ ! \int^{t^{1/(\gamma-1 ) } } \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!dk\,k^{2-\gamma}\right)^{\!\!2}\right ] \nonumber \\[5pt ] & & \!\!\!\!\ ! \left/\left(2m\int^{t^{1/(\gamma-1 ) } } \!\!\!\!\!\!dk\,k^{3-\gamma}\right)\right . , \label{e290}\end{aligned}\ ] ] where @xmath115 , @xmath116 and @xmath117 are constants . this leads to the result : @xmath118 & & \!\!\!\!\!\ ! \sim \!\frac{[c''_1 \ln t\,\,t^{(3-\gamma)/(\gamma-1 ) } + c''_2 t^{(3-\gamma)/(\gamma-1 ) } ] - c''_3 t^{2(3-\gamma)/(\gamma-1)}}{t^{(4-\gamma)/(\gamma-1 ) } } \nonumber \\[5pt ] & & \!\!\!\!\!\ ! \sim - % % \frac{4-\gamma}{(\gamma-3)^2 } t^{-(\gamma-2)/(\gamma-1 ) } , \label{e300}\end{aligned}\ ] ] where @xmath119 , @xmath120 and @xmath121 are constants . in a similar way , we derive the other asymptotics listed in eqs . ( [ e50 ] ) and ( [ e60 ] ) . interestingly , both the terms in eqs . ( [ e260 ] ) and ( [ e280 ] ) give contributions of the same order of magnitude to @xmath44 , see the numerator of eq . ( [ e300 ] ) . note that we suppose that for @xmath2 the form of the asymptotics of @xmath98 is the same as in eqs . ( [ e260 ] ) and ( [ e280 ] ) . to verify our assumption , we inspected the corresponding results for @xmath98 in ref . @xcite and found that the asymptotic behavior should be similar at different @xmath0 ( apart of numerical coefficients ) if exponent @xmath3 is fixed . in addition , we checked that results ( [ e50 ] ) also can be derived by using the asymptotics of @xmath98 from ref . this confirms our conclusions about the asymptotics of @xmath44 . our result , namely zero pearson coefficient of random recursive trees at any @xmath122 , naturally leads to the following questions . what is the class of trees that have zero pearson coefficient ? is pearson s coefficient zero for any infinite recursive tree ? at present , we can not answer the first question . as for the second question , the answer is negative . indeed , as a counter - example , we present the simplest infinite recursive tree with a non - zero pearson coefficient . this is a star , which is a tree of @xmath123 vertices , including @xmath124 leafs and the hub of degree @xmath124 . for star of any size , clearly , @xmath125 . thus we have studied the pearson coefficient in strongly correlated growing networks . they form a representative class of networks with strong structural correlations including pairwise correlations between degrees of the nearest - neighbor vertices . despite these correlations , we have found that in a wide range of infinite correlated networks the pearson coefficient approaches zero . for any infinite random recursive tree whose growth is driven by arbitrary linear preferential attachment of new vertices , we observed zero pearson coefficient . these networks include random recursive trees with rapidly decaying and even exponential degree distribution , where the third moment of the degree distribution is finite . so here zero value of pearson s coefficient demands zero correlation function in the numerator of definition ( [ e10 ] ) . this surprising equality to zero indicates an exact mutual compensation of assortative and disassortative degree degree correlations . in this respect , the recursive trees is a very special case of random recursive networks @xcite . we have investigated the size dependence of pearson s coefficient in the growing networks . we have found that the size effect is significant even for very large networks . we have shown that a growing network during its evolution may demonstrate essential and even non - monotonous variation of pearson s coefficient . due to this marked size dependence , it is hardly feasible to use this integral characteristic of correlations for quantitative comparison of different real - world networks . instead , for this purpose , one has to use more informative characteristics , for example @xmath92 . this work was partially supported by the following projects ptdc : fis/71551/2006 , fis/108476/2008 , and sau - neu/103904/2008 , and also by socialnets eu project . the authors thank m. barroso for his help with administration of computing facilities used for numerical simulations . m. e. j. newman , the structure and function of complex networks , siam review * 45 * , 167 ( 2003 ) . m. . serrano , m. bogu , r. pastor - satorras , and a. vespignani , in _ large scale structure and dynamics of complex networks : from information technology to finance and natural science _ ( world scientific , singapore , 2007 ) , eds . g. caldarelli and a. vespignani , p. 35 . note that the equality @xmath126 is valid for an arbitrary network . s. maslov and k. sneppen , specificity and stability in topology of protein networks , science * 296 * , 910 ( 2002 ) . note that the numerator in definition ( [ e10 ] ) also diverges if @xmath91 . this divergence , however , is slower than that of the denominator , see eq . ( [ e300 ] ) which leads to zero pearson s coefficient in this range . l. krapivsky , s. redner , and f. leyvraz , connectivity of growing random networks , phys . lett . * 85 * , 4629 ( 2000 ) . m. bogu , r. pastor - satorras , and a. vespignani , absence of epidemic threshold in scale - free networks with connectivity correlations , phys . rev . lett . * 90 * , 028701 ( 2003 ) . a. barrat and r. pastor - satorras , rate equation approach for correlations in growing network models , phys . e * 71 * , 036127 ( 2005 ) . s. n. dorogovtsev and j. f. f. mendes , scaling properties of scale - free evolving networks : continuous approach , phys . e * 63 * , 056125 ( 2001 ) . note that there is some difference between asymptotics ( [ e280 ] ) and the corresponding result in ref . the difference is apparently due to the fact that we derived these expression by using the exact asymptotics taken from ref . @xcite while the result of ref . @xcite was obtained by using the continuous approximation for the rate equations . nonetheless , one can easily check that for @xmath107 , the expression for @xmath98 from ref . @xcite leads to the same asymptotic behavior of the pearson coefficient as in the present paper . note that in equilibrium networks , dead ends ( vertices of degree @xmath16 ) , abundant in random trees , also play a special role for degree degree correlations , see p. bialas and a. k. ole , correlations in connected random graphs , phys . e * 77 * , 036124 ( 2008 ) .	we obtained pearson s coefficient of strongly correlated recursive networks growing by preferential attachment of every new vertex by @xmath0 edges . we found that the pearson coefficient is exactly zero in the infinite network limit for the recursive trees ( @xmath1 ) . if the number of connections of new vertices exceeds one ( @xmath2 ) , then the pearson coefficient in the infinite networks equals zero only when the degree distribution exponent @xmath3 does not exceed @xmath4 . we calculated the pearson coefficient for finite networks and observed a slow , power - law like approach to an infinite network limit . our findings indicate that pearson s coefficient strongly depends on size and details of networks , which makes this characteristic virtually useless for quantitative comparison of different networks . the pearson coefficient @xmath5 is used as an integral characteristic of structural correlations in a network . pearson s coefficient characterizes pairwise correlations between degrees of the nearest neighboring vertices in networks . some observable quantities in correlated networks ( e.g. , the size of a giant connected component near the point of its emergence ) are directly expressed in terms of this coefficient @xcite . the pearson coefficient is the normalized correlation function of the degrees of the nearest neighbour vertices @xcite . the coefficient is the ratio : @xmath6 where @xmath7 is the average product of the degrees @xmath8 and @xmath9 of the end vertices of an edge , @xmath10 is the average degree of an end vertex of an edge , @xmath11 is the average square of the degree of an end vertex of an edge @xcite , and @xmath12 is for normalization . here @xmath13 and @xmath14 denote averaging over vertices and edges , respectively [ see eqs . ( [ e150 ] ) , ( [ e170 ] ) , and ( [ e185 ] ) below ] . pearson s coefficient can be positive ( in average , assortative mixing of the nearest neighbors degrees ) or negative ( disassortative mixing ) and takes values in the range from @xmath15 to @xmath16 . the pearson coefficient @xmath5 is a convolution of the joint distribution of nearest neighbor degrees , @xmath17 . this joint distribution is the probability that the ends of a randomly chosen edge have degrees @xmath8 and @xmath9 , @xmath18 . being an integral characteristic of degree degree correlations , the pearson coefficient misses details of these correlations , compared to @xmath19 @xcite . despite this fact , pearson s coefficient is widely used for characterization and comparison of real - world networks @xcite . note that the compared real networks have different sizes . in this paper we show that since @xmath5 is a markedly size dependent quantity , pearson s coefficient may be used for comparison of networks only with a very critical attitude . we calculate pearson s coefficient for the simplest growing complex networks , namely recursive random networks with preferential attachment of new vertices . we describe the size dependence of @xmath5 in these networks , fig . [ f1 ] . remarkably , for all infinite recursive trees of this kind , we find that the pearson coefficient is exactly zero at any value of the degree distribution exponent @xmath3 , although all these networks are strongly correlated . the statement that pearson s coefficient is zero in the range @xmath20 is essentially non - trivial . the point is that zero value of the pearson coefficient in the range @xmath21 where the third moment of the degree distribution diverges is clear . indeed , in this region , the denominator in definition ( [ e10 ] ) diverges in the infinite network limit @xcite . this is not the case at @xmath20 , and for zero @xmath5 , the numerator in eq . ( [ e10 ] ) must be zero . in this range , zero value of pearson s coefficient means a surprising exact mutual compensation of different degree degree correlations in these trees . there are two opposing kinds of degree degree correlations in complex networks : assortative and disassortative . here assortativity is a tendency of high degree vertices to have high degree neighbors and low - degree vertices to have low degree neighbors . in contrast , disassortative mixing means neighborhood of vertices with contrasting ( low and high ) degrees . this tendencies may be opposing in different ranges of degrees , see discussion in sec . [ zz ] , fig . [ f4 ] . that is , assortative and disassortative mixing may coexist . in particular , this is the case for random recursive trees . our results show that in these growing networks , the different kinds of correlations completely compensate each other in the infinite network limit .
we have demonstrated a novel route for the self - organization of arrays of planar nanowires on nominally - flat commercial wafers . it is based on the fabrication of a template of trenches upon the deposition under kinetic limitations of a non - magnetic material along a surface of uniaxial symmetry , here w(110 ) , followed by the filling of the bottom of the trenches by a magnetic material under layer - by - layer deposition conditions , here fe . periods in the range @xmath39 were demonstrated . owing to a self - limiting effect the micro - facets are all of type @xmath40 with an angle of @xmath41 , which is promising for attaining a low distribution of physical properties . concerning magnetism the easy axis of magnetization lies in - the - plane along the wires . we have demonstrated the possibility to tailor its magnitude using interface mae with different capping- or underlayers , bringing the _ mean _ blocking temperature to @xmath36 inserting an ultrathin underlayer of mo . b. b. and a. r. acknowledge financial support from french rgion rhne - alpes ( mobility program ) , and fp6 eu - nsf program ( strp 016447 magdot ) , respectively .	we have self - organized versatile magnetic nanowires , with variable period and adjustable magnetic anisotropy energy ( mae ) . first , using the kinetic roughening of w(110 ) uniaxial templates of trenches were grown on commercial sapphire wafers . unlike most templates used for self - organization , those have a variable period , are demonstrated here . fe deposition then results in the formation of wires in the trenches . the magnitude of mae could be engineered up or down by changing the capping- or underlayer , in turn affecting the mean superparamagnetic temperature , raised to so far . 0.5 in 0.5 in the bottom - up approach is promising for the fabrication of nanostructures at moderate cost , with better resolution and less microscopic defects than with lithography . in epitaxial self - organization ( so ) the building blocks are atoms that aggregate to each other during growth , a process that can be engineered to fabricate wires and dots with lateral dimensions from the micron size@xcite down to the atomic size@xcite . the interest of so for fundamental research is often the narrow size dispersion@xcite , so that macroscopic measurements reveal the properties of one single nanostructure . issues like the increase of orbital momentum and magnetic anisotropy energy ( mae ) at atomic edges or kinks have been addressed@xcite . concerning applications it is sometimes argued that so arrays could be used by addressing single nanostructures one by one , to store one bit of information for example . a more realistic view is the use of the array as a material with specific properties arising from the nanoscale . this is the case of semiconductor quantum dots with lasing properties@xcite . do magnetic so systems meet the requirements of applied materials ? a prerequisite is that some versatility of geometrical as well as physical properties is achieved , like tuning the magnitude of the mae . we focus here on the fabrication and properties of wires , which lie at the background of the fast - developing field of spin electronics making use of the propagation of domain walls in wires for storage or logics devices@xcite . so magnetic wires are often achieved by step - decoration of vicinal surfaces@xcite . this approach is not versatile as a new crystal has to be prepared with a specific miscut whenever the period needs to be changed . templates resulting from kinetic effects are potentially more versatile as the period can be changed with processing parameters , like temperature . this has been explored using ion etching under grazing incidence to create ripples on surfaces , independently of crystalline directions@xcite . however a significant control of the period has not been demonstrated so far . here we explore a new approach , based on the growth of body - centered - cubic ( 110 ) materials on nominally - flat saphir yielding parallel trenches with an adjustable period , which we then use to grow magnetic nanowires . we could tune the magnitude of the mae of the wires using suitably - chosen capping layers or underlayers . the use of commercial wafers while achieving some control on both the period and the mae represents a significant advancement in the versatility of so nanowires . the samples were grown in a set of ultra - high vacuum chambers using pulsed - laser deposition with a nd - yag laser ( @xmath0{nm}$ ] ) . the chambers are equipped with a quartz microbalance , sample heating and a translating mask for the fabrication of wedge - shaped samples . a 10 kev reflection high energy electron diffraction ( rheed ) setup with a ccd camera synchronized with laser shots permits operation during deposition . an omicron room - temperature scanning tunneling microscope ( stm-1 ) and an auger electron spectrometer ( aes ) are available . the metallic films are grown on sapphire _ nominally - flat _ @xmath1 commercial wafers . a detailed description of the chambers and growth procedures can be found in @xcite . the magnetic measurements were performed ex - situ on -mo - capped samples , with a quantum design superconducting quantum interference device ( squid ) magnetometer . ordered arrays of wires were obtained in three steps , consisting of the fabrication of 1 . a non - magnetic flat buffer layer ; 2 . a non - magnetic template displaying a uniaxial array of trenches ; 3 . fe wires by layer - by - layer deposition at the bottom of the trenches . the first step is the preparation of a smooth buffer layer . a seed layer of mo ( nominal thickness @xmath2{nm}$ ] ) followed by w ( @xmath3{nm}$ ] ) are deposited on sapphire at room temperature ( rt ) followed by annealing at @xmath4c . this yields a @xmath5 surface of quality similar to that of metal single crystals , with atomically - flat terrasses of width up to hundreds of nanometers@xcite . the second step consists of the preparation of a non - magnetic template . it was inspired by reports of the kinetic uniaxial roughening of films of the bcc elements fe(110)@xcite and w(110)@xcite during homoepitaxy at moderate temperature , explained by anisotropic diffusion along the steps and the occurrence of an ehrlich - schwoebel barrier . a ( resp . c ) shows the topography of a w(110 ) layer obtained upon deposition at @xmath6 ( resp . following nucleation at @xmath7 and subsequent growth at @xmath6 ) . a uniaxial array of trenches aligned along @xmath8 $ ] is thus obtained , with a period of @xmath9 ( resp . @xmath10 ) . the stm cross - sections shown as insets reveal a depth of @xmath11 ( resp . ) . the structure of the trenches can be probed using rheed with the electron beam along @xmath8 $ ] . the diffraction patterns consist of arrows whose half - angle reveals the slope of the microfacets . the arrows are much sharper for trenches of larger depth because the coherence length along the facets is increased . this allows us to determine accurately the angle of the facets to @xmath12 , in perfect agreement with the triangular - like stm cross - sections of c. this angle is unambiguously ascribed to facets of type @xmath13 , with @xmath14 expected . the monitoring of rheed patterns during growth shows that @xmath15 is a stationary value , reached after a few nanometers of nominal thickness . the occurrence of a stationary angle garanties that the shape of the trenches is not influenced by fluctuations of the width of the trenches , and thus displays essentially no distribution . this is crucial as the dispersion of mae increases tremendously for nanosized systems in relation with the distribution of structural environments@xcite . the existence of a stationary angle was postulated in ref.@xcite , however the analysis concluded to facets of type @xmath16 with @xmath17 for w. the discrepancy may come from the fact that the leed patterns used in ref.@xcite were broad because trenches with a small period were studied . we could produce templates with the same facets for mo , nb , v and ta by deposition at room temperature . w(110 ) templates : ( a , c ) @xmath18 stm images of templates with small ( a ) and large ( c ) period , see text ( b , d ) rheed patterns along the @xmath8 $ ] azimuth for surfaces a and c , resp . the sample lies vertical , on the left of the patterns . , width=306 ] the third step consists of the fabrication of wires . fe is deposited on the templates at @xmath6 , for which layer - by - layer growth occurs with pld on a smooth surface@xcite . for all deposits we have used w templates , which remain perfectly stable up to @xmath19 . for templates of small period the surface gradually smoothes and becomes essentially flat for @xmath20(atomic layers ; not shown here ) . the details of the early stages of growth are easier to investigate for templates of large period . for these stm cross - sections reveal a flat level at the bottom of the trenches , suggesting a progressive filling and thus yielding fe wires with a triangular cross - section . disconnected wires can be formed up to @xmath21 , beyond which percolation sets in for the period . ( a ) @xmath22 stm image of wires fe()/w ( b ) cross - section along the line shown in a ( c ) cross - section along the line shown in c ( same scales).,width=321 ] we report below the magnetic properties of fe wires prepared on w(110 ) templates with a mean period of and an fe nominal thickness of . the mean thickness and width of the wires are and , respectively , as deduced from stm . for mo / fe / w wires the easy axis of magnetization lies in - the - plane along the wires , along [ 001]a . the mae is deduced from loops along [ 1@xmath230 ] like @xmath24{j / m^{3}}$ ] at @xmath25 . @xmath26 originates from the sum of several contributions , among which only the dipolar energy @xmath27 can be estimated reliably . other sources of mae are surface ( nel - type ) expected to favor the [ 1@xmath230 ] direction for the ( bottom ) fe / w interface@xcite and [ 001 ] for the ( top ) mo / fe interface@xcite , step - edge for fe / w , expected to favor [ 1@xmath230 ] @xcite ; magneto - elastic unlike the case of thin films , here the large density of steps is expected to induce a significant out - of - plane mismatch and shear stress , so that no figure or even sign can be reliably predicted ; finally the bulk fe mae is negligible . surprisingly @xmath28 for this sample despite this complex situation , similarly to ( fe , ag ) self - organized arrays of wires@xcite . the loops measured at different temperatures with @xmath29 $ ] have a rather square shape , and at remanence full saturation is observed b. the coercivity continuously decreases with temperature while @xmath30 remains essentially unchanged , suggesting a superparamagnetic behavior . the ultimate blocking temperature determined by the `` zero field cooling / field cooling '' ( zfc - fc ) process is @xmath31{k}$ ] d. the _ mean _ value of @xmath32 is around @xmath33{k}$ ] , defined as half - way up the zero - field cooling remagnetization curve . in the following we report on the use of surface mae to tailor the magnitude of the mae of the wires and move towards features that would be required for applications like recording media , like functionality at room temperature and adjustment of mae . first , the very same wires were fabricated after inserting a @xmath34-thick underlayer of mo on the w trenches . the magnetic easy axis is again along the wires , however its magnitude is increased , which is consistent as fe / mo interfaces favor alignement of magnetization along @xmath8$]@xcite . this brings the mean @xmath32 from @xmath35 without underlayer to @xmath36 with underlayer . the coercive field is increased at all temperatures , compare . on the reverse @xmath32 is lowered to @xmath37 for al / fe / w . magnetization curves of fe ( ) wires with period . mo / fe / w wires ( a ) at in - plane along @xmath8 $ ] and @xmath38 $ ] ( b ) in - plane along @xmath8 $ ] at different temperatures ; ( c ) same as b , however for mo / fe / mo / w wires , see text ; ( d - f ) in - plane field - cooled ( fc ) and zero - field - cooled ( zfc ) magnetization of ( d ) mo / fe / w ( e ) mo / fe / mo / w and ( f ) al / fe / w wires . a schematic illustration of the samples is shown in insets.,width=323 ]
the nucleosynthesis of nuclei above the iron peak ( @xmath7 ) proceeds mainly by neutron capture in the so - called @xmath8- and @xmath9-processes . in principle , neutron - deficient nuclei in this mass region ( the so - called @xmath10 nuclei , see ref . @xcite for a complete list ) can be produced from more neutron - rich seed nuclei either by the removal of neutrons or by the addition of protons . however , due to the coulomb barrier proton capture is strongly suppressed . there is general agreement that heavy neutron - deficient nuclei with masses above @xmath11 have been synthesized by photodisintegration of previously produced nuclides at sufficiently high temperatures of @xmath12k ( @xmath13 , with @xmath14 being the temperature in billion degrees ) . this so - called @xmath4-process or p - process is discussed in detail in @xcite . several astrophysical sites for the @xmath4-process have been proposed , whereby the oxygen- and neon - rich layers of type ii supernovae seem to be good candidates @xcite , but also exploding carbon - oxygen white dwarfs have been suggested @xcite . however , no definite conclusions have been reached yet . nucleosynthesis calculations for the @xmath4-process require a huge number of reaction rates . up to about 1000 nuclei and 10000 reaction rates have been included in previous reaction networks @xcite . recently , the complete network of the first self - consistent study of the @xmath4-process including all relevant nuclei up to bi amounted to about 3000 nuclei and all their respective reactions @xcite . unfortunately , almost none of these reaction rates have been measured and the astrophysical calculations have to rely completely on statistical model calculations ( e.g. refs . @xcite ) . of special importance are ( @xmath4,@xmath1)/(@xmath4,n ) branchings which determine abundance ratios of certain nuclides which , in turn , can in some cases be compared to abundances found in meteoritic inclusions @xcite . it has been stated that the uncertainties for ( @xmath4,@xmath1 ) reaction rates are huge @xcite . the determination of the @xmath1-nucleus potential at astrophysically relevant energies helps to reduce the uncertainties of the calculations significantly @xcite . ( @xmath4,n ) reaction rates have been measured in a recent work using a quasi - thermal photon spectrum , and rough agreement between theoretical predictions and the measured rates was found @xcite . the overall agreement between the calculated and the observed abundance patterns of the @xmath10 nuclei is relatively good . however , the mass region @xmath15 is generally underproduced in the nucleosynthesis calculations @xcite . the production among others depends on the neutron - producing @xmath16ne(@xmath1,n)@xmath17 mg reaction rate ( which may enhance the @xmath8-process seed nuclei for the @xmath4-process @xcite ) and on the photodisintegration rates in the @xmath4-process but it remains unclear whether the underproduction can be cured by a change in those rates @xcite . other explanations for this discrepancy include additional production mechanisms like neutrino - induced nucleosynthesis @xcite and additional production sites like the rapid proton capture ( @xmath18 ) process on accreting neutron stars ( e.g. @xcite ) . however , it is still an open question whether @xmath18-material can be ejected into the interstellar medium in sufficient quantities from these x - ray bursters @xcite . the motivation of the experimental determination of the @xmath1-nucleus potential for @xmath0mo is twofold . the determination of the @xmath1-nucleus potential at energies below the coulomb barrier is limited in general because ( @xmath19 ) the experimental data show only small deviations from the rutherford cross section , and ( @xmath20 ) the optical potentials have ambiguities . an experiment on @xmath0mo allows to extend the systematic study of @xmath1-nucleus potentials @xcite to lower energies . the second motivation refers directly to the production of the @xmath10 nuclei @xmath0mo and @xmath21mo . a possible reaction path leading to the production of @xmath0mo and @xmath21mo is shown in fig . [ fig : gammaproc ] . photodisintegration reactions of the nucleus @xmath3ru can lead to the production of ( @xmath19 ) @xmath0mo by @xmath3ru(@xmath4,@xmath1)@xmath0mo and ( @xmath20 ) @xmath21mo by @xmath3ru(@xmath4,n)@xmath22ru(@xmath4,n)@xmath21ru(2@xmath23 @xmath24)@xmath21mo . if this reaction path were the only production mechanism for @xmath0mo and @xmath21mo , the abundance ratio between @xmath0mo and @xmath21mo would be directly related to the ratio of ( @xmath4,@xmath1 ) and ( @xmath4,n ) reaction rates of @xmath3ru . in this case the ratio @xmath21mo/@xmath0mo could be a thermometer for the @xmath4-process because of the temperature dependence of the ( @xmath4,n)/(@xmath4,@xmath1 ) branching ratio . however , for a quantitative analysis contributions of the @xmath18-process to @xmath0mo and @xmath21mo and the weak @xmath8-process contribution to @xmath21mo have to be known . the choice of the measured energies at about 13 , 16 , and 19mev has the following reasons . the astrophysically relevant energy window for ( @xmath4,@xmath1 ) reactions at @xmath25 is of the order of @xmath26mev corresponding to @xmath27mev for the reverse reaction @xmath0mo(@xmath1,@xmath4)@xmath3ru . scattering experiments at these low energies are possible ; however , a reliable determination of optical potentials is impossible because of the dominating coulomb interaction . the height of the coulomb barrier is about 15 mev . we decided to measure at several energies above and below the coulomb barrier to extract the optical potential and its energy dependence at energies as close as possible to the astrophysically relevant energy range . in the following paper we first present our experimental setup ( sect . [ sec : exp ] ) . the experimental results are analyzed using systematic folding potentials , and discrete and continuous ambiguities are discussed in detail ( sect . [ sec : om ] ) . the optical potential at astrophysically relevant energies is determined by extrapolation using the systematic behavior of @xmath1-nucleus potentials @xcite , and the ( @xmath4,@xmath1 ) reaction rates are calculated ( sect . [ sec : extra ] ) . finally , some conclusions are given ( sect . [ sec : summ ] ) . a preliminary analysis of this experiment was presented already in @xcite . the experiment was performed at the cyclotron laboratory at atomki , debrecen . we used the @xmath28 diameter scattering chamber which is described in detail in ref . @xcite . here we discuss only those properties which are important for the present experiment . a similar setup has been used in our previous @xmath29sm(@xmath1,@xmath1)@xmath29sm experiment @xcite . the @xmath0mo targets were produced by evaporation of 97.33% enriched @xmath0moo@xmath30 on a thin ( 20@xmath31g/@xmath32 ) carbon backing directly before the beamtime at the target laboratory of atomki . the target was mounted on the target holder in the center of the scattering chamber . the surface of the evaporated @xmath0moo@xmath30 turned out to be not very flat leading to relatively broad low - energy tails in the spectra of the elastically scattered @xmath1 particles ( see fig . [ fig : spectra ] ) . the target which was used during the whole experiment had a thickness of about 200@xmath31g/@xmath32 . the target stability was monitored during the whole experiment to avoid systematic uncertainties from changes in the target foil . a remote - controlled target ladder was placed in the center of the scattering chamber . additionally , two apertures were mounted on the target holder to check the beam position and the size of the beam spot directly at the position of the target . the two apertures had a width and height of 2x6 mm@xmath33 and 6x6 mm@xmath33 , respectively . the apertures were placed at the target position instead of the @xmath0moo@xmath30 target before and after each variation of the beam energy and beam current . the beam was optimized until no current could be measured on the larger aperture , and the current on the smaller aperture was minimized ( typically less than 1na compared to about 300na beam current ) . the width of the beam spot was smaller than 2 mm during the whole experiment , which is very important for the precise determination of the scattering angle . note that the relatively poor determination of the height of the beam spot does not disturb the claimed precision of the scattering angle ( see sect . [ subsec : angle ] ) . for the measurement of the angular distribution we used four silicon surface - barrier detectors with an active area @xmath34 and thicknesses between @xmath35 and @xmath36 . the detectors were mounted on an upper and a lower turntable , which can be moved independently . on each turntable two detectors were mounted at an angular distance of @xmath37 . directly in front of the detectors apertures were placed with the dimensions 1.25 mm @xmath23 5.0 mm ( lower detectors ) and 1.0 mm @xmath23 6.0 mm ( upper detectors ) . together with the distance from the center of the scattering chamber @xmath38 ( lower detectors ) and @xmath39 ( upper detectors ) this results in solid angles from @xmath40 to @xmath41 . the ratios of the solid angles of the different detectors were determined by overlap measurements with an accuracy better than 1% . additionally , two detectors were mounted at the wall of the scattering chamber at a distance of @xmath42 mm and at a fixed angle of @xmath43 ( left and right side relative to the beam direction ) . these detectors were used as monitor detectors to normalize the measured angular distribution and to determine the precise position of the beam spot on the target . the solid angle of both monitor detectors is @xmath44 . the signals from all detectors were processed using charge - sensitive preamplifiers ( pa ) , which were mounted directly at the scattering chamber . the output signal was further amplified by a main amplifier ( ma ) and fed into adcs . the data were collected using the commercially available system wintmca which provides an automatic deadtime control which was found to be reliable in a previous experiment @xcite . for the coincidence measurements ( sect . [ subsec : angle ] ) additionally the bipolar signals of the mas were fed into timing single channel analyzers ( tsca ) , and the unipolar outputs of the mas were gated using linear gate stretchers ( lgs ) . the energy resolution of the detectors was tested before the experiment using a mixed @xmath1 source and values better than 20kev were measured . during the experiment the achieved energy resolution is determined mainly by the energy spread of the primary beam and by the thickness and flatness of the target . depending on the measured angle , the achieved energy resolution was between 0.5% and 2% corresponding to @xmath45 at @xmath46 . two typical spectra are shown in fig . [ fig : spectra ] . the relevant peaks from elastic @xmath0mo-@xmath1 scattering are well separated from inelastic and from background peaks . the angular calibration of the setup is of crucial importance for the precision of a scattering experiment at energies close to the coulomb barrier because the rutherford cross section depends sensitively on the angle with @xmath47 . a small uncertainty of @xmath48 in the determination of @xmath49 leads to a cross section uncertainty of 2.0% ( 1.0% ; 0.6% ) at an angle @xmath50 ( @xmath51 ; @xmath52 ) . the following methods were applied to measure the precise scattering angle @xmath49 . the position of the beam on the target was continuously controlled by two monitor detectors . the precise position of the beam spot was derived from the ratio of the count rates in both monitor detectors . typical corrections were smaller than one millimeter , leading to corrections in @xmath49 of the order of @xmath48 . however , because of the minor beam quality at the 16mev measurement , larger corrections had to be applied to this angular distribution leading to larger uncertainties in the determination of the optical potential . the position of the four detectors was calibrated using the steep kinematics of @xmath53h(@xmath1,@xmath1)@xmath53h scattering at forward angles ( @xmath54 ) @xcite . the results of our previous experiment could be confirmed within the uncertainties @xcite . finally , we measured a kinematic coincidence between elastically scattered @xmath1 particles and the corresponding @xmath55c recoil nuclei using a carbon backing without molybdenum as target . one detector was placed at @xmath56 ( left side relative to the beam axis ) , and the signals from elastically scattered @xmath1 particles on @xmath55c were selected by a tsca . this tsca output was used as gate for the signals from another detector which was moved around the corresponding @xmath55c recoil angle @xmath57 ( right side ) . the maximum recoil count rate was found almost exactly at the expected angle ( see fig . [ fig : coin ] ) . in summary , the overall uncertainty of the angles @xmath49 in this experiment is about @xmath48 for the measurements at 13 and 19mev and about @xmath58 for the 16mev measurement . three angular distributions have been measured at energies of @xmath59 , 16.42 , and 13.83mev . the beam current was between @xmath60na and @xmath61na . the experiment covers the full angular range from forward angles of @xmath50 to backward angles of @xmath62 in steps of about @xmath63 at all three energies . the statistical uncertainties of each data point vary from @xmath64 at forward angles to about @xmath65 at backward angles . the count rates @xmath66 in the four detectors have been normalized to the number of counts in the monitor detectors @xmath67 : @xmath68 with @xmath69 being the solid angles of the detectors . these measured cross sections have been transferred to the center - of - mass system . the cross section at the monitor position @xmath70 is given by the rutherford cross section . the relative measurement eliminates the typical uncertainties of absolute measurements which come mainly from changes in the target and from the beam current integration . nevertheless , the beam current was measured by standard current integration in the faraday cup , and the absolute value of the cross section was consistent with the measured relative cross sections . the three angular distributions are shown in fig . [ fig : results ] . the lines are the result of optical model calculations ( see sect . [ sec : om ] ) . the measured cross sections cover five orders of magnitude between the highest ( forward angles at @xmath71mev ) and the smallest cross section ( backward angles at @xmath72mev ) with almost the same accuracy . further details of the experimental set - up and the data analysis can be found in ref . @xcite . the theoretical analysis of the scattering data was performed in the framework of the optical model ( om ) . the complex optical potential is given by @xmath73 where @xmath74 is the coulomb potential , and @xmath75 and @xmath76 are the real and the imaginary part of the nuclear potential , respectively . the real part of the optical potential was calculated by a double folding procedure : @xmath77 where @xmath78 , @xmath79 are the densities of projectile and target , respectively , and @xmath80 is the effective nucleon - nucleon interaction taken in the well - established ddm3y parametrization @xcite . details about the folding procedure can be found in refs . @xcite . the folding integral in eq . ( [ eq : fold ] ) was calculated using the code dfold @xcite . the resulting real part of the optical potential @xmath75 is derived from the folding potential @xmath81 by two minor modifications : @xmath82 firstly , the strength of the folding potential is adjusted by the usual strength parameter @xmath83 with @xmath84 . this leads to volume integrals of the real potential [ see eq . ( [ eq : vol ] ) ] of about @xmath85mev@xmath86 in the analyzed energy range @xcite . secondly , the densities of the @xmath1 particle and the @xmath0mo nucleus were derived from the experimentally known charge density distributions @xcite , assuming identical proton and neutron distributions . for @xmath87 nuclei up to @xmath88zr ( @xmath89 , @xmath90 ) this assumption worked well @xcite . however , to take the possibility into account that the proton and neutron distributions are not identical in the nucleus @xmath0mo ( @xmath91 , @xmath90 ) a scaling parameter @xmath92 for the width of the potential has been introduced , which remains very close to unity . for a comparison of different potentials we use the integral parameters volume integral per interacting nucleon pair @xmath93 and the root - mean - square ( rms ) radius @xmath94 , which are given by : @xmath95 ^{1/2 } \label{eq : rms}\end{aligned}\ ] ] for the real part of the potential @xmath75 , and corresponding equations hold for @xmath76 . @xmath96 and @xmath97 are the nucleon numbers of projectile and target . note that in the discussion of volume integrals @xmath98 usually the negative sign is neglected ; also in this paper all @xmath98 values are actually negative . the values for the folding potential @xmath99 ( with @xmath100 ) are @xmath101 and @xmath102 . the coulomb potential is taken in the usual form of a homogeneously charged sphere where the coulomb radius @xmath103 is chosen identically with the rms radius of the folding potential @xmath99 : @xmath104 . different parametrizations of the imaginary part of the optical potential were chosen . volume and surface woods - saxon ( ws ) potentials are defined by the following equations @xmath105 with @xmath106 the depth @xmath107 , the radius parameter @xmath108 , and the diffuseness @xmath109 have been adjusted to the experimental elastic scattering data . fourier - bessel ( fb ) potentials are given by @xmath110 with the cutoff radius @xmath111 . again , the fourier - bessel coefficients @xmath112 are adjusted to the experimental data . the elastic scattering cross sections were calculated from optical potentials with the computer code a0 @xcite . the code allows a variation of the potential parameters and determines the best - fit values from a @xmath113 test . in the first analysis the potential parameters @xmath83 and @xmath92 of the real part were kept close to the expected values from the systematic study @xcite : @xmath84 and @xmath114 . several parametrizations of the imaginary potential were tested . it was found that different imaginary potentials reproduced the experimental data with similar quality . five different fits are shown in fig . [ fig : ambcont ] and the potential parameters are given in tables [ tab : final ] , [ tab : ws ] and [ tab : fb ] . it turns out that the real potential is practically identical in all these fits , but the shape of the imaginary part shows strong variations . the five imaginary potentials are shown in fig . [ fig : imagpot ] . since we want to determine the optical potential at astrophysically relevant energies we have to extrapolate from the present measurements . because of the oscillating behavior of the fourier - bessel potentials we decided to use the combination of a volume and surface woods - saxon potential as basis for the extrapolation . the @xmath113 obtained with this potential is practically identical to the @xmath113 obtained from the fourier - bessel potentials . the calculations with a pure volume woods - saxon or a pure surface woods - saxon show significantly worse @xmath113 values . in a second analysis the strength parameter @xmath83 and the width @xmath92 of the real potential were varied in a wider range . it was found that it is not possible to get a good fit to the data when the @xmath92 parameter deviates from 1.0 by more than a few per cent . however , a variation of @xmath83 leads to the known so - called `` family problem '' . it is possible to obtain comparable fits to the experimental data with various @xmath83 parameters . this phenomenon was discussed in detail for a similar experiment in @xcite . in fig . [ fig : family ] we present the @xmath113 values which were obtained from the following procedure : ( @xmath19 ) the parameter @xmath83 was varied from about 0.5 to 3.5 ; ( @xmath20 ) the width parameter @xmath92 and the imaginary part of the potential ( consisting of a combination of volume and surface woods - saxon potentials ) were adjusted to the experimental data for each value of the strength parameter @xmath83 . one can clearly see the families 2 , 3 , 4 , and 5 as minima in @xmath113 , corresponding to @xmath83 values of @xmath115 . note that the minima are more pronounced than in the previous @xmath29sm(@xmath1,@xmath1)@xmath29sm experiment @xcite because the ratio @xmath116 between the energy @xmath117 and the coulomb barrier @xmath118 is much higher in this @xmath0mo experiment . it is not possible to extract the correct family from these experimental data only . but together with the systematic behavior of the volume integrals found in @xcite we can decide that `` family no . 4 '' ( @xmath119 ) should be used for the description of the experimental scattering data and for the extrapolation to astrophysically relevant energies ( see sect . [ sec : extra ] ) . as mentioned above , family no . 4 with @xmath120mev@xmath86 corresponds to the values of about @xmath85mev@xmath86 which are expected from the systematics of @xmath1-nucleus potentials and also from other systems @xcite . neither family no . 3 with @xmath121mev@xmath86 nor family no . 5 with @xmath122mev@xmath86 fit into the systematics . numerical problems in the fitting routine showed up for very shallow and very deep real potentials , and a clear determination of families 1 and @xmath123 was not possible . one further interesting fact has to be mentioned . the real potentials corresponding to the families @xmath124 are shown in fig . [ fig : onepoint ] . the potentials from families @xmath125 which are well - defined as minima in @xmath113 ( see fig . [ fig : family ] ) have the same depth @xmath126mev at the radius @xmath127fm . however , not all potentials which have this depth do describe the data equally well ; additionally , one has to find a minimum in @xmath113 in fig . [ fig : family ] . this behavior of a so - called `` one - point potential '' has been observed in several experiments , and the relevant radius has also been called `` sensitivity radius '' ( see e.g. @xcite ) ; however , to our knowledge the additional restriction of a significant minimum in @xmath113 has only been observed in the analysis of the @xmath29sm(@xmath1,@xmath1)@xmath29sm data so far @xcite which has been performed at a similar energy . the procedure described in the previous sect . [ subsubsec:19mev ] was repeated for the lower energies of 13 and 16mev , and similar results were obtained . the imaginary part of the potential for all energies is taken as a combination of volume and surface woods - saxon potentials . the potential parameters are listed in table [ tab : final ] , and the calculations have been compared to the experimental data already in fig . [ fig : results ] . however , the 16mev data do not fit very well into the systematic behavior shown in fig . [ fig : volume ] . the potential extracted from the 16mev data has larger uncertainties than at the other energies because of experimental problems ( see sect . [ subsec : angle ] ) . two angular distributions at energies of about 30mev are available in literature @xcite . unfortunately , both angular distributions show systematic deviations between each other , and both distributions have not been measured in the full angular range , but in the ranges @xmath128 @xcite and @xmath129 @xcite . if one adjusts the potential parameters to these discrepant angular distributions , discrepant optical potentials are obtained . the potential parameters are labeled in table [ tab : liter ] . however , the potentials extracted from the data of @xcite fit into the systematics , whereas the data of @xcite do not fit . in both cases the limited angular range restricts the sensitivity of the potential parameters significantly . in fig . [ fig : volume ] only the volume integrals derived from @xcite are shown . the excitation function for @xmath0mo(@xmath1,@xmath1)@xmath0mo has been measured by eisen _ @xcite at @xmath130 from 7 to 16mev . we have calculated this excitation function from our best - fit potentials at 13 , 16 , and 19mev , and we find excellent agreement between the experimental and the calculated excitation function . the measured and the calculated excitation function at @xmath131 are shown in fig . [ fig : excit ] . the calculation with the 13mev potential underestimates the deviation from the rutherford cross section at higher energies . such a behavior can be expected because of the smaller volume integral of the imaginary part in the 13mev data corresponding to weaker absorption . however , the measured scattering cross sections at one special backward angle do not contain enough information to fix the optical potential and its energy dependence . a set of experimental data corresponding to the reaction @xmath0mo(@xmath1,@xmath132)@xmath22ru is available in @xcite ( sect.iv b , fig.6a ) . we have calculated the cross section from our model in the measured energy range , and found very good conformity between our calculations and the existing experimental data . however , the available data from the different experiments show discrepancies , which make it difficult to fix the energy dependence of the optical potential . also the existing reaction data do not cover the astrophysically interesting energy range ( between 7 and 9mev , see tab . [ tab : gamow ] ) , which would be helpful in order to confirm our predictions ( see sect . [ sec : extra ] ) . volume integrals for various @xmath1-nucleus potentials in a broad range of masses and energies are shown in fig . [ fig : volume ] for the real ( [ fig : volume]a ) and the imaginary part ( [ fig : volume]b and [ fig : volume]c ) of the optical potential . the systematic behavior of volume integrals is also confirmed for lighter target nuclei @xcite and in various other systems which have been analyzed recently @xcite . for the extrapolation of the optical potential to astrophysically relevant energies ( sect . [ sec : extra ] ) parametrizations of the real and imaginary volume integrals are needed . as can be seen from fig . [ fig : volume]a , there is only a weak energy dependence of the real volume integral @xmath93 at energies below the coulomb barrier . as well as in ref . @xcite , a gaussian parametrization is adjusted to the new data @xmath133\ ] ] with @xmath134=337mev@xmath86 , @xmath135=21.55mev and @xmath136=147.01mev , leading to a curve ( full line ) which is somewhat flatter than the one proposed in ref . @xcite ( dotted line ) . the uncertainties for extrapolations to lower energies are of the order of less than 5% corresponding to about @xmath137mev@xmath86 . furthermore , the shape of the real potential is given by the folding procedure ( sect . [ subsec : fold ] ) . this means that the real part of the @xmath1-nucleus optical potential can be determined at energies below the coulomb barrier with relatively small uncertainties because ( @xmath19 ) continuous ambiguities can be avoided using the folding potential and ( @xmath20 ) discrete ambiguities can be resolved from the systematic behavior of @xmath1-nucleus potentials . the situation for the imaginary part of the potential is much worse . the volume integral @xmath138 of the imaginary part depends strongly on the energy because many reaction channels open at energies around the coulomb barrier . different parametrizations have been proposed @xcite . as an example we present the so - called brown - rho ( br ) parametrization @xcite @xmath139 with the excitation energy @xmath135 of the first excited state . the saturation parameter @xmath140 and the rise parameter @xmath136 are adjusted to the experimentally derived values . another fermi - like parametrization of the imaginary volume integral , first introduced in ref . @xcite , reads @xmath141 } } \label{eq : gg}\ ] ] with a similar saturation value @xmath140 and the parameters @xmath142 and @xmath143 . the latter shape was also used for an attempt to derive a global @xmath1-potential @xcite , with @xmath142 and @xmath143 depending on @xmath135 . however , the line derived with the parameters given in @xcite shows clear deviations from the new @xmath0mo data . therefore , we have adjusted this fermi - like function to the experimental data . both parametrizations utilizing our fit parameters are shown in fig . [ fig : volume]b ( brown - rho ) and [ fig : volume]c ( fermi - like ) for our new @xmath0mo data . the parameters are listed in table [ tab : brgg ] . in the following , we will always use the parameters given in that table for the two descriptions unless specified otherwise . in general , the shapes of the br and the fermi parametrizations are quite similar : there is a saturation value @xmath140 and a parameter that describes the steep rise of @xmath138 : @xmath136 for br and @xmath143 for the fermi shape . however , there are also important differences because the br parametrization leads to a somewhat flatter rise of @xmath138 than that of the fermi function , and consequently , the extrapolation to lower energies is lower for the fermi parametrization than for br . consequences of these small differences will be given in sect . [ subsec : rate ] . nevertheless , it should be noted that the br parametrization only contains two free parameters , because @xmath135 is fixed , whereas the fermi shape has three free parameters , in principle . the shown ambiguities do not allow to determine the shape of the imaginary part . these ambiguities reduce the reliability of extrapolations to lower energies . a more stringent determination of the shape of the imaginary part requires extremely precise scattering data over a wide range of energies . a scattering experiment at about 50mev might help to reduce these uncertainties and to find the best parametrization of the imaginary volume integrals . the astrophysical decay rate @xmath144 is given by @xmath145 with the speed of light @xmath146 , the cross section @xmath147 of the ( @xmath4,@xmath1 ) reaction , and the photon density @xmath148 of a thermal photon bath at temperature @xmath149 @xmath150 the integrand of eq . ( [ eq : tau ] ) can be analyzed under the assumption that the astrophysical s - factor of the reverse ( @xmath1,@xmath4 ) reaction is constant : @xmath151 . then the maximum of the integrand in eq . ( [ eq : tau ] ) is found at the energy @xmath152 with @xmath153 and the threshold energy @xmath154 for the ( @xmath4,@xmath1 ) reaction . the most effective energy @xmath155 for ( @xmath4,@xmath1 ) reactions is given by the energy of the well - known gamow window @xmath156 for the inverse ( @xmath1,@xmath4 ) reaction plus the separation energy @xmath154 of the @xmath1 particle . note that the energy @xmath155 is the energy of the photon , whereas the energy @xmath156 is the center - of - mass energy in the system @xmath0mo - @xmath1 . the astrophysically relevant energies for the system @xmath0mo - @xmath1 are listed in table [ tab : gamow ] . in all astrophysical applications reaction rates are input only for reactions with positive @xmath157 value and the inverse rate is then computed by applying detailed balance ( see e.g. @xcite ) . that way , numerical stability of the reaction network is guaranteed and the proper equilibria of forward and reverse rates can be attained for a given channel . the rates in the two directions depend linearly on each other and thus the change of , say , the @xmath1 potential equally influences both , in our case the @xmath1 capture as well as the photodisintegration with @xmath1 emission . this relation is valid provided that stellar rates are used in both directions , accounting for thermal excitation of the respective targets . because of that fact , in the following sections we make use of rate ratios so that the conclusions apply to the forward and inverse rates as well . as stated in sect . [ subsec : disc ] , the extrapolation of the real potential can be performed reliably leading to @xmath158mev@xmath86 at astrophysically relevant energies with an uncertainty of about 5% . the corresponding strength parameter is @xmath159 . the width parameter was fixed at @xmath160 . the extrapolation of the imaginary part was performed as follows . in a first step the volume integral @xmath138 was determined from the br and fermi parametrizations leading to @xmath161mev@xmath86 ( br ) and @xmath162mev@xmath86 ( fermi ) at @xmath163mev ( corresponding to @xmath164 ) . the average of these values is @xmath165mev@xmath86 which was used for the following calculations . the shape of the potential was taken as sum of volume and surface woods - saxon potentials where the radius parameter @xmath108 and the diffuseness @xmath109 were estimated from the experimental data . the contribution of the volume term to @xmath138 is assumed to be 30% , and the surface term contributes to 70% . this ratio is determined from the experimental scattering data at @xmath166 , 19 , and 30mev . the effect of a variation of the relative contributions of volume and surface term to @xmath138 will be discussed below . these and other variations of the potentials allow an estimation of the uncertainties of the calculated reaction rates . the variation of the reaction rates when using various potentials is shown in tables [ tab : rateratio ] and [ tab : variation ] . in table [ tab : rateratio ] the ratios of rates obtained with the different potentials in respect to a standard rate ( taken from refs . @xcite and using an @xmath1 potential from ref . @xcite ) are shown . as can be seen , the rate calculated with the global potential of ref . @xcite is lower by about two orders of magnitude than the standard rate . however , it was already stated above that this potential does not describe the @xmath0mo data at higher energies . a simple equivalent square well potential @xcite also yields a factor of 8 lower cross sections but neither does it describe the data nor is it considered to be reliable for this application @xcite . when using the two potentials with the extrapolated parameters for @xmath5 and @xmath6 from table [ tab : extrapol ] , a reduction of the rate of about @xmath167% is found . case ab explores the dependence on the geometry of the potential . a change in the geometry parameters of only @xmath168% leads to a variation in the ratio of @xmath169% in the rate ratios which underlines the importance of additional scattering experiments to determine the shape of the imaginary optical potential . however , it should be mentioned that case ab is not fully consistent within our approach because it has a slightly different volume integral @xmath138 and rms radius due to the unchanged depths of the volume and surface terms , but the differences are only of the same order of magnitude as those in the geometry parameters . the sensitivity of the rates to variations in the extrapolated volume integral @xmath138 and the relative contributions of volume and surface term are studied in table [ tab : variation ] . here the varied rates are compared to the rate obtained in case a of table [ tab : rateratio ] . the ratios are given in the temperature range @xmath170 in order to show the temperature dependence of those effects although strictly speaking the potential was derived assuming @xmath5 . the contribution of the surface term to @xmath138 was varied within a reasonable range of @xmath171% . this resulted in a variation of the rate of about @xmath172% . another uncertainty is introduced by the fact that we assumed the extrapolated @xmath138 to be the mean between the value obtained by the br and fermi parametrizations . using the higher br value of @xmath173 mev@xmath86 increases the rate by 8% while using the lower value of @xmath174 mev@xmath86 a suppression by about 10% is obtained . thus , the error introduced by the different shapes of the parametrizations used for the extrapolation of @xmath138 to low energies is dominating but still within satisfactory accuracy . closing this section we conclude that the recommended rates are case a for @xmath5 and case b for @xmath6 from table [ tab : rateratio ] with an error of 16% , mainly introduced by the ambiguities of the extrapolation of the imaginary part down to the relevant energies . the recommended rate is roughly a factor of two lower than the standard rate given in previous tabulations @xcite . we measured the elastic scattering cross section of @xmath0mo(@xmath1,@xmath1)@xmath0mo in a wide angular range at energies of @xmath2 13 , 16 , and 19mev . additionally , data from literature have been analyzed @xcite . the real and imaginary parts of the optical potential for the system @xmath0mo - @xmath1 have been extracted from the data and extrapolated down to the astrophysically relevant energies around and below the coulomb barrier . the result fits well into the known systematic behavior of @xmath1-nucleus folding potentials . the extrapolation of the imaginary part is not unique but our study shows that the use of two different energy dependencies introduces an error in the obtained rate of not more than 15% . the derived stellar rates ( for @xmath0mo(@xmath1,@xmath4)@xmath3ru as well as @xmath3ru(@xmath4,@xmath1)@xmath0mo ) are @xmath175% of the rates given in refs . @xcite at stellar temperatures @xmath176 . assuming the @xmath3ru(@xmath4,n)@xmath22ru rate to remain unchanged , this would lead to a corresponding decrease in the abundance ratio @xmath177 with respect to an abundance ratio calculated with the previous rate in the @xmath4-process ( as e.g. in @xcite ) . it is interesting to note that many network calculations for the @xmath4-process @xcite show an overproduction of @xmath0mo relative to @xmath21mo which may be reduced by the results of this work . however , as mentioned in the introduction , a complete analysis has not only to follow the @xmath4-process consistently but also to account for the possible @xmath18- and @xmath8-process contributions . this is beyond the scope of this paper . we would like to thank the cyclotron team of atomki for the excellent beam during the experiment . two of us ( p. m. , m. b. ) gratefully acknowledge the kind hospitality at atomki . f. is a bolyai fellow . t. r. is a profil professor ( swiss nsf grant 2124 - 055832.98 ) . this work was supported by otka ( t034259 ) , the swiss nsf ( grant 2000 - 061822.00 ) , and the u.s . nsf ( contract nsf - ast-97 - 31569 ) . d. l. lambert , astron . astrophys . * 3 * ( 1992 ) 201 . m. arnould and k. takahashi , rep . phys . * 62 * , 395 ( 1999 ) . k. langanke , nucl . phys . * a564 * ( 1999 ) 330c . g. wallerstein _ et al . _ , rev . * 69 * ( 1997 ) 995 . k. ito , prog . * 26 * , 990 ( 1961 ) . r. l. macklin , , 353 ( 1970 ) . s. e. woosley and w. m. howard , suppl . * 36 * , 285 ( 1978 ) . m. rayet , n. prantzos , and m. arnould , astron . astrophys . * 227 * , 271 ( 1990 ) . m. rayet , m. arnould , m. hashimoto , n. prantzos , and k. nomoto , astron . astrophys . * 298 * , 517 ( 1995 ) . n. prantzos , m. hashimoto , m. rayet , and m. arnould , astron . astrophys . * 238 * , 455 ( 1990 ) . w. m. howard , b. s. meyer , and s. e. woosley , , l5 ( 1991 ) . v. costa , m. rayet , r. a. zappal , and m. arnould , astron . astrophys . * 358 * , l67 ( 2000 ) . a. heger , r. d. hoffman , t. rauscher , s. e. woosley , proc . _ 10@xmath178 workshop on nuclear astrophysics _ , w. hillebrandt , e. mller , mpa / p12 , p. 105 , mpa , garching , 2000 . t. rauscher , a. heger , r. d. hoffman , s. e. woosley , , to be submitted ( 2001 ) . t. rauscher and f .- k . thielemann , at . data nucl . data tables * 75 * , 1 ( 2000 ) . s. goriely , proc . _ 10@xmath178 int . capture gamma - ray spectroscopy _ s. wender , aip conference proceedings * 529 * , 287 ( 2000 ) . t.rauscher and f .- k . thielemann , at . data nucl . data tables * 79 * , in press ( 2001 ) . s. e. woosley and w. m. howard , , l21 ( 1990 ) . t. rauscher , f .- k . thielemann , and h. oberhummer , , l37 ( 1995 ) . e. somorjai , zs . flp , a. z. kiss , c. e. rolfs , h .- trautvetter , u. greife , m. junker , s. goriely , m. arnould , m. rayet , t. rauscher , h. oberhummer , astron . astrophys . * 333 * , 1112 ( 1998 ) t. rauscher , proc . _ nuclei in the cosmos v _ , volos , greece , ed . n. prantzos and s. harissopulos , p. 484 , editions frontires , paris , 1998 . p. mohr , t. rauscher , h. oberhummer , z. mt , zs . flp , e. somorjai , m. jaeger , and g. staudt , , 1523 ( 1997 ) . m. gledenov , p. e. koehler , j. andrzejewski , k. h. guber , and t. rauscher , , 042801(r ) ( 2000 ) . p. mohr , k. vogt , m. babilon , j. enders , t. hartmann , c. hutter , t. rauscher , s. volz , a. zilges , phys . b * 488 * , 127 ( 2000 ) . k. vogt , p. mohr , m. babilon , j. enders , t. hartmann , c. hutter , t. rauscher , s. volz , a. zilges , , 055802 ( 2001 ) . s. e. woosley , private communication ( 2001 ) . r. d. hoffman , s. e. woosley , g. m. fuller , and b. s. meyer , , 478 ( 1996 ) . h. schatz , l. bildsten , a. cumming , and m. wiescher , , 1014 ( 1999 ) . r. k. wallace and s. e. woosley , suppl . * 45 * , 389 ( 1981 ) . h. schatz , a. aprahamian , v. barnard , l. bildsten , a. cummings , m. ouellette , t. rauscher , f .- k . thielemann , m. wiescher , , 3471 ( 2001 ) . u. atzrott , p. mohr , h. abele , c. hillenmayer , and g. staudt , , 1336 ( 1996 ) . p. mohr , , 045802 ( 2000 ) . p. mohr , m. babilon , d. galaviz , a. zilges , zs . flp , gy . gyrky , z. mt , e. somorjai , l. zolnai , and h. oberhummer , nucl . * a688 * , 424 ( 2001 ) . z. mt , s. szilgyi , l. zolnai , . bredbacka , m. brenner , k .- kllmann , and p. manngrd , acta phys . hung . * 65 * , 287 ( 1989 ) . p. mohr , c. hutter , k. vogt , j. enders , t. hartmann , s. volz , and a. zilges , europ . j. a * 7 * , 45 ( 2000 ) . d. galaviz , diploma thesis , tu darmstadt , unpublished . h. abele , univ . tbingen , computer code dfold , unpublished . h. de vries , c. w. de jager , and c. de vries , atomic data and nuclear data tables * 36 * , 495 ( 1987 ) . h. abele , univ . tbingen , computer code a0 , unpublished . p. mohr , , 061601(r ) ( 2000 ) . d. t. khoa , w. von oertzen , h. g. bohlen , and f. nuoffer , nucl . * a672 * , 387 ( 2000 ) . a. a. oglobin _ et al . _ , , 044601 ( 2000 ) . c. p. silva _ et al _ , nucl . a679 * , 287 ( 2001 ) . k. matsuda , y. awaya , n. nakanishi and s. takeda , j. phys . jap . * 33 * , 2 ( 1972 ) . e. j. martens and a. m. bernstein , nucl . * a117 * , 241 ( 1968 ) . y. eisen _ et al . _ , nucl . * a236 * , 327 ( 1974 ) . b. strohmaier , m. fabender and s. m. qaim , , 2654 ( 1997 ) . d. t. khoa , , 034007 ( 2001 ) . g. e. brown and m. rho , nucl . * a372 * , 397 ( 1981 ) . c. grama and s. goriely , proc . _ nuclei in the cosmos v _ , volos , greece , ed . n. prantzos and s. harissopulos , p. 463 , editions frontires , paris , 1998 . l. mcfadden and g. r. satchler , nucl . phys . * 84 * , 177 ( 1966 ) . j. holmes , s. e. woosley , w. fowler , and b. zimmerman , atomic data nucl . data tables * 18 * , 305 ( 1976 ) . cccccccccc fit & @xmath117 & @xmath179 & @xmath180 & @xmath181 & & @xmath182 & @xmath183 & @xmath184 & + 1 & 19 & -1.584 & 1.7667 & 0.2659 & & 334.25 & 1.2605 & 0.2073 & + - & 16 & -9.558 & 1.668 & 0.248 & & 308.20 & 1.348 & 0.099 + - & 13 & -5.128 & 1.656 & 0.002 & & 467.06 & 1.369 & 0.071 + + fit & @xmath117 & @xmath83 & @xmath92 & @xmath93 & @xmath94 & @xmath138 & @xmath185 & @xmath186 + & ( mev ) & & & @xmath187 & @xmath188 & @xmath187 & @xmath188 & + 1 & 19 & 1.257 & 1.003 & 337.2 & 4.991 & 86.2 & 5.806 & 2.15 + - & 16 & 1.346 & 0.9974 & 357.8 & 4.976 & 85.9 & 5.992 & 4.84 + - & 13 & 1.352 & 0.9758 & 336.7 & 4.869 & 67.3 & 6.043 & 1.26 + ccccccccc fit & @xmath179 & @xmath180 & @xmath181 & & @xmath182 & @xmath183 & @xmath184 & + 2 & -10.806 & 1.7116 & 0.3601 & & & & & + 3 & & & & & 121.76 & 1.4947 & 0.4206 & + + fit & @xmath83 & @xmath92 & @xmath93 & @xmath94 & @xmath138 & @xmath185 & @xmath186 + no . & & & @xmath187 & @xmath188 & @xmath187 & @xmath188 & + 2 & 1.237 & 1.010 & 341.1 & 5.037 & 57.9 & 6.131 & 3.67 + 3 & 1.188 & 1.021 & 338.9 & 5.095 & 80.6 & 6.957 & 4.28 + cccccccccc fit & @xmath189 & @xmath190 & @xmath191 & @xmath192 & @xmath193 & @xmath194 & @xmath195 & @xmath196 + 4 & 12.8 & 154.94 & -311.98 & 839.41 & -325.97 & 880.41 & - & - + 5 & 12.0 & 167.94 & -329.11 & 1103.43 & -592.10 & 1666.61 & -346.79 & 1118.06 + + fit & @xmath83 & @xmath92 & & @xmath93 & @xmath94 & @xmath138 & @xmath185 & @xmath186 + no . & & & & @xmath187 & @xmath188 & @xmath187 & @xmath188 & + 4 & 1.272 & 0.998 & & 338.8 & 4.979 & 68.2 & 4.524 & 2.23 + 5 & 1.287 & 0.991 & & 336.0 & 4.947 & 53.9 & 4.319 & 2.14 + ccccccc ref . & @xmath179 & @xmath180 & @xmath181 & @xmath182 & @xmath183 & @xmath184 + @xcite & -4.91 & 1.78 & 0.40 & 183.13 & 1.13 & 0.36 + @xcite & -3.51 & 1.72 & 0.95 & 337.06 & 1.26 & 0.23 + + ref . & @xmath83 & @xmath92 & @xmath93 & @xmath94 & @xmath138 & @xmath185 + no . & & & @xmath187 & @xmath188 & @xmath187 & @xmath188 + @xcite & 1.19 & 1.022 & 334.76 & 5.098 & 90.95 & 5.704 + @xcite & 1.15 & 1.014 & 315.49 & 5.056 & 107.91 & 6.014 + ccccccccc @xmath117 & @xmath179 & @xmath180 & @xmath181 & & @xmath182 & @xmath183 & @xmath184 & + 7.6 & -1.466 & 1.717 & 0.268 & & 36.86 & 1.295 & 0.419 + 5.8 & -1.091 & 1.720 & 0.270 & & 27.44 & 1.297 & 0.420 + + @xmath117 & @xmath83 & @xmath92 & @xmath93 & @xmath94 & @xmath138 & @xmath185 & @xmath186 + ( mev ) & & & @xmath187 & @xmath188 & @xmath187 & @xmath188 & + 7.6 & 1.219 & 1.000 & 327.1 & 4.991 & 26.2 & 6.085 & - + 5.8 & 1.209 & 1.000 & 324.3 & 4.991 & 19.6 & 6.095 & - +	the elastic scattering cross section of @xmath0mo(@xmath1,@xmath1)@xmath0mo has been measured at energies of @xmath2 13 , 16 , and 19mev in a wide angular range . the real and imaginary parts of the optical potential for the system @xmath0mo - @xmath1 have been derived at energies around and below the coulomb barrier . the result fits into the systematic behavior of @xmath1-nucleus folding potentials . the astrophysically relevant @xmath3ru(@xmath4,@xmath1)@xmath0mo reaction rates at @xmath5 and @xmath6 could be determined to an accuracy of about 16% and are compared to previously published theoretical rates . 2
quantum continuous variable entanglement is essential for quantum information tasks realized with quantum optical means . a great deal of contributions have been devoted to the entanglement criteria of gaussian states @xcite @xcite @xcite @xcite @xcite which are continuous variable states that can be determined only by their first and second moments . recently , non - gaussian states attract much attentions for its good performance in quantum teleportation @xcite , dense coding @xcite , quantum computation @xcite and nonlocality test @xcite . several entanglement criterion have also been proposed for non - gaussian states @xcite @xcite @xcite @xcite @xcite . however , types of non - gaussian states that have been studied are sporadic and limited by experimental conditions and theoretical interest until now . we will introduce the general ideas of coherent subtraction and addition to describe non - gaussian states and study their entanglement conditions . to generate non - gaussian state from gaussian state , one of the means is to subtract ( or add ) photon . the photon subtraction ( addition ) state is @xmath0 ( @xmath1 ) , where @xmath2 is a gaussian state ( hereafter referred as gaussian kernel ) , @xmath3 and @xmath4 are the annihilation operators of the two modes , @xmath5 and @xmath6 the creation operators . the replacement of @xmath7 with the superposition @xmath8 has been proposed for quantum state engineering @xcite . the most generic definition of bipartite photon coherent subtraction and addition state is @xmath9 where @xmath10 with complex coefficients @xmath11 this is so generic a definition that all two - mode continuous variable states can be expressed in such form . the photon coherent subtraction state is @xmath12 with @xmath13 the photon coherent addition state is @xmath14 with @xmath15 more specifically , if the two modes are symmetry , the symmetric coherent subtraction and addition states are @xmath16 respectively . the symmetric coherent subtraction or coherent addition state set is still large enough to contain many interesting non - gaussian states . in this paper , we will study the entanglement of several symmetric coherent subtraction or addition states . section 2 is to introduce the realignment entanglement criterion for non - gaussian states , with a new form of the criterion in coherent state representation . section 3 is devoted to the entanglement conditions of coherent subtraction state when the gaussian kernel is a two - mode squeezed thermal state ( tmst ) , when the gaussian kernel is in its standard form . in section 4 , we investigate the entanglement conditions of coherent addition states such as photon number entangled states ( pnes ) evolved in thermal noise and amplitude damping environment , coherent addition of tmst . conclusions are drawn in section 5 . the realignment entanglement criterion has been studied for continuous variable system recently @xcite . the derivation is based on fock space representation . we here give the criterion based on coherent state representation . a bipartite continuous variable state @xmath17 is completely specified by its characteristic function @xmath18,$ ] where @xmath19 is the displacement operator , and @xmath20 @xmath21 for @xmath22 two - mode state@xmath23 a gaussian state is completely determined by its first and second moments . the characteristic function of a gaussian state @xmath2 with nullified first moments ( the first moments are not relevant to the entanglement and can be removed by displacement ) is @xmath24,$ ] where @xmath25 is the @xmath26 complex covariance matrix of @xmath27 denote @xmath28 and denote the two mode coherent state as @xmath29 . in coherent state representation , we have @xmath30 \chi _ g(\mu ) \left\langle \alpha \right\| \mathcal{d}(-\mu ) \left\| \beta \right\rangle = $ ] @xmath31 @xmath32 @xmath33 @xmath34 @xmath35.$ ] where @xmath36 @xmath37 here @xmath38are the pauli matrices , @xmath39 is the @xmath40 identity matrix . the realigned density matrix @xmath41 is the interchange of second and third subscripts of @xmath42 namely , @xmath43 thus @xmath44\ ] ] where @xmath45 with @xmath46 and all the other entries of the @xmath26 matrix @xmath47 are zeros . the realignment criterion of entanglement is @xmath48 , from which we can derive a condition of entanglement@xmath49 the later is easy for use but may detect less number of entangled states as the former . here @xmath50 is the realignment of @xmath51 for a gaussian state , we have the realignment entanglement sufficient condition @xmath52 where @xmath53 the non - gaussian state produced by coherent subtraction and addition from gaussian state can be written as derivative of functional of gaussian state . the method to derive the entanglement criterion of non - gaussian states can be found in ref.@xcite . a @xmath54 bipartite gaussian state can be transformed to its first standard form ( denoted as @xmath55 ) by local operations . further , it is always possible to transform @xmath55 to its second standard form @xmath56by proper local squeezing operations@xcite . to prove the necessary condition of entanglement for a gaussian state , it is shown that a separable gaussian state @xmath57 takes the form of @xmath58 where @xmath59 is a gaussian function hence positive definite , @xmath60 is the product of coherent states . photon subtraction from gaussian state @xmath57 will lead to a non - gaussian state @xmath61 more generic photon subtraction is the coherent subtraction , or the superposition of all possible subtraction operations , namely , @xmath62 . the non - gaussian state produced from gaussian kernel @xmath57 is @xmath63 then state @xmath64 is separable when @xmath57 is separable , since we have @xmath65 which is explicitly separable . hence a non - gaussian state prepared by coherent subtraction from second standard form gaussian state is separable if the kernel gaussian state is separable . a gaussian state with first standard form @xmath55 differs from its second standard form by local squeezing , @xmath66 with some properly chosen squeezing parameters @xmath67 and @xmath68 and @xmath69 are the single - mode squeezing operators . keep in mind that the explicit separable decomposition ( [ wee1 ] ) is only for the gaussian state with second standard form . so @xmath55 may not be diagonal in coherent state representation even when it is separable . a tmst is @xmath70 , where @xmath71 $ ] is the two mode squeezing operator , @xmath72 is the thermal state with @xmath73 the average photon number of state @xmath74 is @xmath75 consider the case of symmetric modes , we have @xmath76 the characteristic function of @xmath77 is @xmath78,$ ] where the complex covariance matrix @xmath79 with @xmath80 @xmath81 the characteristic function can also be written as @xmath82,$ ] where @xmath83 and @xmath84 are imaginary and real parts of @xmath85 the real symmetric correlation matrix is @xmath86 @xmath87 notice that the first standard form @xmath88 and the second standard form @xmath89 are equal to @xmath90@xcite for symmetric tmst@xmath23 we have @xmath91 @xmath92 for symmetric tmst . the necessary and sufficient condition of the separability of the symmetric tmst is @xmath93 lets consider coherent subtraction state @xmath94 if ( [ wee4 ] ) is fulfilled , @xmath77 is separable , then @xmath95 is separable since it can be written in the form of eq.(3 ) . the inverse problem is that if @xmath95 is entangled or not when @xmath96 is entangled . we consider the case of symmetric coherent subtraction of photon , the symmetric non - gaussian state is @xmath97 the realignment entanglement criterion for unnormalized @xmath98 is @xcite @xmath99 where @xmath100 , @xmath101 @xmath102,$ ] @xmath103,$ ] with @xmath104,$ ] @xmath105.$ ] then the entanglement condition is @xmath106 so we keep @xmath107 invariant . denote @xmath108 then @xmath109 is a small positive quantity for entangled symmetric tmst near the separable boundary . expanding all the quantities in ( @xcite ) to the first order of @xmath110 we have @xmath111 @xmath112 @xmath113 @xmath114 @xmath115 where @xmath116 notice that up to the first order of @xmath109 , we have @xmath117 , while @xmath118 hence ( [ wee5 ] ) is fulfilled even for infinitive small @xmath119 thus @xmath120 is entangled if @xmath96 is entangled . we have proven that for any symmetric coherent subtraction state @xmath121 produced from symmetric tmst , the necessary and sufficient of separability is ( [ wee4 ] ) . the first standard form gaussian state @xmath55 is completely characterized by its complex covariance matrix @xmath122where @xmath123 $ ] @xmath124,$ ] @xmath125,$ ] @xmath126 . $ ] the real symmetric correlation matrix is @xmath127 @xmath128 the necessary and sufficient condition of the entanglement of @xmath55 is @xmath129 for symmetric coherent subtraction state @xmath130we consider if it is separable at the @xmath131 and @xmath132 the boundary of separability of its gaussian kernel . the entanglement criterion is @xmath133 where @xmath134,$ ] @xmath135,$ ] with @xmath136 @xmath137 then @xmath138 the same expression can be found for @xmath139 with the interchange of @xmath140 and @xmath141 consider the function @xmath142 notice that @xmath143 we only need to consider @xmath144 for odd @xmath145 one of the roots for equation @xmath146 is @xmath147 for even @xmath145 @xmath148 are the roots of equation @xmath149 we have verified numerically that there are no other real roots for all @xmath150 since @xmath151 we have @xmath152 for all @xmath153 from @xmath154 we have @xmath155 , thus @xmath156 as @xmath157 is assumed to be negative ( for positive @xmath157 , the realignment criterion @xmath158 should be modified @xcite ) , then @xmath159 at @xmath154 the realignment entanglement criterion @xmath160 can be rewritten as @xmath161 where @xmath162 if @xmath163 for all @xmath164 and @xmath165 at least for one pair of @xmath164 @xmath166 the coherent subtracted state @xmath167 one of the case is that the real and/or imaginary parts of the coefficients @xmath168 are non negative and at least two of @xmath168 have positive real and/or imaginary parts . coherent subtraction means at least two of the coefficients @xmath168 are non zeros . we conclude that if the coefficients have positive real and/or imaginary parts , the symmetric non - gaussian state produced from the coherent subtraction of an edge separable first standard form of gaussian state is entangled . unlike the coherent subtraction state , the sufficient condition of separability for a coherent addition state is difficult to be found even when the gaussian kernel @xmath2 is in its second standard form . we will consider the sufficient condition of entanglement for symmetric coherent addition state @xmath169 in the following . an initially prepared non - gaussian entangled state will become separable after it interact with the thermal noise and amplitude damping environment . the evolution of the state could be specified by the evolution of its characteristic function @xcite . the realignment entanglement criterion for time evolution state is @xcite @xmath170>\mathcal{o}\exp [ -\frac 12\mathbf{v}\gamma ^{\prime } \mathbf{v}^t ] . \label{wee18}\ ] ] where @xmath171with @xmath172 @xmath173 @xmath174 with @xmath175 @xmath176 and @xmath177 here @xmath178 is the thermal noise , the amplitude damping.coefficient is merged in the unit of time @xmath179 a special case of symmetric coherent addition state is the pnes when the gaussian kernel @xmath2 is the vacuum state . we have @xmath180 where @xmath181 for vacuum gaussian kernel , @xmath182 @xmath183 @xmath184,$ ] where @xmath185 then @xmath186 , with @xmath187.$ ] denote @xmath188 then @xmath189 we also get @xmath190=m!n!\delta _ { mn},$ ] @xmath191 the realignment entanglement criterion for time evolution pnes is @xmath192 where @xmath193 for a two mode squeezed thermal state evolving in amplitude damping and thermal noise channel , the initial state is @xmath194 thus @xmath195 by interchanging the order of summations in ( [ wee19 ] ) , the entanglement criterion of the final state reduces to @xmath196 condition ( [ wee20 ] ) can also be derived from the well know entanglement condition of gaussian state , for the final state is gaussian . this example shows that the choice of gaussian kernel may not be relevant for the entanglement criterion . for a two mode coherently correlated state ( tmc ) with poisson coefficients evolves in thermal noise and amplitude damping channel , the unnormalized initial state is @xmath197 then @xmath198 where @xmath199 the function @xmath200 converges fast than exponential function . the entanglement criterion ( [ wee19 ] ) leads to @xmath201 the critical separable time is shown in fig.1(a ) for different noise @xmath202 comparing with the direct numerical calculation @xcite , we confirm that the numerical calculation of the realignment entanglement criterion of tmc is valid . another example is the evolution of photonic bell - state @xmath203 in thermal noise and amplitude damping environment . the entanglement criterion ( [ wee19 ] ) leads to @xmath204 the critical separable time is shown in fig.1(b ) , where the simon s criterion is also shown for comparison @xcite . the similar curve has been obtained by nha s criterion@xcite . the coherent addition of symmetric tmst is @xmath205 the realignment entanglement criterion is still ( [ wee18 ] ) , where @xmath206 @xmath207\sigma _ 1\otimes i_2-\frac 12[(b_0+c_1)\tau + ( b_0-c_1)+1]i_2\otimes \sigma _ 1,$ ] @xmath208 with @xmath100 . we consider the case that the kernel is a symmetric tmst of critical separable , that is @xmath209 then @xmath210 the entanglement criterion ( [ wee18 ] ) then reads @xmath211 where @xmath212 is defined as in ( [ wee8 ] ) . notice that when @xmath154 we have @xmath213 here @xmath214 is assumed as before . hence @xmath215 as we have argued . so if all @xmath212 are nonnegative ( at least one of them is nonzero thus it is positive ) , the state should be entangled . one of the cases is that if only two of the coefficients are nonzero , say @xmath168 and @xmath216 , the entanglement condition will be @xmath217 denote @xmath218 the entanglement condition reduces to @xmath219 for any @xmath220 thus if the argument difference of the two complex coefficients is an acute angle , the state is entangled . borrowing term from optics , we can say that if the coherent addition constructively interferes , the state is entangled . we have introduced the ideas of coherent subtraction and coherent addition to describe many kinds of bipartite non - gaussian states and investigate their inseparability . the non - gaussian states produced from their gaussian kernels are quite generic . the results of the paper are of several folds . for any coherent subtraction state produced form the second standard form @xcite gaussian state , we prove that it is always separable when the gaussian kernel is separable . for any symmetric coherent subtraction state produced from the symmetric two - mode squeezed thermal state , the necessary and sufficient separable condition is obtained and it is exactly the same as that of its kernel state . a symmetric coherent subtraction state produced from the symmetric first standard form gaussian state ( except two - mode squeezed thermal state ) is entangled when the gaussian kernel is at the boundary of separability and all the superposition coefficients have positive real and/or imaginary parts . thus a proper arranged coherent subtraction produces entanglement from a separable state . for photon number entangled state evolving in thermal noise and amplitude damping channel , we give the analytical entanglement condition . as applications , we present the separable conditions of two mode coherently correlated states with poisson coefficients and photonic bell states evolving in thermal noise and amplitude damping environment . any two term coherent addition state produced form boundary separable symmetric two - mode squeezed thermal state is entangled if the two complex coefficients give rise to constructive interference . further works are desirable for other non - gaussian states or using entanglement criterion other than realignment criterion . this work is supported by the national natural science foundation of china ( grant no . 60972071 ) , zhejiang province natural science foundation ( grant nos .	photon subtraction and addition are experimental means of generating non - gaussian states from gaussian states . coherent subtraction or addition is a combination of photon subtractions or additions . the resultant states are quite general non - gaussian states . the states can be photon number entangled states with arbitrary coefficients . we derive the entanglement conditions for several classes of coherent subtraction or coherent addition bipartite continuous variable states . one of the entanglement conditions is necessary and sufficient . pacs number(s ) : 03.67.mn ; 03.65.ud
let us have a quick look of our main results , the details and the proofs are in the paper . we introduce the hamiltonian system in @xmath0 @xmath1 where @xmath2 is a @xmath3 function near @xmath4 and satisfies @xmath5 and @xmath6 . the hamiltonian function for the system is @xmath7 and another first integral is @xmath8 the @xmath9-plane is invariant and the origin is a center on it , namely an open neighbourhood @xmath10 of @xmath11 in this plane is the union of periodic orbits which enclose @xmath11 . we restrict the phase space to @xmath12 , so we have a global center in the @xmath9-plane . the hamiltonian vector fields @xmath13 in and @xmath14 are both complete on @xmath15 ( with @xmath16 skew - symmetric matrix , see ) . the functions @xmath17 are in involution . the vectors @xmath18 are linearly independent at each point of the set @xmath19 which is invariant for both the hamiltonian vector fields @xmath20 and @xmath21 . a nonempty component @xmath22 of a level set of @xmath17 is diffeomorphic to @xmath23 and there are coordinates @xmath24 mod @xmath25 and @xmath26 on @xmath27 such that the differential equations defined by the vector field @xmath28 on @xmath29 take the form @xmath30 if @xmath31 belongs to the projection of @xmath29 on the @xmath9-plane and @xmath32 is the period of the first component of any integral curve of @xmath28 on @xmath29 , then @xmath33 if and only if @xmath34 . a necessary and sufficient condition for the stability of the origin of is that all periodic orbits in a neighbourhood of @xmath11 in the @xmath9-plane have the same period so the center is ( locally ) isochronous . in this case all orbits of ( 1.1 ) in @xmath0 , with @xmath35 near @xmath11 , are periodic and have the same period . moreover , a further first integral appears as we shall see using a result of barone and cesar @xcite . this happens for instance for the following two functions @xmath36 in section [ isochronous ] we show a theorem on isochronous centers taken from @xcite . we reproduce here the proof to be self - contained . our result relates these functions with the even functions and so it shows that isochronicity in the @xmath9-plane , and lyapunov stability for , is a rare phenomenon among the class . isochronous centers were first studied by urabe @xcite with a different approach , in @xcite a new characterization by means of the so called ` involutions ' @xmath37 was found which permits to construct the functions giving isochronous centers . generally we have instability of the equilibrium for and it is very easy to show explicit functions for it , for instance @xmath38 with @xmath39 . this kind of instability is quite weak since the linearization of the system at the origin in @xmath0 has the double eigenvalues @xmath40 . moreover , there are no asymptotic motions to the equilibrium , namely no motions with the origin as limit point as @xmath41 . in the present paper , instability without asymptotic motions is called _ weak instability_. in our hamiltonian systems , instability occurs because there is a sequence of initial data which converges to the origin whose corresponding solutions are unbounded and the motion is slow , indeed shows that the coordinate @xmath26 is an affine function of time . _ acknowledgement . _ the picture in section 3 was made using the application _ mathematica _ by wolfram research inc . by means of the package curvesgraphics6 by gianluca gorni of the university of udine available on www.dimi.uniud.it/~gorni . let us start from the following equation @xmath42 where @xmath2 is continuous in a neighbourhood of @xmath4 in @xmath43 and differentiable at @xmath4 ( in section [ introduction ] the function @xmath2 was @xmath3 , while in the present section [ isochronous ] the existence of @xmath44 is enough ) . this o.d.e . has the first integral of energy @xmath45 by means of this first integral , we easily see that each cauchy problem for has a unique solution if @xmath46 only at @xmath47 . the potential energy @xmath48 is a @xmath3 function and there exists @xmath49 . we can restrict the domain of @xmath2 to an open interval @xmath50 such that @xmath48 is strictly increasing on @xmath51 , strictly decreasing on @xmath52 , and for each point @xmath53 there is a unique point @xmath54 with @xmath55 where @xmath56 is the sign of @xmath57 . we check at once that the function @xmath58 is a @xmath3 diffeomorphism onto the image @xmath59 which is a symmetric interval @xmath60 moreover , since @xmath61 we have that @xmath37 is also a diffeomorphism : @xmath62 we call @xmath37 the _ involution _ associated with @xmath48 . remark that the graph of @xmath37 is symmetric with respect to the diagonal which intersects at the origin ; indeed @xmath63 has @xmath64 as symmetric point and this coincides with the point @xmath65 of the graph . [ thisochronous ] let @xmath48 be a @xmath3 function near @xmath4 in @xmath43 with @xmath66 , assume there exists @xmath67 , let @xmath68 be an open interval as above and let @xmath69 be the involution associated with @xmath48 . then all orbits of @xmath70 which intersect the @xmath68 interval of the @xmath57-axis in the @xmath71-plane , are periodic and enclose @xmath11 . moreover , they all have the same period if and only if @xmath72 in this case we say that the origin is an isochronous center for @xmath70 . by composition with the inverse of the diffeomorphism , the first integral gives @xmath73 . the first part of the thesis follows at once . now , consider a periodic orbit in the @xmath71-plane which intersects the @xmath57-axis at @xmath74 , @xmath75 , then @xmath76 is the other intersection . by the energy conservation we get the period of the orbit as @xmath77 if @xmath78 , @xmath79 , @xmath80 , then @xmath81 , and the period @xmath82 the change of integration variable @xmath83 gives for @xmath78 , @xmath79 , @xmath84 so @xmath85 all orbits have the same period if and only if @xmath86 equals this limit value for all @xmath78 , @xmath79 . by this is equivalent to the following condition for all @xmath87 , @xmath88 , @xmath89 we multiply both sides by @xmath90 and we integrate from @xmath4 to @xmath91 , @xmath92 , @xmath93 by the change of the integration order , see figure [ triangle ] , we get @xmath94 @xmath95 we deduced this condition from , and now we see at once that it implies . by plugging @xmath96 into we have @xmath97 this condition is equivalent to . formula corresponds to formula ( 6.2 ) in @xcite , the proof is included in the proof of proposition 1 in @xcite as a particular case . in @xcite , section 6 , there is the construction we are going to show and also another one . from the theorem we easily get the following result . [ constructionisochronous ] let @xmath98 be a @xmath3 function on an open interval @xmath99 containing @xmath4 which satisfies the conditions in . let @xmath100 and define @xmath101 then there exists @xmath102 and all orbits of @xmath70 which intersect the @xmath68 interval of the @xmath57-axis in the @xmath71-plane , are periodic and have the same period @xmath103 . to get @xmath37 as above , we can just consider an arbitrary even @xmath3 function on a ( symmetric ) open interval which vanishes at @xmath4 , then a @xmath104 clockwise rotation of its graph gives the curve @xmath105 provided the interval is small enough to get a graph . for instance , starting from @xmath106 we can ( even explicitly ) calculate @xmath107 notice that the original quadratic function must be restricted to a suitable interval to get a function after rotation . another construction of the potential energies of the isochronous centers is given in @xcite ( section 6 from formula ( 6.3 ) ) . finally , a simple example is also @xmath108 now , let us see some necessary conditions to have constant period which can be used to check whether a given function @xmath48 , or its derivative @xmath2 , locally gives an isochronous center or not . in the sequel @xmath48 has as many derivatives as necessary . we saw how @xmath37 is related to an even function . the even functions have vanishing odd derivatives at @xmath4 , so it will not be a surprise to see that the derivatives of the involution @xmath37 are not arbitrary . consider the relation @xmath109 , perform 2 derivations and calculate at @xmath4 taking into account @xmath110 and @xmath111 . in this way we get an identity which shows that @xmath112 can take any value . by 3 derivations and calculating at @xmath4 we get @xmath113 . going further we see that the even derivatives are free while the odd ones are uniquely determined by the preceding derivatives . the first 5 terms in taylor s formula are given by @xmath114 where @xmath115 are free parameters . next , formula shows that the derivatives of the isochronous @xmath48 at @xmath4 are constrained to obey some conditions . by means of we can calculate @xmath116 and @xmath117 by the lower order derivatives and get 2 necessary conditions . of course we can go forward to infinite conditions . [ necessaryconditions ] let @xmath48 admit @xmath117 and satisfy @xmath66 , @xmath67 . moreover , let the origin be an isochronous center for @xmath70 . then @xmath118 to illustrate the necessary conditions let us use it to prove the well known lack of isochronicity of the simple pendulum @xmath119 @xmath120 another example is @xmath121 for which the second condition in implies @xmath122 , and the first condition with @xmath122 gives @xmath123 too . we could also prove the above formulas in using the approach of barone , cesar and gorni @xcite where the first 2 derivatives of the period function @xmath32 at @xmath4 are computed with a procedure which carries on to higher - order derivatives . with some regularity on @xmath48 they find the following formulas for @xmath124 @xmath125 in this section we deal with the following 4-dimensional system defined by the lagrangian function @xmath126 @xmath127 where @xmath128 near @xmath4 in @xmath43 . this system of differential equations has two first integrals @xmath129 and @xmath130 @xmath131 the first differential equation separates and its dynamics was studied in the previous section . we restrict our attention to an open interval @xmath68 as in theorem [ thisochronous ] and to the orbits in the @xmath71-plane which intersect @xmath68 ; their union is an open neighbourhood @xmath10 of @xmath11 . let us fix @xmath74 , @xmath75 , and denote by @xmath132 the periodic solution of @xmath133 with @xmath31 as initial condition at time @xmath4 . next , we plug @xmath134 into the second differential equation in and get the linear equation with periodic coefficient , hill s equation , @xmath135 the partial derivatives of @xmath136 give 2 independent solutions of as one see by derivation of the first equation for @xmath134 , in particular @xmath137 let us define @xmath138 so that @xmath139 let us denote by @xmath140 , briefly @xmath141 , the period of @xmath132 , then @xmath142 is also a solution of , so a linear combination of @xmath24 and @xmath143 , and the previous values at @xmath4 give @xmath144 where we used @xmath145 which comes from the last equality in with @xmath146 if we notice that @xmath147 and @xmath148 . taking the derivative , calculating at @xmath146 , and taking into account @xmath149 , we have @xmath150 for all @xmath151 . this last result is equivalent to @xmath152 if @xmath153 the solution @xmath24 is periodic , otherwise it is unbounded . a necessary and sufficient condition to have the former case is @xmath153 namely @xmath154 to go ahed we need the differentiability of the period function [ pex3.1 ] the period @xmath32 of @xmath132 is a @xmath3 function on @xmath155 . by formula , for @xmath74 , @xmath75 , we have @xmath156 the function @xmath157 is a @xmath3 diffeomorphism as in the previous section . in the stronger hypothesis of this section , @xmath128 , we have that @xmath158 for @xmath159 is @xmath160 as well as @xmath161 and its inverse @xmath162 for @xmath92 . this proves the result . now , we can differentiate @xmath163 with respect to @xmath164 @xmath165 we just remind that @xmath166 for @xmath75 , and have that condition is equivalent to @xmath167 this is the condition in order @xmath24 to be periodic . since the linear combination of @xmath24 and @xmath168 gives the general solution , we have just proved the following [ pex3.2 ] let us fix @xmath74 with @xmath75 , then the solutions to the @xmath32-periodic equation are either all @xmath32-periodic , or all unbounded but those proportional to @xmath169 . a necessary and sufficient condition for the former case is . in figure [ unbounded ] the curve @xmath170 , namely the solution to system of differential equations with @xmath171 , @xmath172 , @xmath173 , in a non - periodic case . the initial point @xmath174 and the point @xmath175 are crossed at any period . now , notice that the origin in @xmath0 is an unstable equilibrium for if and only if we can find @xmath75 arbitrarily close to @xmath4 such that equation has unbounded solutions . so by proposition [ pex3.2 ] we have [ ext3.3 ] let @xmath2 be a @xmath3 function near @xmath4 in @xmath43 with @xmath5 , @xmath6 , then the origin in @xmath0 is a stable equilibrium for the system , if and only if the origin in @xmath176 is a locally isochronous center for @xmath177 , in this case all orbits of with @xmath178 near @xmath11 are periodic and have the same period . if the equilibrium is unstable , then there is a sequence of initial data which converges to the origin whose corresponding solutions are unbounded . the instability is weak , by this we mean that there are no asymptotic motions to the equilibrium , namely no motions which tends to the origin as @xmath41 . we already proved the first part of the statement , for the last , let us only remark that the distance of an orbit from the origin in @xmath0 is greater than the distance of the projection in the @xmath71-plane which is strictly positive unless the projection is @xmath11 and in this case the second differential equation in gives the harmonic oscillator @xmath179 . an explicit function @xmath2 to give such an example is @xmath180 ( see ) . however , as we know from section [ isochronous ] , almost all functions @xmath2 as in the statement of the theorem , give instability . examples of the rare functions which give stable equilibria are ( see and ) @xmath181 @xmath182 for all these functions which give stable equilibria , at least whenever @xmath183 the system admits a ( positive definite ) lyapunov function which is a further first integral ; this additional first integral is smooth in a neighbourhood of the origin in @xmath0 and at least continuous at the origin as proved by barone and cesar in @xcite . in the following statement @xmath184 is the set defined below formula . let @xmath185 be a @xmath186 function on the open interval @xmath187 with @xmath188 , @xmath5 , @xmath6 . suppose that all orbits of @xmath189 which intersect the @xmath68 interval of the @xmath57-axis in the @xmath71-plane , are periodic and have the same period @xmath190 , so the origin in @xmath0 is a stable equilibrium for the system . then there exists @xmath191 continuous and positive definite function , which is a ( global ) first integral for the system . the first integral is obtained by means of the functions in and a suitable inverse function , see @xcite for details . let us also notice that [ pex3.5 ] let @xmath2 be a @xmath3 function near @xmath4 in @xmath43 with @xmath5 , @xmath6 , then the linearization of the system at the origin in @xmath0 has the double eigenvalues @xmath40 . let us remark that we can easily construct unstable cases with some @xmath164 at which holds and so the corresponding orbits are all periodic . if we want the period function composed with @xmath192 to be for instance @xmath193 , whose derivative vanishes at @xmath194 we integrate @xmath195 so we get the relation @xmath196 the symmetric choice @xmath197 gives @xmath198 by inversion we get @xmath199 , then @xmath200 and finally @xmath201 . there is an isolate @xmath164 at which holds . we can also imagine more complicate examples where these @xmath164 accumulate at @xmath4 . finally , let us mention that most of the results in section [ lagrangian ] were found in @xcite by a different ( more complicated ) proof which included other differential equations in a family for which the present system was a particular case . however , in @xcite i had not realized the lagrangian character of the present case which is very important to our purposes and will be exploited in the next section . in this section we write @xmath202 . the legendre transformation takes the lagrangian system into @xmath203 the hamiltonian function corresponds to the first integral @xmath204 while the first integral @xmath205 becomes @xmath206 our phase space is @xmath207 where @xmath184 is an open set mentioned in section [ lagrangian ] between formulas and so to have a global _ center _ in the @xmath9-plane . the hamiltonian vector fields @xmath208 @xmath209 are both _ complete _ on @xmath15 , namely their integral curves are all defined on the whole @xmath43 . indeed , this is clear for since we have a global center on the @xmath9-plane and linear equations in @xmath210 ; for we simply integrate and remind @xmath211 the functions @xmath17 are _ in involution _ , indeed their poisson brackets vanish : @xmath212 the vectors @xmath18 are _ linearly independent _ at each point of the set @xmath213 which is _ invariant _ for both the hamiltonian vector fields @xmath20 and @xmath21 . indeed , the condition @xmath214 implies @xmath215 and @xmath216 , so @xmath217 since @xmath218 ; then it also implies @xmath219 and @xmath220 so @xmath122 . let @xmath22 be a nonempty component of a level set of @xmath221 , then we see at once that it is not compact since it contains the unbounded curve @xmath222 where @xmath223 is the value of @xmath224 and @xmath225 is the one of @xmath226 . by a well known theorem ( see theorem 3 , chapter 4 , in arnold , kozlov and neishtadt @xcite ) , @xmath29 is diffeomorphic to @xmath23 . more precisely [ tex4.1 ] let @xmath2 be a @xmath3 function near @xmath4 in @xmath43 with @xmath5 , @xmath6 . the functions @xmath227 , in and , are in involution on @xmath15 , @xmath228 , and the hamiltonian vector fields @xmath20 and @xmath21 are complete on @xmath15 . the set @xmath229 in is invariant for @xmath20 and @xmath21 and @xmath17 are independent on @xmath229 . let @xmath22 be a nonempty component of a level set of @xmath221 then it is diffeomorphic to @xmath230 and there are coordinates @xmath24 mod @xmath25 and @xmath26 on @xmath27 such that the differential equations defined by the vector field @xmath28 on @xmath29 take the form @xmath231 if @xmath31 belongs to the projection of @xmath29 on the @xmath9-plane and @xmath32 is the period of the first component of any integral curve of @xmath28 on @xmath29 , then @xmath232 and @xmath33 if and only if @xmath34 . we already proved everything in the statement of the theorem but the last sentence which is a straight consequence of @xmath34 being the necessary and sufficient condition for the solutions on @xmath29 to be all periodic ( see proposition [ pex3.2 ] ) , and gives periodic solutions if and only if @xmath33 . for the _ isochronous _ systems , namely whenever the period function is constant and all integral curves of @xmath28 in @xmath15 have the same period , we know from theorem [ tex3.4 ] that a further first integral exist which is quadratic in @xmath210 . of course we do not expect this first integral to be in involution with @xmath226 . in the next section we investigate the ( isochronous ) cases where a third first integral quadratic in the momenta @xmath233 exits and we arrive at explicit formulas for it . let @xmath2 be a @xmath3 function near @xmath4 in @xmath43 with @xmath5 , @xmath6 . suppose that the system has a first integral @xmath234 of the form @xmath235 then the equation @xmath236 give a cubic polynomial in @xmath233 . the coefficients vanish if and only if @xmath237 @xmath238 the conditions in are the same as those in ( 3.2.2 ) in hietarinta @xcite where the celebrated darboux problem is treated ( only the order of the coordinates is changed since @xmath239 , @xmath240 ) . they have the following solution where @xmath241 are arbitrary @xmath242 plugging in we get the following condition for the integrability of @xmath243 @xmath244 for @xmath245 we get @xmath246 which implies @xmath247 , and back to we have @xmath248 for @xmath249 , @xmath5 and @xmath250 , this equation gives @xmath251 where @xmath252 ( for @xmath253 there are no solutions ) . the values @xmath254 and @xmath255 give . the first integral for this @xmath2 is easily obtained with @xmath256 , @xmath249 , @xmath247 @xmath257 where @xmath258 . for @xmath259 , @xmath260 , we have the first integral @xmath226 we already know , while @xmath224 corresponds to the choice @xmath261 , @xmath262 . for @xmath263 , @xmath264 we have the first integral @xmath265 this function is positive definite near the origin of @xmath0 as we check at once . this is why we have lyapunov stability of the equilibrium and compact orbits . we dealt with the particular case @xmath249 . in the sequel @xmath266 and we fix @xmath267 dividing the first integral by @xmath268 . the coefficient of @xmath269 in equation is a quadratic function in @xmath270 with discriminant @xmath271 . let us consider first the case where this vanishes , namely @xmath272 . if @xmath273 then with @xmath5 , @xmath250 , gives the trivial function @xmath274 , while for @xmath275 we get the following solution with @xmath5 , @xmath250 : @xmath276 notice that this formula includes for @xmath255 and @xmath277 . the first integral , which corresponds to , with @xmath278 , @xmath279 , and which vanishes at the origin of @xmath0 , is @xmath280 we can check that at the origin of @xmath0 its gradient vanishes and the hessian matrix is the diagonal matrix @xmath281 , so we have again a ( positive definite near the origin ) lyapunov function for all values of the parameter @xmath282 . now we are ready to deal with the generic case . let @xmath267 and @xmath283 , we also consider @xmath284 and @xmath285 since @xmath273 gives the trivial solution and @xmath253 no solutions at all . then the solution of the equation with @xmath5 and @xmath286 is @xmath287 the first integral with @xmath278 , @xmath288 , divided by @xmath289 , and which vanishes at the origin of @xmath0 , is @xmath290 we can check that at the origin of @xmath0 its gradient vanishes and the hessian matrix is the diagonal matrix @xmath291 , so we have a positive definite function near the origin . vladimir i. arnold , valery v. kozlov & anatoly i. neishtadt , _ mathematical aspects of classical and celestial mechanics_. encyclopaedia of mathematical sciences 3 . dynamical systems iii . springer - verlag ( 1988 ) .	the aim of this paper is to introduce a class of hamiltonian autonomous systems in dimension 4 which are completely integrable and their dynamics is described in all details . they have an equilibrium point which is stable in some rare elements of the class and unstable in most cases . anyhow , the eigenvalues of the linearization at the equilibrium point are imaginary and no motion is asymptotic in the past , namely no solution has the equilibrium as limit point as time goes to minus infinity . in the unstable cases , there is a sequence of initial data which converges to the origin whose corresponding solutions are unbounded and the motion is slow . so instability is quite weak and perhaps no such explicit examples of instability are known in the literature . the stable cases are also interesting since the level sets of the 2 first integrals independent and in involution keep being non - compact and stability is related to the isochronous periodicity of all orbits and the existence of a further first integral .
interstellar dust ( isd ) became a topic of astrophysical research in the early 1930s when the existence of extinction , weakening , and scattering of starlight in the interstellar medium ( ism ) was realised . at that time , astronomical observations provided the only information about the properties of the dust in the ism . with the advent of dust detectors onboard spacecraft , it became possible to investigate dust particles in - situ . about 30 years ago , analysis of the data obtained with the dust instruments flown on a couple of spacecraft suggested that isd grains can cross the heliospheric boundary and penetrate deeply into the heliosphere @xcite . in the 1990s , this was undoubtedly demonstrated with the dust instrument carried by the ulysses spacecraft : the ulysses dust detector , which measures mass , speed and approach direction of the impacting grains , identified isd grains with radius above 0.1@xmath4 sweeping through the heliosphere @xcite . our galactic environment within 500 pc of the sun . currently , the sun is passing through the local interstellar cloud ( lic ) , shown in violet , which is flowing away from the scorpius - centaurus association of young stars . the lic resides in a low - density hole in the interstellar medium called the local bubble , shown in black . nearby , high - density molecular clouds including the aquila rift surround star forming regions , each shown in orange . the gum nebula , shown in green , is a region of hot ionized hydrogen gas . inside the gum nebula is the vela supernova remnant , shown in pink , which is expanding to create fragmented shells of material like the lic ( from p.c . frisch , university of chicago ) . ] the galactic environment of our solar system on a larger scale is shown in figure [ frisch ] . the nearby interstellar medium ( within about 3 pc of the sun ) is dominated by a shell of material , the local interstellar cloud ( lic ) . the solar system currently passes through the lic which is located at the edge of the local bubble . the local bubble was excavated by supernova explosions in the neighbouring starforming regions of the scorpius - centaurus and orion associations . the solar system emerged from the interior of this bubble within the last @xmath5 years . the only direct observation of isd close to the sun is weak polarization observed along the sightline towards 36 oph ( distance about 6 pc ) which is due to magnetically aligned dust grains @xcite . therefore , in - situ sampling of dust from the lic can greatly improve our understanding of the nature and processing of dust in various galactic environments and can cast new light on the chemical composition and homogeneity of the interstellar medium . in this paper we review the results from in - situ isd measurements obtained with the ulysses and other space - borne dust detectors . we review our current knowledge about isd inside the heliosphere and in our local interstellar environment . the ulysses in - situ dust measurements showed that the grain motion through the solar system is parallel to the flow of neutral interstellar hydrogen and helium gas , both gas and dust travelling with a speed of @xmath6 @xcite . the upstream direction of the dust flow lies at @xmath7 ecliptic longitude and @xmath8 latitude @xcite . the interstellar dust flow persists at high ecliptic latitudes above and below the ecliptic plane and even over the poles of the sun , whereas interplanetary dust is strongly depleted at high latitudes @xcite . the interstellar dust flux measured at a distance of about 3 au from the sun is time - dependent , and the mean mass of the grains is about @xmath9 @xcite , corresponding to a grain radius of approximately 0.3 @xmath4 . measurements with the identical dust instrument onboard galileo performed in the ecliptic plane showed that beyond about 3 au the interstellar dust flux even exceeds the flux of micron - sized interplanetary grains . results from the first decade of the ulysses mission showed that the radii of clearly identified interstellar grains range from 0.05@xmath4 to above 1@xmath4 . the data show distance - dependent alteration of the interstellar dust stream caused by gravitational focussing by the sun , solar radiation pressure , and electromagnetic interaction with the time - varying interplanetary magnetic field ( imf ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? radiation pressure and electromagnetic forces strongly depend on grain size , leading to a strong modification of the size distribution and fluxes of grains measured inside the heliosphere @xcite . the size distribution shows a deficiency of small grains below 0.3@xmath4 compared to astronomically observed isd @xcite . in addition , solar radiation pressure deflects grains with sizes of about 0.4@xmath4 and was found to be effective at solar distances below 4 au @xcite . significant differences in the particle sizes were also recorded at different heliocentric distances . in addition to the ulysses measurements which revealed a lack of small 0.3@xmath4 isd grains within 3 au heliocentric distance , measurements by cassini and galileo in the distance range between 0.7 and 3 au showed that the detected interstellar particles were bigger than 0.5@xmath4 , with grain masses increasing closer to the sun @xcite . the flux of these bigger particles did not exhibit temporal variations due to the solar - wind magnetic field like the flux of smaller particles observed by ulysses . the trend of increasing particle masses continues even closer to the sun , as demonstrated by helios which recorded particles of about 1@xmath4 down to 0.3 au @xcite . these facts support the idea that the isd stream is strongly filtered by solar radiation pressure . interstellar particles with optical properties of astronomical silicates or organic refractory materials are consistent with the observed radiation pressure effects @xcite . in addition to studies of the distribution of grain masses , the ulysses dust instrument monitors the flux of the interstellar particles in the heliosphere ( figure [ fig_fit ] ) . in mid 1996 , we observed a decrease of the interstellar dust flux by a factor of 3 from an initial value of @xmath10 down to @xmath11 . this drop was attributed to increased filtering of small grains by the solar wind driven imf during solar minimum conditions @xcite . since early 2000 , ulysses has again detected interstellar dust flux levels above @xmath12 @xcite . monte - carlo simulations of the grain dynamics in the heliosphere showed that the dominant contribution to the dust flux comes from grains with a charge - to - mass ratio of @xmath13 and a radiation pressure efficiency of @xmath14 , corresponding to grain radii of @xmath15 @xcite . particles even bigger ( 40@xmath4 ) than the grains measured in - situ with the spacecraft detectors were reliably identified by meteor radar observations @xcite . the grains were identified by their hyperbolic speeds , and their flow direction varies over a much wider angular range than that of the much smaller grains observed by spacecraft . @xcite identified a general background influx of extra - solar system particles from southern ecliptic latitudes with enhanced fluxes from discrete sources . more sensitive meteor observations with the arecibo radar found micron - sized interstellar meteor particles radiating from the direction of the geminga pulsar @xcite . this is particularly interesting because the supernova that formed the geminga pulsar is a potential candidate which may have created the local bubble . ulysses has monitored the interstellar dust flow through the solar system for more than 15 years now . this time period covers more than two and a half revolutions of the spacecraft about the sun through more than 2/3 of a complete 22-year solar cycle . thus , ulysses measured interstellar dust during solar minimum and solar maximum conditions of the interplanetary magnetic field ( imf ) . the interstellar dust flux modulation due to grain interaction with the rather undisturbed magnetic field during solar minimum could be well explained @xcite . the consideration of sensor side wall effects lead to an improved flux determination @xcite . until early 2005 the approach direction of the interstellar grains was in agreement with the interstellar helium flow direction @xcite . an example is shown in the left panel of figure [ rot ] which shows the impact direction of the interstellar impactors in the period from 1996 to 2000 . six years later , when ulysses was travelling through almost the same spatial region and had an almost identical detection geometry for interstellar grains , the situation was vastly different : first , the range in approach directions of the grains was somewhat wider ( best seen in 2004 ; right panel of figure [ rot ] ; see also @xcite ) ; second , and more noticeable , in 2005/06 the approach direction of the majority of grains was shifted away from the helium flow direction . preliminary analysis indicates that this shift is about @xmath16 away from the ecliptic plane towards southern ecliptic latitudes @xcite . at the moment , we do not know whether it is a temporary shift limited to the time period stated above or whether it continues to the present time . furthermore , the reason for this shift remains mysterious . whether it is connected to a secondary stream of interstellar neutral atoms shifted from the main neutral gas flow @xcite is presently unclear . given , however , that the neutral gas stream is shifted along the ecliptic plane while the shift in the dust flow is offset from the ecliptic , a connection between both phenomena seems unlikely . even though ulysses position in the heliosphere and the dust detection conditions were very similar during both time intervals considered in figure [ rot ] , the configurations of the solar wind driven interplanetary magnetic field ( imf ) , which strongly affects the dynamics of the smallest grains , were completely different . we have to consider that the interstellar grains need approximately twenty years to travel from the heliospheric boundary to the inner solar system where they are detected by ulysses . thus , the effect of the imf on the grain dynamics is the accumulated effect caused by the interaction with the imf over several years : in the earlier time interval ( 1997 - 1999 ) the grains had a recent dynamic history dominated by solar minimum conditions @xcite , while the grains detected during the second interval ( 2002 - 2005 ) had a recent history dominated by the much more disturbed solar maximum conditions of the imf . during the solar maximum conditions the overall magnetic dipole field changed polarity . @xcite predicted that due to this effect in a 22-year cycle , small interstellar grains experience either focussing or defocusing conditions . during these times they are systematically deflected by the solar wind magnetic field either towards or away from the solar magnetic equator plane ( close to the ecliptic plane ) . this latter configuration likely has a strong influence on the dust dynamics and the total interstellar flux in the inner heliosphere but it is not modelled in detail in the presently existing models . an explanation of the grain interaction with the imf at the recent solar maximum conditions is still pending . the fact that the models fit the flux variation by assuming a constant dust concentration in the local interstellar cloud ( lic ) implies that the dust phase of the lic must be homogeneously distributed over length scales of @xmath17 , which is the distance inside the lic travelled by the sun during the measurement period of ulysses from end 1992 to end 2002 @xcite . this conclusion is supported by the more recent ulysses data until the end of 2004 @xcite . the 2005/06 data , on the other hand , put a question mark onto this conclusion because if the observed shift in impact direction turns out to be intrinsic , it would imply that this homogeneity breaks down on larger length scales . isd grains carry information about their past dynamics outside the heliosphere and are thus of strong interest to understand the dynamical processes in the local interstellar cloud ( lic ) . they provide the main reservoir and transport mechanism of heavy elements in the interstellar medium @xcite . the dynamics of the grains is crucial for understanding of nucleation , growth and collisional destruction processes @xcite . these processes strongly depend on the relative velocities of the grains . the most important phenomena responsible for the spread of velocities in the lic are gas drag , interaction with the local interstellar magnetic field , radiation pressure and photoelectric emission @xcite . the relative strengths of the different forces strongly depend on the size and the charge of the grains , together with local conditions of the interstellar medium ( ism ) , like gas or magnetic field turbulences . observations of interstellar material ( ism ) towards nearby stars and inside of the solar system , combined with radiative transfer models , give a self - consistent description of the lic @xcite . the main characteristics of the lic are : atomic neutral hydrogen density @xmath18 , electron and ion density @xmath19 , temperature @xmath20 , and a relative sun - cloud velocity @xmath21 . the physical conditions in the lic are those of the intercloud medium warm , low density , partially ionized gas . an enhancement of refractory elements ( such as fe , mg , mn ) in lic gas , compared to cool interstellar clouds , points to the destruction of interstellar dust grains by interstellar shocks ( velocity 100 to 200 km s@xmath22 ) @xcite . the ism within 10 pc of the sun is highly inhomogeneous . at least 5 distinct cloudlets are found within 5 pc of the sun , with differing compositions and physical properties . temperatures range from 5,400 k ( towards @xmath23 cen ) to 10,000 k ( blue cloud towards @xmath24 cma ) and total densities from @xmath25 ( blue cloud towards @xmath24 cma ) to possibly @xmath26 ( g - cloud towards @xmath23 cen , * ? ? ? * ; * ? ? ? the gas - phase abundance of fe , with respect to undepleted s , varies by @xmath2750% within 3 pc of the sun , evidently due to grain destruction processes @xcite . if the ism is chemically homogeneous , elements absent from the gas phase must be depleted onto dust grains . this argument can be used to evaluate the gas - to - dust mass ratio over the integrated lic column , and can be compared with that of other nearby interstellar clouds . however , the required knowledge of the total chemical composition of the ism is an elusive quantity that has not been reliably determined . a 40%50% variation in / and / for the two clouds towards @xmath24 cma indicates different grain histories for two similar clouds within 3 pc of each other . if atoms not observed in the gas are concentrated in the dust , can be calculated from observations of interstellar absorption lines towards nearby stars . when evidence for 60%70% subsolar abundances is included , @xmath27600 integrated over the diameter of the lic @xcite . gas - to - dust mass ratios calculated from more recent models with improved solar abundances are in the range @xmath27140490 , again depending on solar abundances @xcite . interestingly , determined from comparisons of the ulysses in - situ measurements inside of the solar system , compared to gas densities from these models , yield = 116127 @xcite . it should be emphasised that the obtained from the in - situ measurements is an upper limit , since the smallest interstellar dust grains ( radii 0.1 @xmath4 ) are prevented from entering the heliosphere . overall , the in - situ value is a factor of @xmath2 2 larger than the one derived from astronomical observations , indicating a relative concentration of interstellar dust in the ism close to the sun compared to the @xmath270.5 pc lic cloud length towards @xmath24 cma . the gas - to - dust mass ratio also varies by more than 30% over the nearest 3 pc . if ism abundances are solar , the in - situ and astronomical methods of determining are generally in better agreement , but interstellar absorption line data towards weakly reddened stars remain unexplained . these differences are not yet understood . the chemical composition of interstellar dust grains observed within the solar system thus provides a window on the chemical composition and homogeneity of the ism . the combination of absorption line data toward @xmath24 cma and the modelled photoionization also lead to the conclusion that the lic has a very interesting pattern of gas phase elemental abundances @xcite : c appears to be substantially supersolar while fe , mg and si are subsolar . o and n are close to solar . this indicates that carbonaceous grains have been destroyed in the lic while silicate grains have survived . the extra c in the gas has not been explained but may be evidence for a local enhancement of carbonaceous dust followed by grain destruction in a shock . the masses of interstellar grains measured in - situ with the spacecraft detectors range from @xmath28 to above @xmath1 . if we compare the mass distribution of these interstellar impactors with the dust mass distribution derived from astronomical observations , we find that the in - situ measurements overlap only with the largest masses observed astronomically . this is further supported by the radar measurements which revealed even bigger grains . these measurements imply that the intrinsic size distribution of interstellar grains in the lic extends to sizes much larger than those grains which are detectable by astronomical observations @xcite . there are no direct observations of interstellar dust within 5 pc and outside of the solar system . the observations of very weak starlight polarization towards nearby stars ( @xmath2940 pc ) may originate from magnetically aligned dust grains close to the solar system . the observed polarization strength is consistent with the average interstellar density of @xmath30 over tens of parsecs in the upwind direction @xcite . the in - situ grains have a size distribution consistent with these classical dust grains . interstellar gas and dust couple through collisional processes , and through coupling of ions and charged grains to the interstellar magnetic field . over distance scales of 100500 pc , gas - dust coupling is demonstrated through the correlation of starlight - reddening dust grains ( measured as color excess e(b - v ) ) and interstellar hydrogen ( + 22 ) @xcite . for the multicloud structure observed within 5 pc , gas - dust coupling is not proven . the collisional lifetimes for classical dust grains ( radii @xmath31 m ) in the lic are @xmath32 years , during which time the lic will move @xmath275 pc through local space . the gyroradius is @xmath270.1 pc in a field of 3 @xmath33 g @xcite . the result will be magnetically captured dust grains that are collisionally destroyed over the lifetime of the cloud . grn and landgraf ( 2000 ) suggested that the small `` classical '' grains are replenished by the collisional destruction of larger dust grains . alternatively , silicate grain destruction peaks near shock column densities of @xmath34 @xcite , allowing the breakdown of gas - dust coupling locally over @xmath271 pc length scales . there are important consequences from the existence of the big particle population in the lic . while particles observed by spacecraft couple to the interstellar medium on length scales of less than 1 pc via electromagnetic interactions , more massive grains couple to the gas over much longer scales of 100 to 1000 pc @xcite . therefore , big interstellar meteor particles travel unaffected over much longer distances and may come directly from their source region . ulysses is presently in the inner solar system where interstellar grains can not reliably be separated from interplanetary impactors . after mid-2008 , however , if the ulysses spacecraft remains in good health , the dust instrument will monitor the interstellar grains again . the ulysses mission is presently planned to be extended until at least early 2009 so that additional dust data will hopefully become available . a further mission extension until 2011 is technically feasible and may provide interstellar dust data from the outer heliosphere again . with this latter extension the ulysses measurements would cover an almost entire 22-year solar cycle . it would make the ulysses data a unique data set of dust measurements from interplanetary space for decades to come . together with detailed modelling of the grain interaction with the imf during the highly disordered solar maximum conditions we will hopefully be able to reveal the origin of the observed @xmath16 shift . if the shift turned out to be intrinsic , being potentially connected with a secondary population of interstellar grains , it would put strong constraints on the small - scale structure of the lic . this would also be highly relevant for the interpretation of results from the stardust mission which recently brought a sample of collected interstellar grains to earth ( a. westphal 2006 , priv . comm . ) , and for future dust astronomy space missions aiming at the in - situ analysis and sample return of interstellar dust ( cosmicdune , sarim ; * ? ? ? * ; * ? ? ? we thank the ulysses project at esa and nasa / jpl for effective and successful mission operations . this work has been supported by the deutsches zentrum fr luft- und raumfahrt e.v . ( dlr ) under grants 50 0n 9107 and 50 qj 9503 . support by max - planck - institut fr kernphysik and max - planck - institut fr sonnensystemforschung is also gratefully acknowledged .	in the early 1990s , after its jupiter flyby , the ulysses spacecraft identified interstellar dust in the solar system . since then the in - situ dust detector on board ulysses continuously monitored interstellar grains with masses up to @xmath0 kg , penetrating deep into the solar system . while ulysses measured the interstellar dust stream at high ecliptic latitudes between 3 and 5 au , interstellar impactors were also measured with the in - situ dust detectors on board cassini , galileo and helios , covering a heliocentric distance range between 0.3 and 3 au in the ecliptic plane . the interstellar dust stream in the inner solar system is altered by the solar radiation pressure force , gravitational focussing and interaction of charged grains with the time varying interplanetary magnetic field . the grains act as tracers of the physical conditions in the local interstellar cloud ( lic ) . our in - situ measurements imply the existence of a population of big interstellar grains ( up to @xmath1 ) and a gas - to - dust - mass ratio in the lic which is a factor of @xmath2 2 larger than the one derived from astronomical observations , indicating a concentration of interstellar dust in the very local interstellar medium . until 2004 , the interstellar dust flow direction measured by ulysses was close to the mean apex of the sun s motion through the lic , while in 2005 , the data showed a @xmath3 shift , the reason of which is presently unknown . we review the results from spacecraft - based in - situ interstellar dust measurements in the solar system and their implications for the physical and chemical state of the lic .
nagaoka s theorem @xcite stands almost alone as a rigorous result about itinerant magnetism . it predicts the existence of a fully polarized ferromagnetic state as the unique ground state of the @xmath1 hubbard model , when one hole is doped away half - filling and certain connectivity conditions are satisfied . despite its very restricted validity and its thermodynamic irrelevance , the theorem introduced an interesting idea about quantum magnetism : kinetic magnetism , the possibility of magnetic order driven solely by the motion of the electrons . since the seminal work by nagaoka @xcite , a lot of effort has been dedicated to the study of nagaoka ferromagnetism stability beyond the constraints of the theorem . in particular , some controversy arose as to whether the fully polarized state survives for a finite density of holes ( see @xcite and references therein ) . however , large - scale density matrix renormalization group ( dmrg ) calculations @xcite , among others @xcite seem to have solved the problem , unless for the square lattice , as they predict the existence of nagaoka ferromagnetism up to critical hole density @xmath6 . little is known about the states that supplant the nagaoka ferromagnet beyond @xmath7 @xcite . with respect to the @xmath1 condition , its relaxation leads to the competition between the nagaoka and antiferromagnetic exchange mechanisms . this entails the instability of the nagaoka phase against phase separation : for @xmath8 , a ferromagnetic polaron around the hole moves on an antiferromagnetic background @xcite . lastly , the violation of the connectivity condition @xcite can also destabilize the nagaoka phase . nagaoka s theorem requires that @xmath9 where @xmath10 is the sign of the hopping amplitudes around the smallest loop of the lattice . when this condition is not fulfilled the hole kinetic energy is frustrated . kinetic frustration @xcite is a quantum mechanical phenomenon , without classical analog since it originates in the quantum interferences of different hole paths . a simple way to break the connectivity condition is the consideration of particular signs for the hopping parameters in non - bipartite lattices . as an alternative way , the hopping integrals can be modulated by a staggered magnetic flux @xcite . in 2005 , haerter and shastry @xcite found that the ground state of the @xmath1 triangular hubbard with @xmath11 a kinetically frustrated system where nagaoka s theorem is not valid ( @xmath12 ) , has a @xmath13 nel order . more recently , we found another example of kinetic antiferromagnetism , a @xmath14 nel order as the ground state of the square hubbard model with second - nearest neighbor hopping @xmath15 , and we uncovered the classical nature of these antiferromagnets @xcite . at the same time , we proposed a microscopic mechanism responsible for this kinetic antiferromagnetism , based on the relaxation of the kinetic frustration as the hole moves on an antiferromagnetic background . in order to deepen our understanding of kinetic magnetism , in this work we study the evolution of the nagaoka ferromagnet with kinetically frustrating @xmath0 hoppings . to make it , we solve the @xmath1 hubbard model on the anisotropic triangular lattice and the square lattice with second - nearest neighbor hopping ( see fig . [ redes ] ) , using the numerically exact dmrg . as we vary @xmath0 we can move between the two known limits : the nagaoka state ( @xmath16 ) and the novel kinetic antiferromagnet ( @xmath17 ) @xcite . we find that the classical order extends for all @xmath0 , ferromagnetic below a critical @xmath18 ( that slightly depends on the lattice ) , and antiferromagnetic above . we analyze the characteristics of the transition , the antiferromagnetic structure above @xmath18 , and the physical microscopic mechanism at work for each lattice . beyond the sustained theoretical interest in kinetic magnetism for decades , up to date , there are no clear experimental realization of nagaoka conditions . at the end of this work , we briefly mention some recent experimental proposals . in order to analyze the stability of the nagaoka ferromagnetic state against kinetic frustration , we study the @xmath19 hubbard model , with one hole doped away the half - filled case , on two lattices : the spatially anisotropic triangular lattice and the square lattice with second - nearest neighbor hoppings , as shown in fig . [ redes ] . the hubbard model reads @xmath20 where @xmath21 denote pair of sites on each lattice , @xmath22 are the hopping integrals , and @xmath23 is the onsite coulomb repulsion . in fig . [ redes ] , the solid lines represent @xmath24 ( which we take as the energy scale , @xmath25 ) , while the dashed lines represent @xmath0 , the varying anisotropic hoppings ( second - nearest neighbor hoppings ) for the triangular ( square ) lattice . it should be noticed that , for @xmath26 the hubbard model on the two lattices are equivalent . are the kinetically frustrating hoppings . we take @xmath25 as the energy unit.,scaledwidth=40.0% ] when @xmath27 , the connectivity condition of nagaoka s theorem is fulfilled because the minimal loops for the hole motion are squares with @xmath28 . for finite @xmath0 , the minimal loops are triangles with @xmath29 so , for @xmath30 , the theorem is valid and the ground state is a unique fully polarized ferromagnetic state for both lattices . on the other hand , a positive @xmath0 introduces kinetic energy frustration in the hole motion ( @xmath12 ) , invalidating nagaoka s theorem . in a previous work @xcite , we have shown that , in the special case @xmath17 , the ground states have classical antiferromagnetic orders : a @xmath31 pattern for the ( isotropic ) triangular lattice and the usual @xmath32 for the square one . in this work we will use dmrg @xcite to solve the ground state of the @xmath1 hubbard model , for @xmath33 . we apply dmrg to ladders of dimension @xmath34 , with @xmath35 legs ( enough to describe correctly two - dimensional systems @xcite ) and up to @xmath36 rungs . we impose cylindrical boundary conditions with periodic wrapping in the rung direction and open boundary conditions along the legs . to maintain errors of the dmrg smaller than symbol sizes in each figure , we have kept up to @xmath37 states , with a truncation error less than @xmath38 . as we mentioned above , the @xmath16 ground state of hamiltonian ( [ hubbard ] ) is ferromagnetic , while for @xmath17 it exhibits antiferromagnetic order for each lattice @xcite . hence , there will be a critical value @xmath18 where the nagaoka state is destabilized . to determine @xmath18 , we have resorted to an energy analysis . let @xmath39 be the ground state energy of the @xmath1 hubbard model , for an @xmath40-site lattice ( with @xmath41 electrons ) and a given sector of the spin projection @xmath42 . for a given @xmath0 , we have compared the ( ground state ) energies of the different spin projection sectors , from the maximal @xmath43 corresponding to a full spin polarization in the @xmath44 direction to the minimal @xmath45 . notice that , due to the @xmath46 symmetry of the model , the ferromagnetic nagaoka state is @xmath47-fold degenerate , where @xmath48 is its total spin , and it has projections in all the @xmath42 sectors . we have found that , for small values of @xmath0 and for both lattices , the computed ground state energies of all the @xmath42 sectors are degenerate ( in particular , @xmath49 ) . therefore , we can suspect that these degenerate states belong to the nagaoka ground state manifold . as a check , we have verified that the spin correlations , for different @xmath42 sectors , correspond to a fully saturated ferromagnet ( @xmath50 for @xmath51 ) , discarding then the possibility that the nagaoka state may be degenerate with lower total spin states . on the other hand , for larger values of @xmath0 , we have obtained that the ground state always belongs to the minimal spin projection sector @xmath52 , which would correspond to a total spin @xmath53 . we have not found partially polarized ground states for any @xmath0 , although we must warn that , very close to the transition point @xmath54 the flattening of @xmath55 makes the numerical treatment harder and less precise . with this caveat , we can say that , at @xmath54 there will be a transition from a nagaoka ferromagnet to a minimal spin state which , later on , we will characterize as an antiferromagnetic state . for the triangular lattice as a function of @xmath0 , and for different @xmath40-site clusters . dashed ( solid ) lines correspond to the @xmath56 ( @xmath57 ) sector energy . for smaller @xmath0 both energies overlap . @xmath58 is the thermodynamic limit of the nagaoka state energy . inset : zoom of the critical region , showing the intersection of @xmath59 with the ferromagnetic energy @xmath58.,scaledwidth=45.0% ] to determine the critical @xmath54 firstly we have looked for the @xmath0 value where the degeneracy between @xmath59 and @xmath60 is broken for each lattice . in figs . [ ene - t ] and [ ene - c ] we show these energies as a function of @xmath0 and for different cluster sizes , for the triangular and square lattices , respectively . we can see that the degeneracy is broken in the region around @xmath61 , signaling , as we explained above , that the ground state of the systems moves from the highest - spin nagaoka state to another one with minimal total spin . as the conventional dmrg algorithm uses the @xmath42 quantum number without discriminating between different total spin @xmath62 , we can not compute the energy of the excited @xmath53 state below @xmath18 ( notice that , in this case , the calculated @xmath59 corresponds to the @xmath63 sector energy of the @xmath64 nagaoka ground state ) . therefore , we do not have access to the expected energy level crossing between the highest- and lowest - spin sectors , that would facilitate the determination of @xmath18 . for the square lattice model , the lack of the level crossing is not so important as there is an appreciable kink in the ground state energy at @xmath18 ( see main panel of fig . [ ene - c ] ) . however , for the triangular case , the transition seems to be much smoother , as is shown in the main panel of fig . [ ene - t ] , and , consequently , it is more difficult to estimate the critical point where the degeneracy between @xmath65 and @xmath59 is lost . to avoid this difficulty , we have evaluated @xmath18 extrapolating the level crossing between @xmath59 and the infinite - lattice nagaoka energy , @xmath66 ( see insets of figs . [ ene - t ] and [ ene - c ] ) . @xmath66 can be computed exactly as the problem of one hole moving in a ferromagnetic background is identical to the spinless tight - binding system @xcite .critical @xmath18 values for different @xmath40-site triangular clusters , and its @xmath67 extrapolation limit . [ cols="<,^,^,^,^,^,^ " , ] second , for the square lattice with second - nearest neighbor hoppings , the exact energy of the fully polarized state in the thermodynamic limit is , @xmath68 , ( with @xmath69 ) and it is shown in fig . [ ene - c ] . in table [ table_ene - c ] we present the critical values for different @xmath40-site clusters , leading to @xmath70 in the @xmath71 limit . this value is quite close to the only one that existed in the literature up to the present , @xmath72 obtained within a restricted hilbert space @xcite . it is instructive to compare our results with the solution of the simplest systems that preserve the basics of nagaoka physics . that is , three electrons in hubbard square four - site clusters with nearest - neighbor @xmath24 hoppings and ( a ) @xmath0 hopping along only one diagonal ( triangular lattice ) ; ( b ) @xmath0 hoppings along both diagonals ( square lattice ) @xcite . in both cases , there is an energy level crossing for some @xmath18 . for @xmath73 the ground state has @xmath74 , corresponding to the nagaoka state , while for @xmath75 the ground state has minimal spin @xmath76 for system ( a ) the transition occurs at @xmath77 while for system ( b ) @xmath78 . these values roughly correspond to the gap energy @xmath79 between the @xmath74 and @xmath53 ground states for @xmath16 , that is , @xmath80 . it is remarkable that the critical @xmath18 values for these toy models are very close to the thermodynamic limit ones that we have presented above ( tables [ table_ene - t ] and [ table_ene - c ] ) . from this fact , we can deduce that the relevant quantum interferences for the nagaoka physics are those associated with the hole motion along the smallest lattice loops . we want to draw attention to the fact that the critical hoppings for both lattices are very similar ( @xmath81 , @xmath82 ) , and also they are numerically similar to the critical doping for the destabilization of the nagaoka state for @xmath16 , that is , @xmath6 @xcite . we guess that this agreement is not casual : if the nagaoka ferromagnet , with @xmath16 and one hole doped , is separated by an energy gap @xmath79 of other spin sectors , it is plausible that a `` perturbation '' may destabilize the phase as long as its characteristic energy is of the order of @xmath79 ( in the case of doping , we can associate it with an energy @xmath83 ) . so , if this argument is correct , we expect a gap of the order of @xmath84 for the @xmath16 nagaoka ferromagnet in the square lattice , a value close to the gap for the 4-site cluster system . the magnetic properties of the ground state can be inferred from the static magnetic structure factor @xmath85 defined as @xmath86 where @xmath21 run over all sites . we evaluate @xmath87 for @xmath88 , since all the momenta that belong to the first brillouin zone of each lattice have an equivalent point in this region ; for reference , the edges of the brillouin zones will be displayed in the figures . the @xmath89 component of each momentum is unequivocally determined by the periodic boundary conditions along the @xmath90 direction ; on the other hand , the open boundary conditions along the @xmath91 direction do not impose any restriction for the @xmath92 component . therefore , we can take advantage of this freedom to circumvent the discreteness of @xmath93 in the @xmath91 direction . for the triangular lattice with ( a ) @xmath16 , and ( b ) @xmath17 . the solid lines indicate the edges of the hexagonal brillouin zones . notice the discreteness of @xmath89 values.,scaledwidth=45.0% ] we have observed that @xmath85 exhibits a pronounced peak for a certain momentum ( and equivalent points in the reciprocal space ) , for both lattices , and all @xmath0 , except very close to the critical @xmath54 as we will discuss later . the intensity of the peak increases linearly with the cluster size , pointing out the existence of long - range magnetic order , and its position determines the magnetic wave vector @xmath94 of the magnetic order . therefore , besides the nagaoka state , long - range magnetic order is ubiquitous for the studied systems . next , we analyze the magnetic pattern as a function of the kinetic energy frustrating hoppings @xmath0 . first , we present the results for the triangular lattice . we have chosen the @xmath95 sites cluster for the presentation of @xmath85 as , for @xmath35 , it is expected to be the cluster most representative of the two - dimensional case @xcite . we begin revisiting the two previously known situations : @xmath16 and @xmath17 . for @xmath96 nagaoka s theorem is valid and , consequently , @xmath85 exhibits a sharp maximum at the magnetic wave vector @xmath97 , as it is shown in fig . [ orden - t](a ) . the other known case , @xmath17 @xcite , is presented in fig . [ orden - t](b ) . we can see two @xmath85 maxima at @xmath98 and @xmath99 , both equivalent and corresponding to the @xmath31 nel order . now , we analyze the magnetic order for intermediate @xmath0 between the two limits presented above . as long as @xmath100 we have found ferromagnetic order , that is @xmath101 in agreement with the ground state energy analysis of the previous section . increasing @xmath0 beyond the critical value , the @xmath85 peak locates at @xmath3 , with @xmath102 changing monotonously from @xmath103 at @xmath104 to @xmath105 at @xmath106 , as it is shown in fig . [ qtp ] . as a function of @xmath0 for the triangular lattice . insets : intensity plots of @xmath85 for @xmath107 and @xmath108 . the darker regions correspond to the magnetic wave vector @xmath3 ( and other equivalent point).,scaledwidth=40.0% ] this magnetic wave vector is characteristic of a spiral order with a pitch angle @xmath102 in the @xmath91 direction , and it connects the ferromagnetic and the @xmath13 nel orders . from the figure , it seems that @xmath102 depends continuously on @xmath0 , although it exhibits a sharp rise close to @xmath18 . for small spiral pitch , the period of the spin pattern along the @xmath91 direction ( @xmath109 ) can exceed the cluster length @xmath110 ( @xmath111 in our case ) , preventing a correct description of the spiral order . a clear manifestation of this kind of finite size effects is that , very close to the transition , the peaks of the magnetic structure factor are not so pronounced . for these reasons , we can not state categorically that the transition from the nagaoka state to the spiral one is continuous , although the @xmath102 dependence on @xmath0 suggest it . to underline this point , it is worth remembering that , in the previous section , we have seen the ground state energy for the triangular lattice also behaves rather smooth across the transition . we can speculate that an infinite - order phase transition takes place here , like the one found as a function of doping , for @xmath16 , in ref . @xcite . from fig . [ qtp ] we notice that , due to the sharp rise of @xmath102 , the spiral order has a pervasive antiferromagnetic character ( @xmath112 ) , except near the transition point . in the following , we focus on the square lattice model . in figs . [ orden - c](a ) and [ orden - c](b ) we show the @xmath16 and @xmath17 intensity plot of @xmath85 , respectively . we have chosen the @xmath113 cluster in this case @xcite . we can see in fig . [ orden - c](a ) that the ground state has ferromagnetic order , as it was expected , because of the validity of nagaoka s theorem for @xmath16 . for @xmath17 , fig . [ orden - c](b ) indicates that the ground state has the typical nel order with magnetic wave vector @xmath114 as was found in ref . the magnetic order evolution with second - nearest neighbor hopping @xmath0 is very simple . below @xmath115 the ground state is the nagaoka ferromagnet , while for @xmath116 it is the @xmath4 nel order . therefore , the transition is clearly discontinuous , with no intermediate order between the @xmath117 nagaoka and @xmath4 nel states . it should be remember that , in this case , the ground state energy ( fig . [ ene - c ] ) exhibits an appreciable non - analytical behavior at the transition . similar discontinuous transitions occur when kinetic frustration is due to staggered magnetic fluxes @xcite . for the square lattice with ( a ) @xmath16 , and ( b ) @xmath17 . solid lines refer to the edges of the square brillouin zones.,scaledwidth=45.0% ] summarizing , the addition of rather small @xmath0 hoppings ( @xmath118 ) destabilizes the nagaoka state and it induces long - range antiferromagnetic order , whose characteristics depend on the system : a spiral pattern in the triangular lattice , and the usual nel order in the square case . concerning the physical origin of this kinetic antiferromagnetism , recently we identified its microscopic mechanism @xcite . while the introduction of @xmath0 increases the hole kinetic energy ( the only one involved for @xmath1 ) in a ferromagnetic background , quantum interference effects can release this kinetic energy frustration if the hole moves along a certain antiferromagnetic pattern . this release can occur in two different ways , depending on the lattice geometry and the hopping terms : ( a ) the hole acquires a non - trivial spin berry phase due to the antiferromagnetic texture or , ( b ) the magnetic order leads to an effective vanishing of the hopping amplitude along the frustrating loops . in our work , the spin - berry phase mechanism is operating for the stabilization of the spiral order in the triangular lattice model since , in that order , the hole acquires a @xmath119 phase when it hopes along one elementary triangle . on the other hand , for the square case , the effective hole nearest - neighbor hopping amplitude vanishes as a consequence of the antiparallel spin structure of the @xmath14 nel order , and the hole moves only along the diagonal directions . we refer the reader to ref . @xcite ( especially to the supplemental material ) , for a more detailed discussion of the kinetic antiferromagnetism mechanism . finally , we have calculated the magnetic order parameter , that is , the local staggered magnetization @xmath120 , which can be defined as @xmath121 where @xmath122 denotes the magnetic sublattices . a straightforward calculation shows that @xmath123 . on the other hand , for non - collinear magnetic structures , like the @xmath3 spiral , we use the relation @xmath124 that can be deduced considering semi - classical spin correlations , @xmath125 for @xmath126 . the same procedure recovers @xmath127 for collinear orders . as a function of @xmath0 for ( a ) the @xmath95 triangular lattice without a magnetic field ( open triangles ) and with a magnetic field @xmath128 applied to one @xmath13 sublattice ( solid triangles ) ; ( b ) the @xmath113 square lattice without a magnetic field ( open squares ) and with a magnetic field @xmath128 applied to one of the two sublattices of the @xmath14 nel order . the dashed lines correspond to the classical local magnetization , @xmath129.,scaledwidth=45.0% ] the curve with open triangles ( squares ) in fig . [ ordenqx - t ] shows the evolution of the local magnetization with @xmath0 , for the @xmath95 ( @xmath113 ) triangular ( square ) cluster . the behavior for other clusters is qualitatively similar . in the nagaoka phase , @xmath120 takes the classical value , given by @xmath130 , as it was expected for a fully saturated state . on the other hand , the kinetic antiferromagnetic phases have large values of @xmath120 , especially the triangular model ( upper panel of the figure ) where it is close to @xmath131 for all @xmath0 . in ref . @xcite , we found that , for the particular value @xmath17 and both models , @xmath120 extrapolates to its classical value in the thermodynamic limit . there , we argued that the classical value is not reached for finite clusters due to the @xmath46 symmetry of the hubbard model . one way to confirm this idea , for commensurate magnetic orders , was the finding that the application of a small uniform magnetic field @xmath132 in one sublattice only , was enough to pin the classical order for the finite clusters @xcite . in this work , we wonder if the remarkable classical character mentioned above extends to other values of @xmath0 . hence , we have applied a magnetic field @xmath133 in one of the two ( three ) sublattices of the @xmath134 ( @xmath13 ) structure in the square ( triangular ) lattice model . notice that , in the triangular case , due to the incommensuration of the @xmath135 spiral phase for a generic @xmath102 , the application of the pinning magnetic field matches the order only when @xmath136 ( so , for @xmath137 we do not apply @xmath138 ) . the curves with solid symbols in fig . [ ordenqx - t ] show the @xmath0 dependence of @xmath120 with the magnetic field @xmath138 applied . convincingly , it can be seen that @xmath138 brings out the `` hidden '' classical nature of the kinetic antiferromagnetism in these finite clusters . therefore , we have found that , for all @xmath0 , the ground state of the considered @xmath1 hubbard models has classical magnetism . above @xmath18 , this is a remarkable result as it can be thought as a perfect counterpart of nagaoka s theorem . about the physical reason for the classical nature of the antiferromagnetism , we speculate that the hole motion under the @xmath1 condition generates effective long - range spin interactions , that may favor classical ordering @xcite . to conclude , we have investigated the evolution of the nagaoka state with @xmath0 hopping processes that induce hole kinetic energy frustration . to this purpose , the numerical density matrix renormalization group is used to compute the magnetic ground state properties of the @xmath1 hubbard model , with one hole doped away from half - filling , on two lattices : the spatially anisotropic triangular lattice and the square lattice with second - nearest neighbor hoppings . we have found that the nagaoka ferromagnetic state is destabilized for rather small frustrating hoppings : @xmath139 ( 0.214 ) for the triangular ( square ) lattice . taking into account that these values are comparable to the corresponding ones for a simple 4-site cluster , we can state that nagaoka physics is driven mainly by quantum interferences generated by the hole motion along the smallest lattice loops . the analysis of the magnetic structure factor indicates that the ground state of the @xmath1 hubbard model has long - range magnetic order for all hopping @xmath0 . for @xmath75 the square lattice ground state has a classical @xmath14 nel order . in the triangular case , the magnetic order is a classical spiral pattern with magnetic wave vector @xmath3 , that seems to connect continuously the nagaoka ferromagnet at @xmath140 with the @xmath13 nel antiferromagnet at the isotropic point @xmath17 . therefore , the kinetic antiferromagnetism , first discovered by haerter and shastry @xcite on the triangular lattice and further developed by us @xcite , seems to be a robust phenomenon in kinetically frustrated electronic systems . its classical nature makes this antiferromagnetism the perfect counterpart of the nagaoka s state for frustrated systems . on the other hand , the lattice geometry affects the nature of the transition between the different classical states , as we have seen : continuous for the triangular lattice , discontinuous for the square one . finally , it remains a challenge to find experimental realizations of both , the old nagaoka ferromagnetism and the new kinetic antiferromagnetism that have been studied in this and previous works @xcite . up to date , there is no clear evidence of nagaoka physics in condensed matter systems , being the main obstacle the large onsite coulomb repulsion needed , @xmath141 . however , tunable feshbach resonances in ultracold atoms allow to reach such large @xmath142s , and , on the other hand , the generation of artificial gauge fields in triangular optical lattices @xcite can induce frustrated motion . other proposals involve artificial lattice of quantum dots @xcite , and high - density two - dimensional electron gas at the interface between mott and band insulators @xcite . all this opens the interesting possibility to test experimentally our proposals about kinetic magnetism .	we investigate , using the density matrix renormalization group , the evolution of the nagaoka state with @xmath0 hoppings that frustrate the hole kinetic energy in the @xmath1 hubbard model on the anisotropic triangular lattice and the square lattice with second - nearest neighbor hoppings . we find that the nagaoka ferromagnet survives up to a rather small @xmath2 at this critical value , there is a transition to an antiferromagnetic phase , that depends on the lattice : a @xmath3 spiral order , that continuously evolves with @xmath0 , for the triangular lattice , and the usual @xmath4 nel order for the square lattice . remarkably , the local magnetization takes its classical value for all considered @xmath0 ( @xmath5 ) . our results show that the recently found classical kinetic antiferromagnetism , a perfect counterpart of nagaoka ferromagnetism , is a generic phenomenon in these kinetically frustrated electronic systems . pacs numbers : : 75.10.lp , 71.10.fd
[ [ section ] ] the three - oscillator model for the dipole , the target cavity and the fp cavity can be described by a non - hermitian hamiltonian @xmath12 , \label{eq : hamiltonian_s}\end{aligned}\ ] ] where @xmath13 , @xmath14 and @xmath15 are the angular frequencies of the dipole , the target cavity mode and the fp mode , respectively . @xmath66 and @xmath67 describe the intrinsic linewidths of the uncoupled cavity modes , which are related to the q - factor , @xmath16 . the dipole is assumed to be spatially decoupled from the fp modes and is coupled to the target cavity with a coupling rate of @xmath68 . the coupling rate between the target cavity and the fp cavity is described by @xmath17 . at first we consider the submatrix which contains the components from the target and fp cavity @xmath69 . \label{eq : submatrix_s}\end{aligned}\ ] ] by solving the eigenvalue equation @xmath70 , the complex eigenvalues @xmath71 are obtained @xmath72 the eigenvectors are found as @xmath73= & \left [ \begin{array}{ccc } \dfrac{\eta}{\sqrt{\eta^2+\left\|\tilde{\omega}_1-\left(\omega_t - i\kappa_t\right)\right\|^2 } } \\ \dfrac{\tilde{\omega}_1-\left(\omega_t - i\kappa_t\right)}{\sqrt{\eta^2+\left\|\tilde{\omega}_1-\left(\omega_t - i\kappa_t\right)\right\|^2 } } \\ \end{array } \right ] , \\ \left [ \begin{array}{ccc } \alpha_2 \\ \beta_2 \\ \end{array } \right]= & \left [ \begin{array}{ccc } -\dfrac{\tilde{\omega}_1-\left(\omega_t - i\kappa_t\right)}{\sqrt{\eta^2+\left\|\tilde{\omega}_1-\left(\omega_t - i\kappa_t\right)\right\|^2 } } \\ \dfrac{\eta}{\sqrt{\eta^2+\left\|\tilde{\omega}_1-\left(\omega_t - i\kappa_t\right)\right\|^2 } } \\ \end{array } \right ] , \label{eq : eigenfunction_s}\end{aligned}\ ] ] where @xmath25 , @xmath26 , and @xmath27 . going back to the three - oscillator case , the vectors used to diagonalize the submatrix read @xmath74 , a_2= \left [ \begin{array}{ccc } 0 \\ \alpha \\ -\beta\\ \end{array } \right ] , a_3 = \left [ \begin{array}{ccc } 0 \\ \beta \\ \alpha \\ \end{array } \right ] . \label{eq : basis_s } \end{split}\end{aligned}\ ] ] the transformation matrix can be constructed with the vectors in eq . ( [ eq : basis_s ] ) , which changes the hamitonian to the new bases of coupled cavity modes @xmath75 . \label{eq : tmatrix_s}\end{aligned}\ ] ] thus , the hamiltonian with the new bases as also shown in eq . ( 4 ) can be derived @xmath76 . \label{eq : hamiltonian2_s } \end{split}\end{aligned}\ ] ] we note that the effect of pure dephasing and thus off - resonant cavity feeding are not taken into account in the above simple model . in principle , for large dephasing rates @xmath77 , a similar analysis as in the main paper can be carried out using a generalized purcell formula for a broad emitter , which has been derived elsewhere @xcite . this leads to a similar expression in case of strong coupling for @xmath78 therein the actual q - factor of the cavities is replaced by an effective quality factor @xmath79 taking into account the broad emitter linewidth @xmath80 . in the present case , we believe that even at the temperature of 77k which is used in the experiment , the linewidth of the emitter is still smaller or comparable to the cavity mode . even in the case of strong dephasing the main conclusions of the paper are not affected , though the actual shape of the theoretical curves of the se change in fig.4 may slightly change . in any case the field delocalization and the change of the effective quality factor will determine the change of the decay rate of the emitter . the sample was grown by metal - organic vapor phase epitaxy on a inp ( 100 ) substrate misoriented @xmath81 toward the ( 110 ) facet . a 100 nm inp buffer layer was initially deposited , followed by a 110 nm ingaasp layer with bandedge emission at 1250 nm . a single layer of qds was then grown by depositing 2 monolayers ( mls ) of inas on top of a 1.2ml gaas interlayer ( ref . @xcite of the main text ) . a growth interruption in tertiary butyl arsine of 10 seconds was applied after the dot formation , which was followed by another 110 nm ingaasp layer . a 50 nm inp layer was finally grown as the capping layer . the photonic crystal cavities was fabricated using electron beam lithography and inductively coupled plasma etching with the cl@xmath82/ar / h@xmath82 mixture . the wet etching in hcl / h@xmath82o was employed to remove the inp sacrificial layer and to make a 220 nm ingaasp free standing membrane . a lattice constant of 480 nm and a filling factor of 0.30 were chosen to achieve the target cavity mode near 1550 nm . as seen from the @xmath41pl spectrum of the fp cavity in fig . [ fig : graph2](c ) of the main text , the target cavity mode is detuned 15 nm from the dispersion edge of the waveguide mode supported by the fp cavity ( see also ref . @xcite of the main text for a detailed discussion of the photonic band structure . )	we demonstrate the remote all - optical control of the spontaneous emission ( se ) of quantum dots using coupled photonic crystal cavities . by spectrally tuning a fabry - perot cavity in resonance with a target cavity , the quality factor and the local density of states experienced by emitters in the target cavity are modified , leading to a change in the se rate . from the theoretical analysis of the coupled - cavity quantum electrodynamics system , the se rate change can be higher than the quality factor change due to a reduction of the vacuum field at the emitters position when the two cavities are brought in resonance . both the weak and strong coupling regimes of two cavities have been observed experimentally and the se decay rate has been modified by more than a factor of three with remote optical control . semiconductor quantum dots ( qds ) precisely positioned inside photonic crystal ( phc ) cavities represent a scalable platform for solid - state cavity quantum electrodynamics ( cqed ) @xcite . in general , quantum information should be stored in the charge or spin degree of freedom in the quantum system , and then released to a photonic channel , e.g. through spontaneous emission ( se ) . the se rate can be greatly enhanced by increasing the local density of states ( ldos ) at the emitters position using a wavelength - sized , low - loss cavity @xcite . in order to control the timing of the emission process , a procedure is needed to switch the emitter - cavity coupling on and off . an attractive approach will be the ultrafast control of the cavity characteristics , leading to a dynamic modulation of the ldos and therefore of the se rate . the dynamic tuning of the quality factor ( q - factor ) recently demonstrated in silicon - based phc cavities is an example of such cavity modulation techniques @xcite . a static and quasi - permanent change of the q - factor and of the emission intensity of qds was recently observed by producing a structural modification in the vicinity of the cavity @xcite . however , for the practical application in cqed , it requires a remote and reversible control scheme which produces large changes in the ldos but does not affect the emitter - cavity interaction by local perturbations . the coupled - cavity quantum electrodynamics ( ccqed ) platform proposed by hughes @xcite is an interesting candidate for such remote and reversible control , and several demonstrations of strong coupling between phc cavities have been recently reported @xcite . in this work , we present the experimental observation of the controlled and reversible modification of the se rate in a target cavity by tuning the properties of a remote control cavity . the tuning is implemented using thermo - optic effects , but the concept is readily extendable to the ps time scale by using free - carrier injection , opening the way to the dynamic control of se in solid - state cavities . the enhancement of the se rate @xmath0 in a cavity as compared to the se rate in the bulk @xmath1 is generally described in perturbation theory by @xcite , @xmath2 where @xmath3 is the cavity mode wavelength , @xmath4 is the dipole emission wavelength , @xmath5 is the linewidth of the cavity mode , @xmath6 is the refractive index , @xmath7 is the q - factor of the cavity mode and @xmath8 is the normalized mode function at the dipole position @xmath9 . the first term at the right - hand side of eq . ( [ eq : purcell ] ) is the well - known purcell factor , @xmath10 , which is governed by the q - factor and the mode function distribution or inversely the mode volume . the second term describles the effect of the wavelength detuning . the se rate is conventionally controlled through the wavelength detuning , e.g. by changing the emitter s energy or the cavity mode frequency . instead , in the following we demonstrate a remote optical control of the se rate by a real - time change of the purcell factor @xmath10 using coupled cavities . fig . [ fig : graph1 ] ( a ) shows an ideal emitter - cavity scheme where one of the cavity mirrors is replaced by a fabry - perot ( fp ) etalon . the fp cavity mode has a periodic mode structure sketched in fig . [ fig : graph1 ] ( b ) and it is assumed to have a lower q - factor compared to the target cavity mode . by adjusting the length of the fp cavity or modulating the refractive index of the media inside it , the set of fp modes is spectrally tuned with respect to the target cavity mode . at resonance the coupling causes an additional leakage for the optical field confined in the target cavity resulting in a change of the q - factor and the mode function distribution @xmath11 . hence , the se rate of emitters in the target cavity can be controlled non - locally by adjusting the resonant condition . differently from the structure proposed in ref . @xcite , we use evanescent coupling between two adjacent cavities and an extended fp cavity for the control , whose periodic mode structure facilitates the spectral resonance with the target cavity . the presented system is described by a three - oscillator model including the dipole , the target cavity and the fp cavity . the dressed states of the system are obtained by diagonalizing the non - hermitian hamiltonian @xcite @xmath12 \label{eq : hamiltonian}\end{aligned}\ ] ] where @xmath13 , @xmath14 and @xmath15 denote the angular frequencies of the dipole , the target cavity mode and the fp mode , the corresponding losses are related to the q - factor , @xmath16 , and @xmath17 is the coupling rate between the two cavities . the dipole is placed in the target cavity and spatially decoupled from the fp modes . the weak coupling between the dipole and the target cavity ( @xmath18 ) is always assumed and the terms of strong coupling and weak coupling in the following discussion are only related to the coupling between cavities . we first solve the subsystem of coupled cavities ( see supplementary materials ) , which yields the complex eigenvalues @xmath19 the effect of wavelength detuning @xmath20 on the q factor of the coupled cavity modes is illustrated in fig . [ fig : graph1 ] ( c ) . by tuning a low q cavity mode through a high q target cavity mode a significant change in the imaginary parts of the coupled eigenfrequencies @xmath21 hence in the q - factor can be observed at the crossing point . ] transforming the matrix ( [ eq : hamiltonian ] ) into the basis of the coupled cavity modes in eq . ( [ eq : eigen ] ) yields @xmath22 . \label{eq : hbase } \end{split}\end{aligned}\ ] ] the normalized mode functions , @xmath23 , of the coupled - cavity system are mixtures of the eigenfunctions of the initial cavity modes , @xmath24 , where @xmath25 , @xmath26 , and @xmath27 . consequently , the coupling constants between the dipole and the new eigenmodes are weighted by their target cavity fraction due to the positioning of the dipole . because of the mode coupling , the mode distribution of the state @xmath28 becomes increasingly delocalized for decreasing detuning , leading to a reduction of the vacuum field at the dipoles position . this effect can be intuitively interpreted as a change of the effective mode volume due to the field delocalization . the modulation of the q - factors and the mode function distribution will affect the dynamics of the dipole - cavity interaction . for instance in the case that the two coupled cavities are in the strong coupling regime @xmath29 , due to the energy splitting a narrow dipole will interact with only one of the two eigenmodes , leading to @xmath30 , where @xmath31 is the uncoupled q - factor of the target cavity , @xmath32 is the higher q - factor of coupled modes at resonance , and @xmath33 and @xmath34 are the corresponding se rates . it follows that the se ratio between the off - resonant and on - resonant cases , @xmath35 , is not only affected by the q - factor ratio , @xmath36 , but also enhanced by a factor of two due to the mode delocalization . in contrast , in the weak coupling @xmath37 , the se enhancement has contributions from both coupled modes @xmath38 , which yields @xmath39 in the case of @xmath40 . ) distribution between two cavities when they are out of resonance ( top ) and in resonance ( bottom ) . ( c ) @xmath41pl spectra of ccqed system when focusing the excitation beam on the target cavity ( red ) and fp cavity ( green ) . [ fig : graph2 ] ] fig . [ fig : graph1 ] ( d ) presents simulation results for the ccqed system using matrix ( [ eq : hamiltonian ] ) . we set @xmath42 , @xmath43 , and the dipole is always resonant with the coupled mode which has higher q - factor @xcite . the q - factor ratio , @xmath44 , is plotted as a function of the coupling strength divided by the fp cavity loss , @xmath45 ( red dashed curve ) . the corresponding ratio of the se decay rate , @xmath46 , of the dipole as a function of the coupling strength is also plotted ( green dashed curve ) . while in the weak coupling regime the se change follows the change in the q - factor , in the strong coupling regime the se change is up to a factor of two larger than the q - change as discussed above . in reality , the target cavity mode can couple to more than one fp mode simultaneously in case of high coupling strength . the q - factor change and the se change of the dipole with multiple fp modes are shown by the green and red solid curves in fig . [ fig : graph1](d ) . three equally spaced fp modes with 3 nm separation are taken into account in simulation . due to the more complex interaction of multiple modes the q - factor change and the change in the spontaneous emission rate are modified with respect to the single fp mode calculation . however , the above analysis of the decay rates for the weak and strong coupling still holds . the continuous black lines are fits based on the coupled mode theory . [ fig : graph3 ] in the experiment , our platform for the ccqed is based on phc cavities using qds as the emitter . cavities were realized within a hexagonal phc lattice fabricated in an ingaasp membrane which contains a single layer of self - assembled inas qds with a dot density around @xmath47 ( see supplementary materials ) . the qd ensemble gives a broad emission centered at 1570 nm @xcite . fig . [ fig : graph2 ] ( a ) shows the scanning electron microscope ( sem ) image of the fabricated structure based on two local defects along a w1 phc waveguide . the target cavity is constituted by a high - q double - heterostructure ( dhs ) cavity @xcite with two periods of the lattice constant slightly larger than the original lattice constant , @xmath48 . the fp cavity consists of eighty periods of modulated lattice constant ( @xmath49 ) . fig . [ fig : graph2 ] ( b ) presents the simulated electric field distribution based on 3d fdtd method @xcite . a leakage from the target cavity to the fp cavity can be clearly observed when the two cavities are in resonance . the measurements were performed in a confocal microscope at a temperature of 77 k , in order to reduce the influence from the non - radiative recombination and the homogenous broadening of qds , while allowing the thermo - optic tuning of the phc cavity @xcite . a diode laser emitting at 780 nm was used for exciting the cavity and the @xmath41pl signal collected by the objective is measured with a spectrometer and an ingaas array . fig . [ fig : graph2](c ) shows two typical @xmath41pl spectra taken from single - beam measurements by focusing the laser beam on either the target or the fp cavity . the fp cavity has a quasi - periodic mode structure , which is governed by the modified dispersion relation close to the slow - light frequency of the phc waveguide @xcite . firstly , single - beam measurements have been performed with one laser beam focused on the target cavity and the signal collected from the same position . by increasing the laser power , a red shift of the target cavity mode due to thermo - optic tuning is observed with only a small shift of fp modes caused by heat diffusion @xcite . fig . [ fig : graph3](a ) shows a weak coupling case in a phc with 4.5-period barrier between two cavities and 20-period barriers on the two sides . the main panel in fig . [ fig : graph3](a ) shows the spectral map with normalized intensities for various detunings , while the two small plots present the cavity mode wavelength and q - factor as functions of the wavelength detuning . it exhibits a clear crossing of two coupled modes and indicates the weak coupling regime . the q - factor of the target cavity decreases down from 13800 to 6500 and then increases again to 15100 , whilst the q - factor of the fp cavity increases slightly from 4000 to 4900 and then decreases to 3500 . by fitting the crossing behavior of two coupled modes using eq . [ eq : eigen ] , as shown by the black lines , we extract the parameters @xmath50 , @xmath51 , and @xmath52 . fig . [ fig : graph3](b ) exhibits a different situation where the barrier thickness between two cavities is reduced to 4 periods . in this case , two coupled modes shows anti - crossing , which indicates the strong coupling regime . the q - factor of the mode at shorter ( longer ) wavelength varies from 8500 to 1800 ( from 2600 to 8500 ) , respectively . numerical fitting reveals @xmath53 , @xmath54 , and @xmath55 . two - beam measurements have been carried out to demonstrate the remote control of both q - factor and se rate with one pulsed laser beam ( @xmath56 , pulse width 100 ps ) focused on the target cavity and a cw heating laser beam focused at the center of the fp cavity , which is tilted slightly of the optical axis of the objective lens in order to position it at a distance of a few tens of micrometers from the other beam . fig . [ fig : graph4](a ) presents the spectra with weakly - coupled cavities at different pump powers of the tuning laser . the decay time of qds resonant with the high - q coupled mode has been measured by time - correlated single photon counting with a superconducting single photon detector @xcite , using a narrow bandpass filter ( @xmath57 ) . fig . [ fig : graph4](b ) shows the se decay at the high - q mode frequency with different wavelength detunings between the cavities and the decay of the qd emission in the bulk ( outside the phc ) , in a @xmath58 bandwidth , as a reference ( grey curve ) . by fitting the initial part of the curve with an exponential decay function , the spontaneous emission lifetime modulated by phcs is obtained , while the slow decay component is attributed to dark excitons @xcite . compared to the 2.5 ns decay time of the qd emission , the decay time at target cavity is 0.5 ns and 1.5 ns when two weakly coupled modes are off - resonance and on - resonance , respectively . the strong coupling case has been also studied as shown in fig . [ fig : graph4](d)-(f ) . the se decay time is 1.4 and 2.2 ns for the high - q mode , when two cavity modes are off - resonance and on - resonance , respectively . pl spectra measured at different wavelength detuning . ( b ) and ( e ) se decay curves at different wavelength detuning . the grey curve is the se decay from qds in the bulk . ( c ) and ( f ) the decay time and q - factor at the high - q mode as a function of the wavelength detuning . the dashed curves are simulations based on the coupled mode theory . [ fig : graph4 ] ] to fit the experimental decay times we use the expression @xmath59 , where @xmath60 is the total emission rate into the two coupled modes calculated from diagonalizing the matrix ( [ eq : hamiltonian ] ) , while @xmath61 is the emission rate into the out - of - plane leaky modes which is measured from the decay time of qds in the phc mirror region . the global fit as shown by the dashed curves in fig . [ fig : graph4](c ) and ( f ) closely matches the experimental results . @xmath62 and 1.40 can be identified for the weak and strong coupling , respectively . both are close to the transition regime defined by @xmath63 . the measured decay rate into the cavity modes changes by a factor of 3.7 while the q factor changes by a factor of 2.1 in the weak coupling regime . in the strong coupling case the cavity decay rate changes by a factor of 2.7 while the q factor changes by a factor of 1.6 . both cases confirm that in case @xmath64 the se rate change can be higher than the q - factor change . in summary , we have demonstrated the real - time , remote all - optical control of the se rate by more than a factor of three in a coupled phc cavity system where the control beam is separated more than 20 @xmath41 m from the target cavity . the results have been interpreted in the framework of coupled mode theory , showing that the se rate is affected by the change in both the q - factor and the mode volume . these results open the way to the ultrafast control of the qd - cavity interaction by replacing the thermo - optic tuning with free - carrier injection on the ps time scale . the authors are grateful to b. wang , m. a. d@xmath65ndar , and r. w. van der heijden for fruitful discussions , and to e. j. geluk , e. smalbrugge , and t. de vries for technical support . this research is supported by the dutch technology foundation stw , applied science division of nwo , the technology program of the ministry of economic affairs under project no . 10380 , the fom project no . 09pr2675 , and nanonextnl , a micro and nanotechnology program of the dutch ministry of economic affairs , agriculture and innovation ( el&i ) and 130 partners . 35 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . , , , , , , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , , , * * , ( ) . , , , , * * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . it should be noted that , in the case of a dipole with fixed energy , the se rate change in the strong coupling regime will be even higher due to the spectral decoupling between the dipole and coupled mode as the latter is tuned . however , in the experiments described in this work the se rate of emitters resonant with the coupled mode is measured and the results in fig . 1 apply . , , , , , * * , ( ) . , , , , * * , ( ) . meep is developed by the joannopoulos research group at mit . http://ab-initio.mit.edu/wiki/index.php/meep , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , , , , , , * * , ( ) . , , , , , * * , ( ) .
recent data from the super - kamiokande experiment @xcite have strengthened the interpretation of the solar @xcite and atmospheric @xcite neutrino anomalies in terms of neutrino oscillations . oscillations have also been invoked to describe the appearance of electron neutrinos and antineutrinos in the lsnd experiment @xcite . because confirmation of the lsnd results awaits future experiments and recent measurements in the karmen detector exclude part of the lsnd allowed region @xcite , a conservative approach is to assume that oscillations need only account for the solar and atmospheric data . then the two mass - squared difference scales in a three - neutrino model are sufficient to describe the data . a number of three - neutrino models have been investigated @xcite . one attractive possibility is that both the atmospheric @xmath5 and solar @xmath6 oscillate maximally @xcite or near - maximally @xcite at the @xmath7 and @xmath8 scales , respectively . most detailed oscillation fits have been done separately for the solar and atmospheric data in effective two - neutrino approximations @xcite . in this paper we make full three - neutrino fits to the solar and atmospheric oscillation data to determine the allowed values for the general three - neutrino mixing matrix under the assumption that one mass - squared difference , @xmath9 , explains the atmospheric neutrino oscillations and that the other independent mass - squared difference , @xmath10 , explains the solar neutrino oscillations via the vacuum long - wavelength scenario @xcite . we choose a particular parametrization for the three - neutrino mixing in which the expressions for the atmospheric and solar neutrino oscillations share only a single parameter , and discuss the degree to which the two phenomena decouple . we find that the extension from two to three neutrino species does not improve the separate fits to either the atmospheric or solar data . although pure @xmath11 oscillations of atmospheric neutrinos are favored , there exist three - neutrino solutions with non - negligible @xmath12 oscillations , even with the constraints from the chooz reactor experiment @xcite . hence both @xmath12 and @xmath13 oscillations may be observable in future long - baseline experiments . we also find that future measurements of seasonal variations in the solar neutrino signal can provide an unmistakable sign of vacuum oscillations , and can further constrain the allowed parameter regions . in sec . ii we review the formalism for oscillations of three neutrinos relevant to the atmospheric and solar phenomena . in sec . iii we evaluate the current allowed two - neutrino parameter regions , and briefly review the evidence indicating that two distinct @xmath14 are needed to describe the solar and atmospheric data . in sec . iv we obtain three - neutrino solutions from a combined fit to the atmospheric and solar data . we conclude with a discussion of future tests of vacuum oscillations in sec . the survival probability for a given neutrino flavor @xmath15 in a vacuum is @xcite p ( _ _ ) = 1 - 4 _ k < j mixing matrix ( in the basis where the charged - lepton mass matrix is diagonal ) , @xmath16 , @xmath17 , and the sum is over all @xmath18 and @xmath19 , subject to @xmath20 . the matrix elements @xmath21 are the mixings between the flavor ( @xmath22 ) and the mass ( @xmath23 ) eigenstates , and we assume without loss of generality that @xmath24 . the solar oscillations are driven by @xmath25 and the atmospheric oscillations are driven by @xmath26 . the off - diagonal vacuum oscillation probabilities in a three - neutrino model are @xcite p(_e _ ) & = & 4\|u_e3 u_3^\|^2 ^2_atm -4 re\ { u_e1 u_e2^ u_1^ * u_2 } ^2_sun -2j2_sun , [ pemu ] + p(_e _ ) & = & 4\|u_e3 u_3^\|^2 ^2_atm -4 re\{u_e1 u_e2^ u_1^ * u_2 } ^2_sun + 2j2_sun , [ petau ] + p ( _ _ ) & = & 4\|u_3 u_3^\|^2 ^2_atm -4 re\ { u_1 u_2^ u_1^ * u_2 } ^2_sun -2j2_sun , [ pmutau ] where the @xmath27-violating `` jarlskog invariant '' @xcite is @xmath28 for any @xmath29 , @xmath30 , @xmath31 , and @xmath18 ( e.g. , @xmath32 for @xmath33 , @xmath34 , @xmath35 , and @xmath36 ) . the @xmath27-odd term changes sign under reversal of the oscillating flavors , or if neutrinos are replaced by anti - neutrinos . we note that the @xmath27-violating probability at the atmospheric scale is suppressed to order @xmath37 , with the leading term cancelling in the sum over the two light - mass states ; thus , @xmath38 ( and therefore @xmath39 from @xmath40 invariance ) at the atmospheric scale . without loss of generality we work in the basis where the charged - lepton mass matrix is diagonal . the matrix @xmath41 that relates the flavor eigenstates to the mass eigenstates may be parametrized in terms of three euler angles and one ( three ) phases for dirac ( majorana ) neutrinos . this can be understood as follows . in the dirac case , @xmath41 is analogous to the ckm mixing matrix in the quark sector , in which there are three euler angles and one phase . for majorana neutrinos , the @xmath42 mass matrix in the flavor basis must be symmetric but may be complex @xcite . there are 12 independent parameters , which can be taken as 3 real mass eigenvalues , 3 real euler angles , and 6 phases . then three of the phases can be absorbed into the definitions of the fields . however , only one of the three remaining phases has physical consequences in neutrino oscillations . therefore for either dirac or majorana neutrinos we can choose the following parametrization for @xmath41 @xcite ( c _ e + _ + _ ) = u ( c _ 1 + _ 2 + _ 3 ) = ( ccc c_1 c_3 & c_1 s_3 & s_1 e^-i + - c_2 s_3 - s_1 s_2 c_3 e^i & c_2 c_3 - s_1 s_2 s_3 e^i & c_1 s_2 + s_2 s_3 - s_1 c_2 c_3 e^i & - s_2 c_3 - s_1 c_2 s_3 e^i & c_1 c_2 + ) ( c _ 1 + _ 2 + _ 3 ) , [ u ] where @xmath43 , @xmath44 and @xmath45 is the @xmath27-violating phase . the experimental indications are that @xmath46 ev@xmath4 and that @xmath47 ev@xmath4 for a vacuum oscillation explanation of the solar neutrino data . then for the oscillations of neutrinos in atmospheric and long - baseline experiments with @xmath48 km / gev , the @xmath49 terms are negligible and the relevant oscillation probabilities are p ( _ _ ) & = & 1 - ( c_1 ^ 4 ^22_2 + s_2 ^ 2 ^22_1 ) ^2_atm , [ pmumu2 ] + p(_e_e ) & = & 1 - ^22_1 ^2_atm . [ pee2 ] + p(_e _ ) & = & s_2 ^ 2 ^22_1 ^2_atm . [ pemu2 ] + p(_e _ ) & = & c_2 ^ 2 ^2 2_1 ^2_atm , [ petau2 ] + p ( _ _ ) & = & c_1 ^ 4 ^2 2_2 ^2_atm . [ pmutau2 ] when @xmath50 ( i.e. , @xmath51 ) , these reduce to pure @xmath52 oscillations with amplitude @xmath53 , and @xmath6 does not oscillate in atmospheric and long - baseline experiments . we define the oscillation amplitudes @xmath54 , @xmath55 , @xmath56 , @xmath57 , and @xmath58 as the coefficients of the @xmath59 terms in eqs . ( [ pmumu2])([pmutau2 ] ) , respectively . the neutrino parameters can then be determined from the atmospheric neutrino data by the relations n_/n^o_= , [ rmu ] and n_e / n^o_e = , [ re ] where @xmath60 and @xmath61 are the expected numbers of atmospheric @xmath62 and @xmath63 events , respectively , @xmath64 , @xmath65 is @xmath66 appropriately averaged , and @xmath29 is an overall neutrino flux normalization , which we allow to vary following the superk analysis @xcite . superk presented @xmath67 and @xmath68 for eight different @xmath69 bins @xcite from a 535 day exposure . the data were obtained by inferring an @xmath69 value for each event from the zenith angle @xmath70 and energy of the observed charged lepton @xmath71 and comparing it to expectations from a monte carlo simulation based on the atmospheric neutrino spectrum @xcite folded with the differential cross section . due to the fact that the charged lepton energy and direction in general differ from the corresponding values for the incident neutrino ( or antineutrino ) , the @xmath69 distribution involves substantial smearing . we estimate this smearing by a monte carlo integration over the neutrino angle and energy spectrum @xcite weighted by the deep - inelastic differential cross section@xcite . we generate events with @xmath72 and @xmath73 , and determine the corresponding @xmath71 and @xmath70 for the charged lepton . we bin the events in @xmath74 , using @xmath70 to determine @xmath75 and an estimated neutrino energy inferred from the average ratio of lepton momentum to neutrino energy , @xmath76 , analogous to the superk analysis @xcite . we calculate a value for @xmath77 for each @xmath69 bin for a given value of @xmath7 . we can then fit eqs . ( [ rmu ] ) and ( [ re ] ) to the data and determine the neutrino parameters in eqs . ( [ pmumu2])([pemu2 ] ) and the normalization @xmath29 . without loss of generality we take @xmath7 to be positive . we do not consider matter effects in our analysis . for the @xmath9 favored by our fits , matter effects are small for the sub - gev neutrinos that constitute most of the data @xcite . also , as evidenced by our fits , the dominant oscillation is @xmath78 , which is not greatly affected by matter @xcite . however , matter effects could be important for the neutrino flux with smaller @xmath79 , i.e. , multi - gev data in solutions with @xmath80 , or in long - baseline experiments @xcite . for neutrinos from the sun @xmath81 km / gev , and the @xmath66 terms oscillate very rapidly , averaging to @xmath82 . then the oscillation probabilities are p(_e _ e ) & = & 1 - 12 ^ 22_1 - c_1 ^ 4 ^ 22_3 ^2_sun , [ peesun ] + p(_e _ ) & = & 12s_2 ^ 2 ^ 22_1 + 4 c_1 ^ 2 s_3 c_3 ^2_sun + & & - 2 s_1 c_1 ^ 2 s_2 c_2 s_3 c_3 2_sun , [ pemsun ] + p(_e _ ) & = & 12c_2 ^ 2 ^ 22_1 + 4 c_1 ^ 2 s_3 c_3 ^2_sun + & & + 2 s_1 c_1 ^ 2 s_2 c_2 s_3 c_3 2_sun , [ petsun ] where @xmath83 from @xmath40 invariance and @xmath84 may be found from @xmath85 by changing the sign of @xmath45 . since only the sum of oscillation channels @xmath86 is tested in solar experiments , the @xmath27-violating parameter @xmath87 can not be constrained from solar measurements . to see the effects of @xmath27 violation one must measure p(__e ) - p(_e _ ) & = & p(_e _ ) - p(__e ) + & = & p ( _ _ ) - p ( _ _ ) + & = & 2 s_1 c_1 ^ 2 s_2 c_2 s_3 c_3 2_sun , the corresponding differences for antineutrinos ( which have the opposite sign ) , or combinations that violate @xmath27 explicitly , such as @xmath88 . when @xmath50 ( i.e. , @xmath51 ) , the solar oscillation acts like a simple two - neutrino oscillation with amplitude @xmath89 ; although the oscillations of @xmath6 may involve both @xmath5 and @xmath90 , the individual channels @xmath91 and @xmath13 are not measurable in solar experiments , since the neutrino energies are below the thresholds for @xmath63 and @xmath92 production . the parameter @xmath93 measures the extent to which the solar oscillations have a constant component coming from the atmospheric oscillation scale @xcite . in our solar fits we use the shape of the standard solar model ( ssm ) neutrino fluxes and absorption cross sections in ref . @xcite and the new normalizations in ref . for the @xmath94cl and @xmath0ga cases , we fold the neutrino oscillation probability with the neutrino absorption cross section and expected neutrino flux to obtain the expected number of events , which can then be compared to the expectations of the standard solar model . we also allow an arbitrary normalization factor @xmath30 for the @xmath95b neutrino flux . the expected number of neutrino events is then n = p(_e _ e ) ( _ b + _ non - b ) de _ , [ n1 ] for the super - kamiokande case , for which the interaction is @xmath96 , the events are binned by outgoing electron energy , and we have included the effects of the detector resolution . the number of events per unit of electron energy is = \ { d_ccde_e^ p(_e _ e ) + d_ncde_e^ [ 1-p(_e _ e ) ] } g(e_e^ , e_e ) _ b de_de_e^[n2 ] where @xmath97 ( @xmath98 ) is the charge - current ( neutral - current ) differential cross section for an incident neutrino of energy @xmath72 and @xmath99 is the probability that an electron of energy @xmath100 is measured as having energy @xmath101 ; the electron energy resolution is taken from ref . the events are then put into 0.5 gev bins starting at the 6.5 gev threshhold for the detector , to compare with the super - kamiokande data . we take as the input solar data the homestake @xmath94cl rate @xcite , the gallex and sage @xcite @xmath0ga rates , and 16 bins of the super - kamiokande detected electron energy spectrum @xcite . we then fit eqs . ( [ n1 ] ) and ( [ n2 ] ) to the data and determine the neutrino parameters in eq . ( [ peesun ] ) and the @xmath95b neutrino flux normalization @xmath30 . in our solar fits we combine day and night results , since they should not be appreciably different for vacuum long - wavelength oscillations . we compare the model predictions with the time - averaged data , since with current statistics the super - kamiokande data does not reflect any seasonal variation @xcite . the possibility of detecting a seasonal variation in super - kamiokande and other experiments with improved statistics is discussed in sec . in the limit @xmath50 ( i.e. , @xmath51 ) , the atmospheric and solar oscillations decouple @xcite and effectively reduce to two separate two - neutrino solutions , each with its own oscillation amplitude and mass - squared difference . also , the only oscillation channel for atmospheric neutrinos is @xmath11 . more specifically , for atmospheric neutrinos p ( _ _ ) & = & 1 - ^22_2 ^2_atm , [ pmumu0 ] + p(_e_e ) & = & 1 , [ pee0 ] + p(_e _ ) & = & 0 , [ pemu0 ] and for solar neutrinos p(_e_e ) = 1 - ^22_3 ^2_sun . [ pee ] in the @xmath51 limit , fits to the atmospheric and solar data may be made independently . for the two - neutrino case with @xmath50 ( only @xmath78 oscillations ) our best fit parameters are @xmath102 where @xmath29 is the overall flux normalization in eqs . ( [ rmu ] ) and ( [ re ] ) , with @xmath103 for 13 degrees of freedom ( dof ) . this @xmath104 corresponds to a goodness - of - fit of 90% . the 95% c.l . allowed region for @xmath7 versus @xmath53 is shown in fig . [ 2nuatm ] . our result is very similar to the fit obtained by the superk collaboration . if we set @xmath105 ( i.e. , assume that the theoretical flux normalization is exact ) , the best fit has @xmath106 , which is acceptable only at the 6% c.l . the calculated flux has a normalization uncertainty of about @xmath107 @xcite . as reported by the superk collaboration , the other two - neutrino case with pure vacuum @xmath12 oscillations ( which corresponds to @xmath108 ) does not give a good fit to the data ( possible effects of matter are discussed at the end of sec . we find that the minimum @xmath104 for this case is 81.9/13 , corresponding to a goodness - of - fit of @xmath109 . therefore the @xmath12 scenario for atmospheric neutrinos is strongly disfavored by the superk data . the results of the two - neutrino fits to the atmospheric data are summarized in table [ atmfits ] . large amplitude @xmath110 oscillations are also excluded by the chooz reactor data @xcite for @xmath111 ev@xmath4 . for the effective two - neutrino oscillation formula with @xmath50 , our best fit to the combined solar data yields the parameters @xmath112 with @xmath113 , acceptable at the 16% c.l . the 95% c.l . allowed regions for @xmath7 versus @xmath89 are shown in fig . [ 2nusun ] . we note that our allowed regions are very similar to those obtained in ref . @xcite with a somewhat different analysis . taken at face value , the preferred value of the @xmath95b normalization would suggest that the ssm underestimates the @xmath95b neutrino flux by a sizable amount . however , this seems unlikely because most alternative solar models give a lower @xmath95b neutrino flux . the best fit values for @xmath114 ( the ssm @xmath95b spectrum normalization ) are @xmath115 with @xmath116 , acceptable at the 7% c.l . we have also searched for the best - fit to the solar data in each of the @xmath117 `` finger '' regions in fig . [ 2nusun ] . the results are given in table [ sunfits ] , where these four best fits have been labeled a , b , c , d in order of ascending @xmath117 . these fits correspond to vacuum oscillation wavelengths ( for a typical @xmath95b neutrino energy ) of approximately @xmath118 , @xmath119 , @xmath120 , and @xmath121 , where @xmath122 is the earth - sun distance . we see that the best fits in the higher @xmath8 regions all have a lower @xmath95b normalization . they also tend to fit the super - k data better , but do somewhat worse on the radiochemical experiments . interestingly , the two solutions with the largest @xmath117 have @xmath123 . because the apparent suppression in the @xmath94cl measurement differs from that of the @xmath0ga and superk cases , we consider the possibility that it had an unknown systematic error . the results of fitting to just the @xmath0ga and superk data are listed in the lower half of table [ sunfits ] . the best fit now occurs for a larger @xmath8 , and the goodness - of - fit improves . the allowed regions do not qualitatively change . however , for the larger @xmath117 regions , the @xmath0ga data is more easily accomodated by small changes of the parameters from their values in the global fit . in making our three - neutrino fits we assume that separate mass - squared difference scales are necessary to explain the atmospheric and solar neutrino data . if the @xmath14 scales were not distinct , or if one of the mass - squared differences were used to explain the lsnd data , then either the solar or atmospheric probabilities would be in a region of @xmath69 where the oscillations have averaged , and there would be no energy dependence . an energy - independent suppression due to oscillations would be equivalent to letting the overall normalization vary . oscillation scenarios where there is not a separate mass scale associated with solar neutrinos have been considered @xcite . it has already been demonstrated in the literature that an energy - independent suppression is strongly disfavored by the solar data @xcite . even if one ignores the @xmath94cl data , and assumes that the @xmath0ga and @xmath124 experimental rates are both consistent with an overall suppression by 50% , the spectrum distortion of @xmath95b neutrinos measured in the @xmath124 experiments disfavors an energy - independent suppression . we have updated this analysis , allowing for an overall flux suppression in addition to the variation of the @xmath95b neutrino normalization . the lowest value @xmath125 can achieve with three neutrinos when all oscillations are averaged is @xmath126 . we found the best fit to the solar data with overall depletion between @xmath126 and unity and arbitrary @xmath95b normalization has @xmath127 , which is ruled out at the 99.99% c.l . if the @xmath94cl data are ignored , the best fit has only @xmath128 , ruled out at the 93% c.l . therefore the distortion of the electron energy spectrum present in the solar neutrino data favors the existence of a separate mass scale for the oscillation of the solar neutrinos , which justifies the form of eqs . ( [ pmumu2])([pmutau2 ] ) and ( [ peesun])([petsun ] ) . if one @xmath14 is used to describe the lsnd data and the other is used to describe the solar data , then the oscillation due to @xmath129 will be averaged for the @xmath69 of the atmospheric neutrinos , so that the oscillation probabilities are independent of both energy and zenith angle . we find the best fit in this scenario to have @xmath130 , which is excluded at the 99.8% c.l . the atmospheric data favor maximal mixing of atmospheric @xmath5 with @xmath90 and no mixing with @xmath6 . the solar data also suggest , although not as strongly , that solar neutrinos may also mix maximally , or nearly maximally . if we require both atmospheric and solar oscillations to be maximal , there is a unique three - neutrino solution to the neutrino mixing matrix @xcite , which corresponds to @xmath50 and @xmath131 . the corresponding oscillation probabilities for atmospheric neutrinos are p ( _ _ ) & = & 1 - ^2_atm , [ pmumubi ] + p(_e_e ) & = & 1 , [ peebi ] + p(_e _ ) & = & 0 , [ pemubi ] and for solar neutrinos p(_e_e ) & = & 1 - ^2_sun . [ peebisun ] , + p(_e _ ) = p(_e _ ) & = & 12 ^2_sun [ pembisun ] . one interesting aspect of this solution is that the solar @xmath6 oscillations are 50% into @xmath5 and 50% into @xmath90 , although the flavor content of the @xmath6 oscillation is not observable in solar experiments . further properties of the bi - maximal and nearly bi - maximal solutions are discussed in ref . . a full three - neutrino fit to the atmospheric data with one @xmath9 scale has been made in refs . @xcite . in terms of the oscillation parameters defined in sec . ii , our best fit values for the four parameters are @xmath132 with @xmath133 , acceptable at the 85% c.l . in fig . [ 3nuatm ] we show the 95% c.l . allowed region for @xmath53 versus @xmath7 for various values of @xmath134 when the flux normalization @xmath29 is allowed to vary . although @xmath135 is favored , nonzero values are allowed , which permit some @xmath136 and @xmath13 oscillations of atmospheric neutrinos . in fig . [ 3nuatm2 ] we show the 95% c.l . allowed region for @xmath134 versus @xmath7 when @xmath29 and @xmath53 are allowed to vary . another limit on @xmath134 comes from the chooz reactor experiment @xcite that measures @xmath137 disappearance a^ee_atm = 4 p_e3 ( 1 - p_e3 ) 0.2 , [ choozlimit ] which applies for @xmath138 ev@xmath4 . the exact limit on @xmath139 varies with @xmath7 , and for @xmath140 ev@xmath4 there is no limit at all . for @xmath141 ev@xmath4 and @xmath142 , @xmath134 is constrained to be less than 0.19 . the result of imposing the chooz constraint is shown in figs . [ 3nuatm ] and [ 3nuatm2 ] . in fig . [ 3nuatm2 ] we see that the range of @xmath134 allowed by the fit to the atmospheric neutrino data and the chooz constraint is 0 _ 1 0.29 , [ sinth1atm ] at 95% c.l . a full three - neutrino fit to solar data can be made by varying @xmath93 , @xmath143 , @xmath117 and the @xmath95b flux normalization @xmath30 , and using the expression in eq . ( [ peesun ] ) for the oscillation probabilities in eqs . ( [ n1 ] ) and ( [ n2 ] ) . we find the best fit values for the solar parameters @xmath144 with @xmath145 , acceptable at the 12% c.l . if the @xmath95b normalization is fixed at the ssm value ( @xmath114 ) , the best fit is @xmath146 with @xmath147 , acceptable at the 5% c.l . the addition of the extra parameter @xmath134 does not improve the fit to the solar data . in fig . [ 3nusun ] we show the 95% c.l . allowed region for @xmath89 versus @xmath117 for various values of @xmath134 when @xmath30 is allowed to vary . as in the atmospheric neutrino case , @xmath135 is favored , although non - negligible nonzero values are allowed . in fig . [ 3nusun2 ] we show the 95% c.l . allowed region for @xmath134 versus @xmath117 when @xmath30 and @xmath89 are allowed to vary . the range of @xmath134 allowed by the solar data is 0 _ 1 0.49 , [ sinth1sun ] at 95% c.l . since the atmospheric and solar fits have only one common parameter , @xmath134 , and the best separate fits to the two data sets both have @xmath135 , then the best fit to the combined solar and atmospheric data sets can be identified immediately as given by eqs . ( [ dm2bestatm2])-([alphabest2 ] ) and ( [ dm2bestsun2])-([betabest2 ] ) , with @xmath148 , acceptable at the 43% c.l . to see how the fits vary with the common parameter , in fig . [ chi2 ] we show @xmath149 for the solar and atmospheric data sets versus @xmath134 , as well as the combined @xmath150 . from the figure we see that the dependence of @xmath149 on @xmath134 is relatively weak for @xmath151 for the solar data set . therefore , the parameter regions allowed by the combined solar and atmospheric data sets are essentially the most stringent of those allowed by the solar and atmospheric data sets separately . the minos @xcite , k2k @xcite , icarus @xcite and noe @xcite experiments can test for @xmath152 oscillations for @xmath153 ev@xmath4 , and minos and icarus can also search for @xmath11 . together these measurements could more precisely determine @xmath53 , @xmath134 , and @xmath7 . further measurements of atmospheric neutrinos will also help constrain these parameters . full three - neutrino fits including the solar neutrino data @xcite can then determine one of the remaining two independent parameters in the mixing matrix , e.g. , @xmath89 , and @xmath117 . there is new physics predicted when @xmath154 , i.e. , @xmath155 oscillations with leading probability given by eq . ( [ petau2 ] ) . the allowed range at 95% c.l . for the @xmath156 oscillation amplitude @xmath57 versus @xmath7 is depicted in fig . [ etau ] ; the effect of the chooz constraint is also shown . the maximal @xmath13 amplitude of about 0.15 occurs at @xmath157 ev@xmath4 . these @xmath155 oscillations could be observed by long - baseline neutrino experiments with proposed high intensity muon sources @xcite , which can also make precise measurements of @xmath12 and @xmath78 oscillations . sensitivity to @xmath158 is expected @xcite for the parameter ranges of interest here ; matter effects would not be large for an experiment from fermilab to soudan @xcite . for @xmath159 ev@xmath4 , @xmath57 could be measured down to @xmath160 ; the expected sensitivity versus @xmath7 is shown in fig . precise measurement of @xmath155 and @xmath12 oscillations in such a long - baseline experiment would uniquely specify the @xmath27-conserving part of the three - neutrino model . in this section we discuss how future solar neutrino measurements can distinguish between the different solar vacuum oscillation solutions . we consider only the two - neutrino solutions since the discussion easily generalizes to the three - neutrino case . one interesting feature of vacuum long - wavelength solutions is that they can give a rise in the fraction of surviving @xmath6 s for higher electron energies , in agreement with the superk measurement @xcite . figure [ spectrum ] shows the ratio of the electron energy spectrum to the ssm prediction for two different vacuum long - wavelength scenarios and the superk data . future measurements at superk will improve the statistics in the high @xmath101 bins . however , with an @xmath161 flux that is @xmath162 times greater than the ssm result , this rise at high @xmath101 could be due to contributions of @xmath161 neutrinos @xcite . superk is also planning to lower their threshhold for electron detection to 5 mev ; this is particularly important since the number of events rises for energies just below the current threshhold of 6.5 mev . in fig . [ lowe ] we show predictions for the electron energy spectrum in superk in the range 5 mev @xmath163 10 mev for the four oscillation solutions a , b , c , and d in table [ sunfits ] . it is evident that solutions a and b may be distinguishable from c and d , and from each other , using data at lower @xmath101 . the energy spectrum measured in the sno experiment @xcite will also provide additional information on the energy dependence of the spectrum suppression . another feature of vacuum solar solutions is that they may cause a detectable seasonal variation as the distance between the earth and sun varies @xcite . we define two seasonal asymmetry parameters , a_1 & = & 2(n_w - n_s ) ( n_w + n_s + n_f + n_sp ) , [ asy1 ] + a_2 & = & n_w + n_s - n_f - n_sp n_w + n_s + n_f + n_sp , [ asy2 ] where @xmath164 , @xmath165 , @xmath166 , and @xmath167 are the number of events collected in the time periods from november 20 to february 19 ( winter , earth closest to the sun ) , february 20 to may 21 ( spring ) , may 22 to august 20 ( summer , earth farthest from the sun ) , and august 21 to november 19 ( fall ) , respectively . since @xmath168 ( the same range of distances is covered ) , @xmath169 and @xmath170 are the only independent quantities that may be constructed from the four seasonal measurements . the quantity @xmath169 is similar to the seasonal asymmetry defined in ref . @xcite , and @xmath169 and @xmath170 are similar to the first two harmonics in the analysis of the first paper in ref . @xcite . the parameter @xmath169 has a significant nonzero value when the oscillation probability increases or decreases monotonically as the earth moves from perihelion to aphelion , while @xmath170 has a significant nonzero value when the oscillation probability reaches a local extremum somewhere between perihelion and aphelion . due to the @xmath171 dependence of the solar neutrino flux and the 3.3% change in @xmath172 from perihelion to aphelion , @xmath169 and @xmath170 have the values 0.030 and 0 , respectively , in the absence of oscillations . the asymmetry @xmath169 is strongly correlated with the electron energy spectrum distortion ; such a correlation is a distinctive characteristic of vacuum oscillations @xcite . for @xmath95b neutrinos , due to their relatively higher energies and longer oscillation wavelengths , the seasonal variation is less pronounced than for solar neutrinos of lower energy . the total superk event rate as a fraction of the ssm value versus time of year @xcite is shown in fig . [ skseas ] ; also shown is the prediction for solution a in table [ sunfits ] and the prediction for no oscillations . the results for other oscillation solutions are similar to the prediction for solution a. oscillation solutions produce an enhanced seasonal effect @xcite . the superk experiment has not observed a significant seasonal variation with the current data sample , but could be sensitive to the effects predicted by vacuum oscillations with increased statistics . in order to extract the most information from the superk data it is advantageous to plot the seasonal asymmetries versus the observed electron energy . in fig . [ superkseas ] we show @xmath169 versus @xmath101 for solutions a , b , c , and d of table [ sunfits ] . although the asymmetries are not large , each solution has a characteristic shape , especially solution a. the energy dependence of @xmath169 clearly distinguishes the oscillation scenarios from the asymmetry induced only by the seasonal flux variation . similar measurements can be done in the sno experiment @xcite . the deviation of @xmath170 from the value for no oscillations is at most about 0.003 , and does not provide a good discrimination between models . the gallex and sage @xmath0ga experiments may also exhibit a seasonal variation with increased statistics if there are vacuum oscillations of solar neutrinos @xcite . in fig . [ a1a2ga ] we plot the predictions for @xmath170 versus @xmath169 in the gallex and sage experiments for a range of @xmath117 from each of the four `` finger '' regions in fig . [ 2nusun ] . most of the seasonal asymmetry in the @xmath0ga experiments is due to the monoenergetic @xmath173be neutrinos which constitute about 25% of the signal in the ssm . the borexino experiment @xcite will primarily measure the @xmath173be neutrinos ( @xmath174 mev ) using the process @xmath175 . the final - state electron kinetic energy @xmath176 has a maximum value of 0.665 mev for @xmath173be neutrinos . the @xmath177 neutrinos have a maximum @xmath176 of 0.26 mev , and also there is a considerable background for @xmath178 mev . therefore selecting events in the range @xmath179 mev will preferentially select the @xmath173be neutrino signal . neutrinos from the @xmath180 reaction and the cno cycle will also give final - state electrons in this energy range , but the @xmath173be neutrinos represent more than 80% of the signal , assuming the ssm . predictions for event rate and seasonal variation in the borexino experiment for various vacuum oscillation parameters have previously been discussed in the literature @xcite . here we examine how the seasonal asymmetries in eqs . ( [ asy1 ] ) and ( [ asy2 ] ) may be used to further discriminate the different vacuum neutrino solutions in table [ sunfits ] . in fig . [ a1a2b ] we plot the predictions for @xmath170 versus @xmath169 in the borexino experiment for a range of @xmath8 from each of the four `` finger '' regions in fig . [ 2nusun ] . the seasonal asymmetries are potentially larger than in the @xmath0ga case since the @xmath173be neutrinos are a much larger fraction of the signal . although there are some regions where the predictions of two or more solutions overlap , combining the asymmetry information with the event rate should significantly reduce the allowed parameter regions for the vacuum solutions , and could select the appropriate `` finger '' region . vb thanks fermilab for support as a frontier fellow and for kind hospitality . we thank john learned for stimulating discussions regarding the super - kamiokande atmospheric data , and we thank sandip pakvasa and tom weiler for collaboration on previous related work . we are grateful to todor stanev for providing his atmospheric neutrino flux program . this work was supported in part by the u.s . department of energy , division of high energy physics , under grants no . de - fg02 - 94er40817 and no . de - fg02 - 95er40896 , and in part by the university of wisconsin research committee with funds granted by the wisconsin alumni research foundation . j.g . learned , s. pakvasa , and t.j . weiler , phys . b207 * , 79 ( 1988 ) ; v. barger and k. whisnant , phys . lett . * b209 * , 365 ( 1988 ) ; k. hidaka , m. honda , and s. midorikawa , phys . lett . * 61 * , 1537 ( 1988 ) . kamiokande collaboration , k.s . _ , phys . lett . * b280 * , 146 ( 1992 ) ; y. fukuda _ et al . _ , phys . lett . * b335 * , 237 ( 1994 ) ; imb collaboration , r. becker - szendy _ et al . . suppl . * 38b * , 331 ( 1995 ) ; soudan-2 collaboration , w.w.m . et al . _ , b391 * , 491 ( 1997 ) . liquid scintillator neutrino detector ( lsnd ) collaboration , c. athanassopoulos _ et al . lett . * 75 * , 2650 ( 1995 ) ; _ ibid . _ * 77 * , 3082 ( 1996 ) ; _ ibid . _ * 81 * , 1774 ( 1998 ) ; talk by h. white at _ neutrino-98 _ , takayama , japan , june 1998 . karmen collaboration , k. eitel and b. zeitnitz , at _ neutrino-98 _ , takayama , japan , june 1998 , hep - ex/9809007 ; future tests of the lsnd results will be made by the karmen experiment and also by the mini - boone experiment , e. church _ et al . _ , nucl - ex/9706011 ; see also j.m . conrad , talk at ichep-98 , hep - ex/9811009 j.w . flanagan , j.g . learned , and s. pakvasa , phys . rev . * d57 * , 2649 ( 1998 ) ; m.c . gonzalez - garcia , h. nunokawa , o. peres , t. stanev , and j.w.f . valle , phys . rev . * d58 * , 033004 ( 1998 ) ; m.c . gonzalez - garcia , h. nunokawa , o. peres , and j.w.f . valle , hep - ph/9807305 ; c.h . albright , k.s . babu , and s.m . barr , phys . lett . * 81 * , 1167 ( 1998 ) . ; j. bordes , h .- m . chan , j. pfaudler , and s.t . tsou , phys . rev . * d58 * , 053003 ( 1998 ) ; j.g . learned , s. pakvasa , and j.l . stone , phys . * b435 * , 131 ( 1998 ) ; l.j . hall and h. murayama , phys b436 * , 323 ( 1998 ) ; s.f . king , hep - ph/9806440 . m. drees , s. pakvasa , x. tata , and t. ter veldhuis , phys . rev . * d57 * , 5335 ( 1998 ) ; e.j . chun , s.k . kang , c.w . kim , and u.w . lee , hep - ph/9807327 ; v. bednyakov , a. faessler , and s. kovalenko , hep - ph/9808224 ; b. mukhopadhaya , s. roy , and f. vissani , hep - ph/9808265 ; o.c.w . kong , hep - ph/98084304 ; e.j . chun and j.s . lee , hep - ph/9811201 ; l. clavelli and p.h . frampton , hep - ph/9811326 ; z. berezhiani and a. rossi , hep - ph/9811447 . v. barger , s. pakvasa , t.j . weiler , and k. whisnant , phys . b437 * 107 ( 1998 ) ; a.j . baltz , a.s . goldhaber , and m. goldhaber , hep - ph/9806540 ; m. jezabek and y. sumino , hep - ph/9807310 ; y. nomura and t. yanagida , hep - ph/9807325 ; g. altarelli and f. feruglio , phys . lett . * b439 * , 112 ( 1998 ) ; hep - ph/9809596 ; h. georgi and s. glashow , hep - ph/9808293 ; s. davidson and s.f . king , hep - ph/9808296 . r.n . mohapatra and s. nussinov , hep - ph/9808301 ; hep - ph/9809415 ; s.k . kang and c.s . kim , hep - ph/9811379 . h. fritzsch and z. xing , phys . b372 * , 265 ( 1996 ) ; hep - ph/9807234 ; hep - ph/9808272 ; e. torrente - lujan , phys . lett . * b389 * , 557 ( 1996 ) ; m. fukugita , m. tanimoto , and t. yanagida , phys . rev . * d57 * , 4429 ( 1998 ) ; m. tanimoto , hep - ph/9807283 . e. ma , hep - ph/9807386 ; j. ellis , g.k . leontaris , s. lola , and d.v . nanopoulos , hep - ph/9808251 ; s. lola and j.d . vergados , hep - ph/9808269 ; k. kang , s.k . kang , c.s . kim , and s.m . kim , hep - ph/9808419 ; e. ma , d.p . roy , and u. sarkar , hep - ph/9810309 ; e. malkawi , hep - ph/9810542 ; e. ma and d.p . roy , hep - ph/9811266 ; j. elwood , n. irges , and p. ramond , hep - ph/9807228 . a. acker , s. pakvasa , and j. pantaleone , phys . rev . * d43 * , 1754 ( 1991 ) ; v. barger , r.j.n . phillips , and k. whisnant , phys . 69 * , 3135 ( 1992 ) ; p.i . krastev and s.t . petcov , phys . lett . * b285 * , 85 ( 1992 ) ; phys . . lett . * 72 * , 1960 ( 1994 ) ; phys . rev . * d53 * , 1665 ( 1996 ) . g. barr , t.k . gaisser , and t. stanev , phys . rev . * d39 * , 3532 ( 1989 ) ; m. honda , t. kajita , k. kasahara , and s. midorikawa , phys . rev . * d52 * , 4985 ( 1995 ) ; v. agrawal , t.k . gaisser , p. lipari , and t. stanev , phys . rev . * d53 * , 1314 ( 1996 ) ; t.k . gaisser and t. stanev , phys . rev . * d57 * , 1977 ( 1998 ) . bernstein and s.j . parke , phys . d44 , 2069 ( 1991 ) ; e.k . akhmedov , p. lipari , and m. lusignoli , phys . lett . * b300 * , 128 ( 1993 ) ; q.y . liu and a. yu . smirnov , nucl . b524 * , 505 ( 1998 ) ; q.y . liu , s.p . mikheyev , and a. yu . smirnov , hep - ph/9803415 ; s.t . petcov , phys . lett . * b434 * , 321 ( 1998 ) ; e.k . akhmedov , hep - ph/9805272 ; e.k . akhmedov , a. dighe , p. lipari , and a.y . smirnov , hep - ph/9808270 ; m. chizhov , m. maris , and s.t . petcov , hep - ph/9810501 . the effects of a nonzero @xmath181 on solar oscillation fits are considered by z. berezhiani and a. rossi , phys . b367 * , 219 ( 1996 ) , p. osland and g. vigdel , phys . lett . * b438 * , 129 ( 1998 ) , and the first paper in ref . @xcite . a. acker and s. pakvasa , phys . b397 * , 209 ( 1997 ) ; p.f . harrison , d.h . perkins and w.g . scott , phys . b396 * , 186 ( 1997 ) ; phys . lett . * b349 * , 137 ( 1995 ) ; g. conforto , a. marchionni , f. martelli , f. vetrano , m. lanfranchi and g. torricelli - ciamponi , astrop . physics * 5 * , 147 ( 1996 ) ; g. conforto , a. marchionni , f. martelli and f. vetrano , phys . b427 * , 314 ( 1998 ) ; g. conforto , m. barone and c. grimani , hep - ph/9809501 ; r.p . thun and s. mckee , phys . lett . * b439 * , 123 ( 1998 ) . three - parameter fits to older atmospheric data without explicit @xmath69 dependence were made by s.m . bilenky , c. giunti , and c.w . kim , astropart . phys . * 4 * , 241 ( 1996 ) , using the ratio of muon to electron rates , and the first paper of ref . @xcite , using up / down ratios ; see also c. giunti , c.w . kim , and m. monteno , nucl . * b521 * , 3 ( 1998 ) ; o. yasuda , phys . rev . * d58 * , 091301 ( 1998 ) ; hep - ph/9809205 ; g.l . fogli , e. lisi , a. marrone , g. scioscia , hep - ph/9808205 . sno collaboration , e. norman _ et al . _ , in proc . of _ the fermilab conference : dpf 92 _ , november 1992 , batavia , il , ed . by c. h. albright , p.h . kasper , r. raja , and j. yoh ( world scientific , singapore , 1993 ) , p. 1450 . for recent discussions , see b. faid , g.l . fogli , e. lisi , and d. montanino , hep - ph/9805293 ; v. berezinsky , g. fiorentini , and m. lissia , hep - ph/9811352 ; s.l . glashow , p.j . kernan , and l.m . krauss , hep - ph/9808470 . borexino collaboration , c. arpesella _ et al . _ , `` infn borexino proposal , '' vols . i and ii , edited by g. bellini , r. raghavan _ et al . _ ( univ . of milan , 1992 ) ; j. benziger , f.p . et al . _ , `` proposal for participation in the borexino solar neutrino experiment , '' ( princeton university , 1996 ) .	we determine the three - neutrino mixing and mass parameters that are allowed by the solar and atmospheric neutrino data when vacuum oscillations are responsible for both phenomena . the global fit does not appreciably change the allowed regions for the parameters obtained from effective two - neutrino fits . we discuss how measurements of the solar electron energy spectrum below 6.5 gev in super - kamiokande and seasonal variations in the super - kamiokande , @xmath0ga , and borexino experiments can distinguish the different solar vacuum solutions . = cmssbx10 scaled 2 to * global three - neutrino vacuum oscillation fits + to the solar and atmospheric anomalies * + 0.7 cm v. barger@xmath1 and k. whisnant@xmath2 + @xmath3_fermi national accelerator laboratory , batavia , il 60510 usa _ + @xmath4_department of physics , university of wisconsin , madison , wi 53706 usa _ + @xmath2_department of physics and astronomy , iowa state university , ames , ia 50011 usa _ +
excess infrared flux associated with main sequence stars usually indicates the presence of thermally emitting dust in a circumstellar disk . when the stellar age is long compared to the timescales for removal of pristine ( i.e. , primordial ) material by poynting - robertson ( p - r ) drag or radiation pressure , the observed dust must be produced by collisional cascades of large particles and planetesimals or by cometary deposits @xcite . such disks are referred to as debris disks . the a - type main - sequence star @xmath0 lep ( hd 38678 ) was identified as a debris disk candidate by the iras discovery of its infrared excess @xcite . it is one of several stars where the disk particles are significantly warmer than is typical for debris disks @xcite , implying that the particles are relatively close ( @xmath310 au ) to the star . for example , the @xmath4320 k mid - infrared color temperature of @xmath0 lep ( chen & jura 2001 ; this work ) contrasts with archetypes such as the disk of @xmath5 pic , which has a mid - infrared color temperature of @xmath4160 k for dust 80 au from the star @xcite . furthermore , @xmath0 lep is one of the few such stars with warm disks that are close enough to us ( 21.5 pc , perryman et al . 1997 ) that ground - based observers can reasonably expect to spatially resolve the disk . two recent observations of the disk drew our attention to the @xmath0 lep disk and motivated our program of mid - infrared imaging at gemini : a strong constraint on the size of the disk at 18 @xmath1 m at keck ( chen & jura 2001 , hereafter cj01 ) and a _ spitzer _ mips photometric survey of a - type stars showing that the infrared excess of @xmath0 lep is high among stars of comparable age @xcite . cj01 have noted that the disk radius must be smaller than 9 au , and therefore probably comparable in size to our own asteroid belt , which makes it unique among debris disks discovered so far . as we report here , we have imaged @xmath0 lep with gemini to constrain more tightly the disk size and further examine dust production mechanisms . we obtained mid - infrared images of @xmath0 lep ( hd 38678 ) on 2005 february 3 ( ut ) at gemini south ( program i d gs-2005a - q-2 ) with t - recs ( thermal region camera and spectrograph ) , using the broadband n ( @xmath6 = 10.36 @xmath1 m , @xmath7 = 5.27 @xmath1 m ) and narrowband qa ( @xmath6 = 18.30 @xmath1 m , @xmath7 = 1.51 @xmath1 m ) filters . the total field of view is 29 `` x 22 '' . the data were reduced with the gemini iraf package . the flux standards @xmath8 lep and @xmath9 cmi were observed before and after , respectively , the group of target and psf observations . flux measurements of @xmath0 lep at n ( 2.03 @xmath10 0.19 jy ) and qa ( 0.96 @xmath10 0.06 jy ) are consistent with previous results @xcite . we observed a point - spread - function ( psf ) comparison star , hd 42042 , both before and after the target observations in each filter . hd 42042 , an m star located 6.6@xmath11 away from @xmath0 lep , is @xmath410 times brighter than @xmath0 lep at 10.4 @xmath1 m and @xmath46 times brighter at 18.3 @xmath1 m . we divided the psf and target integrations into pairs of nodsets to examine the potential effect of airmass on the full - width at half maximum ( fwhm ) value , and there is no direct correlation . in the final stacked images , we find that the azimuthally averaged profiles of @xmath0 lep and the psf star are identical at 10.4 @xmath1 m , whereas the source is clearly extended at 18.3 @xmath1 m ( fig . 1 ) . we have partly characterized this extension by comparing the full width half - maximum ( fwhm ) of the profiles determined from a moffat fit generated in iraf . at 10.4 @xmath1 m , the fits to the psf star and @xmath0 lep profiles measure 0.31 `` @xmath10 0.02 '' . at 18.3 @xmath1 m the profile fits give 0.54 `` @xmath10 0.02 '' and 0.60 `` @xmath10 0.02 '' , respectively . even though we consider the fwhm radius of the profiles , we have determined those values from analytic fits to the full profiles . for @xmath0 lep , the indicated uncertainty is the standard deviation of the mean of the series of fwhm measurements made throughout the integration set , as described in greater detail below . for the psf star , we have only six fwhm measurements , and the uncertainty given is half of the excursion between the maximum and minimum values . like cj01 , we do not resolve @xmath0 lep at 10.4 @xmath1 m , but we do resolve the source at 18.3 @xmath1 m ( fig . 1 ) . to test the robustness of the latter result , we examined the fwhm of the psf star and @xmath0 lep throughout the integration sequence . pupil rotation , incorrect guiding correction , and changes in the quality of seeing during long observations on a program object can result in a final image degraded by lower - frequency components that are not accurately represented in the psf determined from generally shorter integration times . thus , we have divided the total 18.3 @xmath1 m integration sequence into bins of 120 s of on - source integration time and have measured the width of the moffat profile fit for each of these subdivided images . the displacement between the mean profile width of the psf and the mean profile width of the source implies that the source is resolved . quadratic subtraction of the derived fwhm value of the psf star from that of @xmath0 lep , as determined from the 18.3 @xmath1 m images , indicates that the average disk radius is 0.14 `` @xmath10 0.02 '' , or 3.0 @xmath10 0.3 au . cj01 estimate a best - fit value of 2.15 au , but they note that the uncertainties in their observations allow for values as large as 9 au ( cj01 ; c. chen , priv . comm . ) . our observations made under nearly diffraction - limited conditions imply that the size of the emitting region is indeed comparable to that of the asteroid belt in our solar system . to constrain the geometry of the @xmath0 lep disk , we generated 2-d models of the expected 18.3 @xmath1 m brightness distribution with a range of disk parameters . subtraction of the psf image normalized to the peak brightness from the image of @xmath0 lep shows a residual ring of emission that , given the asymmetries evident in the psf itself , is consistent with the disk being approximately azimuthally symmetric . we therefore assume a face - on disk geometry for our models and calculations , but this assumption has no effect on our conclusions . ( our observations provide no constraint on the disk orientation . ) our models consist of a central source ( the star ) with either one or two annuli of dust emission each with uniform surface brightness , where the constant total flux determined by observations at 18.3 @xmath1 m was equally divided between the central source and dust emission according to the estimated excess above the photosphere ( cj01 ) . for the single - annulus models , the free parameters of annulus width ( 2 , 4 , or 6 au ) and annulus inner edge distance from the star ( 2 , 3 , 4 , 6 , or 8 au ) are combined to yield fifteen models . we also generated models with two concentric and contiguous annuli with different but uniform brightnesses . in all of these models , the inner annulus extends from 2 to 4 au , whereas the outer annulus extends from 4 au to either 6 au or 8 au . the ratios of the total fluxes from the inner and outer annuli are either 1:1 or 3:1 , respectively , resulting in four double - annuli models . the double - annulus is a simple geometric construct that likely approximates a more complex radially - dependent brightness . the models were then convolved with a 2-d representation of the psf that was generated from the azimuthally averaged psf star observation at 18.3 @xmath1 m . a @xmath12-squared minimization comparison of the resulting model profiles to the source data shows that the source is best approximated by a model of two annuli . in this model , a 2 - 4 au annulus emits 75% of the thermal dust flux , and a 4 - 8 au annulus emits the remaining 25% . our models are not exhaustive but do illustrate the nature of satisfactory fits to the observed profiles . we also note that the dust emitting the unresolved 10.4 @xmath1 m excess must lie interior to the resolved 18.3 @xmath1 m flux and may be migrating inwards by p - r drag from the region of replenishment by collisions through a much lower - density region close to the star . the expected temperature for blackbody dust particles at @xmath13au from the star is 320@xmath14 k. this is comparable to the color temperature of 323@xmath15 k inferred from the two observed mid - infrared flux densities and associated uncertainties . thus , the particles may be blackbody emitters , in marked contrast to mid - infrared emitting grains in other debris disks ; e.g. , dust 50 au from @xmath5 pic is at a temperature of @xmath16140 k compared to the blackbody temperature of @xmath465 k @xcite . blackbody behavior requires particle radii of a few microns or larger . additional evidence that the particles orbiting @xmath0 lep are large is provided by recent mid - infrared spectroscopy of the source taken with the _ spitzer _ infrared spectrograph ( irs ) @xcite . silicates are commonly found in circumstellar disks , but the spectrum of @xmath0 lep shows no evidence of a distinct silicate emission feature . silicate particles as large as 1 - 2 @xmath1 m can result in a broad silicate feature , which is expected to disappear as particle radii increase above a few microns and the particles become optically thick in the mid - infrared ( e.g. , przygodda et al . @xcite conclude that , if silicates are present , the majority of particle radii are larger than 10 @xmath1 m . we estimate the size of the smallest particles expected to remain on bound orbits ( @xmath5 = @xmath17 @xmath3 0.5 , burns et al . 1979 ) in response to the forces of gravity and radiation pressure , assuming 14 l@xmath18 and 1.9 m@xmath18 for the stellar luminosity and mass , respectively , and unity for the radiation pressure efficiency . adopting the standard silicate density value of 2.5 g @xmath19 , we find that the minimum radius is 3.4 @xmath1 m . particles smaller than this are blown out of the system by radiation pressure within a few orbital timescales . the keplerian orbital period at 3 au from @xmath0 lep is @xmath43 years , so grains smaller than the blowout radius of 3.4 @xmath1 m should be removed on a decade timescale . with this minimum particle size threshold , we estimate the number - density - weighted average dust particle radius of 4.7 @xmath1 m by assuming that the incremental number of grains per radius interval @xmath20 is proportional to @xmath21 , which describes the equilibrium size distribution of the dust particle population in a collisional cascade @xcite . assuming that all particles have the average radius 4.7 @xmath1 m and are in a thin shell 3.0 au from the star , we estimate the dust mass from the amount of ultraviolet and optical flux reprocessed as thermal emission ( jura et al . 1995 , eq . 5 ) . with values for the fractional infrared luminosity @xmath22/l@xmath23 = 1.7 @xmath24 @xcite and particle density 2.5 g @xmath19 , we find a total dust mass of 6.7 @xmath25 g , or 0.37% of the total mass of the asteroids in our solar system . this value is @xmath440% of the mass estimate in cj01 due to the smaller disk size ( and therefore hotter dust ) determined with our observations . a change in the particle - size power - law exponent from 3.5 to 6.4 lowers the mass estimate by @xmath420% . large dust particles ( @xmath26 m ) are subject to loss by spiraling into the star under p - r drag @xcite . however , this process is only relevant if the time interval between destructive particle collisions is much longer than the p - r drag timescale @xcite . the characteristic collisional lifetime of the average radius particle at a distance @xmath27 from the star is @xmath28 , where @xmath29 is the keplerian orbital period and @xmath30 is the effective optical depth . we estimate the effective optical depth , defined as the ratio of the total cross - sectional area of dust to the total area of the disk @xcite , from the total area of average - sized ( 4.7 @xmath1 m ) particles necessary to produce the infrared luminosity and the area within the 28 au annulus ; for @xmath0 lep , this value is 2 @xmath24 , approximately equal to the observed value for @xmath22/l@xmath23 . assuming a constant particle density , the ratio of the p - r drag timescale to the collisional timescale is a weak function of distance from the star , so we expect the relative significance of the dominant process to be similar throughout the disk . for example , the expected time for a particle to migrate by p - r drag from 6 au to 3 au ( the center of each annulus ) is @xmath416000 y , several times longer than the collisional timescale at 4 au ( the nominal division between the annuli ) of only @xmath42000 years . therefore , we consider it unlikely that the particles orbiting @xmath0 lep migrate a substantial distance toward the star before being broken up in a collisional cascade . we conclude that initially radiation pressure and , subsequently , collisions are the two dominant processes defining the population and distribution of particles in @xmath0 lep s disk . a plot of infrared excess versus age @xcite of 266 a - type stars shows that @xmath0 lep has a high infrared excess compared to other main - sequence stars of similar age . in this paper we have adopted the @xcite value of 231 myr , estimated with photometry - determined stellar parameters to fit to stellar evolutionary tracks ; this estimate also incorporates the effects of rapid rotation , and a survey of a - type stars shows that @xmath0 lep s position in the h - r diagram is close to that of other stars with comparable age estimates @xcite . however , if the age is 100 myr or less ( cf . barrado y navascus 1998 ; song et al . ) , the excess level for the disk is within the range of values typical for those age bins . therefore , the dust production scenario that we consider most probable depends , to some extent , on the age we assume for @xmath0 lep : if the disk is older than 100 myr ( cf . lachaume et al . 1999 ; song et al . 2001 ) , the excess is indeed anomalously high and may indicate a recent cataclysmic collision , but , if the disk is much younger , we may be observing the dust production and decay associated with a relatively steady collision rate of smaller bodies . we examine both of these scenarios below . recent cataclysmic collisions may account for some of the excess emitting dust around @xmath5 pic @xcite and vega @xcite , and may also contribute to the range of excesses observed for stars of similar ages @xcite . the destruction of one planetesimal of radius @xmath485 km could account for the entire observed mid - infrared - emitting dust mass in @xmath0 lep . to assess the plausibility of a single cataclysmic collision of such a body resulting in a disk resembling our best - fit model , we consider the dynamics of the dust particles immediately following their production in such an event . the initial orbital evolution of the newly formed dust particles is dominated by radiation pressure , which acts to increase the semi - major axes of their orbits . if we assume that these particles are produced by the destruction of a body on a circular orbit at 2 au ( the inner edge of the inner annulus in the best - fit model to our observations ) , we can estimate the radius for particles expected to migrate to a given distance , using standard relations ( e.g. , kortenkamp & dermott 1998 ; wyatt et al . since smaller particles will have larger apastron distances under the influence of radiation pressure , the size of a particle that can migrate to a given apastron distance corresponds to a lower limit on the size of particles interior to that location . the radius of particles expected to reach 4 au ( 5.1 @xmath1 m ) will be both the upper limit for particles in the 48 au annulus and the lower limit for the 24 au annulus , where in the latter case all smaller grains have been blown farther out . the radius of particles expected to reach 8 au ( 3.9 @xmath1 m ) sets the lower limit for particle radii in the outer 48 au annulus . to characterize the particle distribution that would yield the flux ratio ( 3:1 ) between the annuli of the model fit to our observations ( @xmath313.2 ) , we sum over the areas of all the individual emitting particles , as described by a power - law particle distribution , assuming the radii are @xmath325.1 @xmath1 m in the inner annulus and 3.95.1 @xmath1 m in the outer annulus . ( since the radii of the particles that are expected to remain on bound orbits within 8 au are larger than a few microns , we assume that the particles are blackbody emitters . ) to estimate their temperature and flux , we then assume that all of the particles in the inner and outer annuli are at a characteristic distance of 3 and 6 au from the star , respectively . we find that , to match the model of our observations , the size distribution of the dust population resulting from a collision at 2 au must be described by a power law with an exponent of 6.4 , much steeper than the @xcite value of 3.5 . indeed , there is no reason to expect the immediate post - collision population to resemble the dohnanyi distribution that is thought to apply to a population that has evolved to collisional equilibrium @xcite . the post - collision distribution presumably depends heavily on the details of the collision and the composition of the parent body . for example , with regard to the latter , rubble piles " and differentiated asteroids may produce different relative amounts of small and large particles @xcite . a particle size distribution consistent with our double - annulus model therefore suggests that the proportion of small particles produced in the hypothetical collision is larger than that implied by the dohnanyi distribution . in contrast to the catastrophic collision scenario , dust production by relatively steady - state " grinding collisions may be more likely if the age of @xmath0 lep is less than 100 myr and the excess level is not high compared to other stars of comparable age . indeed , cj01 propose that dust is replenished in @xmath0 lep in an asteroid belt undergoing such a process . they estimate the total mass of the asteroids constituting the parent - bodies mass by assuming that the observed dust mass will , on average , be lost and replenished within the p - r lifetime of the particles . if this has occurred continuously throughout the star s life , the required parent - body mass , @xmath33 , is @xmath164 @xmath34 g , which is @xmath45% of the primordial mass of asteroids ( @xmath35 , e.g. , lissauer 1987 ) in our solar system . however , since we contend that collisional cascades and subsequent radiation pressure effects are the dominant regulators of the dust population , we reconsider the estimate above for the more relevant collisional timescale @xmath36 rather than that of p - r drag and determine a mass of @xmath46 @xmath34 g , @xmath410% of the solar system s initial asteroidal mass . there is a growing body of evidence for cataclysmic collisional events as a major , or even predominant , source of dust production in debris disks , as discussed in 4 . a plot of simulated dust production rate in our solar system @xcite shows a slow decrease of dust area over time , punctuated by spikes from dust - producing collisions . this illustrates that the amount of dust alone can not indicate how long it has been since the most recent cataclysmic collision . much greater spatial resolution that permits the particle orbits to be constrained is needed for us to better understand the collisional history of the @xmath0 lep debris disk . the apparently restricted spatial extent of this disk may be due to planetary perturbations such as resonant trapping , a scenario that has also been proposed for the debris disk around the k star hd 69830 . this disk s limited extent was inferred by _ photometry @xcite and may be due to the influence of its triple planet system @xcite . the presence of planets may also assist in redistribution of the dust azimuthally by secular perturbations , resonant interactions , gravitational scattering , or accretion . our key conclusion is that we have resolved the disk of @xmath0 lep at 18.3 @xmath1 m , which has a characteristic size comparable to that of our own asteroid belt . this is the only spatially resolved debris disk of this size , and it presents a new disk archetype that contrasts with kuiper belt - like disks extending tens to hundreds of au from the central star . we thank christine chen and michael jura for helpful comments . m. m. m. acknowledges that this work was performed [ in part ] under contract with jpl funded by nasa through the michelson fellowship program . jpl is managed for nasa by the california institute of technology . c. m. t. acknowledges support from nsf grant ast-0098392 . lep ( diamonds ) and reference psf star ( dots ) ; vertical lines indicate the fwhm values of profile fits to the psf star and @xmath0 lep . see the electronic edition of the journal for a color version of this figure . ]	we present subarcsecond - resolution mid - infrared images of the debris disk surrounding the 230 myr - old a star @xmath0 lep . our data obtained with t - recs at gemini south show the source to be unresolved at 10.4 @xmath1 m but clearly extended at 18.3 @xmath2 m . quadratic subtraction of the psf profile from that of @xmath0 lep implies a characteristic radius for the dust disk of 3 au , which is comparable in size to our solar system s asteroid belt . simple models suggest that the 18 @xmath1 m flux is well approximated by two contiguous annuli of mid - infrared - emitting dust from 24 and 48 au with a 3:1 flux ratio for the annuli , respectively . we consider two scenarios for the collisions that must be resupplying the dust population : ( 1 ) continuous `` steady state '' grinding of planetesimals , and ( 2 ) an isolated cataclysmic collision . we determine that radiation pressure and subsequent collisions are the dominant determinants of the disk morphology in either case , and that poynting - robertson drag is comparatively insignificant .
one of the goals of nuclear physics is to characterize the thermal properties of nuclear matter , described by its equation of state . investigating heavy ion collisions offers one experimental method of probing nuclei away from the typical low temperature and density , and a number of experiments are underway studying the breakup of nuclei , contributing greatly to our understanding @xcite . several theoretical investigations of these properties have been instigated , thermal @xcite , statistical @xcite transport @xcite and percolative @xcite models all offering insights into the properties of fragmenting nuclei . in a previous set of papers @xcite an exactly solvable canonical ensemble model was developed for studying the multifragmentation of nuclei induced by high energy collisions . detailed properties of the cluster distributions such as inclusive yields and correlations were explored . in this paper , we study the statistical and thermal properties of nuclei using the same model , extending results presented in @xcite . this model gives a unified description of multifragmentation phenomena and the thermodynamic properties of hadronic matter , such as its equation of state and the nuclear compressibility . simple expressions for these quantities are obtained which extrapolate or connect between a low temperature nearly degenerate fermi gas to a high temperature ideal maxwell - boltzmann gas of nucleons . importantly , the effects of the fragmentation degrees of freedom on the thermodynamics is manifestly included . questions related to very high temperatures and particle and anti - particle production will be developed in a subsequent paper . given the partition function @xmath0 for a system of @xmath1 nucleons at a given volume @xmath2 and temperature @xmath3 we can determine the thermodynamic functions by taking various partial derivatives , i.e. the internal energy and pressure are given by @xmath4 , @xmath5 . additional partial derivatives lead to such important quantities as the specific heat and isothermal compressibility , @xmath6 , @xmath7 , and thermodynamic potentials such as the entropy and gibbs free energy can be related to these quantities , @xmath8 , @xmath9 . our interest of course is not in the partition function itself , but rather in these derived functions , and as such the partition function is merely a means to that end . this section constructs a partition function for fragmenting nuclear matter . it does so by investigating the thermodynamic functions at both low and high temperatures and densities . at low temperature / high density , the nucleons coalesce into a single nucleus which can be modeled as a degenerate fermi gas with skyrme interactions . at high temperature / low density , the nucleons evaporate into a maxwell - boltzmann gas of individual nucleons . with these results firmly in mind , a statistical fragmentation model is introduced whose parameters depend on the thermodynamics in a way consistent with the high and low temperature limits , yet allowing more complicated fragmentation patterns at intermediate temperatures . rhs always diverges in the infinite limit the remaining freedom in the model is eliminated by assuming that the cluster yields follow a power law . interestingly these minimal assumptions lead to a model surprisingly rich in physics as we shall see in the next section . at zero temperature , the nucleus can be treated as a degenerate fermi gas , with fermi momentum @xmath10 given by @xmath11 where @xmath12 is the spin - isospin degeneracy factor and @xmath13 is the density . at normal density @xmath14 nucleons @xmath15 , nucleons are non - relativistic and the fermi energy and momentum are related by @xmath16 . from the above expression for the fermi momentum , we see that @xmath17 . at low temperatures , the total kinetic energy for such a fermi gas is given by @xcite @xmath18 however , the nucleons do interact . interactions between the nucleons can be taken into account by using a density dependent skyrme interaction @xcite . in this approach , the internal energy of a nucleus is given by @xmath19 where @xmath20 and @xmath21 are skyrme parameters ( for simplicity , temperature independent ) determined by fixing @xmath22 the binding energy per particle at zero temperature @xmath23 to the empirical value for @xmath24 at @xmath25 ( about 8.0 mev / nucleon at zero temperature ) and having this as the maximum of @xmath24 . the density dependent repulsive term should appear with a higher power of @xmath26 than the attractive part , so that the nucleus does not collapse , and thus requires @xmath27 . commonly used values for @xmath28 are @xmath29 ( three body interaction ) and @xmath30 ( finite range term ) @xcite . having specified the internal energy , we can now determine any other thermodynamic function . the specific heat @xmath31 , and entropy @xmath32 are therefore @xmath33 the partition function can now be determined from @xmath34 , which gives @xmath35 where @xmath36 and @xmath37 . > from this partition function it is easy to determine the pressure and compressibility . @xmath38 at high temperature and/or low density , but below meson and particle - antiparticle production thresholds , the hadronic properties are that of an ideal maxwell - boltzmann gas , with @xmath39 where the entropy @xmath40 is given by the sackur - tetrode law and @xmath41 . we expect that nuclei vaporize into individual nucleons at @xmath42 , the binding energy per particle . the partition function consistent with these thermodynamic functions is @xmath43 , where @xmath44 . in the region between the ideal fermi gas limit and the ideal maxwell - boltzmann gas limit , the effects of the fragmentation of the initial nucleus must be taken into account when considering thermodynamic issues . to account for these effects , we employ a model developed initially to study multifragmentation phenomena @xcite . this model has the correct high and low temperature limits , as determined in the previous sections , but has features similar to that of a bose gas in the intermediate range . to describe the fragmentation of a nucleus into all possible modes of breakup , a weight is given to each possibility . the weight chosen is @xmath45 where @xmath46 and @xmath47 are functions of the thermodynamic variables @xmath48 , @xmath49 is the cluster size dependence of the weight and @xmath50 is the fragmentation vector , with @xmath51 the number of fragments with @xmath52 nucleons such that @xmath53 . with this choice of weight the canonical partition function @xmath54 is a polynomial in @xmath55 , given by @xmath56 with @xmath57 the multiplicity . given the partition function , all thermodynamic properties of the model can be obtained , as well as ensemble averages . for example , the mean number of clusters of size @xmath52 is @xmath58 the partition functions themselves can be obtained using a recursive procedure defined by the constraint @xmath59 , which gives @xmath60 where @xmath61 . the whole procedure is easily implemented by computer . the parameters @xmath55 determine the thermodynamic aspects of the models and as such need to be determined . they are most easily determined by considering the partition function at high and low multiplicity . when the multiplicity is large @xmath62 , and @xmath63 . clearly this is the high temperature limit , and @xmath46 is simply the ideal gas @xmath46 introduced in section [ sec : high - temp ] , given by @xmath64 and involves the volume of the system @xmath2 , the thermal wavelength @xmath65 , and the dimensionality of the system @xmath66 . the term @xmath67 in the weight arises from the thermal motion of each fragment . since the overall motion is zero , this should be replaced by @xmath68 to reflect the center of momentum constraint . when the multiplicity is small @xmath69 , and @xmath70 . assuming that the @xmath46 parameter is removed by the conservation of momentum constraint describe above , this shows @xmath47 to be given by eq . ( [ eq : y - nocutoff ] ) . a correction to the internal excitations however needs to be made at high temperatures to reflect their finite lifetimes . koonin and randrup @xcite argue that a simple cutoff is effective , which results in @xmath71 here @xmath24 is the binding energy per particle , @xmath72 is the level spacing parameter ( related to the fermi energy ) and @xmath73 is the cutoff temperature . so the term @xmath74 in the weight arises from the binding and internal excitations of each fragment . the only parameter now to determine is @xmath49 . since the mass yield distributions are often well represented by a power law @xcite , we choose @xmath49 simply to reproduce this important experimental fact . if @xmath75 , then in the grand canonical limit @xmath76 . typically @xmath77 is found between two and three , and we have chosen @xmath78 as representative for calculations made in this paper . in this section we consider the various fundamental thermodynamic functions of fragmenting nuclei , namely the pressure , compressibility and gibbs free energy ( section [ sec : pressure ] ) , the internal energy and specific heat ( section [ sec : energy ] ) , and the entropy ( section [ sec : entropy ] ) . interesting features of the functions are explored , although a detailed examination of their origin is postponed till the next section . a brief accounting of such properties has already been given @xcite . since the partition function can be expressed succinctly as a function of @xmath55 , derivatives of @xmath79 with respect to these variables will appear and reappear in computations of thermodynamic functions . fortunately for this model , derivatives with respect to @xmath55 yield expectation values of the multiplicity , i.e. @xmath80 , @xmath81 , @xmath82 . a general derivation of these identities is included in appendix [ sec : appendix ] . the hadronic equation of state is of much interest since the pressure and incompressibility reflect the behavior of matter at fixed temperatures and varying volumes . a simple expression can be obtained for this equation of state by using the @xmath83 and the partition function of eq . ( [ eq : pf - decomposed ] ) as determined from the weight in eq . ( [ eq : weight ] ) , giving @xmath84 where @xmath85 is the mean multiplicity , @xmath86 , @xmath87 are the low and high temperature limits defined in sections [ sec : low - temp ] , [ sec : high - temp ] and @xmath88 is simply the cutoff corrected low temperature pressure . the above equation shows the importance of the mean multiplicity in the behavior of @xmath89 . for low @xmath90 , @xmath69 , and the pressure reduces to the fermi gas and skyrme pressure @xmath86 . at high @xmath90 , @xmath62 and the ideal maxwell - boltzmann gas pressure @xmath87 in @xmath89 dominates . thus the result of eq . ( [ eq : eos ] ) connects the two extremes in a simple analytic way . figure [ fig : pressure](a ) plots the pressure for several temperatures for @xmath91 . the low temperature component of the pressure is quite interesting . it is responsible for the pressure rising as the volume increases over a range of volumes at low enough temperatures , a feature characteristic of a van der waals gas or a liquid - gas phase instability @xcite . it arises due to the nuclear interactions . to see this phase instability more clearly , we can calculate the isothermal compressibility from the equation of state , since @xmath92 , a little effort gives us @xmath93 at large values of @xmath90 , @xmath62 , @xmath94 and @xmath95 , the ideal gas limit . at low @xmath90 , @xmath69 , @xmath94 and @xmath96 , an ideal fermi gas with skyrme interaction limit . in between , there is an effect due to the fluctuation in the mean number of clusters . the behavior of the compressibility is shown in fig . [ fig : pressure](b ) . from the figure , the phase instability appears first at @xmath97 , @xmath98 mev , the point where @xmath99 and its derivative are both zero . another method of seeing this instability is to consider the gibbs free energy @xmath100 . figure [ fig : gibbs ] shows that @xmath101 vs. @xmath89 is multivalued below a critical temperature , a clear indication of a phase instability . in nuclear physics , one usually refers to the incompressibility @xmath102 , which if @xmath103 mev gives at @xmath104 @xmath105 for @xmath29 , this model @xmath106 mev , consistent with the usually quoted value for the incompressibility . the energy and specific heat are important quantities also , especially when the volume is fixed and the temperature is changing . the internal energy decomposes into low and high temperature components just as the pressure does , namely @xmath107 where @xmath108 is simply the cutoff corrected low temperature energy . figure [ fig : spec - heat](a ) plots this behavior for a number of densities , which reveals little except for the monotonic increasing nature of the energy . the specific heat on the other hand is more revealing . using @xmath6 , one arrives at @xmath109 and fig . [ fig : spec - heat](b ) illustrates the behavior of @xmath110 for @xmath111 , @xmath91 . the ideal gas limit is seen in the first term @xmath112 . the low temperature fermi gas result is contained in the second term @xmath113 . the last term involving the multiplicity fluctuations gives rise to a peak in @xmath114 when @xmath115 . the variance induced peak is quite interesting . as we shall see in section [ sec : bose - einstein ] , it is analogous to condensation in an ideal bose - einstein gas , a critical transition which has been extensively studied . the entropy of fragmenting nuclei can be obtained from the relation @xmath116 which gives @xmath117 when @xmath69 , the entropy is that of a nearly degenerate fermi gas of @xmath1 nucleons confined to a cluster of size @xmath1 , plus a contribution from the motion of this cluster which is coupled to a heat bath in the canonical ensemble . the nearly degenerate fermi gas entropy is given by eq . ( [ eq : entropy - lowt ] ) where cutoff effects are excluded . including it gets @xmath118 the cutoff correction is of the same order as the next higher order correction . when @xmath62 , @xmath40 reduces to the sackur - tetrode law of eq . ( [ eq : sackur - tetrode ] ) . in the intermediate regime , their is a critical point where a latent heat must be overcome , causing a change in volume . figure [ fig : entropy ] is a plot of @xmath119 vs. @xmath120 . in the plot , the contribution from the thermal motion of the largest cluster has been subtracted out so that @xmath121 as @xmath122 . this is equivalent to requiring the total momentum of the system to be zero , a correction consistently made throughout these calculations . the thermodynamic functions strongly suggest a critical point , which we determine explicitly in this section . to start , let us work out some properties of the cluster distributions in the grand canonical ensemble . here the partition function is given by @xmath123 where @xmath75 and @xmath124 is the chemical potential . in this limit @xmath125 is simply @xmath126 since @xmath127 , we have @xmath128 with @xmath129 , @xmath55 as defined earlier and @xmath130 . notice that @xmath131 and that for @xmath132 , @xmath133 where @xmath134 , a result due to london @xcite . the lhs of eq . ( [ eq : cons - of - mass ] ) is always finite in the thermodynamic limit . if @xmath135 , the rhs always diverges in the infinite limit @xmath136 , and the equation can not hold . at @xmath137 , @xmath138 , the riemann zeta function , which is finite only if @xmath139 . if @xmath140 then a @xmath141 can be found that satisfies the last constraint equation including all terms in the sum . otherwise , the grand canonical expression breaks down , which defines a critical point @xmath142 the failure of the grand canonical ensemble is due to the appearance of an infinite cluster , i.e. of size @xmath143 where @xmath144 is nonzero even as @xmath1 tends to infinity . more specifically , when @xmath145 exceeds @xmath146 , an infinite cluster exists , absorbing enough mass so that the remaining mass is distributed grand canonically with @xmath137 . if @xmath145 is less that @xmath146 , an infinite cluster does not exist . as we have already seen , the multiplicity and its fluctuations play an important role in the thermodynamic functions . so how does the critical point affect the multiplicity and its fluctuations ? the multiplicity @xmath85 is given simply by @xmath147 the variance of the multiplicity @xmath148 is given by @xmath149 figure [ fig : mult](a),(b ) plots @xmath150 . vs. @xmath151 and @xmath152 respectively for both the canonical and grand canonical solutions . the discontinuity in the slope of @xmath153 at the critical point is apparent from the figure . the behavior of @xmath153 vs. @xmath46 shows a cusp - like behavior . for finite systems , @xmath153 has a rounded peak . as we shall see in the next section , this behavior is related to the ideal bose - einstein gas condensation . is this critical behavior consistent with the experimental situation in nuclear collisions ? at the critical point @xmath154 , where @xmath153 has a cusp , @xmath155 with @xmath156 the eos collaboration experiment determined that for gold multifragmentation , the charge multiplicity @xmath157 . using the above expression and @xmath158 for @xmath1 , results in a critical exponent @xmath159 . using a percolation theoretic analysis , they arrived at a somewhat different @xmath160 . the two results are sufficiently close to suggest that the connection between the critical multiplicity and critical exponent @xmath77 @xcite may not differ greatly from this simple model . let us now compare these results with the ideal bose gas of @xmath1 particles . the results can be found in huang @xcite for the case @xmath111 dimensions , and generalizing to arbitrary dimensions is fairly straightforward . in the grand canonical ensemble , the ideal bose gas of @xmath1 particles moving in a volume @xmath2 at a temperature @xmath3 has a fugacity @xmath161 determined by @xmath162 the critical temperature for a particular density and temperature occurs when @xmath163 , @xmath164 , which implies @xmath165 since above the critical point , the second term can be neglected in the infinite volume limit . recalling that @xmath166 , this condition is identical to the fragmentation case if we replace @xmath46 by @xmath151 and @xmath77 by @xmath167 . the energy , specific heat , pressure and incompressibility above and below the critical point are given by @xmath168 both above and below the critical point these results agree with the fragmentation case @xmath169 , @xmath170 if we apply eqs . ( [ eq : m - gc ] ) and ( [ eq : varm - gc ] ) . this agreement can be understood by noting that the weight given to the ideal bose gas in @xmath66 dimensions in a mayer cluster expansion @xcite is given by eq . ( [ eq : weight ] ) with parameter vector @xmath171 at large @xmath46 ( i.e above the critical point ) the second term is negligible and the model reverts to the one considered here . below the critical point , the large cluster formed contributes little to the thermodynamic functions , allowing agreement to continue . the nature of the transition is first order as expected for a bose - einstein like condensation . the particles which have accumulated into the zero momentum mode contribute no pressure and have no volume or entropy in the thermodynamic limit . the pressure due to the balance of particles is constant for a given temperature ( i.e. independent of volume ) . a latent heat per particle of @xmath172 accompanies the transition and of course is the heat @xmath173 which must be spent in moving a particle from the vapor phase which has entropy to the condensed phase which has no entropy . the latent heat of eq . ( [ eq : latent - heat ] ) is the result of the change in volume along the constant pressure maxwell line at a fixed @xmath3 . the volume change is @xmath174 the entropy change can be obtained from the clausius - clapeyron equation @xmath175 to give @xmath176 which is consistent with @xmath177 as expected . the fraction of the number of particles in the bose condensate ( zero momentum mode ) is @xmath178 for @xmath179 . in nuclear fragmentation , this condensation is also present , but the clusterization occurs in real space instead of momentum space . rather than the number of particles in the zero momentum mode signaling a phase transition , the number of particles in the largest cluster plays the same role . percolation models @xcite have been used to describe nuclear fragmentation with percolation clusters corresponding to nuclear clusters arising from the collision process . the percolation cluster distribution is given in terms of the percolation probability @xmath180 and percolation threshold probability @xmath181 above which an infinite cluster exists . specifically , @xmath182 where @xmath183 is a scaling function . percolation models have also suggested useful methods of analyzing nuclear fragmentation data such as campi plots @xcite . the model in section [ sec : model - studies ] has some features that are similar with the percolation model . first , only one parameter @xmath129 describes the distribution , the analog of @xmath180 in percolation theory . second , below a critical point @xmath184 , the fraction of particles found in the largest cluster is finite , even if the number of particles tends to infinity , a situation familiar from percolation theory where for @xmath185 an infinite cluster exists , while for @xmath186 it can not exist . for the model of section [ sec : model - studies ] this can be seen by analogy with the bose condensate , where a finite fraction of particles end up in the zero momentum state . in this nuclear fragmentation model , a finite fraction of the mass ends up in the largest cluster . specifically if @xmath187 is the size of the largest cluster in each event , then in the infinite @xmath1 limit @xmath188 this threshold behavior of @xmath189 as expected exactly parallels bose - einstein condensation . figure [ fig : kmax ] shows the behavior of @xmath190 and its variance for some a range of @xmath151 for @xmath91 . due to scaling behavior @xmath189 for various @xmath1 s approach the single curve shown above . at the critical point , @xmath76 where @xmath139 can be chosen when specifying the original weight . the critical exponent @xmath77 for percolation is 2.21 . in this paper , hadronic matter under some extreme conditions was investigated using an exactly solvable model which can describe both the fragmentation of nuclei and the thermal properties in a unified way . three distinct types of behavior characterize hadronic matter . at low temperatures , the fermionic aspects of the system are apparent through such quantities as fermi degeneracy pressure and fermi energy . at high temperature and low density , ideal gas laws are shown to characterize the system . between these two limits , a regime exists in which the behavior of the fragmenting matter resembles an ideal bose gas with its associated bose - einstein condensation . the consequences of bose - einstein condensation on the properties of hadronic matter were studied . simple expressions were obtained for the nuclear equation of state , heat capacity , compressibility , entropy , energy and fragmentation yields which encompass these three regimes . the importance of multiplicity fluctuations in statistical models of nuclear fragmentation and its effect on various thermodynamic properties is stressed . some recent data from the eos collaboration on critical point behavior of hadronic matter is analyzed and simple expressions from the exactly solvable model are obtained which can account for some features of the data . this work supported in part by the national science foundation grant no . nsfphy 92 - 12016 and by the department of energy grant no . de - fg02 - 96er40987 . given a function @xmath191 where @xmath192 is the multiplicity , it is often useful to compute @xmath193 . if @xmath191 is a polynomial , the following identities can be used to compute @xmath194 @xmath195_{k } \rangle & = & \langle m [ m]_{k } \rangle - \langle m \rangle \langle [ m]_{k } \rangle \ ; , \nonumber \\ x { \partial \over \partial x } \langle m^{k } \rangle & = & \langle m^{k+1 } \rangle - \langle m \rangle \langle m^{k } \rangle \ ; , \nonumber \\ x { \partial \over \partial x } \langle m \rangle_{k } & = & \langle m \rangle_{k+1 } - k \langle m \rangle_{2 } \langle m \rangle_{k-1 } \nonumber \;.\end{aligned}\ ] ] where @xmath196_{n}$ ] = x(x-1) .... (x - n+1 ) and @xmath197 = nth cumulant moment of m. the first is easily derived from the fact that @xmath198_{k } \rangle = { x^{k } \over z_{a}(x ) } { \partial^{k } z_{a } \over \partial x^{k}}$ ] . the second identity can be derived from the first by expanding @xmath199_{k } \rangle$ ] , where @xmath200 is the stirling number of the second kind , combined with the additional identity @xmath198_{k } \rangle = \sum_{j=0}^{k } ( -1)^{k - j } { k \atopwithdelims [ ] j } \langle m^{k } \rangle$ ] , where @xmath201 j}$ ] is the unsigned stirling number of the first kind , and the orthogonality relations between the two stirling numbers . the third can be derived from the second using the binomial expansion of @xmath202 .	the thermal and statistical properties of hadronic matter under some extreme conditions are investigated using an exactly solvable canonical ensemble model . a unified model describing both the fragmentation of nuclei and the thermal properties of hadronic matter is developed . simple expressions are obtained for quantities such as the hadronic equation of state , specific heat , compressibility , entropy , and excitation energy as a function of temperature and density . these expressions encompass the fermionic aspect of nucleons , such as degeneracy pressure and fermi energy at low temperatures and the ideal gas laws at high temperatures and low density . expressions are developed which connect these two extremes with behavior that resembles an ideal bose gas with its associated bose condensation . in the thermodynamic limit , an infinite cluster exists below a certain critical condition in a manner similar to the sudden appearance of the infinite cluster in percolation theory . the importance of multiplicity fluctuations is discussed and some recent data from the eos collaboration on critical point behavior of nuclei can be accounted for using simple expressions obtained from the model .
the causal relation @xmath0 was introduced in @xcite to extend elements of standard global analysis concerning @xmath1 lorentzian metrics to those which are only @xmath2 . the main obstruction to such an extension within the framework of standard causal analysis is that the existence of the convex normal neighbourhoods is not guaranteed so that compactness of the space of causal curves can not be proved in the usual way @xcite . the alternate causal relation @xmath0 was used in @xcite to construct an appropriate compactness theorem in the @xmath2 case . one of the original motivations of the work in @xcite was to develop a causal analysis that would work in spacetimes with degeneracies . indeed , while the usual chronological and causal relations are ill - defined at the degeneracies , @xmath0 can be naturally extended to cover these points as claimed in @xcite . this present paper has grown out of our interest in the causal structure of causal lorentzian topology changing spacetimes , which necessarily possess degeneracies @xcite . traditionally , such degeneracies are excised from the spacetime , so that what remains is a globally lorentzian spacetime . this was the procedure followed in @xcite , to examine the causal continuity of morse spacetimes . one way to incorporate the degeneracies in the overall causal structure would be to treat them as the boundaries of the spacetime , in the sense of @xcite . this procedure , however , is not without ambiguities . since @xmath0 can be defined in the degenerate spacetimes directly , it seems to be a promising alternative . in this paper , our main aim is to make sense of @xmath0 in topology changing morse spacetimes which possess isolated degeneracies . along the way , we demonstrate a number of interesting properties of @xmath0 in both globally lorentzian spacetimes and those with isolated degeneracies . an outline of our results is as follows . in the first section we briefly review some standard definitions of causal analysis before introducing the relation @xmath0 . we then demonstrate that in a globally lorentzian spacetime , @xmath0 is stable to certain changes in the chronological relation @xmath7 on which it is defined . for a @xmath1 globally lorentzian spacetime @xmath8 , we show that @xmath0 changes minimally when we remove a set @xmath9 of isolated points from @xmath10 , namely , @xmath11 . @xmath0 thus appears to be `` robust '' to such changes . next , we demonstrate that in a stably causal globally lorentzian spacetime @xmath0 is precisely the seifert relation @xmath12 which provides us with an intuitive picture of @xmath0 . we then show that in globally lorentzian spacetimes , the hawking and sachs criterion for causal continuity @xcite is equivalent to the condition , @xmath13 for all @xmath14 , which in turn is equivalent to the statement that @xmath15 constitute a _ causal structure _ , as defined by penrose and kronheimer @xcite . in the second section , we examine the special case of morse spacetimes . when the degeneracies are excised , we show that in a morse spacetimes for which the morse function has only one critical value , @xmath16 coincides with @xmath17 and @xmath18 , respectively . the condition for causal continuity is shown to be equivalent to inner continuity of @xmath19 . we then reinstate the degeneracies and show that again , @xmath0 is unaltered by changes in the chronological relation . we conjecture that the @xmath0 s in the spacetimes with and without the degeneracies are related in a particular way and propose a definition of `` k - causal continuity '' in the unpunctured spacetime . we then show that the above conjecture implies the k - causal continuity in the spacetime with the degeneracies is equivalent to the causal continuity of the spacetime with the degeneracies excised . we also show that , independently of whether the conjecture holds , index @xmath20 and @xmath21 morse spacetimes are k - causally discontinuous , as expected @xcite . in the final section we speculate on the relationship between our work and the idea that spacetime is fundamentally discrete @xcite . for example , the presence or absence of an isolated point in our continuum approximations should be irrelevant as far as any underlying discrete causal structure goes . because of its apparent robustness , the relation @xmath0 seems a reasonable candidate to bridge the gap between continuous and discrete causal structures . we review some standard definitions of causal structure of spacetimes . in this section , @xmath8 denotes a time orientable globally lorentzian spacetime , with @xmath22 of differentiability class @xmath2 unless otherwise specified . a timelike or null vector @xmath23 is future pointing if @xmath24 and past pointing if @xmath25 , where @xmath26 is a vector field that define the time orientation . a _ future directed timelike path _ in @xmath10 is a @xmath27 function @xmath28 { \rightarrow}m$ ] whose tangent vector is future pointing timelike at @xmath29 for each @xmath30 $ ] . past directed paths are similarly defined using past pointing tangent vectors . we write @xmath31 whenever there is a future directed timelike path @xmath32 with @xmath33 and @xmath34 totally contained in @xmath35 . the image of a future ( past ) directed timelike path is a _ future ( past ) directed timelike curve_. given a point @xmath36 we use the notation @xmath37 for the chronological future of @xmath14 relative to @xmath35 . when @xmath38 we will often simplify to @xmath39 and @xmath40 . the symbol @xmath41 and the set @xmath42 are defined dually . the chronos relation @xmath43 is defined by @xmath44 . the chronology is an open subset of @xmath45 . as a relation , it is always transitive ( i.e. , @xmath46 ) and if the chronology condition holds , namely , if the spacetime contains no closed timelike curve , it is also antireflexive ( i.e. there is no pair @xmath47 with @xmath48 and @xmath49 ) . the image of a future ( past ) directed causal path is a _ future ( past ) directed causal curve_. the _ causal future _ @xmath50 and past @xmath51 are similarly defined through causal paths with tangent vector timelike or null . one has @xmath52 and dual . the set of all pairs @xmath47 with @xmath53 also defines a transitive relation in @xmath54 . given two metrics @xmath55 , @xmath56 in a fixed manifold @xmath10 , we write @xmath57 if at every point @xmath36 , the @xmath56-lightcone is strictly larger than the @xmath55-lightcone . seifert future _ @xmath58 of a point @xmath14 in the spacetime @xmath8 is defined as the intersection : @xmath59 it s not hard to see that @xmath60 . similarly one defines the seifert past . it follows immediately from these definitions that @xmath61 iff @xmath62 , therefore @xmath63 and @xmath64 define the same relation in @xmath45 , which we simply call @xmath65 . we will need the _ common future _ @xmath66 of an open set @xmath35 , which is the interior of the set of all points connected to each point in @xmath35 along a future directed timelike curve , i.e. , @xmath67 it is clear from the definition that if @xmath68 then @xmath69 . in addition the chronological future of a point is contained in the common future of its chronological past , namely @xmath70 ; and a similar set inclusion holds for their duals , @xmath71 ( proposition 1.1@xcite ) . let @xmath72 be a function which assigns to each event @xmath14 in @xmath10 an open set @xmath73 . then we say that @xmath72 is _ inner continuous _ if for any @xmath14 and any compact set @xmath74 , there exists a neighbourhood @xmath35 of @xmath14 with @xmath75 @xmath76 . and @xmath72 is said to be _ outer continuous _ if for any @xmath14 and any compact set @xmath77 , there exists a neighbourhood @xmath35 of @xmath14 with @xmath78 @xmath76 . the chronological past and future are inner continuous in any globally lorentzian spacetime @xcite . a time - orientable distinguishing spacetime , @xmath8 is said to be _ causally continuous _ if it satisfies any of the following equivalent conditions @xcite : \(i ) for all events @xmath14 in @xmath10 we have @xmath79 and @xmath80 . \(ii ) for all events @xmath14 and @xmath81 , @xmath82 _ iff _ @xmath83 ; \(iii ) for all events @xmath84 , @xmath85 and @xmath86 are outer continuous . \(iv ) for all events @xmath84 , @xmath87 and @xmath88 [ cc.def ] _ causality _ holds in a subset @xmath89 of @xmath10 if there are no causal loops based at points in @xmath89 . _ strong causality _ holds in a subset @xmath89 of @xmath10 if for every point @xmath90 any neighbourhood @xmath35 of @xmath91 contains another neighbourhood @xmath92 of @xmath91 that no causal curve intersects more than once . a spacetime @xmath8 is said to be _ stably causal _ if there is a metric @xmath93 , whose lightcones are strictly broader than those of @xmath22 and for which the spacetime @xmath94 is causal . a spacetime is stably causal if and only if it admits a global time function , i.e. , a function which is monotonically increasing along every future directed causal curve . the spacetime @xmath10 is said to be _ causally simple _ if @xmath95 for every point @xmath36 . a spacetime is _ globally hyperbolic _ if it possesses a spacelike hypersurface which every inextendiblecausal curve intersects exactly once . in @xcite a _ causal structure _ is defined axiomatically as follows . given a set @xmath96 and three relations , @xmath97 , @xmath98 and @xmath99 , on @xmath96 , the quadruple @xmath100 is said to define a _ causal structure _ if the following conditions are satisfied for every @xmath14 , @xmath81 and @xmath101 in @xmath96 : ( i ) & & x < x + ( ii ) & & x < yy < z , x < z + ( iii ) & & x < yy < x , x = y + ( iv ) & & x < < x + ( v ) & & x < < y , x < y + ( vi+ ) & & x < yy < < z , x < < z + ( vi- ) & & x < < y y < z , x < < z + ( vii ) & & xy x < y x<<y [ pkaxioms.def ] note that in any globally lorentzian spacetime , with @xmath102 denoting @xmath103 , @xmath104 denoting @xmath105 and @xmath106 denoting @xmath107 , all the axioms above are automatically satisfied , except for ( iii ) , which is the condition of causality . in @xcite the relation @xmath0 is defined from @xmath7 as follows : given a spacetime @xmath8 the relation @xmath0 on @xmath108 is the smallest transitive relation that contains @xmath7 and is closed as a subset of @xmath45 . [ k.def ] @xmath0 is thus the _ transitive closure _ of @xmath7 in @xmath54 . by analogy with @xmath7 , we can define @xmath109 and write @xmath110 if @xmath111 . @xmath112 is defined similarly . we also define @xmath113 . further , if @xmath114 , then the relation @xmath115 is defined with respect to @xmath116 , the chronology relation in @xmath117 . we will use either @xmath118 or @xmath119 to denote a pair @xmath47 in @xmath115 . we note that @xmath120 , where @xmath121 denotes the closure of @xmath7 in @xmath45 and @xmath122 and @xmath112 are closed @xmath123 . also , @xmath124 . a simple example , from @xcite , where both set - inclusions are strict is @xmath125-dimensional minkowski spacetime with the @xmath126 , @xmath127 semiline removed ( see figure [ m2-semiline.fig ] ) . for any compact set @xmath128 the set @xmath129 is closed and @xmath130 . thus we obtain the following , [ kouter.lemma ] the functions @xmath131 and @xmath132 are outer continuous . given a point @xmath14 and a compact set @xmath128 with @xmath133 , @xmath134 and therefore , by closure of @xmath135 , there must be a neighbourhood @xmath136 with @xmath137 @xmath138 . now , @xmath114 is said to be _ @xmath139-causal _ if @xmath110 and @xmath140 simultaneously imply @xmath141 for all points @xmath142 . an open set @xmath143 is said to be _ @xmath139-convex _ if for every pair @xmath144 with @xmath110 , @xmath145 . a @xmath139-convex set @xmath146 is further _ @xmath139-globally hyperbolic _ if for every pair of points @xmath144 the interval @xmath147 is compact . in contrast with @xmath7 and @xmath148 , the relation @xmath0 is defined without direct reference to curves in the spacetime . as a consequence , certain properties that are obvious for the relation @xmath148 , need to be established for @xmath0 . the properties we list here have been proved in @xcite , and we mark these with an `` ( sw ) '' : [ subset.lemma ] given two open sets @xmath149 in @xmath10 and two points @xmath150 then : ( i ) if @xmath119 , then also @xmath151 ; ( ii ) if @xmath151 and @xmath35 is @xmath139-convex with respect to @xmath152 , then also @xmath119 . [ boundary.lemma ] let @xmath153 be set with compact boundary @xmath154 and suppose there are two points @xmath155 and @xmath156 with @xmath110 , then there must be a point @xmath157 such that @xmath158 . [ closure.lemma ] let @xmath153 be an open set with compact closure and the points @xmath159 have @xmath110 but not @xmath160 , then there is a point @xmath161 such that @xmath158 . [ convex.lemma ] if @xmath8 is @xmath139-causal , then every point @xmath36 has arbitrarily small @xmath139-convex neighbourhoods . [ csimplicity.lemma ] in a @xmath1 lorentzian spacetime @xmath162 for every @xmath14 if and only if the spacetime is causally simple . the significance of lemma [ convex.lemma ] ( lemma 16 in @xcite ) in the context of globally lorentzian spacetimes is that strong @xmath139-causality and @xmath139-causality are equivalent @xcite . this implies , in particular , that if a spacetime is @xmath163-causal , but not strongly @xmath163-causal , then it is not @xmath139-causal . indeed , take a causal spacetime with a strong causality violation as described in @xcite , namely there exists a pair of points @xmath164 satisfying ( i ) @xmath165 and ( ii ) @xmath166 . ( i ) @xmath167 and ( ii ) along with the closure of @xmath0 @xmath168 , i.e. , the spacetime is not @xmath139-causal . lemma [ convex.lemma ] leads to another intuitive result , [ kj.corollary ] if the @xmath1 spacetime @xmath8 is @xmath139-causal , then every point @xmath36 has a neighbourhood @xmath35 such that @xmath169 . indeed , let @xmath35 be a @xmath139-convex neighbourhood contained in a @xmath163-globally hyperbolic neighbourhood @xmath92 of @xmath14 , for which @xmath170 . then @xmath171 , where the last set - equality follows because @xmath139-convexity implies @xmath163-convexity . note that not all the properties of @xmath0 given above are valid in degenerate spacetimes . apart from the obvious failure , namely , lemma [ csimplicity.lemma ] , lemma [ convex.lemma ] too is not readily available , since the lorentzian character of @xmath22 is used at every point in its proof @xcite . however , lemmas [ subset.lemma ] , [ boundary.lemma ] and [ closure.lemma ] are valid , since their proofs do not depend on the assumption that the metric @xmath22 is everywhere lorentzian or even that timelike curves exist through every point of the spacetime . intuitively , the seifert relation , @xmath12 , depends on the stability of the lightcones against perturbations . we show that , indeed , in globally lorentzian causally stable spacetimes , @xmath0 and @xmath12 coincide . [ kjs.proposition ] in a globally lorentzian spacetime @xmath172 . if the spacetime is stably causal then @xmath173 . clearly @xmath174 . thus , in order to show that @xmath172 , we need to prove closure and transitivity of @xmath12 . to prove the equivalence for stably causal spacetimes , it then suffices to show minimality of @xmath12 . for this we are assisted by a lemma proved by seifert under the assumption of stable causality @xcite . \(i ) ( a ) we first prove that @xmath12 is closed . consider a pair @xmath175 . since @xmath176 is closed @xcite in @xmath10 , there exists a neighbourhood @xmath177 of @xmath14 disjoint from @xmath176 . pick a point @xmath178 in @xmath179 . again , since @xmath180 is closed , there is a neighbourhood @xmath181 disjoint from @xmath180 . take a point @xmath182 . there is a metric @xmath183 so that @xmath184 . consider the new neighbourhoods @xmath185 and @xmath186 respectively , of , @xmath14 and @xmath81 . for any pair of points @xmath187 we must have @xmath188 , since otherwise we would deduce that @xmath189 . therefore @xmath190 contains no pair in @xmath12 . \(b ) next we prove that @xmath12 is transitive . consider @xmath191 and @xmath192 , so that @xmath193 and @xmath194 @xmath195 . by transitivity of @xmath196 this implies that @xmath197 . \(ii ) lemma 1 in @xcite says that in a stably causal spacetime @xmath12 is a partial order . lemma 2 in @xcite then states that if @xmath12 is a partial order then it is contained in any relation @xmath198 satisfying @xmath199 . since this is satisfied by the relation @xmath0 , it follows that @xmath200 . @xmath138 as stated in the introduction , we would like to introduce @xmath0 as the natural causal relation in a spacetime with degeneracies , specifically topology changing spacetimes . in previous work we have analysed such topology changing spacetimes by excising these degeneracies and using the standard definitions of chronology ( @xmath7 ) and causality ( @xmath148 ) as defined above . one might wonder whether the causal properties of a spacetime depend crucially on the absence of the excised point and whether adding it back in in some way might nt make a big difference . as a first step towards justifying the use of @xmath0 we will show that , at least in the case of a point excised from and then added back into a globally lorentzian spacetime , the causal structure given by @xmath0 is as robust as can be imagined and is only minimally altered by the excising and adding process , no matter how it is performed . this seems physically reasonable too as we mentioned before . we first prove a lemma which is clearly related to the issue of missing points and which we will use later . [ delete.lemma ] let @xmath8 be a globally lorentzian , @xmath1 spacetime and @xmath201 . then @xmath202 , @xmath203 , implies that @xmath204 . there exists a future directed timelike path @xmath205 \rightarrow m$ ] so that @xmath206 and @xmath207 . let @xmath208 be its image in @xmath10 . either @xmath208 does not intersect @xmath209 , in which case we are done , or it does . suppose the latter . let @xmath210 be the first value of @xmath211 for which @xmath212 and let @xmath213 : \gamma(t ) = p\}$ ] . intersect @xmath209 infinitely often but it is easier here to use the supremum construction rather than prove this . ] by continuity @xmath214 , @xmath215 and @xmath216 . consider therefore the new path @xmath217 \rightarrow m$ ] formed by joining the two pieces of @xmath32 on @xmath218 $ ] and @xmath219 $ ] . its image @xmath220 intersects @xmath209 exactly once . now consider a convex normal neighbourhood of @xmath209 , @xmath35 . choose a point @xmath221 such that all points on @xmath222 between @xmath209 and @xmath223 lie in @xmath35 . similarly choose a point @xmath224 such that all the points on @xmath222 between @xmath178 and @xmath209 lie in @xmath35 . define a new neighbourhood of @xmath209 , @xmath225 . choose @xmath226 such that there are no timelike curves in @xmath35 between @xmath227 and @xmath209 . such a @xmath227 is guaranteed to exist because @xmath35 is a convex normal neighbourhood which , roughly speaking , has the causal structure of a convex open set in minkowski spacetime . by construction there is a timelike path @xmath228 from @xmath178 to @xmath227 and one , @xmath229 from @xmath227 to @xmath223 . now we build a new timelike path @xmath230 from @xmath14 to @xmath81 by joining @xmath231 from @xmath14 to @xmath178 , @xmath228 from @xmath178 to @xmath227 , @xmath229 from @xmath227 to @xmath223 and @xmath231 from @xmath223 to @xmath81 . by construction , its image @xmath232 can not intersect @xmath209 . hence the result . @xmath138 a simple corollary is that a timelike curve can be deformed to avoid any finite number of points whilst remaining timelike . consider the @xmath1 spacetime @xmath8 . let @xmath233 be a set of finitely many points . on @xmath10 we can consider a number of different relations , [ isonm.eq ] i^+_m & & i^+ + ^+_m & & i^+_m ( m - a ) ( m - a ) + ^+_m & & \{(x , y ) : x , y a , x i^-(y , m - a ) } by lemma [ delete.lemma ] we have @xmath234 . so we can consider the following , potentially different , relations : @xmath235 the transitive closure of @xmath236 and @xmath237 the transitive closure of @xmath238 . the following lemma shows that these are equal . [ rik.lemma ] let @xmath8 , @xmath239 , @xmath235 and @xmath237 be as above . then @xmath240 . since it is minimal , @xmath235 is the transitive closure of any relation whose closure is equal to the closure of @xmath236 so it suffices to prove that @xmath241 . we have @xmath242 . consider @xmath243 , so that there exists a sequence of points in @xmath7 , @xmath244 . there are four cases depending on whether @xmath14 and/or @xmath81 is in @xmath9 . we will prove two as the remaining cases are similar . ( i ) if @xmath245 then @xmath246 such that @xmath247 implies that @xmath248 and so @xmath249 . ( ii ) if @xmath250 and @xmath251 then either @xmath252 @xmath253 such that @xmath254 @xmath255 , or there s an infinite subsequence @xmath256 and @xmath257 . in the latter case we re done as in ( i ) . in the former case we have @xmath258 . choose a sequence of points @xmath259i , @xmath260 , which converge to @xmath14 and we are done as before . @xmath138 on the punctured spacetime @xmath261 we have the relations @xmath262 and @xmath263 . these two relations are equal as before . the possible corresponding causal relations on n are @xmath264 the transitive closure of @xmath265 and @xmath266 . the next result shows that if @xmath10 is @xmath139-causal @xmath267 , in other words , the absence of a finite @xmath239 alters the @xmath0 relation only to the extent of removing relations involving points in @xmath9 . our proof is based on the set theoretic technique of transfinite induction , which is a generalisation of ordinary induction , to deal with constructions that do not occur through a countable number of steps . transfinite induction is used to verify that a certain property , call it @xmath96 , satisfied by pairs in @xmath7 is preserved as new pairs are successively added in the construction of @xmath0 , because of the requirements of closure and transitivity @xcite . more concretely , one considers at an intermediate stage , the relation , @xmath268 and the hypothesis , @xmath269 , that @xmath96 is satisfied by all points in @xmath270 . to prove this hypothesis , one must first establish that @xmath271 is true , namely that @xmath96 is satisfied by @xmath272 . next , given that @xmath96 is satisfied for all @xmath273 , where @xmath274 , one needs to show that it is true for @xmath275 . at this stage , there are two cases that arise in this transfinite version of induction . if @xmath275 is a limit ordinal the conclusion follows since @xmath276 for @xmath274 . if @xmath275 is not a limit ordinal , then @xmath277 such that @xmath278 , namely , @xmath279 . the new pair @xmath47 has been added either because there are sequences @xmath280 and @xmath281 with @xmath282 or because there is a point @xmath101 such that @xmath283 and @xmath284 . the essence of the proof is to show that in either case @xmath47 also satisfies the property @xmath96 . for more details , and to appreciate the difference between ordinary induction and transfinite induction , see @xcite . [ kingl-a.proposition ] let @xmath8 , @xmath239 , @xmath261 , @xmath264 and @xmath285 be as above and let @xmath8 be in addition @xmath139-causal . . we will adopt the convention in this proof that any relation , @xmath7 , @xmath0 , @xmath287 , _ etc . _ , that has no subscript labeling which manifold it is on will be understood to be on @xmath10 . first assume that @xmath9 contains a single point @xmath209 , so that @xmath288 . @xmath285 defines a closed transitive relation on @xmath289 which includes @xmath265 . thus , all we need to show is that @xmath290 since @xmath264 is minimal . in other words , if a pair @xmath291 is such that @xmath292 , we want @xmath293 . it is convenient to establish this property by dividing the pairs @xmath47 into two sets , namely ( i ) @xmath294 and ( ii ) @xmath295 \(i ) here , we employ the methods of transfinite induction discussed above . let @xmath269 be the hypothesis that at stage @xmath275 , if the pairs @xmath47 satisfy @xmath296 and @xmath297 , with @xmath294 , then @xmath293 . this is true at the initial step @xmath272 , since @xmath298 , so that any timelike curve from @xmath14 to @xmath81 must be contained in @xmath289 . now , consider the stage @xmath275 and let @xmath299 hold for every @xmath274 . if @xmath275 is a limit ordinal , @xmath269 follows . otherwise , if @xmath300 , we can assume that the pair added in going from @xmath301 to @xmath275 , is @xmath47 , and @xmath302 is trivially satisfied , so that we only need to consider the case , @xmath303 . first , suppose the new pair has been added because of closure , i.e. , there are sequences @xmath280 and @xmath281 with @xmath304 . the @xmath305 are in @xmath289 for large enough @xmath306 . moreover , by closure of @xmath0 , we must since @xmath294 , for sufficiently large @xmath306 , @xmath307 . in other words , there is an integer @xmath308 such that the hypothesis @xmath299 applies to the tail of the sequences @xmath309 and @xmath310 , so that @xmath311 . thus , by closure of @xmath264 , @xmath312 . if the new pair has been added because of transitivity , then there is a point @xmath101 with @xmath313 . since @xmath303 , @xmath314 , and for the same reason @xmath315 and @xmath316 . it follows from @xmath317 that that @xmath318 and , by transitivity of @xmath264 , we see that @xmath293 . thus , we have shown that for all @xmath319 satisfying @xmath320 and @xmath294 , @xmath321 . note that we do not require the @xmath1 condition for this part of the proof . \(ii ) now we prove the property for the pairs @xmath47 , such that @xmath322 . let @xmath35 be a @xmath139-convex neighbourhood such that @xmath323 , which exists by virtue of corollary [ kj.corollary ] . requirement makes its appearance at this crucial stage in the proof . ] choose @xmath324 such that @xmath325 and @xmath326 . that such @xmath327 and @xmath328 exist is guaranteed by applying lemma [ boundary.lemma ] to the compact boundary of a subneighbourhood @xmath329 . now since @xmath330 , we have @xmath331 and @xmath332 . we claim ( see below ) that this implies @xmath333 , which means , by lemma [ delete.lemma ] , @xmath334 . also , by @xmath139-causality we have @xmath335 so that part ( i ) implies @xmath336 . similarly @xmath337 . thus we have @xmath338 , and since @xmath264 is transitive we conclude @xmath293 . if @xmath9 contains more than one point , note that because @xmath339 , for any @xmath340 , it follows that [ step.eq]k^+(x,m-(p_1p_2 ) ) & = & k^+(x , m - p_1)((m - p_1 ) - p_2 ) + & = & ( k^+(x)(m - p_1 ) ) ( m-(p_1p_2 ) ) + & = & k^+(x)(m-(p_1p_2 ) ) , for any @xmath341 with @xmath342 . @xmath138 it remains to prove the claim : [ transi.claim ] if @xmath343 and @xmath344 then @xmath345 . there exist sequences @xmath346 and @xmath347 with @xmath348 and @xmath349 . consider @xmath350 for each @xmath306 . there s a neighbourhood of @xmath81 contained in @xmath350 and so there s an @xmath351 such that @xmath352 . hence @xmath353 . @xmath138 for any finite set @xmath9 in a @xmath139-causal spacetime , @xmath8 , the above result makes it easy to show that the causal future and past of the points in @xmath9 are determined by the relation @xmath264 on @xmath261 simply by taking the closure in @xmath45 , _ i.e. _ , @xmath354 . [ krel.lemma ] let @xmath8 , @xmath9 , @xmath289 , @xmath264 and @xmath235 be as in proposition [ kingl-a.proposition ] . then @xmath354 , where the closure is taken in @xmath45 . we know from proposition [ kingl-a.proposition ] that @xmath355 . so we have @xmath356 . on the other hand suppose @xmath357 . then ( i ) @xmath358 , or ( ii ) @xmath359 , or ( iii ) @xmath360 , or ( iv ) @xmath361 . in each case there exists a sequence in @xmath264 which converges to @xmath47 . we will show this explicitly in case ( i ) only , as the other cases are similar and simpler . choose a sequence of points @xmath362 , @xmath363 . similarly choose a sequence @xmath364 , @xmath365 . then @xmath366 is a sequence that converges to @xmath47 and @xmath367 . so @xmath368 and hence result . @xmath138 another natural way to construct the causal relation on @xmath10 from knowledge of @xmath289 would be to consider @xmath265 , the usual chronological relation on @xmath289 , as a relation on @xmath10 , and construct the transitive closure of it in @xmath45 . in other words take the transitive closure of what we called @xmath369 . that this gives @xmath235 is the statement of lemma [ rik.lemma ] . we conclude that the relation @xmath0 is robust against the subtracting and adding of isolated points . the definition of causal continuity due to @xcite concerns globally lorentzian spacetimes . however , causal continuity appears to be an important property in the study of topology changing morse spacetimes , which contain isolated degeneracies @xcite . in order to extend the notion of causal continuity to spacetimes with degeneracies , we seek alternative characterisations that involve @xmath0 alone , since , as we shall see shortly , this can be defined at the degeneracies . first we prove that two conditions based on the pair @xmath15 are equivalent to the standard definition . [ kccgl.proposition ] in a @xmath139-causal spacetime @xmath8 the following three conditions are equivalent . \(a ) the spacetime ( m , g ) is causally continuous . \(b ) for every @xmath36 , @xmath370 . \(c ) the relations @xmath15 define a causal structure in @xmath10 . we use the definition [ cc.def].ii of causal continuity , namely , for any @xmath371 , @xmath372 . \(a ) implies ( b ) . when causal continuity holds the subsets @xmath373 and @xmath374 are identical in @xmath45 and hence define the same relation . this relation is transitive by claim [ transi.claim ] and contains @xmath7 in a minimal way , so we only need prove that it is closed in @xmath45 . suppose there are sequences @xmath280 and @xmath281 with @xmath375 . we show that causal continuity implies that @xmath376 . for any @xmath377 , there is an integer @xmath378 such that @xmath379 . therefore we have a sequence @xmath380 with @xmath381 , which is a closed set in @xmath10 , hence @xmath382 and , by causal continuity , @xmath383 . since there is such an @xmath178 in every neighbourhood of @xmath14 we conclude , by closure of @xmath384 in @xmath10 , that @xmath385 , or equivalently @xmath386 . \(b ) implies ( a ) . if @xmath370 , then for every pair of points @xmath14 , @xmath81 we have @xmath387 iff @xmath388 iff @xmath111 iff @xmath389 . \(b ) implies ( c ) . all but axioms ( ii ) and ( vi ) in the definition of a causal structure [ pkaxioms.def ] are satisfied by @xmath15 in any globally lorentzian spacetime . axiom ( ii ) is guaranteed by our requirement of @xmath139-causality , so we only need to consider ( vi ) . consider @xmath390 . there exists neighbourhood @xmath391 of @xmath81 , such that @xmath392 . now , @xmath393 , so that there is a point @xmath394 with @xmath395 . hence , by transitivity of @xmath7 , @xmath396 . similarly for ( vi- ) . \(c ) implies ( b ) . let @xmath397 for all @xmath398 . i we have @xmath399 . let us assume that there is an @xmath36 for which @xmath400 . if every @xmath81 in @xmath401 is in the boundary of @xmath122 , we would have @xmath402 and therefore , from the paragraph before lemma [ kouter.lemma ] , @xmath403 , thus contradicting the assumption . this means that there is a point @xmath81 with a neighbourhood @xmath391 contained in @xmath401 . hence for any @xmath404 in @xmath391 we have @xmath405 and not @xmath406 , which is again , a contradiction . notice that `` ( a ) implies ( b ) '' can be immediately inferred from definition [ cc.def](iv ) and theorem [ kjs.proposition ] . we have nevertheless given a proof in terms of @xmath0 because it is simple enough and avoids the mention of @xmath12 . we now propose two characterisations of causal continuity involving only @xmath0 . in the next section , we show that these characterisations are valid for morse spacetimes , while in general , we can partly prove their equivalence to the conditions in proposition [ kccgl.proposition ] . in a @xmath139-causal spacetime @xmath8 consider the conditions : \(d ) @xmath131 and @xmath132 are inner continuous . \(e ) for any @xmath36 and @xmath139-convex neighbourhood @xmath289 of @xmath14 , @xmath407 and @xmath408 are connected . we have put forward ( d ) in analogy to definition [ cc.def](iii ) . remember that in any spacetime @xmath409 and @xmath410 are inner continuous , while @xmath131 and @xmath132 are outer continuous . since outer continuity of @xmath411 is one characterisation of causal continuity , it seems reasonable to expect that so must be inner continuity of @xmath412 . this `` duality '' stems from the fact that @xmath7 is open as a relation in @xmath45 , while @xmath0 is closed . for example , let us look at the past @xmath410 . for any point @xmath14 in a spacetime , @xmath413 becomes arbitrarily small as @xmath81 approaches @xmath14 from its past this is the inner continuity of @xmath410 , while @xmath414 becomes arbitrarily small as @xmath101 approaches @xmath14 from its future outer continuity of @xmath132 . an abrupt change of the past occurs at a point @xmath14 if @xmath415 contains a fixed open set as @xmath101 approaches @xmath14 from its future , or , as we propose , if @xmath416 contains a fixed open set as @xmath81 approaches @xmath14 from its past means roughly that it converges from `` below '' and outer continuity , that it converges from `` above '' , one may envisage unifying the two conditions in a single notion of continuity , using an appropriate topology . it has been suggested @xcite that the vietoris topology @xcite may be suitable for this purpose . ] . as for ( e ) , we have arrived at this condition by looking at different causally discontinuous spacetimes , all of which have in common a local disconnectedness of the @xmath139-pasts or @xmath139-futures . the details in the condition are as follows . we require @xmath289 to be convex in order to avoid any spurious disconnectedness associated with the boundary of @xmath289 ( figure [ connectivity.fig](a ) ) . we impose the condition to hold for arbitrary @xmath289 because often the past or future are disconnected only locally and become connected again in large enough neighbourhoods ( figure [ connectivity.fig](b ) ) . finally we express it in terms of @xmath412 rather than @xmath417 itself so that our condition extends to degenerate spacetimes : degeneracies will sometimes connect otherwise disconnected pasts and futures , but not their interiors ( figure [ connectivity.fig](c ) ) . 0.75 cm [ intk.lemma ] if the @xmath139-causal spacetime @xmath8 is causally continuous , then \(i ) @xmath131 and @xmath132 are inner continuous . ( ii ) for any @xmath36 and @xmath139-convex neighbourhood @xmath289 of @xmath14 , @xmath407 and @xmath408 are connected . \(i ) is immediate , since from proposition [ kccgl.proposition ] , we know that causal continuity implies that @xmath418 , which is inner continuous @xcite . for ( ii ) , suppose that causal continuity holds , so that @xmath13 for each @xmath14 , and let @xmath289 be @xmath139-convex , we show for example that @xmath419 . this suffices , since @xmath420 is clearly connected . now any @xmath421 is also in @xmath122 and therefore in @xmath422 . hence there exists a sequence of points @xmath281 with @xmath423 and we can assume that all the @xmath424 lie in @xmath289 , since this is an open set . let @xmath425 be a sequence of timelike curves from @xmath14 to @xmath424 . then @xmath426 , which by convexity of @xmath289 implies @xmath427 . therefore @xmath428 , as desired . @xmath138 with this we have shown that causal continuity implies conditions ( d ) and ( e ) above . later we show that in morse spacetimes ( d ) implies causal continuity . the following lemma provides a convenient way of expressing condition ( d ) . [ kcc.lemma ] in a @xmath139-causal spacetime @xmath8 , @xmath131 and @xmath132 are inner continuous if and only if for every @xmath371 , @xmath429 . suppose @xmath131 and @xmath132 are inner continuous . if @xmath430 , then there must be a neighbourhood @xmath391 of @xmath81 such that @xmath431 . therefore @xmath432 and @xmath433 . similarly @xmath434 implies @xmath435 . conversely , suppose that for every @xmath371 , @xmath429 . consider any @xmath436 and the compact set @xmath437 . for every @xmath438 , the condition implies that we can find points @xmath439 and @xmath440 such that @xmath441 and therefore neighbourhoods @xmath442 of @xmath14 and @xmath443 of @xmath81 so that @xmath444 . the cover @xmath445 of @xmath446 must have a finite subcover , so @xmath447 . let @xmath448 , then @xmath449 @xmath450 , so that @xmath132 is inner continuous . similarly one proves the inner continuity of @xmath131 . our motivation to consider the relation @xmath0 is that it provides a framework in which to discuss the causal properties of degenerate spacetimes . as mentioned in the introduction , such degeneracies are a necessary feature of compact , causal lorentzian topology changing spacetimes . in particular , one can construct a rather general class of such spacetimes which contain only isolated degeneracies . these `` morse spacetimes '' were studied in detail in @xcite within the sum - over - histories approach to quantum gravity , and can be defined in any topological cobordism , as follows . morse spacetimes are defined on a compact cobordism @xmath451 , with @xmath452 , a riemannian metric @xmath453 on @xmath454 and a morse function @xmath455 $ ] , with @xmath456 , @xmath457 which has @xmath458 critical points , @xmath459 , in the interior of @xmath454 . the critical points of a differentiable function @xmath460 are those where @xmath461 . what characterises a morse function is that the hessian @xmath462 of @xmath460 is an invertible matrix at each of its critical points , which we call morse points . the morse index @xmath463 of the critical point @xmath464 is the number of negative eigenvalues of the matrix @xmath465 . from @xmath453 , @xmath460 and an arbitrary real constant greater than @xmath20 , denoted @xmath466 , we can define a _ morse metric _ in @xmath454 : [ morsemetric.eq ] g_=(h^_f_f)h _ -_f _ f , the vector field @xmath467 is time - like with respect to the metric @xmath22 , so that we can think of @xmath460 as a time - function on @xmath8 . we refer the reader to @xcite for a more expository account . the metric ( [ morsemetric.eq ] ) is lorentzian everywhere , except at the critical points @xmath459 of @xmath460 . if , as is standard practice in causal analysis , we excise these critical points from the spacetime , the resulting _ morse spacetime _ , @xmath8 , is the globally lorentzian spacetime induced in @xmath468 . a general cobordism @xmath454 can be decomposed into a finite sequence of elementary cobordisms @xmath469 , each containing one critical point . if @xmath470 is an elementary cobordism , we will refer to a morse spacetime @xmath471 , defined in @xmath472 , as an elementary morse spacetime . the morse lemma tells us that there exists a neighbourhood of a critical point , @xmath209 , with local coordinates @xmath473 , in which the morse function @xmath460 takes a canonical quadratic form at any point in this neighbourhood , @xmath474 where @xmath475 is the index of the morse function . in what follows we will refer to these neighbourhood spacetimes as neighbourhood morse spacetimes , as opposed to the full morse spacetimes , defined above . we now proceed to explore the @xmath15 structure for morse spacetimes both when the degeneracies are excised to give a globally lorentzian spacetime and when they are left in . in the case of a morse spacetime with a single critical surface , we can show that when causal continuity fails , then at least one of @xmath131 and @xmath132 is not inner continuous . in other words , the converse of lemma [ intk.lemma].(i ) is now true . we know that any causally discontinuous morse spacetime @xmath8 contains a causally discontinuous elementary cobordism @xmath476 . it suffices to show that then there must be @xmath477 such that @xmath478 and yet @xmath479 , for this means that @xmath480 and , by the @xmath139-convexity of @xmath481 ( lemma [ convex.lemma ] ) , that @xmath479 . [ kinelcob.lemma ] for any point @xmath14 in a morse spacetime @xmath8 where the morse function @xmath460 has a single critical value @xmath482 and @xmath483 . let us call @xmath484 the relation consisting of pairs @xmath47 such that @xmath485 . and let @xmath486 be the critical value . \(i ) first we check that @xmath485 if and only if @xmath487 . if the former holds , there are sequences @xmath280 and @xmath488 with @xmath489 and @xmath490 . clearly , it suffices to show that @xmath491 . for each @xmath492 , there is another @xmath493 and a neighbourhood @xmath494 of @xmath492 such that @xmath495 . now for any point @xmath496 , there is an @xmath497 with @xmath498 . this means that @xmath499 , as desired . the converse implication is proved analogously . \(ii ) the relation @xmath484 is closed . let @xmath500 with @xmath280 and @xmath281 . it suffices to show that any @xmath496 is in @xmath501 . given any such @xmath178 , we can find neighbourhoods @xmath136 of @xmath14 and @xmath502 of @xmath178 with @xmath503 . moreover , there is an integer @xmath378 such that @xmath504 . therefore @xmath136 contains points @xmath505 . now for every @xmath506 , there is a @xmath492 with @xmath507 to the past of @xmath223 . hence , @xmath508 , that is @xmath509 . \(iii ) the relation @xmath484 is transitive . it is here that we need the assumption that there is only one critical value . suppose @xmath510 and @xmath511 . we wish to show that @xmath512 . we need to consider three cases : case 1 . @xmath387 and @xmath513 implies , by claim [ transi.claim ] , that @xmath514 . case 2 . @xmath387 and @xmath515 . take a sequence @xmath280 contained in @xmath516 and a sequence @xmath281 in @xmath517 . for each @xmath497 there is a @xmath518 and hence a neighbourhood @xmath519 . it follows that for every @xmath520 , @xmath521 . thus the sequence @xmath522 is also contained in @xmath517 . case 3 . @xmath523 and @xmath513 . we must have @xmath524 , because the subcritical region is globally hyperbolic . and so is the supercritical region , whence , @xmath525 . consider a sequence @xmath280 with @xmath526 , that is , each @xmath497 has a neighbourhood @xmath527 contained in the past of every @xmath528 . take any @xmath529 . it follows from axiom ( vi+ ) in definition [ pkaxioms.def ] that @xmath530 so that there exists a point @xmath531 . thus @xmath532 , i.e. , @xmath533 . @xmath138 with this , we can now prove : [ kinnermorse.lemma ] a morse spacetime @xmath8 is causally continuous if and only if the functions @xmath131 and @xmath132 are inner continuous . we saw in lemma [ intk.lemma ] that causal continuity implies inner continuity of @xmath131 and @xmath132 . conversely suppose @xmath8 is causally discontinuous and so contains a causally discontinuous elementary subcobordism , @xmath481 @xcite . this implies the existence of points @xmath534 @xmath535 . thus @xmath536 by the previous lemma . however , @xmath537 . this is because @xmath538 and @xmath539 ( otherwise , if @xmath81 was in @xmath540 , there would be a sequence @xmath541 converging to @xmath14 and we would have @xmath542 , a contradiction ) . lemma [ kcc.lemma ] @xmath138 to incorporate the critical points of a degenerate spacetime @xmath454 to the @xmath0 causal order we choose a natural chronology in @xmath454 and find its transitive closure . more exactly , the definition that we propose is : let @xmath543 be lorentzian everywhere , except for a set @xmath544 of isolated points where @xmath22 vanishes , @xmath468 be the regular part of the spacetime and @xmath545 be the usual chronology defined through timelike curves in @xmath454 . so @xmath546 and the chronological past and future of each degeneracy are empty . we define the relation @xmath547 in @xmath454 to be the transitive closure of @xmath545 in @xmath548.[kinmorse.definition ] the question of uniqueness arises , since one can think of other natural chronologies in a degenerate spacetime . to justify definition [ kinmorse.definition ] we should demonstrate that the transitive closures of the different chronologies are the same . strictly , a curve passing through a degeneracy @xmath209 is neither timelike , null or spacelike , and in this sense we would say that @xmath549 is empty . on the other hand , there are timelike curves that run off towards @xmath209 , and it would seem reasonable to say that all the points in such curves belong to @xmath550 . this leads us to two natural definitions of chronology in a degenerate spacetime @xmath543 : the first one is @xmath551 is that used in definition [ kinmorse.definition ] . this is the chronology defined through timelike curves in @xmath454 if we are strict about the conditions on the tangent vector of a causal curve . because no tangent vector at a degenerate point @xmath464 can be deemed as past or future directed does not admit a time orientation , because any vector field in @xmath454 will be trivially null at each point @xmath464 . we can still define future and past directed causal curves using the time - orientation available in @xmath10 . ] there are no timelike or causal curves through the degeneracies . thus for each @xmath464 , we have @xmath552 , i.e. , the chronological past and future of each degenerate point is empty . the second one is @xmath553 , which we define by allowing causal curves to go through the degeneracies . to do so , we redefine a future - directed causal curve @xmath208 in @xmath454 as a piecewise smooth curve whose tangent vector @xmath554 obeys @xmath555 and is future - directed at all points in @xmath556 . similarly for past - directed causal curves . in this way we define a causal relation @xmath557 for which @xmath558 will in general be non - trivial for a degeneracy @xmath209 . the chronological relation @xmath553 is then defined as @xmath559 for every point @xmath560 . in figure [ genij.fig ] , we illustrate the definitions of @xmath561 and @xmath553 for the trousers spacetime of @xcite . one can verify by inspection that @xmath562 and @xmath563 are both equal to @xmath557 , as in the closed shaded regions in figure [ genij.fig](b ) . we show below that in fact , this is true for any morse spacetime , i.e. , that although in general , @xmath564 , they lead to exactly the same relation @xmath0 in @xmath8 . [ i1i2.lemma ] let @xmath543 be a morse geometry with the chronologies @xmath561 and @xmath553 in @xmath454 defined as above . if @xmath562 and @xmath563 are their respective transitive closures in @xmath10 , then @xmath565 . notice first that @xmath566 implies @xmath567 . the converse will follow from minimality of @xmath563 if we show that @xmath568 . what we demonstrate is that in fact @xmath569 . let @xmath460 be the morse function in @xmath10 . for any @xmath570 , there is a piecewise smooth path @xmath571{\rightarrow}{\mathcal{m}}$ ] with @xmath572 future - directed null or timelike except possibly for a finite number of degeneracies @xmath573 , through which @xmath208 passes in order of increasing @xmath460 . add to the list the points @xmath574 and @xmath575 and let @xmath576 and @xmath577 . for each @xmath578 between @xmath579 and @xmath458 there are sequences @xmath580 and @xmath581 with @xmath582 such that the associated sequences @xmath583 and @xmath584 consist of regular points in @xmath10 . therefore these points must satisfy @xmath585 , which implies @xmath586 , since @xmath562 contains @xmath587 and is closed . and hence , by transitivity , @xmath588 . @xmath138 in order to relate @xmath0 in spacetimes with and without the degeneracies , one would need a generalised version of proposition [ kingl-a.proposition ] . in other words , if @xmath589 is a spacetime with an isolated degeneracy at @xmath209 and @xmath8 the globally lorentzian spacetime with @xmath590 , then we expect that the causal relations @xmath591 on @xmath543 and @xmath235 on @xmath8 should be related as follows , @xmath592 the latter a consequence of the former as in lemma [ krel.lemma ] . we do not , however , have a complete proof of this statement , due to a difficulty in proving transitivity when the degenerate point is sandwiched between two regular points of the manifold . indeed , the proof of proposition [ kingl-a.proposition ] already gives a hint of the nature of the problem , where in proving transitivity , one requires the spacetime to be @xmath1 at the excised points . this is too strong a requirement to be used here . for the index @xmath579 and @xmath593 elementary morse spacetimes , however , since the degenerate point can not be sandwiched between two regular points of the manifold , there is no problem with transitivity , and one may prove the validity of the condition . from this , and our experience with @xmath594 morse spacetimes , we suspect that the condition is in fact satisfied for all morse spacetimes . we will therefore restrict our attention to those spacetimes satisfying this condition , and investigate its consequences in what follows . we state the condition as : [ krobustinelcob.condition ] for a morse spacetime @xmath8 with causal order @xmath235 and @xmath547 the extended causal order on @xmath595 , with @xmath209 the morse point , we require @xmath596 . we would like to know whether the characterisations of causal continuity for globally lorentzian morse spacetimes , in proposition [ kccgl.proposition ] and lemma [ kinnermorse.lemma ] can be extended to morse spacetimes in which the degeneracies are left in . to define causal continuity in @xmath454 we use the extended causal order @xmath547 . a first possibility is to replace @xmath235 by @xmath547 in condition ( b ) of proposition [ kccgl.proposition ] and we propose [ kc.definition ] the degenerate spacetime @xmath543 , with regular part @xmath10 , is said to be k - causally continuous if every @xmath36 satisfies @xmath597 . here the closure of @xmath85 is taken in @xmath454 instead of @xmath10 . under the assumption that condition [ krobustinelcob.condition ] holds , we can show that the above definition is simply a restatement of condition ( b ) in proposition [ kccgl.proposition ] , i.e. , that it agrees with causal continuity of the globally lorentzian spacetime @xmath8 . let @xmath589 be a spacetime with isolated degeneracies and @xmath8 the associated globally lorentzian spacetime which results from excising the degeneracies . suppose @xmath8 satisfies condition [ krobustinelcob.condition ] , then @xmath543 is k - causally continuous iff @xmath8 is causally continuous . [ kccandcc.lemma ] \(1 ) let @xmath8 be causally continuous , i.e. , @xmath598 , @xmath599 , where the superscript @xmath600 indicates that the closure is taken in @xmath10 . we want to prove that a point @xmath601 must also satisfy @xmath602 . \a ) if @xmath436 , then condition [ krobustinelcob.condition ] says that @xmath603 . \b ) if @xmath604 , that is if @xmath81 is one of the degeneracies . then from lemma [ boundary.lemma ] we can find a sequence of regular points @xmath605 such that @xmath606 . it follows from condition [ krobustinelcob.condition ] that @xmath607 . and therefore @xmath608 , as required . \(2 ) let @xmath543 be k - causally continuous . suppose @xmath609 , then @xmath610 . now , since @xmath436 , a sequence @xmath605 must exist with @xmath423 and such that @xmath611 . therefore @xmath612 , which means @xmath8 is causally continuous . @xmath138 note that part ( 2 ) of the proof does not depend on condition [ krobustinelcob.condition ] and therefore holds in general . in other words , if @xmath8 is causally discontinuous then @xmath543 is k - causally discontinuous . from @xcite we know that @xmath8 is causally discontinuous if it contains index @xmath20 or @xmath21 morse points , which means that it is k - causally discontinuous as well . to extend definition [ kc.definition ] to the degeneracies of a morse spacetime , we first have to choose an appropriate definition of chronology @xmath613 for each morse point @xmath209 . by contrast , conditions ( d ) and ( e ) in the previous section are expressed in terms of @xmath0 alone , so that one may expect them to extend directly to the whole degenerate spacetime by simply replacing @xmath235 by @xmath591 . are these conditions equivalent to causal continuity of the punctured morse spacetimes ? the extension of condition ( d ) is : @xmath614 and @xmath615 are inner continuous as functions on @xmath454 . if the spacetime were to satisfy the condition , @xmath616 ( condition [ krobustinelcob.condition ] ) , then inner discontinuity of @xmath617 would imply that of @xmath614 ( similarly for @xmath132 . ) thus causal discontinuity of the morse spacetime implies inner discontinuity of the extended @xmath618 or @xmath619 . however , the converse is not true : causal continuity does not always imply inner continuity of the extended @xmath620 and @xmath621 . a counterexample that illustrates this is a two - dimensional yarmulke , _ i.e. _ a morse geometry with @xmath622 , @xmath623 , corresponding to the topology change @xmath624 . let @xmath625 be the coordinates in which the morse function takes the form @xmath626 , let @xmath627 , let the riemannian metric @xmath453 be given by the interval @xmath628 in these coordinates and set @xmath629 . studying the null - equations of the morse metric @xmath22 constructed from @xmath460 , @xmath453 and @xmath466 , we obtain the light - cone structure drawn in figure [ badyarm.fig ] . from @xcite we know that the morse spacetime @xmath630 is causally continuous . however , it is easy to see that @xmath618 is not inner continuous . let @xmath9 be the closed region shown in figure [ badyarm.fig ] bounded by the lines @xmath631 and @xmath632 . now , although @xmath633 , for any @xmath634 , @xmath635 . , we had only defined condition ( d ) to hold at the regular points of @xmath454 , then the above yarmulke would in fact satisfy it . however , our interest is in a definition which does nt single out degenerate points ] thus , although the spacetime @xmath543 is causally continuous , @xmath618 is not inner continuous at @xmath209 . it is therefore perhaps inappropriate to use the extended version of condition ( d ) as the definition of k - causal continuity when the degenerate points are included . for condition ( e ) , the extension in terms of @xmath547 on @xmath454 is : for any @xmath636 and @xmath637-convex neighbourhood @xmath289 of @xmath14 , @xmath638 and @xmath639 are connected . we know of no counterexample to the possibility that condition ( e ) for @xmath454 is equivalent to causal continuity of the spacetime with the degeneracies excised , but neither do we have a proof . we have presented preliminary results on the causal structure , and causal continuity in particular , in terms of the natural and robust causal relation @xmath0 on spacetimes with isolated degeneracies . the case of most interest to us is that of topology changing morse spacetimes . to make further progress , it is necessary for us to prove that causal continuity of a degenerate spacetime can be characterised entirely in terms of @xmath0 . ultimately , topology changing processes can only be analysed with rigour in a theory of quantum gravity . any seemingly ad hoc rules governing topology change should be justified by appealing to the more basic theory . underlying our analysis , then , is the idea that perhaps the causal order of a spacetime is its most fundamental aspect . this is the basis of the causal set hypothesis @xcite that the fundamental discrete ( quantum ) substructure of spacetime is described by a causal set . the question thus arises : can any sense be made at the discrete level of a continuum notion like causal continuity ? in a sense we have been working towards this in this paper framing conditions in terms of the @xmath0 relation alone and independently of @xmath7 , for example , could be seen as moving in this direction , since a causal set possesses a causal relation but not a chronological one . basic to the causal set hypothesis is the notion of `` manifold - like '' causal sets which can be obtained by sprinkling points at random , with a fixed density , into a lorentzian manifold and using the spacetime causal order to define the partial order on the set of sprinkled points . we can pursue the question of whether causal continuity makes sense in a causal set by asking what a causal set that arises from sprinkling into causally discontinuous spacetimes looks like . a preliminary investigation suggests that causal discontinuity might manifest itself in the causal set in the existence of a point @xmath14 in the set and a single link from @xmath14 , the severing of which would result in a big change in the causal future or past of @xmath14 in the causal set . what is meant by big change and whether this can be made more concrete is left for future work . we would like to thank rafael sorkin and eric woolgar for valuable discussions .	the causal relation @xmath0 was introduced by sorkin and woolgar to extend the standard causal analysis of @xmath1 spacetimes to those that are only @xmath2 . most of their results also hold true in the case of spacetimes with degeneracies . in this paper we seek to examine @xmath0 explicitly in the case of lorentzian topology changing morse spacetimes containing isolated degeneracies . we first demonstrate some interesting features of this relation in globally lorentzian spacetimes . in particular , we show that @xmath0 is robust and that it coincides with the seifert relation when the spacetime is stably causal . moreover , the hawking and sachs characterisation of causal continuity translates into a natural expression in terms of @xmath0 for general spacetimes . we then examine @xmath0 in topology changing morse spacetimes both with and without the degeneracies and find further characterisations of causal continuity . @[email protected] , @[email protected] , @[email protected]
the calculation of the multiloop integrals is important for the applications of the perturbative quantum field theory . a number of the calculational methods has been developed for this purpose . they roughly fall in to two categories : direct and indirect methods . the former include feynman parameterization technique , mellin - barnes representation method , gegenbauer polynomials expansion in coordinate space . the latter methods make use of the equations which are satisfied by the integrals . the construction of these equations relies on the possibility to make the ibp reduction of the integrals under consideration to some finite set of the integrals , called `` master integrals '' . using this technique one can derive the differential equation for the integrals with several scales , or difference equation for the integrals with one scale . these equations determine the integrals up to the solution of their homogeneous parts . in ref . @xcite a method of multiloop integrals evaluation based on @xmath0 recurrence relations @xcite and @xmath0-analyticity was suggested ( dra method ) . since then it was applied to the calculation of various complicated integrals @xcite . in this contribution we review this method . assume that we are interested in the calculation of the @xmath1-loop integral depending on @xmath2 linearly independent external momenta @xmath3 . there are @xmath4 scalar products depending on the loop momenta @xmath5 : @xmath6 where @xmath7 , @xmath8 , and @xmath9 . the loop integral has the form@xmath10 where the scalar functions @xmath11 are linear polynomials with respect to @xmath12 . the functions @xmath11 are assumed to be linearly independent and to form a complete basis in the sense that any non - zero linear combination of them depends on the loop momenta , and any @xmath12 can be expressed in terms of @xmath11 . the integral @xmath13 can be considered as a value of function of integer @xmath14-dimensional vector @xmath15 . the set of points in @xmath16 ( and the corresponding set of integrals ) , having the same set of positive coordinates , form a _ sector_. thus the whole space is split into @xmath17 different sectors which can be conveniently labeled by their simplest elements . e.g. , @xmath18 denotes the sector in which all points have exactly two first coordinates positive . the points ( and the corresponding integrals ) in @xmath16 and the sectors form a partially ordered set with respect to the following relation . we say that the point @xmath19 is _ simpler _ than the point @xmath20 iff all coordinates , positive in @xmath19 , are also positive in @xmath20 , but not vice versa . it is convenient @xcite to introduce the operators @xmath21 which act on such functions as @xmath22 these operators respect the notion of sectors in a sense that the values of functions @xmath23 and @xmath24 in the point of some sector are expressed via the values of function @xmath25 in the same or simpler sectors . using these operators , we can express the integral in @xmath26 dimensions via the integrals in @xmath0 dimensions @xcite : @xmath27 ^{-1}}{\left ( \mathcal{d}-e - l+1\right ) _ { l } } p\left ( b_{1},\ldots , b_{n}\right ) j^{\left ( \mathcal{d}\right ) } \left ( \mathbf{n}\right)\ , , \label{eq : lowering}\\ j^{\left ( \mathcal{d}-2\right ) } \left ( \mathbf{n}\right ) & = ( \mu/2)^{l}q\left ( a_{1},\ldots , a_{n}\right)j^{\left ( \mathcal{d}\right ) } \left ( \mathbf{n}\right)\ , , \label{eq : raising}\ ] ] where @xmath28 for the euclidean / pseudoeuclidean case , and polynomials @xmath29 and @xmath30 are determined as @xmath31 these relations are the most useful when the integral in the left - hand side is master . then , making the ibp reduction of the right - hand side of eqs . , , one obtains difference equations for the master integrals . after this , we obtain a linear combination of the integrals of the same , or simpler , sectors as the integral in the left - hand side of eqs . , . we assume that the integrals of the the simpler sectors are already known at this stage , either by the same , or by some other method . the ibp reduction makes eqs . , equivalent , so in what follows we consider only the lowering dimensional recurrence relation , obtained from eq . we also change notations , writing @xmath32 where @xmath33 . if a sector of the integral in the left - hand side contains only one master integral , the general form of the equation is the following : @xmath34 where @xmath35 is some rational function of @xmath36 and @xmath37 is a non - homogeneous part constructed of the simpler master integrals in @xmath38 dimensions . if a sector contains several master integrals , it is convenient to consider them as a column - vector integral @xmath39 and to write the equations for them in a matrix form : @xmath40 where @xmath41 is some rational matrix ( i.e. , the matrix with coefficients being rational functions of @xmath36 ) . in order to write the general solution of this equation , we introduce the notion of indefinite sum and indefinite product . given the function @xmath42 , the _ indefinite sum _ @xmath43 is any function , satisfying the relation @xmath44 the indefinite sum @xmath43 is defined up to the addition of arbitrary periodic function of @xmath36 . if the function @xmath42 decreases faster than @xmath45 when @xmath46 and/or @xmath47 , one can write @xmath48 and/or @xmath49 where @xmath50 is arbitrary periodic function . the _ indefinite product _ @xmath51 is any function , satisfying the relation @xmath52 @xmath51 is defined up to the multiplication by arbitrary periodic function of @xmath36 . note that for any rational function @xmath42 one can express @xmath53 as a product of @xmath54-functions multiplied by arbitrary periodic function , see ref . we will determine the indefinite product @xmath55 also for the matrix functions @xmath56 as a matrix function , satisfying the relation @xmath57 obviously , this defines @xmath58 up to the multiplication from the right by arbitrary periodic matrix . in these notations one can write down the general solution of eq . in the form @xmath59 where @xmath60^{-1}$ ] is called a _ summing factor_. if the summand in eq . decreases faster than @xmath45 when @xmath46 and/or @xmath47 , one can use eqs . , to write @xmath61 note that the second term in the right - hand side is invariant with respect to the choice of the summing factor . thus , in order to find the integral , we need to fix the function @xmath62 for any choice of the summing factor . here we note the essential difference between the general solutions of _ differential _ and _ difference _ equations . the homogeneous part of the general solution of the differential equation is parameterized by finite number of constants , while the homogeneous part of the general solution of the difference equation is parameterized by the periodic function(s ) . while fixing the finite number of constants is a relatively simple task , fixing the periodic function may be quite difficult . the key point of the dra approach is the usage of the analytical properties of the integral in order to fix this function . let us rewrite eq . as @xmath63 suppose that we can show that the right - hand side of this equation is a meromorphic function of @xmath64 in the riemann sphere . then @xmath62 is determined , up to a constant , by the principal parts of its laurent series around singular points . the analysis of the analytical properties of the right - hand side of eq . can be performed in any vertical stripe of unit length in the complex plane of @xmath36 , as demonstrated in fig . 1 . there is no cut on the ray @xmath65 because the right - hand side is necessarily a periodic function of @xmath36 . the proper choice of this stripe ( _ basic stripe _ ) can simplify the evaluation of the integral . ( 340.60 , 149.59)(0,0 ) ( 0,0),title="fig : " ] ( 316.74,128.61 ) ( 121.56,134.57)@xmath36 ( 73.17,77.42 ) ( 98.70,77.36)@xmath66 ( 302.89,87.93)@xmath67 usually , one is interested in the analytical properties of the integral as a function of some external invariant . here we are interested instead in the analytical properties of the integral as a function of @xmath33 . it appears that the latter properties are much simpler than the former . let us consider the parametric representation of some @xmath1-loop integral in euclidean space with @xmath68 internal lines ( see , e.g. , ref . @xcite):@xmath69 ^{\nu l - i}}{\left [ p\left ( x\right ) \right ] ^{\nu\left ( l+1\right)-i}}.\label{eq : parametric}\ ] ] the polynomials @xmath70 and @xmath71 are determined by the graph . for our consideration it is important only that both these functions are nonnegative in the whole integration region,@xmath72 suppose that the parametric representation converges for all @xmath36 in some interval @xmath73 . then it is easy to see that it converges on the whole stripe @xmath74 . indeed , we can estimate @xmath75 therefore , eq . determines @xmath76 as a holomorphic function on the whole stripe @xmath77 . in fact , in ref . @xcite a more general fact was proved : the representation determines the meromorphic function on the whole complex @xmath36-plane . the determination of poles position can be automatized thanks to ` fiesta ` @xcite . as @xmath78 tends to @xmath79 while @xmath80 is fixed , eq . shows that @xmath81 thus , the parametric representation can be used for the determination of the analytical properties of the first term in the right - hand side of eq . . since the second term in the right - hand side of eq . is constructed of the integrals already known at this stage , we , indeed , can determine the position of the poles of function @xmath82 . in order to demonstrate how the representations obtained within the dra method can be treated numerically , let us consider the three - loop master integrals depicted in fig . [ fig : masters ] . direct application of the dra method gives the following results : @xmath83 the above results for the integrals @xmath84 and @xmath85 contain repeated sums of the form : @xmath86 from the viewpoint of numerical calculation , it is important that the dependence on the summation variables is factorized in the summand . we note that it is not the case for the mellin - barnes multiple sums . let us show that the calculation of such factorized sums can be performed without nested loops . indeed , let us define @xmath87 sequences @xmath88 as follows : @xmath89 it is easy to check that the last sequence , @xmath90 , converges to @xmath91 . obviously , the iterative calculation of these quantities can be organized in one loop . therefore , the complexity of the numerical calculation grows moderately with the complexity level of the integrals . the dependence of the calculation time on the order of the expansion , required precision , and complexity level of the integrals is shown in fig . [ fig : timing ] . we see that for all three cases the method shows quite moderate growth of the evaluation time . we have briefly reviewed the method of calculation of multiloop integrals based on the @xmath0-recurrence and @xmath0-analyticity . the method appears to be powerful enough to deal with the most complicated cases . this work was supported by rfbr ( grants nos . 10 - 02 - 01238 , 11 - 02 - 01196 , 11 - 02 - 00220 , 11 - 02 - 08379 ) and by federal special - purpose program `` scientific and scientific - pedagogical personnel of innovative russia '' . i appreciate the organizers support for the participation in the acat conference . i am grateful to y. schrder for drawing my attention to typos in the previous version of this paper . 99 lee r n 2010 _ nucl . _ b * 830 * 474 . [ arxiv:0911.0252 ] . tarasov o v 1996 _ phys . _ d * 54 * 6479 , hep - th/9606018 . lee r n , smirnov a v and smirnov v a 2010 _ jhep _ * 1004 * , 020 . [ arxiv:1001.2887 [ hep - ph ] ] . lee r n , smirnov a v and smirnov v a 2010 _ nucl . suppl . _ * 205 - 206 * , 308 . [ arxiv:1005.0362 [ hep - ph ] ] . lee r n and smirnov v a 2011 _ jhep _ * 1102 * , 102 . [ arxiv:1010.1334 [ hep - ph ] ] . lee r n and terekhov i s 2011 _ jhep _ * 1101 * , 068 . [ arxiv:1010.6117 [ hep - ph ] ] . lee r n , smirnov a v and smirnov v a 2011 _ eur . j. _ c * 71 * , 1708 . [ arxiv:1103.3409 [ hep - th ] ] . lee r n , smirnov a v and smirnov v a 2012 _ nucl . _ b * 856 * 95 . [ arxiv:1108.0732 [ hep - th ] ]	we review the method of the calculation of multiloop integrals suggested in ref.@xcite .
it is quite often that the @xmath7 transition - metal oxides with orbital degeneracy display an orbital ordering associated with lattice distortions ( of the jahn - teller type).@xcite in contrast , @xmath6 and @xmath8 transition - metal compounds rarely show this , @xcite mainly due to the delocalized character of the @xmath9 electrons . it was therefore a surprise when the @xmath6-orbital ordering and the orbitally driven spin - singlet dimerization were found in the ruthenate la@xmath1ru@xmath0o@xmath10.@xcite moreover , the fascinating superstructure of the spinel cuir@xmath0s@xmath1 ( ref . ) was explained in terms of the concept of orbitally - induced peierls state ( associated with the partially occupied ir @xmath8 orbitals).@xcite very recently,@xcite the magnetic behavior of the layered cs@xmath0agf@xmath1 was investigated , and the in - plane ferromagnetism ( fm ) of the @xmath11=1/2 ag@xmath4 ions was evidenced by magnetization and inelastic neutron scattering measurements , in sharp contrast to the structurally analogous high-@xmath12 cuprates . moreover , analysis of the structural data suggests that cs@xmath0agf@xmath1 is @xmath6 orbitally - ordered , and it was proposed that such orbital ordering may be the origin of the observed fm . cs@xmath0agf@xmath1 was first synthesized in 1974 by odenthal , paus , and hoppe,@xcite and it received , as well as other silver fluorides , attention due to pursuit of superconductivity in transition - metal compounds other than the cuprates.@xcite cs@xmath0agf@xmath1 is a structural analog of the parent compound of cuprate superconductors , la@xmath0cuo@xmath1 ( agf@xmath0 @xmath13 cuo@xmath0 sheet ) , and both have the @xmath11=1/2 divalent transition - metal cations ( ag@xmath4 @xmath5 @xmath13 cu@xmath4 @xmath14 ) and isoelectronic anions ( f@xmath15 @xmath13 o@xmath16 ) . it is well known that la@xmath0cuo@xmath1 is a mott insulator of charge - transfer type with two - dimensional heisenberg antiferromagnetism ( af ) . upon hole doping ( like ba substitution for la ) it is tuned into a superconductor , and , its high - temperature normal state being a paramagnetic metal , has the holes predominantly on the antibonding @xmath17 band.@xcite silver fluorides have similar electronic band structure , namely the ag - f @xmath17 antibonding character in the vicinity of the fermi level.@xcite moreover , the ag - f bonding is substantially covalent in the ag@xmath4 and ag@xmath18 fluorides , and thus holes might be doped into the f @xmath19 band , which is unprecedented in the chemistry of transition - metal compounds . as a result , an interesting new perspective is opened for the design of a novel family of metallic , perhaps even superconducting materials based on silver fluorides.@xcite subsequent to the finding of fm in cs@xmath0agf@xmath1,@xcite its electronic structure was studied by kasinathan @xmath20 @xmath21@xcite and dai @xmath20 @xmath22@xcite both using density - functional theory ( dft ) . kasinathan @xmath20 @xmath21 found the fm , due to importance of the ag - f covalency , to be itinerant in character , and substantial magnetic moment on the fluorine ions as result of a stoner instability enhanced by hund s coupling.@xcite dai @xmath20 @xmath21 suggested that the fm originates from the spin polarization of the doubly occupied @xmath23 band , which is induced by the @xmath24 orbital interaction through the ag - f - ag bridges.@xcite both groups of authors dealt with the tetragonal lattice@xcite and obtained a nearly half - metallic solution in the fm state , and used it to explain the origin of fm in cs@xmath0agf@xmath1.@xcite however , in reality cs@xmath0agf@xmath1 is an insulator with a relatively large energy gap ( lilac color ) and has a distorted orthorhombic lattice,@xcite and thus the above studies of the fm are insufficient . more recently , kan @xmath20 @xmath21@xcite performed pseudopotential dft calculations for the orthorhombic structure and they obtained an orbitally ordered solution ( in contrast to refs . and ) . however , their calculations underestimated the in - plane distortion and strongly overestimated both the in - plane and out - of - plane magnetic coupling strengths . moreover , they did not investigate the tetragonal structure . therefore , their results do not allow to reach solid conclusions . in the present work , we re - investigate theoretically the electronic structure of cs@xmath0agf@xmath1 and the origin of its fm , by carrying out a comparative study of the orthorhombic and tetragonal lattices of cs@xmath0agf@xmath1 using dft total - energy calculations . we find that the orthorhombic phase is more stable and it is insulating with suppressed ag - f bandwidths . the inherent lattice distortion is accompanied by the ag @xmath6-orbital ordering , and this orbital ordering readily accounts for the observed fm ( similar results were also obtained recently by hao @xmath25@xcite ) . this picture holds in the presence of a moderate correlation of the ag @xmath6 electrons . the resulting picture of an orbital ordering and of a ferromagnetic insulating state of cs@xmath0agf@xmath1 is thus very similar to that of an isoelectronic and isostructural compound k@xmath0cuf@xmath1.@xcite our electronic structure calculations are performed by using the all - electron full - potential augmented plane wave plus local orbital method.@xcite the generalized gradient approximation@xcite ( gga ) to density - functional theory is adopted.@xcite we took the neutron diffraction structure data measured at 6 k.@xcite as seen below , the electronic structure and magnetic properties of cs@xmath0agf@xmath1 are determined by an orbital ordering , and the orbital ordering is the most relevant to the local distortion of the agf@xmath26 octahedra . to account for a possible lattice distortion connected with an orbital ordering , we carried out an optimization of the atomic positions , keeping the unit - cell parameters fixed and relaxing the atomic coordinates . the obtained values of atomic positions and bond - lengths are in good agreement with the experimentally determined values , see below . based on that , in the further calculations we use the experimental structure data . the cs @xmath27 ( @xmath28 ) , ag @xmath29 ( @xmath30 ) , and f @xmath31 ( @xmath32 ) are treated as valence ( semicore ) states . the muffin - tin sphere radii are chosen to be 2.8 , 2.2 , and 1.7 bohr for cs , ag , and f atoms , respectively . the cut - off energy of 16 ryd is set for the plane - wave expansion of the interstitial wave functions . the sufficiently dense @xmath33 mesh , e.g. , 10@xmath3410@xmath344 is used for integration over the brillouin zone of the primitive cell with the experimental orthorhombic lattice constant @xmath35=6.4345 , @xmath36=6.4390 , and @xmath37=14.1495 .@xcite the present set - up ensures a sufficient accuracy of the calculations . agf@xmath1 in the fm orthorhombic phase from gga . this insulating solution has a small band gap of 0.2 ev . the fermi level is set at zero energy . solid red ( dashed blue ) lines depict the spin - up ( down ) states . the panels show , from top to bottom , the total dos per formula unit and orbital - resolved dos . the state around 7 ev comes from the cs @xmath38 orbital . for the ag @xmath6-resolved dos , the local coordinate system is chosen in such a way that the @xmath39 ( @xmath40)-axis is along the in - plane longer ( shorter ) ag - f bond . the ag@xmath4 @xmath2 hole state , and the corresponding @xmath3 hole state of the nearest neighbor ag@xmath4 ions form the in - plane orbital ordering ( see also fig.2).,width=302 ] /@xmath41 ) of the spin density in the the @xmath42 ( upper panel ) and @xmath43 ( lower panel ) planes , through the agf@xmath26 octahedra of cs@xmath0agf@xmath1 in the fm insulating orthorhombic phase obtained by gga . it evidences the @xmath2/@xmath3 hole - orbital ordering of the ag@xmath4 ions in the @xmath44 basal plane . see also fig . 1 . note also a relatively strong spin polarization of the f atoms ( see more in the text ) . , width=226 ] fig . 1 shows the total and the orbital - resolved density of states ( dos ) for the fm orthorhombic phase of cs@xmath0agf@xmath1 obtained in gga . it is an insulator with a gap of about 0.2 ev , but neither a metal ( nor a half - metal ) as predicted before for the tetragonal phase@xcite ( see also below ) nor a semimetal as obtained for the orthorhombic phase.@xcite the strong hybridization between ag @xmath6 and f @xmath19 levels ( apical f@xmath45 and planar f@xmath46 ) , both centered around 3 ev below the fermi level , is obvious in both the valence and the conduction bands , and this gives rise to the large bonding antibonding splitting of about 2 ev for the ag @xmath47 ( @xmath42 , @xmath43 , and @xmath44 ) orbitals and even a larger one for the @xmath48 orbitals . the @xmath47 bands are fully occupied and the @xmath48 bands are three - fourth occupied ( with one hole left on them ) as expected for the formal ag@xmath4 ( @xmath5 ) ions . actually , the one hole spreads , due to the strong ag f covalency , over the six fluorine atoms of the agf@xmath49 octahedron . as a result , the hole state consists of approximately 60% of the ag @xmath6 and 40% of the @xmath19 of the four f atoms ( see the doss above the fermi level in the second , third , and fourth panels of fig . 1 ) , according to the stoichiometry of cs@xmath0agf@xmath1.@xcite this is also reflected by the magnitude of local spin magnetic moments within each muffin - tin sphere , 0.560 @xmath50/ag , 0.105 @xmath50/f@xmath45 , and 0.112 @xmath50/f@xmath46 , see also table i. the hole states have alternating @xmath2 and @xmath3 symmetry shown in figs . 1 and 2 . [email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected] + & & & & gga&&&&&gga+@xmath51 & & + & & @xmath52 & gap&ag&f@xmath45&f@xmath46&@xmath52 & gap&ag&f@xmath45&f@xmath46 + orthorhombic&fm&0&0.2&0.560&0.105&0.112&0&1.0&0.600&0.099&0.097 + & af&34&0.2&@xmath530.457&@xmath530.098&@xmath530.102&13&1.0&@xmath530.560&@xmath530.094&@xmath530.092 + tetragonal&fm&27&@xmath54&0.558&0.103&0.117&83&@xmath54&0.584&0.102&0.084 + & af ( nm)&39 ( 36)&@xmath54&@xmath530.053&@xmath530.015&0&90&0.4&@xmath530.547&@xmath530.111&0 + agf@xmath1 in the fm tetragonal phase from gga . it is nearly half - metallic , and the wide in - plane @xmath23 band and the small @xmath48-level splitting are obvious.,width=302 ] thus , even the gga calculations , not including correlation effects , already give an orbital ordering of the same type as that in k@xmath0cuf@xmath1,@xcite which in a picture of localized electrons would immediately give in - plane ferromagnetism , according to goodenough - kanamori - anderson rules . this orbitally - polarized hole state in the insulating solution is associated with the large distortion of the ag - f bonds of 0.33 ( 2.44 along the @xmath39-axis versus 2.11 along the @xmath40-axis , and 2.11 along the @xmath55-axis , all in a local coordinate system),@xcite which leads to a pronounced crystal - field splitting and suppressed bandwidths . as a result , the antibonding @xmath2 band lies about 1 ev above the antibonding @xmath56 band , and then an exchange splitting of the @xmath2 band gives rise to the small band gap and the @xmath2 hole state . thus , such a cooperative distortion accommodates the @xmath2/@xmath3 hole - orbital ordering in the agf@xmath0 basal plane , giving the observed fm . also , our gga calculations confirm this by showing that the in - plane ( intralayer ) fm state is indeed more stable than the in - plane af state by 34 mev per formula unit ( f.u . ) , see table i. we note that the interlayer magnetic coupling is too weak ( 0.004 mev in energy scale@xcite ) to be captured by the present calculations , which show practically degenerate interlayer fm and af states within the error bar of 1 mev / f.u . actually , the experimental distorted lattice turns out to be in the vicinity of the theoretical equilibrium state after atomic relaxation . the structural optimization shows that cs would shift by only 0.005 and the ag f@xmath45 bond would be elongated by 0.041 and the in - plane longer ( shorter ) ag f@xmath46 bond would be shortened ( elongated ) by 0.018 , as compared with the experimentally observed structure . as a result , the large in - plane distortion of about 0.3 , found experimentally , @xcite is well reproduced theoretically . moreover , the fm orthorhombic phase turns out to be more stable than a fm tetragonal phase by 27 mev / f.u . ( which , as well as the following gga+@xmath51 result , implies a stability of the orthorhombic phase above room temperature as observed@xcite ) . we checked it by performing the calculations for the average tetragonal structure . the tetragonal phase is constructed in such a way that the lattice volume , @xmath37-axis constant , and positions of cs and f@xmath45 atoms are the same as in the orthorhombic phase we calculated above , and that the basal plane is a regular agf@xmath0 square lattice . this fm tetragonal phase shows the same electronic - structure features ( see , e.g. , near half - metallicity in fig . 3 ) as previously found,@xcite since the structural parameters used in refs . and are very close to what we used for this hypothetical phase . the regular agf@xmath0 square sheet would produce a wide @xmath23 band with a bandwidth of about 3 ev , and the compressed distortion of the agf@xmath26 octahedron ( out - of - plane ag f bond length of 2.12 @xmath13 in - plane one of 2.29 ) is not large enough to generate a crystal - field splitting in excess of the bandwidths to open a gap . ( note also that the compressed octahedron is extremely rare in jahn - teller insulators with @xmath48 degeneracy.@xcite ) actually , the crystal field splitting would make the @xmath23 level lie below @xmath57 by only about 0.2 ev . as a result , both the minority - spin @xmath23 and @xmath57 bands cross the fermi level . the total spin moment of 0.98 @xmath50/f.u . ( close to integer 1 @xmath50 ) , in the tetragonal phase would spread over ag ( 0.558 @xmath50 ) , f@xmath45 ( 0.103 @xmath50 each ) , and f@xmath46 ( 0.117 @xmath50 each ) . agf@xmath1 in the fm orthorhombic phase from gga+@xmath51 with @xmath51=3 ev . the insulating gap is increased up to 1.0 ev , compared with gga ( see fig . 1 ) . see other notes in fig . 1 caption.,width=302 ] as we saw above , cs@xmath0agf@xmath1 is insulating in the fm - ordered orthorhombic phase , even in gga , see fig . 1 . a moderate electron correlation may be present for the ag@xmath4 @xmath6 electrons , and thus we also carried out calculations using gga+@xmath51 method@xcite to include the on - site coulomb interaction for both the orthorhombic and tetragonal phases . in fig . 4 we show the dos results for the actual fm orthorhombic structure , given by the gga+@xmath51 calculation with an effective @xmath51=3 ev for the ag@xmath4 @xmath6 electrons . the insulating gap is increased up to 1.0 ev ( 1.4 ev when @xmath51=4 ev ) . the hubbard @xmath51 pushes the occupied @xmath6 levels downwards so that their center of gravity is now lower in energy than the f @xmath19 levels . as a result , the lower - lying bonding states have larger ag @xmath6 character than the higher - lying antibonding states , in contrast to the gga results shown in fig . 1 . again , the @xmath2/@xmath3 hole - orbital ordering is obvious , and the orbital polarization of the @xmath48 states is enhanced . the local spin moments are 0.600 @xmath50/ag , 0.099 @xmath50/f@xmath45 , and 0.097 @xmath50/f@xmath46 . note that in the gga+@xmath51 with @xmath51=3 ev ( @xmath51=4 ev ) , the in - plane fm state is more stable than the in - plane af state by 13 mev ( 10 mev ) per formula unit . using a simple heisenberg model for the spin-1/2 square lattice , @xmath58 ( counting twice the magnetic exchange per spin pair ) , the nearest - neighbor intersite fm exchange integral is estimated to be 6.5 mev ( 5 mev ) when @xmath51=3 ev ( @xmath51=4 ev ) . this value , reduced from the gga estimate of 17 mev , is well comparable to the experimental one of 3.85 mev.@xcite the gga+@xmath51 ( @xmath51=34 ev ) calculations for the tetragonal structure still give a metallic solution for the in - plane fm state with the broad @xmath48 bands , but open an insulating gap of 0.40.7 ev for the in - plane af state with suppressed bandwidths ( not shown ) . with the gap opening , this insulating af state becomes stable with a large ag spin moment , in contrast to the gga results ( see table i ) . note however that also in gga+@xmath51 both the af and fm tetragonal phases have much higher energy than the orthorhombic phases , by more than 80 mev / f.u .. thus , once again these calculations show that the orthorhombic phase is indeed more stable than the tetragonal phase , with the orbitally ordered fm state being the ground state . all this confirms the recent experimental finding that the layered cs@xmath0agf@xmath1 is indeed stabilized in the insulating orthorhombic phase.@xcite summarizing , by means of gga and gga+@xmath51 band structure calculations , we find that the layered cs@xmath0agf@xmath1 , a structural analog of cuprates , is stabilized in an insulating orthorhombic phase rather than in a metallic tetragonal phase . the intrinsic lattice distortion of the orthorhombic phase is accompanied by the @xmath2/@xmath3 hole orbital ordering , which readily accounts for the observed in - plane ferromagnetism . the present calculations confirm the recent experiments@xcite and lead us to the conclusion that cs@xmath0agf@xmath1 is indeed an orbitally - ordered ferromagnetic insulator , in strong contrast to the antiferromagnetic nature of the parent high-@xmath12 cuprates , but in close analogy to k@xmath0cuf@xmath1 . _ acknowledgments : _ we are grateful to d. a. tennant for useful discussion , and to x. hao for informing us of their results.@xcite this work was supported by the deutsche forschungsgemeinschaft through sfb 608 and by the european project comephs . h. wu , z. hu , t. burnus , j. d. denlinger , p. g. khalifah , d. g. mandrus , l .- y . jang , h. h. hsieh , a. tanaka , k. s. liang , j. w. allen , r. j. cava , d. i. khomskii , and l. h. tjeng , phys . * 96 * , 256402 ( 2006 ) . note that the f @xmath19 content of the hole state would be even larger if one would count the number of holes within the volumes with the effective ionic radii of f@xmath15 ( 1.3 ) and ag@xmath4 ( 0.94 ) [ r. d. shannon , acta cryst . a32 , 751 ( 1976 ) ] instead of the present set of the muffin - tin sphere radii .	we found , using density - functional theory calculations within the generalized gradient approximation , that cs@xmath0agf@xmath1 is stabilized in the insulating orthorhombic phase rather than in the metallic tetragonal phase . the lattice distortion present in the orthorhombic phase corresponds to the @xmath2/@xmath3 hole - orbital ordering of the ag@xmath4 @xmath5 ions , and this orbital ordering leads to the observed ferromagnetism , as confirmed by the present total - energy calculations . this picture holds in the presence of moderate @xmath6-electron correlation . the results are compared with the picture of ferromagnetism based on the metallic tetragonal phase .
yield stress materials denote assemblies of mesoscopic constituents such as colloids , droplets or bubbles that display macroscopic properties intermediate between those of fluid and solid @xcite . at rest , or under low stresses , yield stress materials display a solid - like behaviour whereas they flow like liquids above a certain threshold referred to as the yield stress @xcite . their solid - like behaviour at rest originates either from a densely packed microstructure composed of soft objects @xcite such as in dense emulsions @xcite , jammed microgels @xcite , etc . , or from the existence of weak attractive interactions that bind the constituents together and result in the formation of a sample - spanning network @xcite . in the latter case , solid - like properties emerge even at low packing fractions . this defines a category referred to as _ attractive gels _ @xcite which encompasses various systems such as clay suspensions @xcite , carbon black gels @xcite and colloid - polymer mixtures @xcite . all sorts of colloids and larger particles of different shapes and attraction potential also comply with this definition @xcite . over the past ten years , the rheological behaviour of attractive gels has proved to be by far one of the most challenging to understand among non - newtonian fluids . in particular , their mechanical properties at rest are strongly time - dependent : attractive gels display a reversible aging dynamics driven by the weak attractive forces between its constituents @xcite and that can be reversed by shear . as a result , the gel elastic properties display a slow logarithmic or weak power - law increase with time @xcite , up to complete demixing of the system , which is historically referred to as syneresis @xcite . to make matter worse , particles are often denser than the surrounding fluid which fosters syneresis and may trigger the collapse of the gel @xcite . such density mismatch is further suspected to influence the behaviour of these systems under flow , although clear experimental evidence is still lacking @xcite . nonetheless , it is now well established that the behaviour of attractive gels under external shear involves heterogeneous flows that are highly sensitive to preshear history , boundary conditions , and/or finite size effects . for instance , one can emphasize the case of laponite suspensions which steady - state flow properties were shown to be influenced by the nature of the boundaries : under external shear , smooth walls lead to the complete fluidization of the gel and to linear velocity profiles , while rough boundary conditions allow the growth of a steady shear band @xcite . moreover , as an illustration of the impact of confinement on flows of attractive gels , one can mention the spectacular shear - induced structuration observed at moderate shear rates and reported in silica suspensions @xcite , gels of anisotropic particles @xcite , attractive emulsions @xcite and carbon black gels @xcite . in such cases , the gel rearranges to form a striped pattern of logrolling flocs aligned along the vorticity direction , which origin and formation mechanism are still highly debated @xcite . beyond the effects of bounding walls and confinement , attractive interactions alone are also responsible for long - lasting transients under external shear . on the one hand , experiments performed under constant @xcite and oscillatory @xcite stress reveal that the fluidization process of attractive gels initially at rest , may take up to tens of hours . experiments coupled to velocimetry have revealed that such a process is mostly activated , as evidenced by the arrhenius - like dependence of the fluidization time with the applied shear stress @xcite , and that it is strongly heterogeneous in both the vorticity and the flow directions @xcite . on the other hand , attractive gels show an overshoot in the stress response to shear startup experiments . such behaviour , which strongly depends on the preshear history , corresponds to the orientation and subsequent rupture of the gel microstructure into clusters @xcite . beyond the yield point , attractive gels either display homogeneous or shear - banded flows depending on the applied shear rate and on the boundary conditions @xcite . however , only a handful of studies have investigated the influence of the sample age , i.e. the duration separating preshear from the start of the experiment , on such shear - rate - induced fluidization scenario @xcite . recently , martin and hu have shown on laponite suspensions @xcite that aged - enough samples tend to exhibit long - lasting though transient shear - banding , and that shear - banding may also be trapped by the rapid aging of the non - flowing band and become permanent . the latter scenario is remarkable as it strongly differs from the classic shear - banding scheme which relies on the mechanical instability of the sample under scrutiny @xcite . this study highlights the interplay between the sample age and the shear and strongly urges to investigate the impact of sample age on the shear - induced fluidization scenario in other attractive gels . to summarize , attractive gels neither entirely behave as solids nor as fluids over a wide range of timescales . in the landscape of non - newtonian fluids , they define a rather singular category of materials rightfully referred to as the twilight zone " in the early classification established by l. bilmes @xcite , that was recently adapted to complex fluids by g.h . mckinley @xcite . we barely start to understand the ( transient ) rheology of attractive gels and more experiments are obviously needed to shed some light on the heterogeneous flows of such highly time - dependent materials . in the present manuscript we report spatially resolved data on the fluidization dynamics of an attractive gel composed of weakly attractive non - brownian particles . velocimetry performed in a concentric cylinder geometry simultaneously to shear startup experiments reveals that the steady - state behaviour is a subtle function of both the time @xmath0 during which the system was left to age before the beginning of the experiment , and of the value @xmath1 of the applied shear rate . extensive experiments allow us to build a state diagram of steady - state flows in the ( @xmath1 , @xmath0 ) plane . two distinct regions roughly emerge : ( @xmath3 ) homogeneous flows for shear rates larger than a critical value @xmath2 that weakly increases with the aging time , and ( @xmath4 ) steady - state shear banding elsewhere . as a key result of this work , the complete fluidization observed in the upper region of the state diagram involves a transient shear band that is progressively eroded through a series of dramatic fluidization events . these avalanche - like processes show as large peaks in the temporal evolution of both the global shear stress and the slip velocity measured at the moving wall of the shear cell . as further confirmed by two - dimensional ultrasonic imaging , this fluidization process is spatially heterogeneous and may occur at different locations along the vorticity direction . finally , for a range of parameters ( @xmath1 , @xmath0 ) in the vicinity of the boundary between the two main regions of the state diagram , we observe large fluctuations in the stress and slip velocity signals , although the system does not reach complete fluidization . such avalanche - like events are strongly coupled to variations in both the width of the shear band and the slip velocity . the present work offers to our knowledge among the first experimental evidence of local avalanche - like fluidization events in a weak attractive gel under shear . it also provides an extensive data set to test the relevance of the flow stability criteria for shear banding @xcite and stands as a new challenge for spatially resolved models @xcite . the experimental setup allows us to perform time - resolved velocimetry simultaneously to standard rheometry . the rheological measurements are performed in a polished taylor - couette ( concentric cylinder ) cell made of plexiglas ( gap @xmath5 mm ) in which the inner mooney - couette cylinder of angle @xmath6 , height 60 mm and radius 23 mm is connected to a stress - controlled rheometer ( arg2 , ta instruments ) and positioned at @xmath7 m from the bottom of the outer cylinder . a solvent trap located at the top of the rotor and a homemade lid are used to prevent evaporation efficiently up to about 9 hours . the taylor - couette cell is embedded in a water tank connected to a thermal bath which allows us to keep the sample at a constant temperature @xmath8c . local velocity profiles across the gap of the taylor - couette cell are recorded simultaneously to the global rheology by means of two different ultrasonic probes immersed in the water tank , which also ensures acoustic coupling to the shear cell . the first ultrasonic probe is a single high - frequency focused transducer that allows us to record the azimuthal velocity @xmath9 as a function of the radial position @xmath10 across the gap at the middle height of the shear cell i.e. at 30 mm from the bottom . full technical details on this one - dimensional ultrasonic velocimetry ( 1d - usv ) technique can be found in a previous publication @xcite . the second ultrasonic probe consists of a linear array of 128 transducers placed vertically at about 15 mm from the cell bottom . this transducer array is 32 mm high and gives access to images of the azimuthal velocity as a function of the radial position @xmath10 and vertical position @xmath11 over about 50% of the cell height . this two - dimensional ultrasonic velocimetry ( 2d - usv ) technique is thoroughly described in ref . both devices can be used simultaneously and roughly face each other in the water tank , i.e. they are separated by an azimuthal angle of about @xmath12 . while the 1d - usv setup has the advantage of a better spatial resolution ( about 40 @xmath13 m against 100 @xmath13 m ) , only the 2d - usv setup allows us to detect and monitor the presence of flow heterogeneities along the vorticity direction . both velocimetry techniques require that the ultrasonic beam crossing the gap of the shear cell is backscattered either by the fluid microstructure itself or by acoustic tracers added during sample preparation when the system is acoustically transparent @xcite . here , we shall emphasize that the microstructure of the sample further detailed below conveniently backscatters ultrasound in the single scattering regime , which allows us to monitor the fluid velocity in a fully non - invasive way . vs time @xmath14 after a preshear at @xmath15 s@xmath16 for 2 min . inset : same data set in semilogarithmic scales . measurements performed under oscillatory shear stress in the linear regime with frequency 1 hz and stress amplitude 0.05 pa . ( b ) flow curve @xmath17 vs @xmath1 obtained by first decreasing @xmath1 continuously from 10@xmath18 to 10@xmath19 s@xmath16 in 75 logarithmically spaced points of duration @xmath20 s ( @xmath21 ) and then increasing @xmath1 over the same range ( @xmath22 ) . ] our ludox gels are prepared following the same recipe as that used by mller _ . @xcite . as we shall show below , the system is composed of non - brownian particles made of permanently fused silica colloids . these particles are themselves reversibly aggregated in brine due to van der waals forces leading to a space - spanning gel with solid - like properties at rest . a stable suspension of silica colloids ( ludox tm-40 , sigma - aldrich , 40% wt . in silica colloids ) of typical diameter 20 nm [ see fig . [ fig1](a ) for a scanning electron microscopy ( sem ) image of a dilute sample ( supra 55 , vp zeiss ) ] and of @xmath23 is first poured without any further purification into a 10% wt . deionized aqueous solution of sodium chloride ( merck millipore ) up to a mass ratio ludox : nacl of 6:13 , corresponding to a final volume fraction of 7% in silica colloids and a final ph of @xmath24 . the mixture , which instantaneously becomes white and optically opaque , is then shaken intensely by hand for 2 min and left at rest for at least 15 h before being studied . such a drastic change in the sample turbidity strongly suggests the rapid formation of aggregates at the micron - scale . indeed , direct observations using different techniques confirm the existence of a much coarser microstructure than the initial nanometric silica colloids . on the one hand , sem images of a dried droplet extracted from a fresh sample that has been previously diluted in a nacl solution unveil the presence of particles which size ranges from a few microns up to a hundred microns [ fig . [ fig1](b ) ] . on the other hand , bright - field microscopy images ( eclipse ti , nikon ) of the sample neither altered nor diluted further confirm the existence of these micron - sized particles [ fig . [ fig1](c)-(e ) ] , which are stable in time and robust to repeated shear , as confirmed by similar observations performed on samples submitted to different shear histories ( images not shown ) . to account for the formation of such a large scale microstructure that to our knowledge has not been reported in the literature previously @xcite , we propose the following scenario . above ph=7 , silica colloids are negatively charged and bear silanol ( si@xmath25oh ) and dissociated silanol groups that are poorly hydrated @xcite . in most previous works , nacl is added in a relatively small amount ( typically 0.050.5 m ) such that electrostatic repulsion is screened leading to the slow reversible aggregation of individual colloids up to the formation of a colloidal gel @xcite . here , we add a much larger amount of salt ( 1.2 m ) to the colloidal suspension , which leads to an ion exchange where protons are replaced by sodium ions @xcite . the loss of hydrogen bonding between the colloids and the solvent triggers the fast and irreversible aggregation of the silica colloids through the formation of interparticle siloxane bonds @xcite , resulting in the formation of the non - brownian particles described above . finally , these non - brownian particles also aggregate reversibly due to van der waals forces and form a space - spanning network , i.e. a gel , which mechanical behaviour is studied below . note that such microstructure scatters ultrasound efficiently , allowing us to use both 1d- and 2d - usv without requiring any seeding of the sample with tracer particles . the rheological features of the gel are displayed in fig . a strong preshear of @xmath15 s@xmath16 applied during 2 min fully fluidizes the system , which quickly rebuilds once the preshear is stopped and subsequently shows pronounced aging . indeed , small amplitude oscillatory shear reveals that the elastic modulus @xmath26 becomes larger than the viscous modulus @xmath27 within about 20 s [ see fig . 1 in the esi ] and that @xmath26 further increases logarithmically over more than 2 h [ fig . [ fig2](a ) ] . such aging dynamics are reproducible for a given preshear protocol and lead to the formation of a solid - like gel . the latter shows an elastic modulus that is nearly frequency independent and a critical yield strain of about 7% that weakly depends on the sample age [ see figs . 2 and 3 in the esi ] . such a solid - like behaviour also reflects in the nonlinear rheology of the gel . [ fig2](b ) shows the flow curve @xmath17 vs @xmath1 measured by sweeping down the shear rate from @xmath28 to @xmath29 s@xmath16 and back up . the flow curve shows an apparent yield stress of a few pascals and displays a complex non - monotonic shape together with strong hysteresis . velocity profiles recorded simultaneously to the up and down shear rate sweeps shown in fig . [ fig2 ] reveal that wall slip and heterogeneous flows are involved over large ranges of shear rates , below 20 s@xmath16 during the decreasing ramp and up to about 200 s@xmath16 during the increasing ramp [ see fig . 4 in the esi ] . in particular , the large stress peak observed in fig . [ fig2](b ) at @xmath30 s@xmath16 while ramping up the shear rate is the signature of the partial fluidization of the sample which moves as a plug and totally slips at both walls for @xmath31 s@xmath16 , before finally starting to flow homogeneously on the reversible branch at high shear rates i.e. above about 200 s@xmath16 . furthermore , in fig . [ fig2](b ) , the shear stress @xmath17 shows a minimum in between @xmath32 s@xmath16 and 40 s@xmath16 on the decreasing shear rate sweep , which hints at the existence of a critical shear - rate @xmath33 below which no homogeneous flow is possible @xcite . the latter result is confirmed by performing steady - state measurements at constant applied shear rate starting from the liquid state , i.e. on a fully fluidized sample , in order to avoid long - lasting transients that go along with shear startup experiments on a sample at rest , and which are at the core of section [ results ] . here , discrete shear rates of decreasing values ranging from 500 s@xmath16 down to 0.1 s@xmath16 are successively applied for at least 300 s each . the flow remains homogeneous down to @xmath34 s@xmath16 below which the sample exhibits an arrested band [ fig . [ fig2b](a ) ] . such a value of @xmath33 is comparable to the one reported in a previous work on very similar ludox gels @xcite . moreover , as the shear rate is decreased below @xmath33 , the size of the fluidized band @xmath35 decreases roughly linearly with @xmath1 [ fig . [ fig2b](b ) ] in agreement with the classical `` lever rule '' @xcite . the deviation of @xmath35 from a straight line could be related to the wall slip that is present at the rotor . , 30 , 20 , 10 , 5 and 1 s@xmath16 from top to bottom . the parameter @xmath10 denotes the distance to the rotor . these data have been recorded starting from large shear rates ( 500 s@xmath16 ) and decreasing @xmath1 by successive steps of long enough duration to achieve a steady state at every imposed shear rate . ( b ) size of the fluidized band @xmath35 normalized by the gap width @xmath36 vs the applied shear rate @xmath1 . the dashed line corresponds to the standard `` lever rule '' with @xmath37 s@xmath16 . ] to conclude this subsection , complex cycles of hysteresis as the one reported in fig . [ fig2 ] have been also reported for numerous other attractive gels including carbon black gels @xcite and clay suspensions @xcite . although significant progress has been made in extracting quantitative information from hysteresis loops @xcite , we rather chose to focus our study on shear startup experiments in order to fully decouple the fluid dynamics from any time - dependent effect related to the experimental protocol . the present experiments are thus all performed on a sample prepared in a solid - like state , so as to investigate the shear - induced fluidization scenario of the non - brownian particle gel . in practice , after preparing the gel in a well - defined and reproducible initial state thanks to preshearing , we monitor the stress response of the material to a constant shear rate over long durations . the results are discussed in section [ results ] . prior to any shear startup experiment , the sample is presheared at @xmath15 s@xmath16 for 2 min in order to erase any previous shear history @xcite . we check systematically that the velocity profiles during the preshear step become homogeneous across the gap within a duration shorter than 2 min . the system is then left to rebuild for a duration @xmath0 that ranges between 1 and 100 min , and during which we apply small amplitude oscillatory shear stress ( stress amplitude @xmath38 pa , frequency @xmath39 hz ) to monitor the evolution of the gel linear viscoelastic properties and make sure that for a given value of @xmath0 , the gel reaches an initial state that is reproducible from one experiment to another . finally , a constant shear rate @xmath1 is applied to the material over a long duration ( @xmath40 s ) while we monitor the stress response together with local velocity profiles . we first fix the duration between the preshear and the start of the experiment to @xmath41 min and discuss the material response for shear startup experiments performed at different shear rates . for shear rates lower than about @xmath42 s@xmath16 , the stress relaxes smoothly towards a constant value as illustrated in figure [ fig3](a ) for an experiment performed at @xmath43 s@xmath16 . velocity profiles acquired simultaneously reveal that the material remains at rest near the stator , while it flows in the vicinity of the rotor where it shows strong wall slip [ fig . [ fig3](b ) ] : the sample displays steady - state shear banding . for larger shear rates , one observes a completely different scenario , both from global and local measurements . the stress displays a series of spike - like relaxation events , during which stress increases of small amplitude are followed by large drops [ fig . [ fig3](a ) ] . local measurements show that the gel flows heterogeneously . however , the shear band is only transient and the system always ends up being homogeneously sheared at steady state [ fig . [ fig3](c)(e ) ] . qualitatively , the width of the shear band increases while strong wall slip is observed at the rotor , until the whole gap is fluidized and wall slip becomes negligible . this scenario is robustly observed at different shear rates and full fluidization occurs sooner for larger shear rates [ fig . [ fig3](a ) ] . a quantitative analysis of the velocity profiles is further proposed in section [ localanalysis ] . , applied shear rate @xmath44 ) plane . in steady state , the gel may either be fully fluidized ( , blue region ) or display shear banding . steady shear banding is represented in ( ) symbol which size encodes the portion of the gap that is being sheared . unsteady banding , which denotes flows where the band width and the slip velocity at the rotor display significant fluctuations , is represented by ( ) . ] to evidence the impact of the aging time @xmath0 on the material response , figs . [ fig3](f)(j ) show similar experiments performed for different @xmath0 at the same shear rate @xmath43 s@xmath16 . for long waiting times between the preshear and the start of the experiments , e.g. @xmath45 min , one observes a smooth stress relaxation here again associated with steady shear - banded velocity profiles [ fig . [ fig3](f ) and ( g ) ] . decreasing @xmath0 to 30 or 5 minutes leads to a gel of lower elastic modulus . local measurements further reveal that these weaker gels go through a transient shear - banding regime and that in both cases the steady state is homogeneous [ fig . [ fig3](h ) and ( i ) ] . here , unlike the case of transient shear banding reported in fig . [ fig3](a ) for large shear rates in samples left to age for @xmath41 min where fluidization corresponds to a long series of successive stress relaxations , fluidization proceeds in a single stress drop together with small noisy fluctuations [ fig . [ fig3](f ) ] . finally , one observes that very young gels ( @xmath46 min ) barely show any heterogeneous velocity profile during startup flow and reach a homogeneous steady state within a few tens of seconds [ fig . [ fig3](j ) ] . in summary , the longer the sample ages after preshear , the more likely it is to exhibit a long - lasting heterogeneous fluidization scenario or to display steady shear banding . s@xmath16 and @xmath47 min . ( a ) shear stress @xmath17 ( ) and slip velocity @xmath48 ( ) vs time . ( b ) width @xmath35 of the fluidized shear band normalized by the gap width @xmath36 vs time . ( c ) local shear rate within the shear band ( ) and global shear rate ( ) vs time . the horizontal dotted line indicates the shear rate applied by the rheometer . in ( a)-(c ) , the vertical dashed line indicates the fluidization time @xmath49 . ] s@xmath16 and @xmath47 min . ( a ) shear stress @xmath17 ( ) and slip velocity @xmath48 ( ) vs time . ( b ) width @xmath35 of the fluidized shear band normalized by the gap width @xmath36 vs time . ( c ) local shear rate within the shear band ( ) and global shear rate ( ) vs time . the horizontal dotted line indicates the shear rate applied by the rheometer . ] to go one step further and get an overall picture of the material response , we have performed extensive shear startup experiments by varying systematically both the aging time ( @xmath50 min ) and the applied shear rate ( @xmath51 s@xmath16 ) . the entire data set is summarized in the flow state diagram pictured in fig . [ fig4 ] , where the different symbols code for the three different scenarios that we may distinguish at the end of each startup experiment lasting typically 1500 to 3600 s. in steady state , the attractive gel may either fully fluidize after a transient phase possibly involving strong fluctuations ( in fig . [ fig4 ] ) , as observed in the upper part of the state diagram for large shear rates quite independently of the sample age , or exhibit shear banding . the boundary between these two regimes defines a critical shear rate @xmath2 that depends in a nonlinear fashion on the sample age @xmath0 . furthermore , we can discriminate between two types of behaviour when shear banding is observed . steady _ shear banding ( in fig . [ fig4 ] ) , for which both the width @xmath35 of the shear band and the slip velocity @xmath48 at the rotor display negligible fluctuations in steady state , is observed in particular at low aging times @xmath52 min and under low enough shear rates , typically below 100 s@xmath16 . in that case , the width of the arrested band may decrease ( @xmath46 s ) or remain constant ( @xmath53 s and 100 s ) for increasing shear rates . second , we have also observed _ unsteady _ shear banding ( in fig . [ fig4 ] ) . in this case both global and local measurements display significant fluctuations in steady state . these fluctuations are strikingly similar to those observed during the transients leading to complete fluidization in the upper part of the diagram . however here the material never gets entirely fluidized and the shear band width @xmath35 does not show a systematic evolution towards @xmath54 so that an unsteady shear band persists in steady state at least within the finite duration of the experiments . these fluctuations are investigated in more details in the next section . in this section , we focus on the strong fluctuations that are observed ( i ) during transient regimes leading to complete fluidization and ( ii ) during unsteady shear - banded flows at steady state . global rheological data together with the detailed analysis of the corresponding local measurements are reported in figs . [ fig5 ] and [ fig6 ] respectively ( see also corresponding movies 1 and 2 in the esi ) . local data are analyzed as follows : linear fits of the velocity profiles in the fluidized part of the gap are used to estimate the local shear rate and to extrapolate the sample velocity @xmath55 at the rotor . the width @xmath35 of the shear band is obtained as the abscissa of the intersection between the fit and the zero velocity axis , while the slip velocity is given by @xmath56 , where @xmath57 is the rotor velocity . finally , the global shear rate is defined as @xmath58 . we first discuss fig . [ fig5 ] which shows a case of full fluidization for @xmath59 s@xmath16 on a gel left to age for @xmath47 min . the stress relaxes by successive jumps up to full fluidization which occurs at @xmath60 s and after which @xmath17 remains roughly constant [ fig . [ fig5](a ) ] . the temporal evolutions of @xmath35 [ fig . [ fig5](b ) ] and of the local shear rate [ fig . [ fig5](c ) ] show that fluidization occurs by successive spatial `` avalanches '' that are directly correlated to the stress drops . during the whole fluidization process , the slip velocity at the rotor @xmath48 decreases towards a negligible value that is reached at @xmath61 , diminishing by jumps that are in phase with the stress evolution . figure [ fig6 ] focuses on spatiotemporal fluctuations observed during steady - state shear - banded flows for the same aging time ( @xmath47 min ) as in fig . [ fig5 ] but sheared at a lower shear rate ( @xmath62 s@xmath16 ) . the stress exhibits periods of slow increase followed by rapid drops [ fig . [ fig6](a ) ] . within an hour , about half the gap gets fluidized and the size of the fluidized band shows pronounced increase during short period of times , that are synchronized with the stress drops [ fig . [ fig6](b ) ] . the dynamics of the fluid at the rotor is strongly correlated to the global fluctuations , as evidenced by the sudden peaks of the slip velocity when the stress drops . note that , although the slip velocity decreases in average over the whole duration of the experiment , it remains at a high level of about 20% , in noticeable contrast with the fully fluidized scenario described in the previous paragraph . such oscillatory dynamics in the vicinity of the rotor are reminiscent of stick - slip . indeed , the fluidized band shows time interval during which the local shear rate remains constant [ fig . [ fig6](c ) ] , while the band width slowly decreases . as a result , the stress slowly builds up , until the fluidized band experiences a large slip event at the moving wall and gets rejuvenated . these dynamics look similar to the stick - slip motion reported in clays in the pioneering work of pignon _ et al . however , in the latter case , stick - slip occurs along fracture planes located in the bulk sample , while here slip at the wall appears to play a key role . finally , we shall emphasize that in both types of scenarios , the peaks in both the stress and the slip velocity are only seen in the presence of shear banding . although it is not clear which of the shear banding or of the slip at the rotor is the cause of the oscillations , these peculiar dynamics hint toward a subtle flow microstructure coupling that certainly deserves more investigation . vs time during a single stress drop event extracted from shear startup experiment during which the material is fully fluidized ( @xmath63 s@xmath16 , @xmath53 min ) . inset : stress vs time for the whole experiment . the signal pictured in the main graph appears in red . ( b ) spatiotemporal diagram of the velocity data @xmath64 as a function of position @xmath10 and time @xmath14 . the radial position @xmath10 is measured from the rotating inner wall . data obtained with 1d - usv . ( c ) spatiotemporal diagram of the velocity data @xmath65 as a function of the vertical position @xmath11 and time @xmath14 at @xmath66 mm . data obtained with 2d - usv . the vertical position @xmath11 is measured from the top of the transducer array . in both ( b ) and ( c ) , the fluid velocity is color coded in mm.s@xmath16 . ] vs time during a single stress relaxation episode extracted from shear startup experiment at the end of which the material is fully fluidized ( @xmath67 s@xmath16 , @xmath46 min ) . ( b ) spatiotemporal diagram of the velocity data @xmath65 as a function of the vertical position @xmath11 and time @xmath14 at @xmath68 mm . data obtained with 2d - usv . the vertical position @xmath11 is measured from the top of the transducer array . the fluid velocity is color coded in mm.s@xmath16 . ] vs time during a single stress relaxation episode extracted from shear startup experiment at the end of which the material is fully fluidized ( @xmath69 s@xmath16 , @xmath46 min ) . ( b ) spatiotemporal diagram of the velocity data @xmath65 as a function of the vertical position @xmath11 and time @xmath14 , at @xmath68 mm . data obtained with 2d - usv . the vertical position @xmath11 is measured from the top of the transducer array . the fluid velocity is color coded in mm.s@xmath16.,width=325 ] to get better insight on the avalanche - like fluidization scenario , we now focus on the single stress drop shown in fig . [ fig7](a ) . this event is extracted from a shear startup experiments performed at @xmath70 s@xmath16 for @xmath53 min and the steady - state flow corresponds to a fully fluidized sample [ inset of fig . [ fig7](a ) ] . the velocity profiles recorded simultaneously with 1d - usv are plotted as a spatiotemporal diagram in fig . [ fig7](b ) . quite surprisingly the one - dimensional flow profiles before and after the stress drop are very similar as they all show shear localization over about half the gap . by the time the stress reaches its maximum value , about 80% of the gap is sheared . then shear abruptly localizes again over about 1 mm close to the rotor at the beginning of the stress relaxation . therefore , although the strong fluctuations of @xmath64 observed before the stress peak appear to be correlated to those of @xmath71 , the drop of about 15% in the stress value can not be explained by these one - dimensional data in a straightforward manner . we emphasize that the 1d - usv measurements are performed at a given height of the taylor - couette cell so that they may reflect the evolution of global rheology only if the flow is homogeneous along the vorticity direction . as a matter of fact , velocity profiles @xmath72 recorded simultaneously over the whole height of the taylor - couette cell through 2d - usv demonstrate that the flow is strongly heterogeneous in the vertical direction @xmath11 . figure [ fig7](c ) shows a spatiotemporal plot of the fluid velocity along the vertical axis @xmath11 at a fixed radial position @xmath66 mm close to the rotor ( see also movie 3 in the esi ) . depending on the position along the @xmath11-axis , the material can simultaneously be either solid - like as evidenced by areas of very low velocities in fig . [ fig7](c ) ( see for instance at the top of the region of interest for @xmath73 mm ) or fluid - like and flow at high velocities as observed for @xmath74 and @xmath75 mm . note that the 1d - usv measurements are performed at @xmath76 mm and are fully consistent with the 2d - usv data . our measurements further reveal that the particular avalanche - like event analyzed here only corresponds to partial fluidization : at the end of the stress drop , the flow is still heterogeneous , even though arrested regions for @xmath77 s has given way to flowing regions arranged in a vertically banded structure for @xmath78 s. in fact , the inset of fig . [ fig7](a ) shows that several subsequent stress drop events still have to take place before complete fluidization of the sample . to test the existence of a local fluidization scenario that would be generic to all avalanche - like events , we analyze additional experiments in the full fluidization region of the state diagram at @xmath79 and 200 s@xmath16 . the temporal evolution of the shear stress is reported respectively in figs . [ fig8](a ) and [ fig9](a ) and the corresponding 2d - usv measurements @xmath65 are displayed as spatiotemporal diagrams in figs . [ fig8](b ) and [ fig9](b ) for a radial position @xmath68 mm ( see also movies 4 and 5 in the esi ) . here again , the temporal evolution of the velocity field is strongly heterogeneous along the vertical direction in both experiments . furthermore , despite these two experiments are performed at comparable shear rates , the local fluidization scenario is strikingly different . in the shear startup experiment reported in fig . [ fig8 ] , the lower part of the region of interest is fully fluidized after the first stress relaxation while the upper part of the sample necessitates two supplemental avalanches to in turn fully fluidize . on the other hand , the fluidization process shown in fig . [ fig9 ] starts from the upper part of the region of interest before extending to the lower part of the taylor - couette cell . here , each avalanche - like event involves large pieces of the sample with a typical vertical extension of about 5 mm [ see events marked by white dotted lines in fig . [ fig9](a ) ] . some of these events only show up in the stress response as very small peaks while their local signature is much more impressive ( see , e.g. , the event at @xmath80 s ) . for @xmath81 to 500 s , the sample even appears to fluidize , or at least to be set into motion , through regular steps occuring from top to bottom with a characteristic time of 1020 s. the experiments shown in figs . [ fig8 ] and [ fig9 ] allow us to conclude that the peaks in the stress signal may encompass very different local scenarios . since both experiments were performed with the same aging time @xmath46 min , these results illustrate the high sensitivity to `` initial conditions , '' i.e. to the possibly different arrangement of the heterogeneous microstructure after the aging process , and the subtle interplay between aging and shear banding with no systematic failure scenario along the cell height . on the one hand , we conclude from section [ sampleprep ] that the gel under scrutiny displays a discontinuous yielding transition that is very similar to the one reported in other attractive gels @xcite . indeed , one observes that , when the initial condition is a completely fluidized state , applying a series of shear rates of decreasing values leads to the growth of a steady - state arrested band below a critical shear rate @xmath33 . the value of @xmath33 is inherent to the sample and results from the flow instability at low shear rates this is fully consistent with previous results on ludox gels . @xcite on the other hand , the shear startup experiments reported in section [ results ] and performed on the sample prepared in the solid state allow us to identify a second critical shear rate @xmath82 . for applied shear rates such as @xmath83 , the sample gets only partially fluidized and exhibits steady - state shear banding , while long - lasting yet transient shear banding is observed for @xmath84 . such a behaviour can be interpreted in the framework recently described by martin and hu on laponite samples @xcite . under an external shear , the sample in the vicinity of the rotor that is in initially at rest gets fluidized , while the sample close to the stator remains at rest and keeps aging . depending on the intensity of the shear rate , such heterogeneous velocity profile may either be trapped by the sample aging and become permanent as observed below @xmath85 , or become homogeneous after a transient shear - banding regime which duration decreases for increasing values of @xmath1 . such an argument allows us to understand that @xmath85 will be sensitive to the gap size and to the cell geometry , which we have verified by performing identical shear startup experiments with different gap widths . these experiments show that @xmath2 decreases as the gap size is reduced [ see fig . 6 in the esi ] . therefore , steady - state shear banding trapped by aging should be distinguished from shear banding resulting from intrinsic flow instability which should not depend on the geometry . finally , we have shown in section [ pd ] that @xmath85 is a nonlinear function of the sample age @xmath0 , again in contrast with @xmath33 . regarding the transient fluidization dynamics , our work unravels the existence of very characteristic peaks in the global rheological data . these peaks correspond to local avalanches associated with the abrupt fluidization of shear - banded velocity profiles . an avalanche proceeds in two steps . first , the sample ages as evidenced by the slow increase of the stress indicative of progressive consolidation , while the shear band remains roughly fixed . second , the sample suddenly fluidizes before localizing again , at least partially , while the stress drops and the shear rate increases . this scenario strikingly recalls the transient shear banding reported in laponite clay suspensions @xcite and fits well with the stability criterion recently proposed by fielding _ _ @xcite . in practice though , the physical origin of the avalanche - like and successive stress relaxation events remains to be determined . aging is more pronounced in our system than in the attractive gels that have been studied previously @xcite as evidenced from the large values of @xmath86 after preshear [ see inset of fig . [ fig2](a ) ] . therefore , syneresis driven by the aggregation and/or sedimentation of the colloidal flocs due to their density mismatch with the surrounding fluid could also play a role and interfere with the traditional picture of a competition between aging and shear rejuvenation @xcite . as syneresis is negligible in most of the colloidal gels that have been the topic of rheophysical studies so far , it could also explain why such stress oscillations have , to our knowledge , never been reported in the literature . an alternative explanation for the stress oscillations could be related to confinement . indeed , as the size distribution of the fused silica aggregates is wide and extends up to 100 @xmath13 m , the sample mesostructure most likely involves aggregates which size becomes comparable to that of the gap , at least for long enough rest durations @xmath0 . in this framework , stress oscillations would result from a competition between shear - induced structuration as described in the introduction and the strong aging of the sample . such an interpretation would also account for the stick - slip like motion of the fluidized band at the moving wall . nonetheless , despite systematic monitoring of the sample during shear startup experiments , no structuration or spatial pattern could be observed . moreover , supplemental shear startup experiments in narrower gaps show that for a given shear rate , stronger confinement leads to the disappearance of the stress oscillations and to homogeneous velocity profiles [ see fig . 6 in the esi ] . this last result strongly suggests that confinement alone can not account for stress oscillations . finally , the present study has focused on experiments performed under smooth boundary conditions , revealing the presence of strong wall slip associated with heterogeneous , shear - banded flows while fully fluidized states show negligible wall slip . yet the roughness and/or chemical nature of the walls are known to have a crucial influence not only on rheological measurements but also on local flow both close to the walls and in the bulk @xcite . therefore the influence of boundary conditions on the complex fluidization scenario reported here appears as the next key question to address in future work . as a first step , we report preliminary tests on the role of boundary conditions on the above results in the esi . the flow curve measured under `` rough '' boundary conditions in a sand - blasted plexiglas taylor - couette cell with typical roughness of 1 @xmath13 m ( to be compared to a few tens of nanometers for the `` smooth '' cell used so far ) shows a smaller , yet significant hysteresis [ see fig . 1(b ) in the esi ] as well as wall slip at low shear rates ( see fig . 5 in the esi ) . furthermore , one can see a strong difference between the velocity profiles recorded simultaneously to the flow curve respectively with the rough and smooth boundary conditions ( compare fig . 4 and 5 in the esi ) although the surface roughness of the rough boundary does not match exactly the size of the microstructure . in particular the transient fluidization episode reported around 7 s@xmath16 with smooth boundary conditions [ fig . [ fig2](b ) and fig . 4 in the esi ] is absent with rough walls ( see fig . 5 in the esi ) . these preliminary results illustrate the strong impact of the boundary conditions and urge for systematic experiments so as to quantify the impact of the boundary conditions especially on the state diagram reported in fig . we have investigated the local scenario associated with the shear - induced fluidization of an attractive gel made of non - brownian particles . we have identified a critical shear rate @xmath2 that separates steady shear - banded flows from full fluidization and that exhibits a nonlinear dependence with sample age @xmath0 . this critical shear rate @xmath2 is much larger that the one signaling flow instability in experiments starting from fluidized states and depends on the shear cell geometry as well as possibly on the preshear intensity imposed prior to the shear startup experiment . as a key result , we have shown that for shear rates larger than @xmath2 , the fluidization of the sample involves successive local avalanche - like events that are heterogeneously distributed along the cell height and which dynamics is strongly coupled to both the slip behaviour at the wall and the width of the shear band . such avalanches appear in the stress signal as peaks , which individual properties are strongly reminiscent of stick - slip phenomena . future work will focus on the early stage of shear startup experiments , and in particular on the stress overshoot that occurs before the stress relaxation , as well as on the influence of the boundaries and confinement on the fluidization scenario . the authors thank j .- chapel , b. keshavarz and g. ovarlez for stimulating discussions , t. gibaud for allowing us to use his microscope , j. laurent for her precious assistance with the sem experiments as well as two anonymous referees for their constructive remarks on our manuscript . this work was supported by jsps and cnrs under the japan - france research cooperative program ( prc cnrs / jsps rheovolc ) . a.k . acknowledges support by jsps kakenhi grant no . s.m . acknowledges funding from institut universitaire de france and from the european research council under the european union s seventh framework programme ( fp7/2007 - 2013 ) and erc grant agreement no . 258803 . 74[1]#1 [ 1 ] [ 2 ] [ 3 ] [ 3 ] p. coussot , _ j. non - newtonian fluid mech . _ , 2015 , * 211 * , 3149 d. bonn , j. paredes , m. denn , l. berthier , t. divoux and s. manneville , _ arxiv:1502.05281 _ , 2015 p. c. f. mller , a. fall and d. bonn , _ europhys . _ , 2009 , * 87 * , 38004 g. ovarlez , s. cohen - addad , k. krishan , j. goyon and p. coussot , _ j. non - newtonian fluid mech . _ , 2013 , * 193 * , 6879 n. balmforth , i. frigaard and g. ovarlez , _ annu . rev . fluid mech . _ , 2014 , * 46 * , 121146 r. bonnecaze and m. cloitre , _ adv . polym . _ , 2010 , * 236 * , 117161 t. g. mason , j. bibette and d. a. weitz , _ j. colloid interface sci . _ , 1996 , * 179 * , 439448 m. cloitre , m. borrega , f. monti and l. leibler , _ c.r . , 2003 , * 4 * , 221230 j. seth , m. cloitre and r. bonnecaze , _ j. rheol . _ , 2006 , * 50 * , 353376 m. siebenbrger , m. fuchs and m. ballauff , _ soft matter _ , 2012 , * 8 * , 40144024 p. lu , e. zaccarelli , f. ciulla , a. b. schofield , f. sciortino and d. a. weitz , _ nature _ , 2008 , * 453 * , 499503 p. lu and d. weitz , _ annu . rev . condens . matter phys . _ , 2013 , * 4 * , 217233 a. mourchid , a. delville , j. lambard , e. lcolier and p. levitz , _ langmuir _ , 1995 , * 11 * , 19421950 f. pignon , a. magnin , j .- m . piau , b. cabane , p. lindner and o. diat , _ phys . e _ , 1997 , * 56 * , 32813289 e. paineau , i. bihannic , c. baravian , a .- m . philippe , p. davidson , p. levitz , s. funari , c. rochas and l. michot , _ langmuir _ , 2011 , * 27 * , 55625573 v. trappe and d. a. weitz , _ phys . rev _ , 2000 , * 85 * , 449452 v. trappe , v. prasad , l. cipelletti , p. n. segre and d. a. weitz , _ nature _ , 2001 , * 411 * , 772775 k. pham , s. egelhaaf , p. pusey and w. poon , _ phys . e _ , 2004 , * 69 * , 011503 e. mock and c. zukoski , _ j. rheol . _ , 2007 , * 51 * , 541559 n. reddy , z. zhang , m. lettinga , j. dhont and j. vermant , _ j. rheol . _ , 2012 , * 56 * , 11531174 l. cipelletti , s. manley , r. c. ball and d. a. weitz , _ phys . _ , 2000 , * 84 * , 22752278 c. derec , g. ducouret , a. ajdari and f. lequeux , _ phys , 2003 , * 67 * , 061403 g. ovarlez and x. chateau , _ phys . e _ , 2008 , * 77 * , 061403 a. negi and c. osuji , _ phys . e _ , 2009 , * 80 * , 010404 n. koumakis and g. petekidis , _ soft matter _ , 2011 , * 7 * , 24562470 g. scherer , _ j. non - cryst . solids _ , 1989 , * 108 * , 1827 s. manley , j. skotheim , l. mahadevan and d. weitz , _ phys . rev . lett . _ , 2005 , * 94 * , 218302 r. buscall , t. h. choudhury , m. a. faers , j. w. goodwin , p. a. luckham and s. j. partridge . , _ soft matter _ , 2009 , * 5 * , 13451349 g. brambilla , s. buzzaccaro , r. piazza , l. berthier and l. cipelletti , _ phys . rev . _ , 2011 , * 106 * , 118302 l. j. teece , m. a. faers and p. bartlett , _ soft matter _ , 2011 , * 7 * , 13411351 p. barlett , l. teece and m. faers , _ phys . rev . e _ , 2012 , * 85 * , 021404 r. buscall and i. mcgowan , _ faraday discuss . _ , 1983 , * 76 * , 277290 t. gibaud , c. barentin and s. manneville , _ phys . rev _ , 2008 , * 101 * , 258302 t. gibaud , c. barentin , n. taberlet and s. manneville , _ soft matter _ , 2009 , * 5 * , 30263037 j. j. v. degroot , c. w. macosko , t. kume and t. hashimoto , _ j. colloid interface sci . _ , 1994 , * 166 * , 404413 r. navarrete , l. scriven and c. macosko , _ j. colloid interface sci . _ , 1996 , * 180 * , 200211 a. montesi , a. pe@xmath87a and m. pasquali , _ phys . lett . _ , 2004 , * 92 * , 058303 c. o. osuji , c. kim and d. a. weitz , _ phys . e _ , 2008 , * 77 * , 060402(r ) a. negi and c. osuji , _ rheol . acta _ , 2009 , * 48 * , 871881 v. grenard , n. taberlet and s. manneville , _ soft matter _ , 2011 , * 7 * , 39203928 j. vermant and m. j. solomon , _ j. phys . : condens . , 2005 , * 17 * , r187r216 v. gopalakrishnan and c. zukoski , _ j. rheol . _ , 2007 , * 51 * , 623644 t. gibaud , d. frelat and s. manneville , _ soft matter _ , 2010 , * 6 * , 34823488 j. sprakel , s. lindstrm , t. kodger and d. weitz , _ phys . _ , 2011 , * 106 * , 248303 v. grenard , t. divoux , n. taberlet and s. manneville , _ soft matter _ , 2014 , * 10 * , 15551571 a. stickland , a. kumar , t. kusuma , p. scales , a. tindley , s. biggs and r. buscall , _ rheol . acta _ , 2015 , * 54 * , 337352 h. chan and a. mohraz , _ phys . , 2012 , * 85 * , 041403 c. perge , n. taberlet , t. gibaud and s. manneville , _ j. rheol . _ , 2014 , * 58 * , 13311357 p. varadan and m. solomon , _ langmuir _ , 2001 , * 17 * , 29182929 a. mohraz and m. solomon , _ j. rheol . _ , 2005 , * 49 * , 657681 b. rajaram and a. mohraz , _ soft matter _ , 2010 , * 6 * , 22462259 t. divoux , m .- a . fardin , s. manneville and s. lerouge , _ annu . rev . fluid mech . _ , 2016 , * 48 * , 81103 j. s. raynaud , p. moucheront , j. c. baudez , f. bertrand , j. p. guilbaud and p. coussot , _ j. rheol . _ , 2002 , * 46 * , 709732 j. martin and y. hu , _ soft matter _ , 2012 , * 8 * , 69406949 g. ovarlez , s. rodts , x. chateau and p. coussot , _ rheol . acta _ , 2009 , * 48 * , 831844 a. fall , j. paredes and d. bonn , _ phys . rev . lett . _ , 2010 , * 105 * , 225502 l. bilmes , _ nature _ , 1942 , * 150 * , 432433 g. mckinley , _ rheology bulletin _ , 2015 , * 84 * , 1417 r. moorcroft and s. fielding , _ phys lett . _ , 2013 , * 110 * , 086001 j. colombo , a. widmer - cooper and e. del gado , _ phys . lett . _ , 2013 , * 110 * , 198301 j. colombo and e. del gado , _ j. rheol . _ , 2014 , * 58 * , 10891116 s. fielding , _ rep . _ , 2014 , * 77 * , 102601 s. manneville , l. bcu and a. colin , _ eur j. ap _ , 2004 , * 28 * , 361373 t. gallot , c. perge , v. grenard , m .- a . fardin , n. taberlet and s. manneville , _ rev . instrum . _ , 2013 , * 84 * , 045107 p. c. f. mller , s. rodts , m. a. j. michels and d. bonn , _ phys . rev . e _ , 2008 , * 77 * , 041507 m. kobayashi , f. juillerat , p. galletto , p. bowen and m. borkovec , _ langmuir _ , 2005 , * 21 * , 57615769 w. heston , r. iler and g. sears , _ j. chem _ , 1960 , * 64 * , 147150 j. trompette and m. clifton , _ j. colloid interface sci . _ , 2004 , * 276 * , 475482 x. j. cao , h. z. cummins and j. f. morris , _ soft matter _ , 2010 , * 6 * , 54255433 d. truzzolillo , v. roger , c. dupas , s. mora and l. cipelletti , e - print cond - mat/1411.2265 l. allen and e. matijevi , _ j. colloid interface sci . _ , 1969 , * 31 * , 287296 l. allen and e. matijevi , _ j. colloid interface sci . _ , 1970 , * 33 * , 420429 j. depasse and a. watillon , _ j. colloid interface sci . _ , 1970 , * 33 * , 430438 j. depasse , _ j. colloid interface sci . _ , 1997 , * 194 * , 260262 e. drabarek , j. bartlett , h. hanley , j. woolfrey and c. muzny , _ int . j. thermophys . _ , 2002 , * 23 * , 145160 p. coussot , j. s. raynaud , f. bertrand , p. moucheront , j. p. guilbaud , h. t. huynh , s. jarny and d. lesueur , _ phys . rev . lett . _ , 2002 , * 88 * , 218301 a. ragouilliaux , b. herzhaft , f. bertrand and p. coussot , _ rheol . acta _ , 2006 , * 46 * , 261271 a. ten brinke , l. bailey , h. lekkerkerker and g. maitland , _ soft matter _ , 2007 , * 3 * , 11451162 t. divoux , v. grenard and s. manneville , _ phys . lett . _ , 2013 , * 110 * , 018304 v. viasnoff and f. lequeux , _ phys . _ , 2002 , * 89 * , 065701 f. pignon , a. magnin and j .- m . piau , _ j. rheol . _ , 1996 , * 40 * , 573587 p. coussot , q. d. nguyen , h. t. huynh and d. bonn , _ j. rheol . _ , 2002 , * 46 * , 573589 p. coussot , h. tabuteau , x. chateau , l. tocquer and g. ovarlez , _ j. rheol . _ , 2006 , * 50 * , 975994 g. ovarlez and p. coussot , _ phys e _ , 2007 , * 76 * , 011406 p. c. f. mller , j. mewis and d. bonn , _ soft matter _ , 2006 , * 2 * , 274283 j. seth , c. locatelli - champagne , f. monti , r. bonnecaze and m. cloitre , _ soft matter _ , 2012 , * 8 * , 140148 v. mansard , l. bocquet and a. colin , _ soft matter _ , 2014 , * 10 * , 69846989 [ [ section ] ] ) and viscous moduli ( ) vs time after a preshear at @xmath15 s@xmath16 for 2 min ( @xmath39 hz , @xmath38 pa ) . same data as in fig . 2(a ) in the main text . ( b ) flow curve , shear stress @xmath17 @xmath88 applied shear rate @xmath44 , obtained by decreasing ( @xmath89 ) and increasing ( @xmath90 ) shear rate in order from 10@xmath19 to 10@xmath18 s@xmath16 with a waiting time of @xmath91 s per point . black ( resp . red ) symbols correspond to results in a smooth ( resp . rough ) geometry . ] five movies are provided as supplemental materials to figs . 59 of the main text . all movies illustrate the temporal evolution of the global rheology and of the 1d and/or 2d velocity profiles recorded simultaneously to the rheology during successive avalanche - like events . supplemental fig . [ supfig1](a ) displays the evolution of the elastic and viscous moduli of the sample after preshear . the gel rebuilds quickly as evidenced by the fact that the elastic modulus is larger than the viscous modulus about 20 s after preshear has been stopped . supplemental fig . [ supfig1](b ) shows flow curves @xmath17 vs @xmath1 measured by rapidly sweeping down the shear rate from @xmath28 to @xmath29 s@xmath16 and back up . the flow curves have complex non - monotonic shapes together with strong hysteresis under both smooth and rough boundary conditions . supplemental fig . [ supfig2 ] illustrates the result of a frequency sweep performed at a constant strain ( @xmath92 % ) from 10 to 0.02 hz , and for different aging times @xmath47 , 60 and 100 min after stopping the preshear . the elastic modulus @xmath26 is independent of the frequency and increases with the aging time @xmath0 in agreement with supplemental fig . [ supfig1 ] . the viscous modulus shows a broad minimum with no sign of increasing part at low frequencies , a feature that is typical of soft glassy systems @xcite . ) and viscous ( @xmath89 ) moduli vs frequency @xmath93 . the frequency sweep is performed at a constant strain @xmath92 % and at different aging times @xmath0 after preshear . [ color , @xmath0 ( min ) ] : ( , 30 ) ; ( , 60 ) ; ( , 100 ) . ] supplemental fig . [ supfig3 ] shows the results of a strain sweep performed on the gel at a constant frequency @xmath39 hz for different aging times @xmath0 after preshear . the gel yields at a constant strain of about 7 % that is roughly independent of the aging time @xmath0 . ) and viscous ( @xmath94 ) moduli vs the strain amplitude . the strain sweep is performed at a fixed frequency @xmath39 hz with a waiting time of 8 s per point , and at different times @xmath0 after the preshear . [ color , @xmath0 ( min ) ] : ( , 30 ) ; ( , 60 ) ; ( , 100 ) . the vertical dashed line at @xmath95 % emphasizes the yield point defined as the point at which @xmath96 . ] supplemental figs . [ supfig4 ] and [ supfig5 ] show the rheological and 2d - usv data obtained in smooth and rough boundary conditions respectively , by first decreasing the shear rate and then increasing it back again . the corresponding flow curves are shown in fig . 1(b ) of the present esi . here we also show spatiotemporal plots of the velocity profiles as a function of the radial position @xmath10 at a given height ( @xmath76 mm ) in the taylor - couette cell [ supplemental figs . [ supfig4](b ) and [ supfig5](b ) ] , and as a function of the vertical position @xmath11 , at a given radial position inside the gap ( @xmath97 mm ) [ supplemental figs . [ supfig4](c ) and [ supfig5](c ) ] . velocity data are shown only for @xmath98460 and 340 s@xmath16 in smooth and rough boundary conditions respectively , due to the limitation in the ultrasonic pulse repetition frequency to 20 khz , which sets an upper bound on the measurable velocities @xcite . and shear stress @xmath17 vs time @xmath14 obtained by first decreasing continuously @xmath1 from @xmath99 to @xmath29 s@xmath16 in 75 logarithmically spaced points of duration @xmath100 s each , and then increasing @xmath1 over the same range . ( b ) spatiotemporal diagram of the velocity data @xmath101 as a function of the distance @xmath10 to the rotor and time @xmath14 , at @xmath76 mm . ( c ) spatiotemporal diagram of the velocity data @xmath65 as a function of the vertical position @xmath11 and time @xmath14 , at @xmath97 mm . the vertical position @xmath11 is measured from the upper boundary of the 2d - usv probe . the fluid velocity is color - coded in mm.s@xmath16 . ] for smooth boundary conditions , the flow curve exhibits a large hysteresis , mainly localized in a narrow range of shear rates [ supplemental fig . [ supfig1](b ) ] . during the decreasing ramp of @xmath1 , the sample which is first sheared homogeneously at large shear rates , enters a total wall slip regime at @xmath102 s , i.e. , for @xmath103 s@xmath16 [ supplemental fig . [ supfig4](b ) ] . this regime persists until @xmath1 is increased back up again . the subsequent fluidization of the material is abrupt and occurs in a narrow range of shear rates , @xmath104 s@xmath16 , which coincides with the brutal increase of the global stress . supplemental fig . [ supfig4](b , c ) reveal that the subsequent drop in the stress corresponds to flow arrest , i.e. total wall slip , until the sample fluidizes again for @xmath105 s@xmath16 and that the local behavior of the sample is roughly homogeneous along the whole height of the cell during both ramps . supplemental fig . [ supfig5 ] illustrates the exact same protocol as that reported in supplemental fig . [ supfig4 ] , but performed in a rough taylor - couette cell . as also seen in supplemental fig . [ supfig1](b ) , the rheological hysteresis is far less pronounced in comparison to the data obtained with smooth boundary conditions . here the flow remains homogeneous along the vertical axis during both ramps [ supplemental fig . [ supfig5](c ) ] . the velocity profiles at a given height in the taylor - couette cell also display very different behavior from the smooth case . in the decreasing ramp of @xmath1 , the stress plateaus in between @xmath106 s@xmath16 and @xmath107 s@xmath16 , while shear localizes at the rotor . below @xmath107 s@xmath16 , the stress displays a kink towards a second plateau , while the sample exhibits a plug flow . increasing @xmath1 back up , the plug flow persists up to shear rates of a few 10 s@xmath16 . above the latter value , the fluidization of the sample proceeds from the rotor and involves transient banding . full fluidization is not available in supplemental fig . [ supfig5](b ) for the technical reason mentioned above . supplemental fig . [ supfig6 ] shows the influence of the gap size on the fluid response to a shear startup experiment performed at @xmath62 s@xmath16 on a sample left at rest for @xmath53 min . in a gap of size @xmath5 mm , the same value as in the main text , the stress response displays a large number a peaks together with unsteady shear banding over 2400 s [ supplemental fig . [ supfig6](b ) ] . decreasing the size of the gap changes radically the sample behavior . in a gap twice as small ( @xmath108 mm ) , the stress response exhibits less peaks and the sample fluidizes entirely in about @xmath109 s [ supplemental fig . [ supfig6](c ) ] . decreasing the gap by a factor of 4 ( @xmath110 mm ) leads to a smooth stress response and to homogeneous velocity profiles from the beginning of the experiments , with significant slip at the moving boundary [ supplemental fig . [ supfig6](d ) ] . t. g. mason , j. bibette and d. a. weitz , _ phys . lett . _ , 1995 , * 75 * , 20512054 t. gallot , c. perge , v. grenard , m .- a . fardin , n. taberlet and s. manneville , _ rev . instrum . _ , 2013 , * 84 * , 045107	we report on the fluidization dynamics of an attractive gel composed of non - brownian particles made of fused silica colloids . extensive rheology coupled to ultrasonic velocimetry allows us to characterize the global stress response together with the local dynamics of the gel during shear startup experiments . in practice , after being rejuvenated by a preshear , the gel is left to age during a time @xmath0 before being submitted to a constant shear rate @xmath1 . we investigate in detail the effects of both @xmath0 and @xmath1 on the fluidization dynamics and build a detailed state diagram of the gel response to shear startup flows . the gel may either display transient shear banding towards complete fluidization , or steady - state shear banding . in the former case , we unravel that the progressive fluidization occurs by successive steps that appear as peaks on the global stress relaxation signal . flow imaging reveals that the shear band grows up to complete fluidization of the material by sudden avalanche - like events which are distributed heterogeneously along the vorticity direction and correlated to large peaks in the slip velocity at the moving wall . these features are robust over a wide range of values of @xmath0 and @xmath1 , although the very details of the fluidization scenario vary with @xmath1 . finally , the critical shear rate @xmath2 that separates steady - state shear - banding from steady - state homogeneous flow depends on the width on the shear cell and exhibits a nonlinear dependence with @xmath0 . our work brings about valuable experimental data on transient flows of attractive dispersions , highlighting the subtle interplay between shear , wall slip and aging which modeling constitutes a major challenge that has not been met yet .
this work is a sequel to the stacs paper @xcite and its journal version @xcite . however , we try to keep it self - contained by reiterating everything of importance from the two preceeding papers . we extend their results to a more general version of power circuits . as a consequence , we can apply them to larger classes of groups . the group @xmath10 was introduced by higman in 1951 and served to provide the first known example of a finitely generated infinite simple group @xcite . it belongs to a family @xmath11 ( @xmath12 ) of groups with @xmath13 generators and @xmath13 relators : @xmath14 for @xmath15 these groups are trivial , which is easy to see for @xmath16 , but suprisingly hard for @xmath17 . the latter case was proven by hirsch @xcite , see also 23 in @xcite and @xcite . if @xmath6 , then @xmath11 is infinite , see @xcite , section 1.4 . since @xmath11 has no non - trivial normal subgroups of finite index , taking a minimal non - trivial quotient results in a finitely generated infinite simple group . until recently , higman s group @xmath10 was a candidate for a `` natural '' group with non - elementary word problem . this was suggested by the huge compression that this group allows . in fact , there are words of length @xmath18 over the generators @xmath19 and their inverses , which in the group are equal to @xmath20 , where @xmath21 is the tower function ( also called `` tetration '' ) defined by @xmath22 and @xmath23 . however , in @xcite it was shown that the word problem of @xmath10 is decidable in @xmath24 time and @xcite improved this bound to @xmath8 . both results rely on a data structure called `` power circuit '' which was introduced by miasnikov , ushakov , and won in @xcite . power circuits had already proven useful in algorithmic group theory . in fact , their invention was entailed by the wish to efficiently solve the word problem in the baumslag - gersten group @xmath25 ( see @xcite and section [ sec : bg ] ) which shares with @xmath10 the property of huge compression . ( in @xcite the group @xmath25 is called `` baumslag group '' . we use the equally common name `` baumslag - gersten group '' to avoid confusion with the baumslag - solitar group . ) for @xmath2 , power circuits have been implemented @xcite and there is a computer program solving the word problem in @xmath25 . this paper builds on the work of diekert , ushakov , and the author conducted in @xcite and @xcite . its contributions are twofold : in section [ sec : pc ] , we extend power circuits to allow arbitrary bases @xmath0 . this necessitates changes in the reduction procedure , the core component of power circuits . in section [ sec : higman ] , we generalize higman s group @xmath10 by replacing the underlying group @xmath4 by @xmath26 . power circuits with base @xmath1 are naturally suited for computations in this group . furthermore , with the help of rewriting systems , we give a constructive method of treating any group @xmath11 ( @xmath6 ) rather that just @xmath10 . combining this with base @xmath1 power circuits leads to an algorithm for the word problem in generalized higman groups @xmath7 which retains the @xmath8 time bound proved in @xcite for @xmath2 and @xmath27 . [ [ sec : notation ] ] notation and preliminaries + + + + + + + + + + + + + + + + + + + + + + + + + + algorithms and decision problems are classified by their time complexity on a random - access machine ( ram ) . we use the notion of amortized analysis with respect to a potential function , see section 17.3 in @xcite . with regard to group theory , we use standard notation and facts that can be found in any textbook on the subject , e.g. @xcite . in particular , we apply the technique of britton reductions for solving the word problem in hnn extensions and amalgamated products . rewriting systems are of particular importance for this work . we assume that the reader is familiar with the basic notions of ( local ) confluence and termination . see for example the textbook @xcite for a quick introduction . power circuits were introduced by miasnikov , ushakov , and won in @xcite . while the underlying ideas presented in this chapter originate from their work , there are some important differences . first , we use a different ( and hopefully more accessible ) notation , following @xcite and @xcite . we also allow multiple markings in one circuit . the most important modification , which also distinguishes this paper from @xcite and @xcite , is the generalization from base @xmath28 to arbitrary bases @xmath0 . for the rest of this paper , we fix an integer @xmath0 for the base and the interval @xmath29 ( the set of digits ) . we start with a directed acyclic edge - labelled graph without multi - edges , given by @xmath30 . here , @xmath31 is a finite set which will act as the set of nodes ( or vertices ) . the labelled edges ( or arcs ) are given by the map @xmath32 where @xmath33 means that there is no edge from @xmath34 to @xmath35 and @xmath36 implies an edge from @xmath34 to @xmath35 labelled with the number @xmath37 . in other words , the edge set is @xmath38 , the support of the map @xmath39 . in addition , we require that the directed graph @xmath40 is acyclic . we shall make this assumption throughout this paper without mentioning it again . for any operation on graphs introduced in this chapter , it will be obvious that acyclicity is preserved . a marking of @xmath30 is a mapping @xmath41 . often we regard @xmath42 as a labelled subset of the nodes of @xmath43 , where the subset is @xmath44 and the labels are given by @xmath45 . in this sense , @xmath46 ( the constant zero marking ) and @xmath47 ( the empty marking ) are the same . each node @xmath48 induces in a natural way its successor marking @xmath49 defined by @xmath50 intuitively , the successor marking of a node @xmath34 consists of the target nodes of edges starting at @xmath34 and their labels are given by those of the edges . the evaluation function @xmath51 assigns a real number to each node and each marking of a graph @xmath43 . as @xmath43 is acyclic we can give an inductive definition of @xmath51 : @xmath52 note that this implies @xmath53 for all leaves @xmath34 ( nodes without outgoing edges ) . for every node @xmath48 we have @xmath54 figure [ fig : graph_eval ] shows an example of such a graph for @xmath55 . the set of nodes is @xmath56 and @xmath39 is given by @xmath57 the nodes evaluate to @xmath58 [ lem : pc_condition ] let @xmath30 be as described above . the following statements are equivalent : [ prop : pc_condition : node ] @xmath59 for every node @xmath48 [ prop : pc_condition : succ ] @xmath60 for every node @xmath48 [ prop : pc_condition : mark ] @xmath61 for every marking possible marking @xmath42 in @xmath43 this is easily seen by noetherian induction with respect to a topological order of @xmath31 ( i.e. , an order compatible with the edges ) . a _ power circuit _ is a finite acyclic edge - labelled graph @xmath30 without multiple edges that meets the equivalent conditions of lemma [ lem : pc_condition ] . the graph in figure [ fig : graph_eval ] is not a power circuit due to @xmath62 . in contrast , figure [ fig : pc_mark_eval ] depicts a power circuit for @xmath2 . the values of the nodes are given for illustrative purposes only . in general , these number become too large to be computed . the marking @xmath42 evaluates to @xmath63 . in corollary [ cor : test_graph_pc ] we will show that it can be efficiently tested whether a given graph is a power circuit . in @xcite , a power circuit does not have to satisfy the criteria of lemma [ lem : pc_condition ] , but if it does , it is called proper . in this sense , we only deal with proper power circuits . figure [ fig : pc_tower ] shows that a power circuit of linear size can contain markings with values the magnitude of the tower function . \(4 ) at ( 4.5,0 ) ; let @xmath30 be a power circuit and @xmath48 a node . the operation clonewith result @xmath64 creates a new node @xmath35 with the same successor marking as @xmath34 , but no incoming arcs . we extend this operation to markings @xmath42 , by cloning every single node in @xmath44 . the resulting marking @xmath65 is defined as the marking consisting of all these clones , and the signs are copied from @xmath42 : @xmath66 in figure [ fig : pc_clone ] , the marking @xmath42 consisting of two nodes is cloned . now we can define arithmetic operations . let @xmath30 be a power circuit and let @xmath67 and @xmath42 be markings in @xmath43 . if the supports of @xmath67 and @xmath42 are disjoint , the mapping @xmath68 defined by @xmath69 is a marking with @xmath70 . in general , however , the operands @xmath67 and @xmath42 will not be disjoint . in this case we have nodes @xmath71 with @xmath72 , hence @xmath68 is not a valid marking . we solve this problem by cloning : for every node @xmath34 with @xmath72 , we create a clone @xmath73 and modify @xmath68 by putting @xmath74 and @xmath75 . we obtain a valid marking in the ( now enlarged ) circuit with value @xmath76 . in figure [ fig : pc_add ] , @xmath77 and @xmath78 are added . in the resulting marking , the node with value @xmath79 cancels out , whereas both the original node with value @xmath80 and its newly created clone are included . the second operation that we need is multiplication by a power of @xmath1 . we observe that @xmath81 so in principle we would just have to introduce new edges from each node in @xmath82 to each node in @xmath44 . the label of such an edge would be the value that @xmath42 assigns to the respective target node . this operation works as long as 1 . no cycles are introduced into the circuit , 2 . no multi - edges between two nodes are introduced , 3 . there are no edges between nodes in @xmath82 , and 4 . no other marking in the circuit is affected . ( note here , that the original value of @xmath67 is lost in any case . ) again , the solution is cloning . create clones @xmath83 and @xmath84 and introduce new edges by putting @xmath85 for all @xmath86 , @xmath87 . being clones , nodes in @xmath88 and @xmath89 have no incoming edges , which prevents cycles and multi - edges . also , no other marking in the circuit depends on @xmath90 or @xmath91 directly ( by containing these nodes ) or indirectly ( by containing nodes that are topologically above any node in @xmath90 or @xmath91 ) . an example ( in which no further cloning is necessary ) is shown in figure [ fig : pc_mult_power ] . finally , note that the operation @xmath92 which negates the value of @xmath42 is easy to conduct without any complications or the need for cloning . the operations @xmath68 and @xmath93 introduced in the previous section are quite efficient . assuming that the graph is stored using adjacency lists , the time they take depends only on the size of the markings @xmath42 and @xmath67 , not on the size of the circuit . the price for this efficieny is that the structure of a power circuit can quickly become rather intransparent . in particular , it is unclear how ( in)equality of the values of two markings can be determined in an arbitrary circuit . again , note that evaluating the nodes or markings is not an option , due to the vast growth permitted by power circuits . for this reason , we restrict ourselves to a subclass of circuits and augment them with some additional data : [ def : reduced_pc ] a _ reduced power circuit _ is a power circuit @xmath30 together with a list @xmath94 of its nodes and a bit vector @xmath95 such that different nodes evaluate to different numbers , i.e. , for all @xmath96 with @xmath97 , @xmath98 , the list of nodes is sorted by value , i.e. , @xmath99 , @xmath100 if and only if @xmath101 . [ prop : compare_in_red_pc](cf . prop . 5 in @xcite and section 2.1 in @xcite ) given a reduced circuit and two markings @xmath67 and @xmath42 , the values @xmath102 and @xmath103 can be compared ( yielding @xmath104 , @xmath105 , or @xmath106 as the result ) in @xmath107 time . the algorithm can also determine whether @xmath108 . assume that we want to determine whether for a sum @xmath109 with @xmath110 we have @xmath111 , @xmath112 , @xmath113 , @xmath114 or @xmath115 . we can do this inductively using the following procedure : if @xmath116 , use induction on @xmath117 . if @xmath118 , then @xmath119 and the sign of @xmath120 is the same as the sign of @xmath121 . if @xmath122 , look at @xmath123 . if @xmath124 or if it has the same sign as @xmath121 , then @xmath125 since @xmath0 . again , @xmath120 has the same sign as @xmath121 . if @xmath123 has the opposite sign of @xmath121 , use induction on @xmath126 , where @xmath127 . the answer to the original question can be found by applying this algorithm to the mapping @xmath128 given by @xmath129 . note , that the absolute indices @xmath130 of the @xmath131 are not actually needed . instead one can use the information provided by the reduced circuit . [ cor : divisibility_in_red_pc ] for two markings @xmath67 and @xmath42 in a reduced circuit , it can be tested in @xmath107 time whether @xmath132 divides @xmath103 . let @xmath34 be the node of minimal value in @xmath44 . then @xmath133 if and only if @xmath134 . using proposition [ prop : compare_in_red_pc ] , we can check the equivalent condition @xmath135 . power circuits arising from a sequence of arithmetic operations are usually far from being reduced . every cloning creates a pair of nodes with the same value . therefore , we need an algorithm that given an arbitrary circuit produces an equivalent reduced circuit . in this context , equivalence means that for each node and each marking in the old circuit , there is one with the same value in the reduced circuit . before giving the algorithm , we need some preparations . [ def : chain ] a list @xmath136 of nodes in a power circuit is called a _ chain _ ( starting at @xmath137 ) , if @xmath101 for all @xmath138 . it is called a maximal chain ( starting at @xmath137 ) , if it is not part of a longer chain ( starting at @xmath137 ) . note that chains have nothing to do with paths in the graph . in arbitrary power circuits , chains are difficult to spot . however , in a reduced power circuit , they can be easily identified using the bit vector . in a reduced circuit , the maximal chain starting at the unique node of value @xmath79 is of particular interest . it is called the _ base chain _ of the power circuit . for later use , we define in algorithm [ alg : prolong_base_chain ] a procedure prolongbasechainprolonging this chain by one node without destroying the reducedness property of a circuit . the procedure prolongbasechaintakes @xmath107 time . [ alg : prolong_base_chain ] let @xmath139 be the ordered list of the nodes of the reduced circuit @xmath31 . using this list and the bit vector , find the smallest @xmath140 such that @xmath141 . write @xmath130 as a @xmath1-ary number @xmath142 and use this to define the marking @xmath143 with value @xmath144 . insert a new node @xmath34 with @xmath145 into the circuit . place @xmath34 in the ordered list of nodes between @xmath146 and @xmath147 . check whether @xmath148 by applying proposition [ prop : compare_in_red_pc ] to @xmath49 and @xmath149 ( both are contained in the reduced circuit @xmath43 ) . set the bit vector for @xmath34 accordingly . now we can give an algorithm that reduces power circuits . reduction is done node by node . this means that at any point during the reduction procedure , the circuit consists of a reduced part and a part that is not yet reduced . the nodes in the non - reduced part are processed in topological order . in this way , the procedure only has to work for nodes all of whose successors are already in the reduced part . this approach allows us to generalize the reduction procedure . instead of reducing the entire circuit , we can take into account that parts of it might already be reduced . this will turn out to be useful in applications . the procedure extendreductiondescribed in algorithm [ alg : extend_reduce ] takes as input not only the power circuit but also a list @xmath150 of markings that need to be adjusted during reduction in order to preserve their value . [ alg : extend_reduce ] compute a topological order of @xmath151 , i.e. , @xmath152 such that @xmath153 implies @xmath154 . [ prop : extend_reduce](cf . @xcite , thm . 6 ) the procedure extendreductionis correct and takes @xmath155 time . the circuit growth @xmath156 is bounded by @xmath157 . at first , a topological order is computed . the time for this is bounded by the size of the subgraph @xmath151 ( nodes and edges ) , which is @xmath158 . in the main loop starting at line [ alg : extend_reduce : main_loop ] , the nodes are eliminated from @xmath151 one by one . let @xmath159 be the initial size of the whole graph . since @xmath31 grows by @xmath160 during the procedure ( although we keep calling it @xmath31 for convenience ) , @xmath161 is the correct bound for the size of @xmath31 . for each node @xmath162 , its position in the ordering of @xmath31 has to be found in step [ alg : extend_reduce : bin_search ] . since @xmath163 is chosen to be topologically minimal , the successor marking @xmath49 is contained in the reduced circuit @xmath31 , so @xmath34 can be compared to any node @xmath164 in @xmath161 time . using binary search , @xmath165 comparisons are sufficient , taking @xmath166 time in total . for the insertion of @xmath167 in @xmath31 , we distiguish two cases . in the first one ( step [ alg : extend_reduce : no_collision ] ) , there is no node in @xmath31 with the same value as @xmath167 . in this case , @xmath167 is moved from @xmath151 to @xmath31 without any modification . markings containing @xmath167 ( including successor markings , i.e. , edges with target @xmath167 ) are not affected either . the second case ( step [ alg : extend_reduce : collision ] ) , where there is a node @xmath168 with the same value as @xmath167 is more difficult . figure [ fig : pc_reduction ] shows an example . the idea is to delete @xmath167 and replace it in all markings @xmath42 ( both markings from @xmath42 and successor markings of nodes in @xmath151 ) by @xmath168 . this may cause @xmath168 to by `` overmarked '' by @xmath42 , i.e. , @xmath169 . for example , in the simplest case @xmath2 , if @xmath170 , then @xmath171 after the replacement . the solution is inspired by the idea of carry digits used when adding two @xmath1-ary numbers : if @xmath172 , subtract the appropriate number @xmath173 and add @xmath174 to the value that @xmath42 assigns to the next node @xmath175 in the chain , which has @xmath1 times the value of @xmath168 . the carry might propagate to the end of the chain , which is why we preventively prolonged it by one node @xmath176 . note that the time bound for one execution of step [ alg : extend_reduce : adjust_markings ] is not @xmath177 , but rather @xmath178 . since this is not sufficient to prove the claimed bound , instead we count the total amount of time spent in step [ alg : extend_reduce : adjust_markings ] during the whole procedure . the key observation is that for every carry that has to be moved to the next node in the chain , the number @xmath179 decreases . initially @xmath180 , so the total time complexity of step [ alg : extend_reduce : adjust_markings ] is @xmath181 . ( -2,-0.5 ) rectangle ( 3,6.5 ) ; ( 1,-0.5 ) ( 1,6.5 ) ; at ( 1,6.5 ) @xmath31 ; at ( 1,6.5 ) @xmath151 ; ( -2,-0.5 ) rectangle ( 3,6.5 ) ; ( 1,-0.5 ) ( 1,6.5 ) ; at ( 1,6.5 ) @xmath31 ; at ( 1,6.5 ) @xmath151 ; + ( -2,-0.5 ) rectangle ( 3,6.5 ) ; ( 1,-0.5 ) ( 1,6.5 ) ; at ( 1,6.5 ) @xmath31 ; at ( 1,6.5 ) @xmath151 ; ( -2,-0.5 ) rectangle ( 3,6.5 ) ; ( 1,-0.5 ) ( 1,6.5 ) ; at ( 1,6.5 ) @xmath31 ; at ( 1,6.5 ) @xmath151 ; not all markings need to be included in @xmath150 . since @xmath43 remains a subcircuit of @xmath182 and the values of nodes in @xmath31 do not change , all markings whose support is completely contained in @xmath31 are automatically preserved . only markings using nodes in @xmath151 have to be put into @xmath150 . in most applications , @xmath150 consists only of a constant number of markings . the bound for the circuit growth given in proposition [ prop : extend_reduce ] is a rather crude one . a more detailed analysis shows that calling prolongbasechainis only necessary once every time @xmath183 grows by a factor @xmath1 . if one does some `` cleaning up '' in the circuit ( for instance delete unmarked nodes with no incoming edges ) , @xcite shows that the growth during reduction is even bounded by @xmath79 . however , this bound is of no importance in our applications since cloning during arithmetic operations increases the size by @xmath160 anyway . in practice , the circuit size rarely ever increases at all during reduction . usually , the cicuit even shrinks . [ thm : reduce](cf . @xcite , cor . 7 ) there is a procedure reducewhich given a power circuit @xmath30 and a list @xmath184 of markings in @xmath43 , returns a reduced circuit @xmath185 and a list @xmath186 of markings in @xmath182 such that @xmath187 ( @xmath188 ) . reducetakes @xmath189 time and the size of @xmath190 is bounded by @xmath191 . invoke extendreductiontaking @xmath192 as the reduced part and the whole circuit as @xmath151 . step [ alg : extend_reduce : abort ] in extendreductiontests whether @xmath60 . this is one of the equivalent conditions specified in lemma [ lem : pc_condition ] for a graph to be a power circuit . therefore , reduction is at the same time a test whether a graph is a power circuit : [ cor : test_graph_pc](@xcite , cor . 8) given a dag @xmath30 it can be determined in @xmath193 time whether @xmath43 is a power circuit . in this section , we will show that using a richer data structure for reduced circuits , the time complexity of extendreductioncan be reduced to @xmath194 . this eliminates the logarithmic factors both for reduceand for the test presented in corollary [ cor : test_graph_pc ] . we start by taking a closer look at power sums . a power sum is a formal sum @xmath195 whose coefficients @xmath196 are in @xmath197 and only finitely many of them are non - zero . the value @xmath198 of the power sum @xmath199 is defined in the obvious way . on the set of all power sums , we define a rewriting system @xmath200 generated by the rules @xmath201 and for @xmath202 @xmath203 none of these rules changes the value of the power sum . we omit the proof for the following lemma , since it consists of a long but simple enumeration of cases . [ lem : power_sum_system_confluent ] the rewriting system @xmath200 is locally confluent . [ lem : power_sum_system_terminating ] the rewriting system @xmath200 is terminating ( noetherian ) and hence confluent ( @xcite , thm . 1.1.13 ) . let @xmath204 be a sequence of rewritings . since none of the rules of @xmath200 increases the number of non - zero coefficients , this number must eventually reach a minimum . thus , ignoring a finite number of terms , we can assume that the number of non - zeros is constant within the sequence . no rule in @xmath200 moves a non - zero coefficient to the left ( in the direction of smaller exponents ) . as the value of the @xmath205 is fixed , non - zeros can not be moved indefinitely to the right either . again , by disregarding a finite prefix of the sequence , we assume that the positions of the non - zero coefficients are fixed . at this point , no application of @xmath206 or @xmath207 is possible any more . finally , rules of type @xmath208 and @xmath209 move pairs of consecutive coefficients with opposite signs to the left ( or remove them ) , which can also occur only finitely often . thus , the sequence @xmath210 is eventually constant . we call power sums that are irreducible with respect to @xmath200 compact . if @xmath211 is compact , then so is @xmath212 . [ prop : power_sum_properties ] [ prop : power_sum_properties : unique ] for each power sum there is a unique compact power sum of the same value . [ prop : power_sum_properties : min ] compact power sums have the minimal number of non - zero coefficients among all power sums of the same value . [ prop : power_sum_properties : inc ] if @xmath199 and @xmath213 are compact power sums , then @xmath214 if and only if @xmath215 for some power sum @xmath151 , @xmath216 , and for each @xmath217 either @xmath218 and @xmath219 or @xmath220 and @xmath221 . [ prop : power_sum_properties : lex ] the usual order on @xmath197 gives rise to a lexicographical order on power sums ( where coefficients of higher powers of @xmath1 are compared before those of lower powers ) . restricted to compact power sums , this lexicographical order coincides with the order by values . for [ prop : power_sum_properties : unique ] it suffices to show that for two power sums @xmath199 and @xmath213 of the same value , we have @xmath222 . this is true , since applying the rules of @xmath200 forward or backward , one can turn any power sum into an ordinary @xmath1-ary number with coefficients from @xmath223 . ( for instance , for positive values of @xmath199 , use rules of type @xmath208 backward and rules of type @xmath209 forward to push negative coefficients to the right . ) claim [ prop : power_sum_properties : min ] follows from the fact that no rule increases the number of non - zero coefficients . for the `` if '' part of [ prop : power_sum_properties : inc ] , we observe that component - wise subtraction yields @xmath224 which evaluates to @xmath79 . for the `` only if '' part , let @xmath225 be compact . consider @xmath226 . if @xmath227 is a valid power sum ( i.e. @xmath228 ) and @xmath227 is compact , it already has the desired form ( for @xmath229 ) . otherwise we have one of the following cases : @xmath227 is not a valid power sum , since @xmath230 . let @xmath231 be the maximum number such that @xmath232 . we transform @xmath227 into @xmath233 and use induction on @xmath234 . a rule of type @xmath208 can be applied . we have @xmath235 and @xmath236 . applying the rule gives @xmath237 use induction on @xmath238 . a rule of type @xmath206 can be applied . we have @xmath235 and @xmath239 and @xmath240 for some @xmath241 . this yields @xmath242 and induction applies to @xmath243 . finally , [ prop : power_sum_properties : lex ] is a consequence of [ prop : power_sum_properties : inc ] . the notion of compactness was introduced in @xcite for @xmath2 and subsequently used in @xcite . our definition is a generalization that inherits most of the original characteristics . there is , however , one important difference : it is much less obvious for @xmath244 how to make a power sum compact in linear time . in the case @xmath2 it suffices to apply the rules of @xmath200 from left to right . yet , if for instance @xmath55 , the application of rule @xmath208 to @xmath245 turns the previously compact prefix @xmath246 into the @xmath200-reducible sum @xmath247 . [ prop : compactify_power_sum ] any power sum @xmath248 can be transformed into a compact power sum with the same value in @xmath161 time . any two power sums @xmath199 and @xmath213 with @xmath249 can be transformed into each other using only replacements of the form @xmath250 moreover , at most one application of @xmath251 or @xmath252 is needed for each @xmath130 . in fact , whether @xmath251 or @xmath252 or neither is needed , depends only on @xmath199 and @xmath213 . this can be seen using induction on @xmath130 . let @xmath253 be the compact power sum with @xmath249 . define @xmath254 ( @xmath255 ) depending on whether the replacement @xmath251 or @xmath252 or neither of them occurs in the sequence @xmath256 . for notational conveniance , we define @xmath257 and @xmath258 . then we have @xmath259 for @xmath260 . it remains to compute the values @xmath261 ( @xmath262 ) . these are the unique solution of the following system of conditions ( @xmath260 ) : @xmath263 for @xmath264 and for all @xmath265 we define @xmath266 $ ] to be the set of possible values for @xmath261 , provided that @xmath267 and @xmath268 : @xmath269:=\{j_i\>:\>&\text{there are $ j_1,\ldots , j_{i-3}$ such that}\cr & \text{$j_0=0,j_1,\ldots , j_{i-3},j_{i-2}=j , j_{i-1}=k , j_i$ satisfy $ ( \meddiamond_0),\ldots,(\meddiamond_{i-1})$}\}\end{aligned}\ ] ] since @xmath270 only affects @xmath271 , the sets @xmath272 $ ] can be computed in constant time using @xmath273 $ ] . after this , the solution @xmath274 can be read from right to left . [ ex : compactify_power_sum ] suppose that @xmath55 and we want to make @xmath275 compact . we get : lcccccc@xmath266 $ ] & @xmath276 & @xmath277 & @xmath278 & @xmath279 & @xmath280 & @xmath281@xmath282 & @xmath192 & @xmath192 & @xmath192 & @xmath283 & @xmath192 & @xmath192@xmath284 & @xmath192 & @xmath192 & @xmath192 & @xmath192 & @xmath285 & @xmath192@xmath286 & @xmath192 & @xmath192 & @xmath192 & @xmath192 & @xmath192 & @xmath192@xmath287 & @xmath192 & @xmath192 & @xmath288 & @xmath192 & @xmath192 & @xmath192@xmath289 & @xmath283 & @xmath288 & @xmath192 & @xmath192 & @xmath192 & @xmath283 @xmath290 & @xmath291 & @xmath192 & @xmath192 & @xmath192 & @xmath192 & @xmath192@xmath292 & @xmath192 & @xmath192 & @xmath192 & @xmath192 & @xmath192 & @xmath192@xmath293 & @xmath192 & @xmath192 & @xmath192 & @xmath192 & @xmath192 & @xmath192@xmath294 & @xmath192 & @xmath283 & @xmath192 & @xmath192 & @xmath192 & @xmath192 since @xmath295 , we deduce from the last column that @xmath296 as well . the only @xmath297 in the column for @xmath280 with @xmath298 yields @xmath299 and so on . we end up with @xmath300 which results in @xmath301 . property [ prop : power_sum_properties : lex ] of proposition [ prop : power_sum_properties ] is the main motivation for the following definition . [ def : treed_pc](cf . @xcite , def . 5 ) a power @xmath30 circuit is called _ treed _ ( a convenient shorthand for `` reduced and equipped with additional data in the form of a tree '' ) , if [ def : treed_pc : red ] @xmath43 is reduced ( i.e. , different nodes evaluate to different values and it is equipped with an ordered list @xmath302 and a bit vector , see definition [ def : reduced_pc ] ) [ def : treed_pc : tree ] @xmath43 is equipped with a directed tree @xmath213 in which each node has up to @xmath303 outgoing edges , labelled with pairwise distinct values from @xmath197 ( and ordered from left to right by increasing values ) . for each leaf in @xmath213 , the unique path from the root to this leaf must consist of exactly @xmath304 edges . the sequence of labels @xmath305 of such a path ( read from the leaf to the root ) corresponds to a marking @xmath306 . for all leaves , the power sum @xmath307 ( which evaluates to @xmath103 ) must be compact . any marking represented as a leaf in this way is called compact . [ def : treed_pc : comp ] all successor markings @xmath49 ( @xmath48 ) are compact . [ def : treed_pc : mark ] any node @xmath164 that is contained in the support of some marking ( compact or not ) is not the top node of a maximal chain . [ def : treed_pc : lev ] for each level in the tree , there is a list containing the nodes of this level . figure [ fig : treed_circuit ] shows an example of a treed circuit ( for @xmath2 ) alongside the tree containing its markings . note that the order of the nodes is implicitly given by the lowest level of the tree . \(e ) at ( 0,0 ) ; ( 0 ) at ( 0,-1 ) ; ( 00 ) at ( 0,-2 ) ; ( 00 m ) at ( -1.5,-3 ) ; ( 000 ) at ( 0,-3 ) ; ( 00p ) at ( 3.5,-3 ) ; ( 00m0 ) at ( -1.5,-4 ) ; ( 0000 ) at ( 0,-4 ) ; ( 000p ) at ( 2,-4 ) ; ( 00p0 ) at ( 3.5,-4 ) ; ( 00m0p ) at ( -1,-5 ) ; ( 00000 ) at ( 0,-5 ) ; ( 0000p ) at ( 1,-5 ) ; ( 000p0 ) at ( 2,-5 ) ; ( 00p0 m ) at ( 3,-5 ) ; ( 00p0p ) at ( 4,-5 ) ; ( e ) edge node[right ] @xmath297 ( 0 ) ( 0 ) edge node[right ] @xmath297 ( 00 ) ( 00 ) edge node[above left=-0.1 ] @xmath308 ( 00 m ) ( 00 ) edge node[right ] @xmath297 ( 000 ) ( 00 ) edge node[above right=-0.1 ] @xmath309 ( 00p ) ( 00 m ) edge node[right ] @xmath297 ( 00m0 ) ( 000 ) edge node[right ] @xmath297 ( 0000 ) ( 000 ) edge node[above right=-0.1 ] @xmath309 ( 000p ) ( 00p ) edge node[right ] @xmath297 ( 00p0 ) ( 00m0 ) edge node[right ] @xmath309 ( 00m0p ) ( 0000 ) edge node[left ] @xmath297 ( 00000 ) ( 0000 ) edge node[right ] @xmath309 ( 0000p ) ( 000p ) edge node[left ] @xmath297 ( 000p0 ) ( 00p0 ) edge node[left ] @xmath308 ( 00p0 m ) ( 00p0 ) edge node[right ] @xmath309 ( 00p0p ) ; ( u1 ) at ( -2,-4.5 ) @xmath137 ; ( u2 ) at ( -2,-3.5 ) @xmath310 ; ( u3 ) at ( -2,-2.5 ) @xmath311 ; ( u4 ) at ( -2,-1.5 ) @xmath312 ; ( u5 ) at ( -2,-0.5 ) @xmath313 ; ( li0 ) at ( -2.5,0 ) ; ( li1 ) at ( -2.5,-1 ) ; ( li2 ) at ( -2.5,-2 ) ; ( li3 ) at ( -2.5,-3 ) ; ( li4 ) at ( -2.5,-4 ) ; ( li5 ) at ( -2.5,-5 ) ; ( lt0 ) at ( 5,0 ) ; ( lt1 ) at ( 5,-1 ) ; ( lt2 ) at ( 5,-2 ) ; ( lt3 ) at ( 5,-3 ) ; ( lt4 ) at ( 5,-4 ) ; ( lt5 ) at ( 5,-5 ) ; ( li0 ) edge ( e ) ( e ) edge ( lt0 ) ( li1 ) edge ( 0 ) ( 0 ) edge ( lt1 ) ( li2 ) edge ( 00 ) ( 00 ) edge ( lt2 ) ( li3 ) edge ( 00 m ) ( 00 m ) edge ( 000 ) ( 000 ) edge ( 00p ) ( 00p ) edge ( lt3 ) ( li4 ) edge ( 00m0 ) ( 00m0 ) edge ( 0000 ) ( 0000 ) edge ( 000p ) ( 000p ) edge ( 00p0 ) ( 00p0 ) edge ( lt4 ) ( li5 ) edge ( 00m0p ) ( 00m0p ) edge ( 00000 ) ( 00000 ) edge ( 0000p ) ( 0000p ) edge ( 000p0 ) ( 000p0 ) edge ( 00p0 m ) ( 00p0 m ) edge ( 00p0p ) ( 00p0p ) edge ( lt5 ) ; ( m ) at ( -1,-5 ) @xmath42 ; ( lv1 ) at ( 0,-5 ) @xmath314 ; ( lv2 ) at ( 1,-5 ) @xmath315 ; ( lv3 ) at ( 2,-5 ) @xmath316 ; ( lv4 ) at ( 3,-5 ) @xmath317 ; ( lv5 ) at ( 4,-5 ) @xmath318 ; the most time consuming step in extendreductionwas to find the position of a new node in the sorted list of @xmath31 . using binary search , this took @xmath319 time . if @xmath31 is treed and @xmath49 is compact , we can improve this to @xmath161 : the position of the leaf corresponding to @xmath49 already tells the position of @xmath34 in @xmath31 . in order to adjust the bit vector , we have to read the paths from the root of the tree to the respective leaves and check the condition given by proposition [ prop : power_sum_properties ] [ prop : power_sum_properties : inc ] . yet , making a circuit treed is more complicated than just reducing it . for the time analysis , we use amortization with respect to a potential function @xmath320 , mapping power circuits to numbers , see 17.3 in @xcite . an algorithm on power circuits is said to run in amortized time @xmath321 , if the real running time is bounded by @xmath322 , where @xmath43 is the input and @xmath182 is the resulting circuit . thus , an algorithm may take longer than its indicated amortized time , as long as it decreases the potential by the same amount . the potential can be thought of as a debt which is accumulated whenever the algorithm does not terminate in time . for a power circuit @xmath30 , the number of maximal chains is denoted @xmath323 . the _ potential _ of @xmath43 is @xmath324 . in a situation where @xmath325 is a graph with an embedded power circuit @xmath326 , we define @xmath327 and @xmath328 . [ lem : insert_node ] there is a procedure insertnode , which takes a treed power circuit @xmath30 and a compact marking @xmath42 in @xmath43 and which inserts a new node @xmath34 with @xmath145 into @xmath43 . insertnoderuns in amortized time @xmath107 . since @xmath42 is compact , the position of @xmath35 in the ordered list of nodes is determined by the position of the leaf corresponding to @xmath42 in the tree . the bit vector can be adjusted by comparing @xmath42 to the successor markings of the nodes immediately before and after @xmath35 . the tree has to be `` stretched '' by inserting a new level corresponding to @xmath35 . all edges on this level are labelled @xmath297 , since no marking contains the new node @xmath35 . using the lists of nodes of the same level , this takes @xmath107 time . the insertion of @xmath35 increases @xmath183 and in some cases also @xmath323 . thus , the potential grows by up to @xmath329 . incrementing a compact marking by @xmath79 is easier than doing the same with an arbitrary marking . let @xmath42 be a compact marking in a treed circuit , in which @xmath330 are the unique nodes with values @xmath79 and @xmath28 respectively ( assuming they exist ) . if @xmath331 , @xmath332 can be incremented directly . if @xmath333 , then @xmath334 thanks to compactness . set @xmath335 and increment @xmath336 . however , the resulting marking is not always compact . [ lem : compactify_marking ] there is a procedure compactifymarkingwhich , given an arbitrary marking @xmath42 in a treed circuit @xmath30 , computes a compact marking @xmath91 with @xmath337 in @xmath107 time . this is based on the following observation : if @xmath338 is a power sum , then in the corresponding compact power sum @xmath213 only the coefficients of @xmath339 can be non - zero . hence , for each individual chain @xmath340 , we can apply the algorithm from proposition [ prop : compactify_power_sum ] to the power sum defined by @xmath341 . note that a node for @xmath342 exists due to property [ def : treed_pc : mark ] of the treed circuit @xmath43 . strictly speaking , we have not proved that the resulting circuit is treed . in fact , condition [ def : treed_pc : mark ] demanding an unmarked node at the top of every chain might have been lost during compactification . we will fix this shortly . for now , we content ourselves with the observation that the problem does not arise when compactifymarkingis called during prolongbasechain . in this special case , the successor marking of the new node with value @xmath343 only uses nodes with much smaller values from the base chain ( approximately the first @xmath344 ) . this shows : [ cor : prolong_base_chain_treed ] the procedure prolongbasechaingiven in algorithm [ alg : prolong_base_chain ] can be adapted for treed power circuits . the amortized time complexity is @xmath107 . [ cor : increment_compact_marking ] there is a procedure incrementmarkingwhich , given a compact marking @xmath42 in a treed power circuit @xmath30 , computes a compact marking @xmath91 with @xmath345 and leaves @xmath43 treed . incrementmarkingtakes @xmath107 amortized time and increases the circuit size by @xmath346 . as discussed above , incrementing @xmath42 by @xmath79 only affects the two nodes with values @xmath79 and @xmath28 . as a consequence , the compactification process of the incremented marking is limited to the base chain of @xmath43 . in order to fulfill condition [ def : treed_pc : mark ] of definition [ def : treed_pc ] , we invoke prolongbasechainwhich creates a new node @xmath34 . if @xmath34 is the new top node of the base chain , we are done . otherwise the insertion of @xmath34 has linked two maximal chains , decreasing @xmath323 by one . we use the released potential to pay for the @xmath107 time used so far and repeat . for treed circuits , the procedure extendtreegiven in algorithm [ alg : extend_tree ] replaces extendreduction . [ alg : extend_tree ] compute a topological order @xmath152 . [ prop : extend_tree ] the procedure extendtreeis correct and runs in amortized time @xmath347 . the circuit growth @xmath156 is bounded by @xmath348 . the basic structure of extendtreeis the same as that of extendreduction , so we focus on the differences . let @xmath159 . between cycles of the main loop , we keep up the following invariants for all markings @xmath349 : [ prop : extend_tree : inv_comp ] if the support of @xmath42 is completely contained in @xmath31 , @xmath42 is compact . [ prop : extend_tree : inv_top ] @xmath31 is treed . in particular , condition [ def : treed_pc : mark ] holds , which means that for all nodes @xmath350 , the top node of the maximal chain starting at @xmath35 is not marked by @xmath42 . at the beginning , both invariants are true by definition . the time complexity for finding @xmath168 in step [ alg : extend_tree : search ] is reduced to @xmath161 due to the representation of the markings in the tree . if @xmath31 contains no node with the same value as @xmath167 , we can insert it as we did in extendreduction . remember that @xmath351 is compact due to [ prop : extend_tree : inv_comp ] . the case when @xmath352 also ressembles extendreductionand has the same ( now amortized ) time bound . all markings @xmath42 whose support is completely contained in @xmath31 after the processing of @xmath167 must be made compact in order to regain [ prop : extend_tree : inv_comp ] . this is done in step [ alg : extend_tree : comp_marks ] . after that , we have to restore condition [ def : treed_pc : mark ] . using the same argument as in corollary [ cor : increment_compact_marking ] , step [ alg : extend_tree : restore_inv_top ] takes @xmath161 amortized time and causes the circuit to grow by at most @xmath353 nodes . together with step [ alg : extend_tree : prolong_chain ] , the overall circuit growth during extendtreeis bounded by @xmath348 . [ thm : make_tree](treed analogon of theorem [ thm : reduce ] ) there is a procedure maketreewhich given a power circuit @xmath30 and a list @xmath184 of markings in @xmath43 , returns a treed circuit @xmath185 and a list @xmath186 of compact markings in @xmath182 such that @xmath187 ( @xmath188 ) . maketreetakes @xmath354 time and the size of @xmath190 is bounded by @xmath355 . [ rem : pc_strategy](working with treed power circuits ) + the usual strategy when solving a problem using power circuits is to create one power circuit and keep all integers as markings in this circuit . the power circuit is kept in treed form in order that comparisons can be done efficiently at any time . for each arithmetic operation , the markings corresponding to the operands are cloned , and the operation ( addition or multiplication by a power of two ) is performed on the clones . finally , extendtreeis called with the set of clones as @xmath151 to regain a treed circuit . this takes @xmath194 time and this time bound also absorbs everything else done during the operation . for an estimate of the time complexity of an entire algorithm , we need to keep track of the circuit size @xmath183 as well as the size @xmath356 of the supports of the markings . the latter is usually called the `` weight '' of the ciruit and determines the growth during each operation . the cost for one operation is @xmath357 . if we start with a comparatively small circuit and @xmath356 remains constant during the algorithm ( which is normally the case ) , then after @xmath358 operations the circuit size is bounded by @xmath359 and the time by @xmath360 . in our main application the solution of the word problem in higman s group @xmath358 will turn out to be quadratic and @xmath356 linear in the input size @xmath18 , leading to an @xmath8 time algorithm . seen from the outside , extendreductionand extendtreeas well as reduceand maketreebehave very much alike . in applications , all four procedures are used as `` black boxes '' and of the resulting circuits only the weaker property of reducedness is used . therefore , in order to simplify nomenclature , we will speak of `` reduction '' and `` reduced '' circuits , subsuming both concepts . the reader may then choose whether to use the simpler reduction concept at the cost of logarithmic factors or to go through the more complicated procedures for treed power circuits with better asymptotic time complexity . although the main goal of this paper is to solve the word problem in higmans groups , we sidetrack briefly to present another generalization that is made possible by power circuits with arbitrary base @xmath1 . the baumslag - gersten group is defined as @xmath361 this is an hnn extension of @xmath4 generated by @xmath362 and @xmath321 . replacing @xmath4 by @xmath3 ( for @xmath0 ) , we get a family of generalized baumslag - gersten groups : @xmath363 [ thm : wp_bg ] the word problem for the generalized baumslag - gersten group @xmath364 is solvable in @xmath365 time . the proof of theorem [ thm : wp_bg ] is literally the same as that for @xmath25 which was given in @xcite and @xcite , except that the new base @xmath1 power circuits from section [ sec : pc ] are used . we generalize the groups @xmath11 defined in the introduction by replacing the underlying baumslag - solitar group @xmath4 by @xmath3 . [ def : higman_group ] the ( generalized ) higman group @xmath366 is defined as @xmath367 while @xmath368 ( @xmath369 ) retains all the important properties of @xmath10 ( infinite , huge compression , no non - trivial normal subgroup of finite index ) , this is not entirely true for @xmath366 in general . for example , for all @xmath12 , the homomorphism given by @xmath370 sends @xmath371 onto a finite non - trivial group . in this section , we will prove : [ thm : wp_hig ] let @xmath372 . the word problem for the generalized higman group @xmath366 can be solved in @xmath8 time . the key observation for the solution of the word problem is the decomposition of @xmath366 into a series of amalgamations of @xmath13 copies of the baumslag - solitar group @xmath373 , see @xcite . @xmath374 where @xmath375 and in both cases @xmath376 is the subgroup generated by @xmath377 and @xmath378 , which in fact freely generate @xmath376 ( if @xmath6 ) . furthermore , we can break @xmath379 and @xmath380 down to @xmath381 where @xmath382 and the indices are read in @xmath383 . each group @xmath384 is a copy of the baumslag - solitar group @xmath3 and thus isomorphic to the semidirect product @xmath385\rtimes{\ensuremath{\mathbb{z}}}$ ] which consists of pairs @xmath386\rtimes{\ensuremath{\mathbb{z}}}$ ] . the isomorphism is given by @xmath387 and @xmath388 . in @xmath385\rtimes{\ensuremath{\mathbb{z}}}$ ] , we have the following formulae for multiplication and inversion : @xmath389 when dealing with more than one group @xmath384 , we add @xmath130 as a subscript to those pairs designating an element of @xmath384 . in order to solve the word problem for @xmath366 , we first need a solution for the subgroup membership problem of @xmath390 in @xmath391 ( with @xmath392 ; this covers both @xmath379 and @xmath380 ) . furthermore , we have to do this in an effective way , i.e. , given a sequence of pairs @xmath393 which represents an element of @xmath390 , we have to find a corresponding sequence of pairs of the form @xmath394 and @xmath395 . we start by giving a reduction system @xmath396 for @xmath391 : @xmath397 the system @xmath396 is not confluent in general , but the following property holds : [ prop : britton_g_1_f ] if @xmath398 is an @xmath396-reduced word that equals @xmath79 in @xmath391 , then @xmath398 is the empty word . this ressembles britton s lemma for hnn extensions . in fact , bass - serre theory provides a unifying notion ( and proof ) for both phenomena . we give no further proof here , but instead apply the system @xmath396 to the subgroup membership problem . let @xmath399 be the system @xmath396 extended by the rules @xmath400 where all @xmath401 . the new rules respect the group structure , and hence proposition [ prop : britton_g_1_f ] holds for @xmath399 as well . the new rules are not length - increasing , since @xmath392 . starting with an arbitrary sequence @xmath398 of pairs @xmath393 representing an element in @xmath391 , one can compute an equivalent @xmath399-reduced word @xmath402 with linearly many operations : first , compute an @xmath396-reduced word @xmath403 , then apply rules @xmath404 and @xmath405 . note that the latter leave @xmath403 @xmath396-reduced . only the second pair generated by either of these rules can be part of another application of @xmath404 or @xmath405 . therefore , @xmath403 can be @xmath399-reduced with one pass from left to right . let @xmath398 be a sequence of pairs @xmath393 which represents an element of the subgroup @xmath406 . if @xmath398 is @xmath399-reduced , then @xmath398 is already an alternating sequence of pairs of types @xmath394 and @xmath395 . let @xmath407 . we assume that @xmath408 and that @xmath403 contains no trivial pairs @xmath409 which makes @xmath403 @xmath396-reduced . the case where @xmath403 starts with @xmath410 is similar . the sequence @xmath411 equals @xmath79 in @xmath391 and must therefore @xmath396-reduce to the empty sequence . note that both @xmath412 and @xmath398 are @xmath396-reduced , so any @xmath396-reduction can only occur at the border between the two words . clearly , we can not have @xmath413 or else @xmath414 would be @xmath396-reduced . if @xmath415 , then a reduction of type @xmath209 is possible if @xmath416 and @xmath417 , in which case we get @xmath418 . but after that , the sequence is @xmath396-reduced since @xmath416 and @xmath417 imply @xmath419 . hence , we are left with @xmath420 . in that case , we get @xmath421 . if this is @xmath422 , we have @xmath423 and we proceed inductively with the remaining sequence . otherwise , we must have @xmath424 in order to continue applying rules . if @xmath425 , the next rule can only apply to @xmath426 , so @xmath427 and we get @xmath428 . again , the sequence is @xmath396-reduced unless @xmath429 . we iterate this argument until we arrive at @xmath430 on the way , we have found @xmath431 , and @xmath432 for @xmath433 . one further reduction of type @xmath207 brings us to @xmath434 since @xmath435 , the next reduction requires @xmath436 . thus , rule ( 6 ) of @xmath399 can be applied to the prefix @xmath437 of the original word @xmath398 . for the amalgamated product @xmath438 , a property similar to [ prop : britton_g_1_f ] holds : [ prop : britton_h ] let @xmath439 be a non - empty sequence with @xmath440 or @xmath441 , alternatingly . if @xmath398 equals @xmath79 in @xmath366 , then @xmath442 for some index @xmath130 . [ alg : wp_hig ] rewrite the input @xmath398 by replacing each @xmath443 by @xmath444.[alg : wp_hig : init ] break @xmath398 into subsequences @xmath445 such that in each @xmath446 the subscripts of all pairs are either in @xmath447 or in @xmath448 . let @xmath449 . whether @xmath450 . from this proposition , we can derive algorithm [ alg : wp_hig ] which solves the word problem in @xmath366 . in this algorithm , the word @xmath445 is split into @xmath451 and @xmath452 . the first part is an @xmath399-reduced alternating sequence of group elements from @xmath379 or @xmath380 with no @xmath453 ( @xmath454 ) being in the subgroup @xmath376 . in each loop cycle either @xmath321 increases , or @xmath321 decreases by one and at the same time , some @xmath453 is merged with @xmath455 , which means that @xmath358 decreases . thus , the loop is executed only linearly often . in order to get a time bound for algorithm [ alg : wp_hig ] , it remains to show how to perform the tests arithmetic operations on the pairs @xmath393 efficiently . since power circuits are designed to work with integers , we have to avoid fractions for the first components of pairs @xmath456\rtimes{\ensuremath{\mathbb{z}}}$ ] . therefore , we use the triple notation introduced in @xcite . for @xmath457 with @xmath458 , let @xmath459_i:=(u\cdot q^x , x+k)_i\in g_{i , i+1}.\ ] ] if @xmath151 , @xmath460 , and @xmath67 are markings in a base @xmath1 power circuit , we call @xmath461_i$ ] a triple marking and define its value by @xmath462_i\in g_{i , i+1}$ ] . any element of @xmath384 can be written as a triple , but not in a unique way . for instance , @xmath463_i=(2,0)_i=[4,-1,1]_i$ ] , if @xmath2 . the group operations translate to formulae for multiplication and inversion of triples : @xmath464\cdot[v , y,\ell]&=[u\cdot q^{-y}+v\cdot q^k , x+y , k+\ell]\cr [ u , x , k]^{-1}&=[-u ,- k ,- x]\end{aligned}\ ] ] furthermore , @xmath465_i\in{\ensuremath{\langle a_i\rangle}}\le g_{i , i+1}$ ] if and only if @xmath466 and @xmath467 . similarly , @xmath465_i\in{\ensuremath{\langle a_{i+1}\rangle}}\le g_{i , i+1}$ ] if and only if @xmath468 , and finally @xmath465_i$ ] is the group identity if and only if @xmath468 and @xmath466 . at the beginning of algorithm [ alg : wp_hig ] , we create a power circuit with base @xmath1 consisting of a single node @xmath34 with @xmath53 . we represent each pair @xmath393 by a triple marking . of the three markings in each initial triple , two are zero ( empty ) and the third has either value @xmath309 or @xmath308 and can be created using @xmath34 . let @xmath356 be the sum of the sizes of ( the supports of ) all these markings . we call @xmath356 the weight of the circuit . from the multiplication formula for triples we see that @xmath356 never increases during the algorithm , keeping in mind that after an operation we can `` forget '' the operand and just keep the result . the initial value of @xmath356 is exactly @xmath469 . after step [ alg : wp_hig : init ] , we reduce the circuit , which takes @xmath470 time . from now on , we keep @xmath43 reduced following the strategy proposed in remark [ rem : pc_strategy ] . the swapping operation @xmath471 works in the following way for triples : @xmath464&\mapsto \begin{cases } [ 0,0,u\cdot q^x]&\text{if $ x =- k$ and $ q^x\mid u$ and $ u\ge 0$}\cr [ 0,u\cdot q^x,0]&\text{if $ x =- k$ and $ q^x\mid u$ and $ u<0 $ } \end{cases}\cr [ 0,x , k]&\mapsto[x+k,0,0]\end{aligned}\ ] ] the whole algorithm computes @xmath161 many times an @xmath399-irreducible word . each of these computations necessitates @xmath161 arithmetic operations ( and subsequent calls to extendreduction ) . the circuit size remains bounded by @xmath472 . thus , one call of extendreductiontakes @xmath473 time . we get a total time bound of @xmath8 . this concludes the proof of theorem [ thm : wp_hig ] . we have shown that the word problem for the generalized higman groups @xmath366 is solvable in polynomial time . an important ingredient for this result was the extension of the power circuit data structure to arbitrary bases @xmath0 . from an algorithmic point of view , this is an interesting result in itself and may provide a useful tool in group theory as well as other areas . the techniques used in this paper do not apply to the even more general groups @xmath474 and @xmath475 , where the underlying baumslag - solitar group @xmath3 is replaced by @xmath476 for some @xmath477 . this is because @xmath478 is not a semi - direct product when @xmath479 . the word problem for these groups is open . note that even for @xmath15 the group @xmath474 can be non - trivial if @xmath479 .	this paper continues the 2012 stacs contribution by diekert , ushakov , and the author . we extend the results published in the proceedings in two ways . first , we show that the data structure of power circuits can be generalized to work with arbitrary bases @xmath0 . this results in a data structure that can hold huge integers , arising by iteratively forming powers of @xmath1 . we show that the properties of power circuits known for @xmath2 translate to the general case . this generalization is non - trivial and additional techniques are required to preserve the time bounds of arithmetic operations that were shown for the case @xmath2 . the extended power circuit model permits us to conduct operations in the baumslag - solitar group @xmath3 as efficiently as in @xmath4 . this allows us to solve the word problem in the generalization @xmath5 of higman s group , which is an amalgamated product of four copies of the baumslag - solitar group @xmath3 rather than @xmath4 in the original form . as a second result , we allow arbitrary numbers @xmath6 of copies of @xmath3 , leading to an even more generalized notion of higman groups @xmath7 . we prove that the word problem of the latter can still be solved within the @xmath8 time bound that was shown for @xmath9 . keywords : data structures ; compression ; algorithmic group theory ; word problem .
low - temperature crystal solids are full of wonderful and surprising phenomena , that are of both fundamental and practical importance but often difficult to observe owing to the imperfect nature and inflexibility of solid state systems . free of these shortcomings , cold atoms , offer a pristine and highly controllable environment for creating novel quantum matter . experimental realization of the superfluid mott insulator transition , a quintessential condensed matter phenomenon , in cold atom lattices @xcite as exactly predicted by theory @xcite serves as one of many remarkable examples . it is experimental developments like this , along with the ability to create cold atom lattices that meet virtually any need , that has motivated tremendous activity in the past decade in using cold atom lattice models as an excellent platform for exploring and mimicking complicated condensed matter phenomena @xcite . a subject of profound importance in condensed matter physics is superconductivity , a coherent phenomenon arising from condensation of correlated electron pairs , also known as bardeen - cooper - schriffer ( bcs ) pairs , on a macroscopic scale . a bcs pair , however , is not energetically favored unless there exists an attraction between its constituents , which appears impossible in an electron system where coulomb repulsion dominates . the effort to solve this mystery led , through the insights of fllich , to the phonons of ionic crystal lattice oscillations and the eventual understanding that the interaction they induce between electrons is the source of the attraction responsible for superconductivity in most materials @xcite . in cold atom physics , an analogous situation can arise from a bose - fermi mixture , where phonons can exist as low energy excitations of a bose - einstein condensate ( bec ) , describing the density fluctuation of the bec @xcite . the same phonons in the bose gas can then induce an attractive interaction between fermions in the fermi gas @xcite . this , along with the unprecedented ability to tune parameters in such systems , e.g. , two - body interaction and dimensionality , allows for the possibility of clean and controllable realization ( or quantum simulation ) of superfluid pairings in cold atom bose - fermi mixtures @xcite . remarkable experimental progress has been made in creating quantum - degenerate bose - fermi mixtures , e.g. , @xmath5li-@xmath6li @xcite , @xmath7na-@xmath6li @xcite , @xmath8rb-@xmath9k @xcite , @xmath6li-@xmath8rb @xcite , and @xmath7na-@xmath9k @xcite . most recently , a mixture where both bosons and fermions are in the superfluid state has been realized @xcite . these developments together with the recent upsurge of experimental efforts in achieving dipolar quantum gases , in systems with @xmath9k@xmath8rb @xcite , @xmath6li@xmath9k @xcite , and @xmath7na@xmath9k @xcite heteronuclear molecules , @xmath8rb spinor condensates @xcite , and @xmath10cr @xcite , @xmath11dy @xcite and @xmath12er @xcite atoms , has opened up the exciting possibility of creating dipolar bose - fermi mixtures with intriguing and unique properties . of particular relevance to the present work is a bose - fermi mixture involving a two - component fermi gas , examples of which include the mixture of bosonic @xmath13k with two fermionic species @xmath9k@xmath14and @xmath6li @xcite , feshbach molecules with two lowest hyperfine states of the same fermionic species @xmath6li @xcite , and ground state @xmath6li@xmath9k molecules with two fermionic species @xmath6li@xmath15and @xmath9k @xcite . the last one is particularly motivating since the ground @xmath6li@xmath9k molecules are bosons with a large dipole moment of 3.6 debye . in a dipolar bec @xcite , the bosons interact not only via the short - range s - wave but also the long - range dipole - dipole interaction , and the dipolar interaction constitutes an additional control knob inaccessible to a nondipolar condensate and may thus open up new possibilities for engineering quantum gases with novel properties . motivated by this new prospect , we consider a dipolar bose - fermi mixture loaded in a quasi two - dimensional ( 2d ) trap , forming a square optical lattice , where a two - component fermi gas [ fig . [ fig : mixtureinlattice](b ) ] is mixed with a single - component dipolar bec [ fig . [ fig : mixtureinlattice](a ) ] . a plethora of interesting phenomena have been predicted to occur in bose - fermi mixtures in square lattices , including quantum phases of composite particles involving pairing of fermions with boson particles or boson holes @xcite , superfluid - mott insulator and metal - insulator transition @xcite and coexistent phases @xcite within a slave - rotor mean - field approximation , competition between phase separation and supersolid states driven by a combination of van hove instabilities and fermi surface nesting @xcite , superfluid pairings @xcite and superfluid - mott insulator transition @xcite in the presence of artificial staggered magnetic field , and an interplay between density waves and superfluids with unconventional pairing symmetries based on both a mean - field study @xcite and functional renormalization group analysis @xcite . however , thus far these studies have considered nondipolar bosons that interact only through the repulsive hard - core potentials . the focus of the present work is on the last topic competition between density waves and unconventional superfluids , which are the cold atom analogs of the unconventional superconductors in electron systems . conventional superconductors consist of spin - singlet s - wave bcs pairs @xcite where both spin and orbital angular momenta vanish , and it has been well established that the conventional ( low-@xmath16 ) superconductors are those where the pairing mechanism is dominated by the phonon - induced electron - electron interaction . the study of unconventional pairing states , where the gap parameters possess symmetries different from the usual s - wave symmetry , began with the work by anderson and moral @xcite and that by balian and werthamer @xcite , which led to the discovery of a- and b - phase with @xmath2-wave symmetries in superfluid @xmath17he @xcite . it has remained to this day an active area of research , in large part , because of the discovery of high-@xmath16 superconductors in cuprate compounds by bednorz and mller in 1986 @xcite , which are believed to possess unconventional order parameters with @xmath1-wave symmetry @xcite . in this paper , for simplicity , we shall restrict our attention to superfluids where the spatial symmetry of a bcs pair can be classified according to the irreducible representations of @xmath0 , the point group of a square . an important early work by micnas _ @xcite investigated , within mean - field theory , superconducting phases of the same point group , based on an extended hubbard model with on - site repulsive and intersite attractive interactions . it was shown explicitly that when the intersite interaction is restricted to nearest neighbors , symmetrized two - body interactions in momentum space can be cast into separable forms that support parings with @xmath18- , @xmath1- and @xmath2-wave symmetries but not @xmath3-and @xmath4-wave symmetries . in a bose - fermi mixture , however , the effective fermi - fermi interaction , which is mediated by the phonons of a bec , has an interaction range that depends on , among other things , the effective healing length . this healing length is tunable by changing , e.g. , the boson hopping amplitude . in principle , as the healing length increases , the effective fermi - fermi interaction may reach over many lattice sites @xcite . thus , classification by micnas _ @xcite holds only in the limit where the healing length is much smaller than the lattice constant , which is precisely the argument used by bukov and pollet in their recent work @xcite to limit superfluids to those with @xmath18- , @xmath1- , and @xmath2-wave symmetries . in contrast , we do not place such a constraint on our model and we therefore include superfluids with @xmath3- and @xmath4-wave symmetries , which can not be excluded a priori in the presence of a long - range interaction @xcite . the inclusion of these two phases allows us to take into consideration a wider choice of possibilities when analyzing phase diagrams , making our studies , to some extent , more rigorous than those without them . this paper is organized as follows . in sec . ii , we introduce a set of 2d two - body interactions including onsite and offsite ( nearest - neighbors ) dipolar ones , reducing our 3d trap model to a quasi 2d system in a square optical lattice and describe it with the single - band bose - fermi hamiltonian in both real and reciprocal lattice space . in sec . iii , we introduce phonons that obey the dispersion spectra of a homogeneous dipolar bec . in the same section , we eliminate the phonon degrees of freedom and arrive at an effective hamiltonian where fermions interact not only with the hard - core potential but also the phonon - induced interaction . in sec . iv , we take the hartree - fock bogoliubov mean - field approach and construct , in the space spanned by the base functions of the irreducible representations of @xmath0 , the matrix representation of the linearized gap equation . the main complexity in this endeavor comes from evaluating the related matrix elements which are 4d integrals in momentum space involving symmetrized interactions that depend on momenta in a nontrivial way , something absent in the simplest bcs theory . the gap equation in the simplest bcs theory ( for the s - wave superfluid pairing in continuous 3d models ) can be reduced to eq . ( [ 3d s ] ) involving a 1d integral , which is the origin of the well - known formula , eq . ( [ 3d critical temperature ] ) , that one uses to estimate the critical temperature . a major contribution of the present work is that by using a combination of algebra involving elliptical integrals and symmetry considerations , we are able to reduce the 4d integrals in our model into 1d integrals . this reduction results in a set of 1d integral equations analogous to eq . ( [ 3d s ] ) for all the possible superfluid phases classified under the irreducible representations of group @xmath19 . an example derivation is given in the appendix to highlight the main techniques we employed to achieve such simplifications . in sec . v , we derive the conditions for the onset of the spin - density and charge - density waves , two possible orderings arising from the instability of the fermi gas against the density - density fluctuation . in sec . vi , we apply the formulas that we developed in sec . iv and v and perform a detailed investigation of the effect that various system parameters ( e.g. the healing length , the fermi filling factor , and the dipolar interaction ) have on the critical temperatures for the onset of various competing orders . finally , in sec . vii , we summarize the results and highlight the significance of our work . [ ptb ] fig1 . let us now describe our model in some detail . the mixture under consideration consists of a single - component bec made up of dipolar bosons of mass @xmath20 [ fig . [ fig : mixtureinlattice](a ) ] and a two - component fermi gas made up of spin up @xmath21 and spin down @xmath22 ( nondipolar ) fermions of mass @xmath23 with population balance [ fig . [ fig : mixtureinlattice](b ) ] in square optical lattices . an external field is introduced to align all dipoles along its direction @xmath24 , a unit vector specified by elevation and azimuthal angles , @xmath25 , which are defined with respect to the 2d lattice plane as illustrated in fig . [ fig : mixtureinlattice ] . the system is assumed to operate in the quasi-2d regime where atoms still experience 3d interactions which are divided into a short- and long - range part . the short - range part is made up of the s - wave scatterings characterized with the strength @xmath26 , between two bosons , @xmath27 $ ] , between a boson and a fermion , and @xmath28 , between two fermions of opposite spin , where @xmath29 are the relevant s - wave scattering lengths . ( the s - wave interaction between two fermions of identical spins is prohibited by the pauli exclusion principle . ) the long - range part is the dipole - dipole interaction between bosons , which takes the form @xmath30 /\left\vert \mathbf{r}\right\vert ^{3}$ ] in position @xmath31 space , with @xmath32 the induced dipole moment . the quasi-2d optical lattice is created by three pairs of laser fields of wavelength @xmath33 propagating along three orthogonal directions @xcite where for simplicity we have assumed that the laser fields for bosons have the same wavelength as those for fermions . this arrangement creates a trap potential , @xmath34 , for bosons ( @xmath35 ) and fermions ( @xmath36 ) , where@xmath37 . \label{v2alpha}\ ] ] @xmath38 in eq . ( [ v1alpha ] ) , with @xmath39 being much stronger than @xmath40 , is to provide atoms with a tight confinement along the axial dimension @xmath41 while @xmath42 in eq . ( [ v2alpha ] ) is to provide atoms with a qausi-2d square lattice of lattice constant @xmath43 in the radial or transverse dimensions @xmath44 . the parameters are chosen in such a way that @xmath40 , while much weaker than @xmath39 , is sufficiently high compared with the photon recoil energy , @xmath45 , and all other energy scales are sufficiently small compared with the energy band gap ( in the order of @xmath40 ) . as such , atoms reside essentially in the lowest bloch band and the tight binding remains a fairly good approximation . in our calculation , we assume that the single - particle wannier functions located at site @xmath46 are in a separable form , @xmath47 , where @xmath48 are the ground state wavefunctions of the harmonic oscillators , with @xmath49 the harmonic oscillator lengths and @xmath50 the harmonic oscillator frequencies . the bose - fermi mixture under these approximations is described by the single - band bose - fermi hamiltonian,@xmath51 \nonumber\\ & + \sum_{i}\left [ \frac{u_{b}}{2}\hat{n}_{i , b}\left ( \hat{n}_{i , b}-1\right ) + u_{bf}\hat{n}_{i , b}\hat{n}_{i , f}+u_{ff}\hat{n}_{i,\uparrow}\hat { n}_{i,\downarrow}\right ] \nonumber\\ & -\sum_{i}\left [ \mu_{b}\hat{b}_{i}^{\dag}\hat{b}_{i}+\mu_{f}\left ( \hat{a}_{i,\uparrow}^{\dag}\hat{a}_{i,\uparrow}+\hat{a}_{i,\downarrow}^{\dag } \hat{a}_{i,\downarrow}\right ) \right ] , \label{hamiltonian}\ ] ] where @xmath52 stands for the summation over nearest neighbors , @xmath53 and @xmath54 are the field operators for bosons and fermions of spin @xmath55 at site @xmath46 , respectively , @xmath56 and @xmath57 @xmath58 are the corresponding particle number operators , and @xmath59 and @xmath60 are the chemical potentials for bosons and fermions , respectively . the hamiltonian ( [ hamiltonian ] ) has been expressed in terms of various effective 2d coefficients whose explicit expressions we now present . the kinetic energy part of the hamiltonian is described by the terms proportional to the hopping amplitudes between nearest neighbors @xcite @xmath61 the on - site interactions consist of @xmath62 ^{-1},\\ u_{bb } & = u_{bb}^{3d}\left ( \sqrt{2\pi}d_{bz}\right ) ^{-1}\left ( 2\pi d_{b\perp}^{2}\right ) ^{-1},\end{aligned}\ ] ] which are 2d analogs of 3d short - range s - wave interactions , @xmath63 , introduced in the beginning of the section . as to the long - range dipole - dipole interaction , it contributes both to the onsite interaction and the off - site interaction . the former is given by @xcite @xmath64 , \label{chi_dd}\ ] ] where @xmath65 measures the dipolar interaction relative to the s - wave boson - boson interaction , and @xmath66 is a unitless factor fixed purely by the geometry of the trapping potential , where @xmath67 . the first term in the second line of eq . ( [ hamiltonian ] ) contains @xmath68 which represents the total onsite interaction for bosons . here , the use of the single - subscripted @xmath69 on @xmath70 is to stress that @xmath70 has contributions not only from the short - range s - wave interaction @xmath71 but also from the on - site dipole - dipole interaction @xmath72 . the off - site dipole - dipole interaction , owing to their rapid decay with distance ( @xmath73 ) , is assumed to be limited to the nearest - neighbor ( nn ) dipole - dipole interaction @xcite , well approximated by the formula @xcite @xmath74 , where @xmath75 is the angle between the dipole and the displacement from site @xmath46 to nearest site @xmath76 . to study the superfluid pairings , we move to the lattice momentum space ( with a total of @xmath77 sites ) in which @xmath78 and @xmath79 are field operators and @xmath80 and @xmath81 @xmath82 are the corresponding particle number operators , and transform the hamiltonian in eq . ( [ hamiltonian ] ) into @xmath83 \hat{\rho}_{\mathbf{k},b}\hat{\rho}_{-\mathbf{k},b}\nonumber\\ & + \sum_{\mathbf{k\sigma}}\left ( \epsilon_{\mathbf{k},f}-\mu_{f}\right ) \hat{a}_{\mathbf{k,}\sigma}^{\dag}\hat{a}_{\mathbf{k},\sigma}\nonumber\\ & + \frac{1}{n_{l}}\sum_{\mathbf{k}}\left ( u_{bf}\hat{\rho}_{\mathbf{k},b}\hat{\rho}_{-\mathbf{k},f}+u_{ff}\hat{\rho}_{\mathbf{k},\uparrow}\hat{\rho } _ { -\mathbf{k},\downarrow}\right ) , \label{h1}\ ] ] where the sum over the momentum @xmath84 ( or @xmath85 to be precise ) is limited to the first brillouin zone , @xmath86 is the single - particle dispersion and @xmath87 is the nn dipole - dipole interaction in momentum space , with @xmath88 and @xmath89 being given by @xmath90 , \\ u_{y } & = \frac{d^{2}}{a^{3}}\left [ 1 - 3\sin^{2}\theta_{d}\sin^{2}\phi _ { d}\text { } \right ] .\end{aligned}\ ] ] we have used the unit convention in which @xmath91 and @xmath92 are scaled to the inverse lattice constant @xmath93 . an ultracold bose gas can be prepared in ground states of rather different natures , e.g. , superfluid and mott insulator phases , depending on the filling factor and strength of the two - body particle interaction relative to the hopping energy . a number of papers studied bose - fermi mixtures in which the bose gas operates in the parameter regime close to the mott insulator - superfluid transition with a relatively low boson filling factor ( see , e.g. , references @xcite ) . a dipolar bose gas is more complex than its nondipolar counterpart and thus may enter additional phases . in our model here , owing to the dipoles being directed away from the axial direction , the dipole - dipole interaction inherits both side - by - side repulsion and head - to - tail attraction . this , along with the ability to tune the ratios @xmath94 and @xmath95 , makes not only the superfluid and mott insulator but also supersolid phases accessible to the present 2d dipolar system . we focus on the bose - fermi mixture in which the bose gas operates in the superfluid region where the dipolar superfluid phase is characterized with a uniform order parameter @xmath96 , a chemical potential @xmath97 , and a bogoliubov excitation spectrum @xmath98 \right ) } , \label{ekb}\ ] ] where @xmath99 has been defined in eq . ( [ u(k ) ] ) . this superfluid phase exists in the parameter space where @xmath100 is real ; when @xmath100 changes from real to imaginary , the dipolar gas may undergo a dynamical instability towards other competing phases . the parameter region that supports a stable homogeneous superfluid in the @xmath101 space is marked in fig . [ fig : regionofdipolarbec ] as a square bordered by the four lines @xmath102 @xmath103 @xmath104 and @xmath105 . inside the square , the excitation spectra in the long wavelength limit , at the mode of the checkerboard supersolid phase @xmath106 , and at the modes of the striped supersolid phase @xmath107 and @xmath108 , are all real @xcite . [ ptb ] fig2 . as is well - known , in a bose - fermi mixture , phonons of the condensate can induce an effective attraction between two fermions @xcite . as a result , in addition to interacting through the s - wave interaction ( between fermions of opposite spins ) with strength @xmath109 , two fermions can interact indirectly by exchanging virtual phonons . exchanging virtual phonons in this fashion induces between two fermions an effective spin - independent attractive interaction , which , in momentum space , is given by @xmath110 } , \label{uind}\ ] ] where @xmath111 and @xmath112 is the healing length ( in the unit of lattice constant @xmath113 ) of the bosons interacting only under the short - range s - wave interaction . in arriving at eq . ( [ uind ] ) , we have ignored the retardation effect . this implies that eq . ( [ uind ] ) holds when the characteristic energy scale of fermions is small in comparison to that of phonons , which translates into the fast phonon limit . in this limit , the phonon velocity is much larger than the fermi velocity so that the phonon degrees of freedom can be adiabatically eliminated , leading to eq . ( [ uind ] ) . however , the same formula is often used in studies where the fast phonon limit is not quite fulfilled . the rational behind this is that the retardation effect affects the pre - exponential factor from the fermi energy scale to some characteristic bosonic frequency scale @xcite . thus , we expect it to be true for our dipolar bose - fermi mixture as well including the retardation effect will not affect the ( qualitative ) predictive power of the formulas we aim to derive in this work . the effective hamiltonian for fermions after integrating out the phonon degrees of freedom becomes , apart from a c - number , @xmath114 where @xmath115 is the total effective fermi - fermi interaction . we shall make two important comments concerning our assumptions in arriving at this effective hamiltonian . first , we have taken into account only the effect of bosons on fermions in the form of the phonon - mediated interaction . in the presence of bose - fermi interaction , fermions will always disturb and therefore modify the bose gas . this will result in a correction to the bose gas proportional to the bose - fermi interaction . this correction will , in turn , affect the fermi gas , which , in terms of the bose - fermi interaction , is a higher order effect than the effect of the undisturbed bose gas on the fermi gas . therefore , one can ignore the former effect in comparison to the latter one in the limit of weak bose - fermi interaction . second , we have neglected the fermion - induced interaction , in which a fermion interacts with a second fermion via polarization of the fermionic background by the second fermion . gorkov and melik - barkhudarov @xcite found this interaction to be responsible for an order unity correction to the critical temperature of the superfluid with s - wave symmetry in fermi gases with bare attractive interactions ( also see @xcite ) . the same induced - interaction was proposed by kohn and luttinger @xcite as a pairing mechanism ( kohn - luttinger mechanism ) for realizing superfluids with high - partial wave symmetries in fermi gases with bare repulsive interactions . as such , this induced interaction plays an important role in the exploration of pairing physics in pure fermi gases . in a bose - fermi mixture , however , not only background fermions but also phonons of the bec can mediate an interaction between fermions . the former interaction can be ignored in situations where it is much weaker than the latter interaction @xcite , which will be our case here . additionally , we may ignore the fermion interaction since the kohn - luttinger effect in 2d systems with rotational invariance is exponentially suppressed compared with its 3d counterpart @xcite , and our model takes place in the setting of a 2d lattice . at any rate , our results in this paper , like those in many others ( see , for example , @xcite ) which also ignore the kohn - luttinger mechanism , hold in the parameter space where the phonon - induced interaction dominates . at temperatures below the fermi temperature , fermions tend to form the normal ground state ( fermi sea ) , in which states below the fermi energy are all occupied with each state accommodating one particle . as in crystal solids , the phonon - mediated interaction in our model , being attractive , may serve as a catalyst , causing the normal state to undergo dynamical instability towards the bcs state , in which fermions are divided into highly correlated pairs with opposite momenta . this normal - to - superfluid phase - transition is accompanied with a lowering of symmetry and is therefore of second - order in nature . it is characterized by a symmetry - breaking order parameter , known as the gap , that represents the extent of macroscopic phase coherence . to study the bcs state of a fermionic system described by the effective hamiltonian ( [ hf_eff ] ) , we use the self - consistent hartree - fock - bogoliubov mean - field approach @xcite , in which we associate the bcs paring with the gap parameter @xmath116 , direct pairing with the hartree potential @xmath117 , and exchange paring with the fock potential @xmath118 . a further application of the bogoliubov diagonalization procedure leads to the gap equation@xmath119 and the particle number equation@xmath120 where @xmath121 , and @xmath122 is the fermionic quasi - particle spectrum . in eq . ( [ gap 1 ] ) , we have adjusted the chemical potential according to @xmath123 $ ] , where @xmath124 is the total density of fermions , which is also known as the fermion filling factor . note that in this paper , we absorb the boltzmann constant @xmath125 into the temperature @xmath126 so that @xmath126 has the units of energy . the gap parameter @xmath127 in eq . ( [ gap 1 ] ) , according to the landau theory of second order phase transitions @xcite , must transform according to one of the irreducible representations of the symmetry group of the high - temperature disordered phase . the symmetry group to which the effective hamiltonian @xmath128 in eq . ( [ hf_eff ] ) belongs is dictated , in our model where the hopping is assumed to be isotropic , by the symmetry underlying the fermi - fermi interaction . in this work , we explore the superfluid phases of the effective hamiltonian @xmath128 [ eq . ( [ hf_eff ] ) ] that preserves the symmetry of a square lattice . hence , we limit our study to systems that operate , within the ( shaded ) region in the @xmath101 space of fig . [ fig : regionofdipolarbec ] , along the + 45@xmath129(red ) line , where @xmath130 and the nn interaction is isotropic . this can be achieved by fixing the azimuthal angle of the dipole to @xmath131 while changing either the elevation angle @xmath132 or the induced dipole moment @xmath32 . for such systems , the phonon - mediated fermi - fermi interaction , eq . ( [ uind ] ) , simplifies to@xmath133 } , \label{uind 1}\ ] ] where@xmath134 , \label{b}\ ] ] and @xmath135 the table in fig . 3 lists the irreducible representations of the point group @xmath0 along with the simplest possible base functions ( for a square lattice ) that transform according to the corresponding representations @xcite . we stress that for each irreducible representation , there corresponds infinitely many other base functions . however , near the fermi surface , the quasiparticle energy is determined by the gap parameter ( which is a function of @xmath136 ) . it is thus energetically favorable for the gap parameter to have as few nodes as possible on the fermi surface , suggesting that the pairing phases that will most likely materialize are those with the simplest possible base functions @xcite . this may explain why well - known works ( see , for example , review articles @xcite and @xcite ) use the simplest base functions listed in fig . 3 to classify superfluid order parameters . as can be seen , the function @xmath137 in eq . ( [ uind 1 ] ) transforms according to the total symmetric representation , @xmath138 . hence , as anticipated , the interaction in eq . ( [ uind 1 ] ) exhibits the same symmetry as that of the underlying optical lattice which , for a square lattice , is described by the point group @xmath0 . [ ptb ] fig3 . in the spirit of the landau theory of second - order phase transitions , we now expand the gap parameter in the space spanned by the base functions of the point group @xmath0 in fig . [ fig : d4 ] , @xmath139 where @xmath140 has been defined previously in eq . ( [ gamma k ] ) and@xmath141 the gap parameter behaves very much like the wave function for two identical fermions . it is made up of two parts : a spin part and a spatial part . the first two lines in eq . ( [ decomposition ] ) are devoted to the superfluids in the spin singlet sector where the spin part is in the singlet state antisymmetric with respect to spin exchange and the spatial part must then transform according to the four parity - odd 1d representations @xmath142 and @xmath143 in order to meet the fermi statistics . in contrast , the last line in eq . ( [ decomposition ] ) is devoted to the superfluids in the spin triplet sector where the spin part is in the triplet state symmetric with respect to spin exchange and the spatial part must then transform according to the parity - even 2d representation @xmath144 . accordingly , @xmath145 , @xmath146 , @xmath147 , @xmath148 and @xmath149 in eq . ( [ decomposition ] ) are the gap parameters for the superfluids in the spin singlet sector with on - site @xmath18- , extended @xmath18- , @xmath1- , @xmath3- , and @xmath4-wave symmetries , respectively , while @xmath150 and @xmath151 in eq . ( [ decomposition ] ) are the gap parameters for the superfluids in the spin triplet sector with @xmath2-wave symmetries . the gap parameter is vanishingly small near the critical temperature . in order to determine the critical temperature for each superfluid state , we insert the ansatz ( [ decomposition ] ) into the gap equation ( [ gap 1 ] ) linearized around @xmath152 and replace @xmath153 either with the singlet potential , @xmath154 or with the triplet potential , @xmath155 depending on whether the gap equation falls into the singlet or triplet sector of superfluid pairings . this procedure is equivalent to finding , in the space spanned by the base functions of the irreducible representations of the point group @xmath0 , the matrix representation of the linearized gap equation , which , due to the inherent symmetry of our model , is expected to be block diagonal . for organizational purposes , we divide the gap equation in the spin singlet sector into one for the s - wave pairing , @xmath156{cc}s_{0}^{0}-1 & s_{0}^{1}\\ s_{1}^{0 } & s_{1}^{1}-1 \end{array } \right ) \left ( \begin{array } [ c]{c}\delta_{s0}\\ \delta_{s1}\end{array } \right ) = 0,\ ] ] where [ s]@xmath157 and one for the @xmath1- , @xmath3- , and @xmath4-wave pairings , @xmath156{ccc}d_{0}-1 & 0 & 0\\ 0 & d_{1}-1 & 0\\ 0 & 0 & g-1 \end{array } \right ) \left ( \begin{array } [ c]{c}\delta_{d_{x^{2}-y^{2}}}\\ \delta_{d_{xy}}\\ \delta_{g}\end{array } \right ) = 0,\ ] ] where [ d]@xmath158 with @xmath159 being defined as @xmath160 the gap equation in the spin triplet sector is also diagonalized with the form given by@xmath156{cc}p_{x}-1 & 0\\ 0 & p_{y}-1 \end{array } \right ) \left ( \begin{array } [ c]{c}\delta_{px}\\ \delta_{py}\end{array } \right ) = 0,\ ] ] where , due to symmetry , the coefficients @xmath161 , with @xmath2 being given by @xmath162 the coefficients in eqs . ( [ s ] ) , ( [ d ] ) and ( [ triplet ] ) are nothing but the matrix elements [ of the linearized gap equation ( [ gap 1 ] ) , or , to be more precise , the right - hand side of eq . ( [ gap 1 ] ) ] between various base functions of the irreducible representations of @xmath0 . a major complication in evaluating these coefficients is that they are functions of the fock potential @xmath163 and chemical potential @xmath60 , which , in principle , are themselves unknowns to be determined self - consistently . we take two measures to reduce this complexity . first , we estimate the fock potential on the fermi surface and absorb it into the chemical potential . second , we compute the chemical potential from the number equation for the @xmath164 normal fermi gas,@xmath165 where @xmath166 @xmath167 is the scaled chemical potential and @xmath168 is the complete elliptic integral of the first kind . ( all elliptical integrals in this paper follow the convention of ref . @xcite . ) it is to be stressed that only in the weak interacting limit where the characteristic interaction energy is much weaker than the fermi energy ( which is roughly equal to @xmath169 when near half filling ) do both approximations hold . thus , the expressions we will show below remain quantitatively accurate only in the weak coupling regime , but , nevertheless , are expected to help us gain some qualitative insights into the superfluid pairings under more general conditions . note that these were the same approximations micnas and ranninger @xcite used when they needed to gain quick insights into their critical temperatures . in spite of these assumptions , these matrix elements remain complicated integrals in 4d momentum space involving functions , @xmath170 , @xmath159 , etc . , that have nontrivial dependences on momenta . nevertheless , as the example in the appendix illustrates , we are able to reduce each integral into a product between a 1d integral and an analytical function . the 1d integral ( to be defined shortly ) is with respect to @xmath171 and involves @xmath172 where @xmath173 is @xmath159 of eq . ( [ sq(t ) ] ) in the scaled form . the analytical function will be one of the following functions : [ i]@xmath174 , \label{i1}\\ i_{2 } & = \frac{\bar{u}_{0}\left [ k\left ( \chi^{2}\right ) -\frac{\pi}{2}\right ] } { \pi^{3}b\xi^{2}},\label{i2}\\ i_{3 } & = \frac{8}{3\pi^{3}}\frac{\bar{u}_{0}}{b\xi^{2}}\left [ \left ( \frac{4}{\chi}-2\chi\right ) k\left ( \chi^{2}\right ) -\frac{4}{\chi}e\left ( \chi^{2}\right ) \right ] , \label{i3}\\ i_{4 } & = \frac{32}{15\pi^{3}}\frac{\bar{u}_{0}}{b\xi^{2}}\left\ { 2\pi-\left ( \frac{4}{\chi}\right ) ^{2}e\left ( \chi^{2}\right ) \right . \text { \ } \label{i4}\\ & \left . + \left [ \left ( \frac{4}{\chi}\right ) ^{2}-12\right ] k\left ( \chi^{2}\right ) \right\ } , \end{aligned}\ ] ] where @xmath168 and @xmath175 are , respectively , complete elliptical integrals of the first and second kind , and @xmath176 is a function defined as @xmath177 here , the scaled variables , @xmath178 , @xmath179 , @xmath180 , are similarly defined as the scaled chemical potential @xmath166 introduced in eq . ( [ nf ] ) . the conditions for the onset of various superfluid pairings can then be expressed in terms of various 1d integrals that we now introduce . the critical temperature is determined from @xmath181{cc } s_{0}^{0}\left ( \bar{t}\right ) -1 & s_{0}^{1}\left ( \bar{t}\right ) \\ s_{1}^{0}\left ( \bar{t}\right ) & s_{1}^{1}\left ( \bar{t}\right ) -1 \end{array } \right\vert = 0 , \label{s - wave superfluid}\ ] ] for the pairing with the @xmath18-wave symmetry , where [ s integtrals ] @xmath182 from @xmath183 for the @xmath2-wave pairing , where @xmath184 s_{+}\left ( x , t\right ) dx , \label{p integrals}\ ] ] and finally from@xmath185 for the pairings with @xmath1- , @xmath3- , and @xmath4-wave symmetries , respectively , where @xmath186 -e\left [ \left ( \frac{1-x}{1+x}\right ) ^{2}\right ] \right\ } , \label{dd}\\ d_{1 } & = i_{3}\int_{0}^{1}dxs_{+}\left ( x,\bar{t}\right ) \left ( 1+x\right ) \left\ { \left ( 1+x^{2}\right ) e\left [ \left ( \frac{1-x}{1+x}\right ) ^{2}\right ] -2xk\left [ \left ( \frac{1-x}{1+x}\right ) ^{2}\right ] \right\ } , \label{ee}\\ g & = i_{4}\int_{0}^{1}dxs_{+}\left ( x,\bar{t}\right ) \left\ { \left ( 1 + 14x^{2}+x^{4}\right ) \left ( 1+x\right ) e\left [ \left ( \frac{1-x}{1+x}\right ) ^{2}\right ] -2x\left ( 1+x\right ) \left ( 1 + 6x+x^{2}\right ) k\left [ \left ( \frac{1-x}{1+x}\right ) ^{2}\right ] \right\ } . \label{gg}\ ] ] the reduction of 4d integrals into 1d integrals , which constitute the main results of the present work , involves an extensive use of algebra containing elliptical integrals and symmetry considerations . in the appendix , we show how to simplify the matrix element for the d - wave pairing from eq . ( [ d0 ] ) into eq . ( [ dd ] ) , using it as an example to showcase the techniques that help us simplify the matrix elements for all other pairings . the reasons why we consider these results significant will be presented in the final section ( sec . vii ) where we conclude this work . the subject of fermionic superfluids studied in the previous section is founded on an instability discovered by cooper in the presence of an attractive interaction , irrespective of its weakness , two fermions on a fermi surface find it energetically favorable to form a bound state , thereby causing a normal fermi gas to undergo dynamical instability towards the bcs state . this , however , is not the only instability which a normal fermi gas may experience . in our model , fermions are subject to the density - density interaction , @xmath187 , due to the s - wave scattering between bosons and fermions . this interaction induces a change in the fermion density , which , within linear response theory , equals @xmath188 for a two - component fermi gas , where @xmath189 is the well - known lindhard function ( in the static limit ) with @xmath190 ^{-1}$ ] the fermi distribution function . for continuous models , no drastic response of this function occurs at any @xmath191 in 3d and 2d , but does in 1d , where it is found to diverge logarithmically at @xmath192 . the fermi surface of a 1d fermi gas consists of two points , one at @xmath193 and the other at @xmath194 . this along with the dispersion relation being linear near the fermi points results in a perfect nesting , where the two fermi points can be mapped to each other via the nesting condition , @xmath195 , thereby giving a divergent contribution to the lindhard function in eq . ( [ lindhard ] ) @xcite . for discrete ( lattice ) 2d models , owing to the reduced symmetries ( relative to the continuous 2d models ) , similar nesting effects can arise . in particular , nesting is known to occur at the nesting vector @xmath196 for a half - filled fermi gas in a square lattice , leading to the divergence of @xmath197 at half filling @xcite . this can also be understood from the zero temperature lindhard susceptibility , @xmath198 , as a function of the chemical potential @xmath60 . replacing @xmath199 with @xmath200 at @xmath164 , we can simplify eq . ( [ lindhard ] ) into @xmath201 .\ ] ] the singularity , @xmath202 , due to nesting and the van hover singularity , @xmath203 , in the density of states combine to yield a divergent response of the susceptibility , in the limit of half filling where @xmath204 , according to @xmath205 this divergence implies that at very low temperatures and when operating close to half filling , a normal fermi gas is unstable against spatially varying density perturbations and may evolve spontaneously into phases that may rival the superfluid phases studied in the previous section . we focus on two such possibilities , the spin - density wave ( sdw ) characterized with order parameter @xmath206 where the spin density varies in space with a commensurate wavevector @xmath207 ( the case of an incommensurate wavevector is beyond the scope of this paper ) , and the charge - density wave ( cdw ) characterized with order parameter , @xmath208 where the charge density modulates in space with a commensurate wavevector @xmath207 . in the same spirit as the hartree - fock bogoliubov mean - field approach that we employed previously in the study of superfluid pairings , we arrive at the mean - field equation for the order parameter @xmath209 @xmath210 and that for the particle number density@xmath211 where @xmath212 the same set of equations become the ones for the cdw state with the substitution of @xmath209 with @xmath213 and @xmath109 with @xmath214 $ ] . in the weak interacting regime where one can ignore the fock potential and fix the chemical potential according to the number equation ( [ nf ] ) for a normal gas at @xmath164 , the critical temperatures for the sdw and cdw phase are determined from the equation,@xmath215 where @xmath216 with @xmath217 and @xmath218 where @xmath219 is the scaled induced interaction at the nesting wavevector . before presenting numerical results in the next section , a comment is in order . our results for the onset of superfluid pairings and density waves was derived within the usual mean - field approach , in which superfluid instability is treated independently from density - wave instability . a more rigorous treatment for systems operating at ( or near ) half filling is the parquet method @xcite , which couples the two instabilities . the parquet method calculates the sum of an infinite series of diagrams using flow equations from1 renormalization group theory , a system of nonlinear differential equations . in 1d near half filling , the parquet prediction is drastically different from the mean - field prediction , but in 2d near half filling the predictions qualitatively agree @xcite . a quantitative comparison of the two approaches requires adapting the parquet method to our 2d model , which is beyond the scope of the current work and which we leave for future study . in this section , we apply the formulas outlined in previous sections to numerically determine critical temperatures and use these results as a guide to understanding the phase structures and competing orders of the system . the critical temperatures depend on several parameters . consider first @xmath220 , @xmath221 , and @xmath222 , which are functions of and thus controlled by @xmath223 , @xmath224 , and @xmath225 . in fig . [ fig : parameters ] , we plot ( a ) @xmath220 as a function of @xmath223 under different ratios of the axial to radial trap potential , and ( b ) @xmath221 as a function of @xmath224 and ( c ) @xmath222 as a function of @xmath225 for different ratios of the axial to radial trap potentials as well as different ratios of the fermion to boson mass . here and throughout what follows , we fix the wavelength to @xmath226 nm and assume that @xmath227 and @xmath228 . it is to be stressed that we do not intend to limit our investigation to particular systems in view of the rich existence of atomic elements and their isotopes in nature ; the particular choices here in fig . [ fig : parameters ] are nothing more than to establish the orders of magnitude for the ranges in which @xmath220 , @xmath221 , and @xmath222 can vary when the corresponding control parameters are tuned within some realistic domains . as can be seen , they do not change significantly with the ratio of the axial to radial trap potential ( so long as they are within the quasi-2d regime ) , but the healing length @xmath222 is sensitive to the fermion - boson mass ratio and the larger this ratio , the higher the healing length . [ ptb ] fig4 . the next parameter in line is @xmath229 , defined in eq . ( [ chi_dd ] ) , a control knob inaccessible to nondipolar bose - fermi mixtures . as can be seen , by adjusting the polar angle of the dipole @xmath132 and the dipolar interaction @xmath230 , one can change @xmath229 continuously from positive values where the dipole is oriented close to the @xmath41 axis with @xmath231 to negative ones where the dipole is tilted away from the @xmath41 axis with @xmath232 . finally , we consider @xmath233 defined in eq . ( [ b ] ) . it can be shown that @xmath233 is determined by the trap geometry according to@xmath234 where @xmath235 has been defined previously in eq . ( [ beta ] ) . as can be seen from eq . ( [ g ] ) , @xmath233 becomes small when @xmath236 and @xmath237 become large in comparison with the photon recoil energy @xmath238 . thus , it is not surprising that for a quasi-2d trap where @xmath239 , @xmath233 is found to be less than 1 for typical values of @xmath237 , which for our interests lie within @xmath240 as illustrated in fig . [ fig : parameters](d ) . from now on , without loss of generality , we fix @xmath233 to @xmath241 for all calculations . for systems with @xmath242 , a straightforward analysis indicates that the phonon spectrum near @xmath243 acquires an imaginary component when @xmath229 @xmath244 $ ] and as a result , only when @xmath245 , does the homogeneous dipolar bec become stable against collapse . figure [ fig : phononspectrum ] displays , within the first brillouin zone , the phonon dispersion spectra for different dipolar interaction @xmath229 . the middle surface corresponds to the spectrum without the dipolar interaction . evident in fig . [ fig : phononspectrum ] is that increasing @xmath246 on the positive side ( top surface ) raises the phonon energy relative to the middle one . contrarily increasing @xmath246 on the negative side ( bottom surface ) lowers the phonon energy . this latter ability provides us with the opportunity to enhance the phonon - induced fermi - fermi interaction by tuning @xmath229 on the negative side ( but keeping it less than @xmath247 ) . [ ptb ] fig5 . having discussed the key parameters , we now seek to gain qualitative understanding of the critical temperatures as functions of these parameters . we do so from effective interactions which we now define . the critical temperatures are determined by matrix elements such as those in eq . ( [ s ] ) , each of which can be decomposed into a 1d integral , which is a function of @xmath126 and @xmath60 ( and is always positive ) , and a coefficient such as @xmath248 in eq . ( [ i2 ] ) , which is a function of the key parameters just discussed above . this latter quantity is what we regard as the effective interaction strength for the corresponding phase where the symmetry of the phase has been taken into consideration . there is a total of seven phases , which we divide into two categories . the first one consists of @xmath1- , @xmath3- , @xmath4- , and @xmath2-wave pairing phases , whose critical temperatures are independent of @xmath109 . the second one consists of the @xmath18-wave pairing state , sdw ordering , and cdw ordering , whose critical temperatures can be either enhanced or suppressed by adjusting @xmath109 . in the first column of fig . [ fig : coefficients ] , we plot the ratios , @xmath249 , @xmath250 , and @xmath251 , which are responsible for the phases in the first category , as functions of the healing length @xmath222 under different dipolar interactions @xmath229 . let us first turn our attention to the middle curve in each plot in the first column , which is produced in the absence of the dipolar interaction ( @xmath252 ) . ( a ) @xmath248 , ( b ) @xmath253 , and ( c ) @xmath254 reach their peak when @xmath222 is set at @xmath255 , @xmath256 and @xmath257 , respectively . this implies that in the weak coupling limit , the corresponding critical temperature is expected to reach its maximum at the corresponding peak healing length , independent of @xmath258 . it is interesting to observe that the maximum healing length @xmath259 @xmath255 for @xmath248 , which we obtain here in a semi - analytical fashion , almost equals to what wang _ et al . _ @xcite found from their numerical analysis of the critical temperature for the @xmath1-wave pairing . next , we turn to the curves produced using non - zero @xmath229 . the bottom curves of fig . [ fig : coefficients](a ) , ( b ) and ( c ) in the first column indicate that increasing the dipolar interaction on the positive side reduces the peak value and shifts the peak towards high healing length while the top curves show that increasing the dipolar interaction on the negative side increases the peak value and pushes the peak position towards low healing lengths . thus , in our study of the critical temperature in the next section , we will always use negative @xmath229 . [ ptb ] fig6 . let us now turn our attention to the effective interactions for the phases in the second category . consider , first , the s - wave pairing , which , in general , is a coherent superposition between the onsite and extended s - wave pairings . its critical temperature , the root to eq . ( [ s - wave superfluid ] ) , depends on the interplay between the @xmath260 and @xmath248 coefficients in a non - intuitive fashion . it is thus quite difficult to single out , under general circumstances , the effective interaction responsible for the s - wave pairing . however , in the case of half filling , the off diagonal elements , @xmath261 and @xmath262 , vanish so that eq . ( [ s - wave superfluid ] ) reduces to the diagonal form @xmath263{cc}s_{0}^{0}\left ( \bar{t}\right ) -1 & 0\\ 0 & s_{1}^{1}\left ( \bar{t}\right ) -1 \end{array } \right\vert = 0,\ ] ] which allows us to identify , straightforwardly , that the effective " interactions are @xmath260 for the onsite s - wave pairing and @xmath248 for the extended s - wave pairing . as illustrated in the plots in the second column of fig . [ fig : coefficients ] , in contrast to @xmath248 , @xmath253 , and @xmath254 , the coefficient @xmath260 is a monotonically decreasing function of @xmath222 . this is nothing more than a simple reflection of the well established fact that fermions favor the s - wave pairing more than unconventional pairings in the limit of a short healing length . as expected , in comparison with fig . [ fig : coefficients](e ) where @xmath264 , attractive @xmath109 enhances the onsite s - wave pairing [ fig . [ fig : coefficients](f ) ] while repulsive @xmath109 suppresses or even excludes ( in the region where @xmath260 is negative ) the onsite s - wave pairing [ fig . [ fig : coefficients](d ) ] . in addition to the s - wave pairing , the second category also contains the sdw and cdw ordering . in the sdw state , fermions with opposite spins tend to be distributed in different sites , which is only possible for repulsive fermions . the phonon - induced interaction is always attractive . this explains why the effective interaction for the sdw phase , @xmath265 , is completely determined by the s - wave interaction between fermions of opposite spins @xmath109 [ eq . ( [ sdw ] ) ] and exists only when @xmath266 . in the cdw phase , fermions of opposite spins tend to aggregate on the same site , which is only possible for attractive fermions . the phonon - induced interaction is thus the main source of the interaction responsible for the cdw phase . in addition , as shown in the last column of fig . [ fig : coefficients ] , compared to fig . [ fig : coefficients](h ) where @xmath264 , attractive @xmath109 [ fig . [ fig : coefficients](g ) ] shall help increase the effective interaction for the cdw phase , @xmath267 @xmath268 , and hence enhance cdw , while repulsive @xmath109 [ fig . [ fig : coefficients](i ) ] shall help decrease @xmath267 and hence suppress cdw . in fig . [ fig : coefficients](i ) , cdw does not exist in the region where @xmath269 , and dominates sdw in the region on the left side of the intercept between @xmath267 and @xmath270 in which @xmath271 . [ ptb ] fig7 . figures [ fig : tc1](a ) and ( b ) show @xmath16 as a function of the healing length @xmath222 at half filling . the most prominent feature is that when @xmath229 is tuned negative in fig . [ fig : tc1](b ) , the critical temperatures for unconventional pairings become significantly higher than their counterparts in fig . [ fig : tc1](a ) where @xmath252 ; even the @xmath4-wave pairing , which is absent in fig . [ fig : tc1](a ) for the temperatures plotted , enters fig . [ fig : tc1](b ) . the features exhibited in fig . [ fig : tc1 ] are precisely what we expect based on theeffective interaction analysis summarized in fig . [ fig : coefficients ] , examples of which include that both @xmath1- and @xmath2-wave pairings peak at the same healing length @xmath272 since both are determined by @xmath248 , that there is an intercept between the sdw and cdw phase ( which takes place at @xmath271 ) , and that the peaks shift toward the lower healing length when a negative @xmath229 is turned on . figures [ fig : tc1](c ) and ( d ) display @xmath16 , again at half filling , as a function of @xmath273 . as @xmath258 increases , the system , shown in fig . 7(c ) where @xmath252 , changes its phase from sdw to the @xmath1-wave pairing , and then to the s - wave pairing . there is a substantial increase in @xmath16 for pairings different from @xmath1- and @xmath18-wave pairings when @xmath229 is tuned negative in fig . [ fig : tc1](d ) , but this increase in @xmath16 will not alter the sequence of phases exhibited in fig . [ fig : tc1](c ) . the conclusion is that at half filling , the @xmath1-wave pairing always dominates all other unconventional pairings in spite of the fact that the critical temperatures for the latter may increase more dramatically than the @xmath1-wave pairing in response to an increase in @xmath246 and @xmath258 . [ ptb ] fig8 . finally , we consider the effect of the filling factor @xmath274 on @xmath16 for various competing orders . in fig . [ fig : tc - nf](a ) , we display @xmath16 as a function of @xmath275 for the phases in the first category when @xmath252 ( solid lines ) and @xmath276 ( dashed lines ) . it can be seen from solid lines@xmath15(@xmath252 ) , the @xmath2-wave pairing dominates the @xmath1-wave pairing away from half filling in the region where @xmath275 exceeds around 0.4 . in the presence of @xmath229 (= @xmath277 ) ( dashed lines ) , the crossing from the @xmath2- to @xmath1-wave pairing shifts ( slightly ) towards a higher @xmath275 while the critical temperature for the @xmath2-wave pairing experiences a significant increase . note that since both @xmath2-wave [ eq . ( [ p integrals ] ) ] and @xmath1-wave [ eq . ( [ dd ] ) ] pairings share the same effective interaction strength @xmath248 , in the limit of weak interaction , the @xmath2- to @xmath1-wave crossing becomes independent of @xmath248 and hence is fixed essentially by the pairing symmetries . this is to be contrasted to other crossings such as from @xmath1- to @xmath3-wave as shown by the dashed lines , which depends on the ratio of @xmath248 to @xmath253 and hence is highly susceptible to changes in system parameters . the system is not in the @xmath2-wave state unless the @xmath2-wave pairing also has a higher @xmath16 than the phases in the second category . in fig . [ fig : tc - nf](b ) , we show @xmath16 as a function of @xmath278 for the phases in the second category under different @xmath109 @xmath279 with the solid lines for @xmath264 , the dashed lines for @xmath280 @xmath281 , and the dotted lines for @xmath282 . in the absence of @xmath109 ( solid lines ) , sdw does not exist due to the lack of repulsive interaction , cdw ( solid green line ) exists close to half filling due to the phonon - induced attraction , and the s - wave pairing ( solid black line ) has a high @xmath16 throughout . increasing @xmath109 suppresses cdw ( not shown due to a low temperature ) while enhancing and expanding sdw ( dashed and dotted red lines ) , but most importantly it suppresses the s - wave pairing in the region of low @xmath275 ( dashed and dotted black lines ) . this suppression is strong enough that , as shown by the dotted lines in [ fig : tc - nf](b ) , all phases in the second category are strongly suppressed in the same region where the @xmath2-wave pairing exceeds the @xmath1-wave pairing displayed in fig . [ fig : tc - nf](a ) . this , in turn , creates a window of opportunity for the @xmath2-wave superfluid to come out a winner . two such examples are shown , respectively , in fig [ fig : tc - nf](c ) and ( d ) with and without dipolar interaction . in particular , fig [ fig : tc - nf](d ) demonstrates that the @xmath2-wave pairing with a critical temperature in the order of a hundredth of @xmath169 is possible . the conclusion is that in contrast to the @xmath1-wave pairing , which may dominate close to half filling , the @xmath2-wave may dominate away from half filling in a window centered roughly around @xmath283 , while all the other unconventional pairings , in spite of the possibility of being substantially enhanced by the use of the dipolar interaction , are unable to emerge as dominant phases . we have considered a cold atom mixture in a square optical lattice where the mixture is made up of a two - component fermi gas and a single - component dipolar bose gas in the state of a homogeneous bec . by eliminating the phonons of the dipolar bec , we have constructed an effective lattice fermi system which preserves the symmetry of point group @xmath0 . focusing on this effective model , we have explored , within the hartree - fock - bogoliubov mean - field theory , the competition among density waves and superfluids with both conventional and unconventional parings . we have constructed , in the weak coupling regime , the matrix representation of the linearized gap equation in the irreducible representations of @xmath0 . we have simplified each matrix element from a 4d integral into a separable form involving a 1d integral , which is only a function of @xmath126 and @xmath60 , and a pairing - specific effective interaction , which is an analytical function of the parameters that characterize the fermi - fermi interactions in our system . to help appreciate the significance of these 1d integral equations , we remind that for a continuous 3d two - component fermi gas , there exists a similar 1d integral equation @xcite @xmath284 dx=1,\label{3d s}\ ] ] which is the origin of the well - known formula , @xmath285 that one uses to estimate the critical temperature @xmath286 of s - wave superfluid pairing , where @xmath287 is the fermi wave number , @xmath288 the s - wave scattering length , and @xmath289 euler - mascheroni constant . the 1d integral equations we highlighted in sec . iv are analogs of eq . ( [ 3d s ] ) for fermions in a square lattice that interact via both a contact and a long - range interaction . the long - range interaction here is in the form of an attractive ( lattice - version ) yukawa potential , which occurs ubiquitously across a broad spectrum of physics , e.g. , nuclear physics and condensed matter physics . thus , in analogy to eq . ( [ 3d s ] ) , we expect that our 1d equations will find applications beyond the current model of interest . further , the ability to reduce our gap equations into ones analogous to eq . ( [ 3d s ] ) may hint that similar reductions may exist for interaction potentials with mathematical forms different from the yakawa type . the various techniques we developed in this work may then offer a good starting point for future studies to build toolboxes for simplifying gap equations associated with such interactions . as eq . ( [ 3d s ] ) , our 1d integral equations enjoy the advantage of allowing many features of the critical temperatures to be determined in a semi - analytical manner . in particular , we have applied them to analyze the critical temperatures of various competing orders as functions of the healing length and filling factor in both the absence and presence of the dipolar interaction . we have found that tuning dipolar interaction @xmath229 on the negative side lowers the phonon energy and can significantly enhance the unconventional pairings . we have found that close to half filling , the @xmath1-wave pairing with a critical temperature in the order of a fraction of @xmath169 may dominate all the other phases , and at a higher filling factor , the @xmath2-wave pairing with a critical temperature in the order of a hundredth of @xmath169 may emerge as a winner . our theory also includes the pairings with @xmath3- and @xmath4- wave symmetries . in spite of dramatic enhancements of their critical temperatures , we have found that tuning a dipolar interaction will not be able to make the pairings with @xmath3- and @xmath4- wave symmetries to dominate the other phases . it is to be stressed that the phase diagram based on the critical temperature analysis does not correspond necessarily to what is realizable at zero temperature . the actual ground state may exist as a superfluid with a pure symmetry or a coherent superposition of those with pure symmetries . [ the latter dominates the phase diagram when the dipolar interaction is tuned away from the diagonal line ( but stays within the square ) in fig . [ fig : regionofdipolarbec ] where the system breaks d4 symmetry . ] to construct the phase diagram at zero temperature requires the computational machinery of energy minimization , which can be a time - consuming task for a dipolar bose - fermi mixture where the parameter space is unusually large . the 1d integrals developed in this paper allow us to gain not only quick insights into various critical temperatures but also , very importantly , a sense for the role each parameter plays in this large parameter space . thus , these integrals , just as with their 3d continuous analog in eq . ( [ 3d s ] ) , serve as a valuable theoretical tool that one can use to perform a preliminary analysis before embarking on more complex tasks such as energy minimization . b. k. is grateful to itamp and the harvard - smithsonian center for astrophysics for their hospitality while completing this work . h. y. l. was supported in part by the us army research office under grant no.w911nf-10 - 1 - 0096 and in part by the us national science foundation under grant no . phy11 - 25915 . in this appendix , we outline the main steps that we use to reduce @xmath290 , the matrix element for the d - wave pairing , from a 4d integral in eq . ( [ d0 ] ) to a 1d integral in eq . ( [ dd ] ) , and use it as an example to showcase the techniques that we employed to simplify the matrix elements for all other pairings from 4d integrals in eqs . ( [ s ] ) , ( [ d ] ) and ( [ triplet ] ) to 1d integrals in eqs . ( [ s integtrals ] ) , ( [ p integrals ] ) , ( [ ee ] ) , and ( [ gg ] ) . to begin with , we organize eq . ( [ d0 ] ) , a 4d integral over the first brilliouin zone in momentum space , into a 2d integral involving @xmath159 @xmath291 in terms of another 2d integral involving the fermi - fermi interaction @xmath292@xmath293 as a first step to solving eq . ( [ kd ] ) , we substitute @xmath294 in eq . ( [ us ] ) , where @xmath295 is given by eq . ( [ uind 1 ] ) , into eq . ( [ kd ] ) , and change eq . ( [ kd ] ) into @xmath296 where we have used the fact that the integral involving @xmath109 is zero and the one involving @xmath297 is the same as the one involving @xmath298 . for notational simplicity , we put eq . ( [ kai d ] ) in a more compact form @xmath299 in terms of @xmath300 and @xmath301 defined as @xmath302 next , we make the change of variables @xmath303 , and by taking advantage of the integrand being a periodic function of momentum , we transform eq . ( [ kai d ] ) into@xmath304 where @xmath305 $ ] or@xmath306 by virtue of symmetry considerations , the integrals involving the sine functions in the second line of eq . ( [ eta k+q ] ) vanish while those involving the cosine functions in the first line of eq . ( [ eta k+q ] ) can be put into a separable form @xmath307 to proceed , we rewrite eq . ( [ k q 1 ] ) as @xmath308 , \label{kq}\ ] ] where @xmath309 is the integral @xmath310 which we now focus on . integrating first with respect to @xmath91 from @xmath311 to @xmath312 , we change eq . ( [ l ] ) into @xmath313 where@xmath314{c}\frac{1 + 4\tilde{\xi}^{2}+2\left\vert b\right\vert \tilde{\xi}^{2}}{2b\tilde{\xi}^{2}}\text { , if } b>0\\ \frac{1 + 4\tilde{\xi}^{2}-2\left\vert b\right\vert \tilde{\xi}^{2}}{2b\tilde{\xi}^{2}},\text { if } b<0 \end{array } \right . , \label{r+}\ ] ] and@xmath315{c}\frac{1 + 4\tilde{\xi}^{2}-2\left\vert b\right\vert \tilde{\xi}^{2}}{2b\tilde{\xi}^{2}}\text { , if } b>0\\ \frac{1 + 4\tilde{\xi}^{2}+2\left\vert b\right\vert \tilde{\xi}^{2}}{2b\tilde{\xi}^{2}},\text { if } b<0 \end{array } \right . . \label{r-}\ ] ] we next integrate eq . ( [ l 1 ] ) and express the result in terms of the elliptical integral as @xmath316 } { \sqrt{\left ( r_{+}-1\right ) \left ( r_{-}+1\right ) } } , \label{m k}\ ] ] which , when @xmath317 and @xmath318 are replaced with eq . ( [ r+ ] ) and eq . ( [ r- ] ) becomes @xmath319 .\ ] ] finally , inserting this result into eq . ( [ kq ] ) , we find@xmath320 expressed in terms of @xmath248 defined in eq . ( [ i2 ] ) , where the use of eq . ( [ new u_0 and ksi ] ) has been made to revert back to the original notations . having solved @xmath321 , we now turn to @xmath290 in eq . ( [ d0 a ] ) , which , when @xmath159 is replaced with its scaled form in eq . ( [ s(x , t ) ] ) becomes @xmath322 where @xmath323 . to solve this integral , we change variables from @xmath324 to @xmath325 via the transformation@xmath326 which maps the first brillouin zone , a square in momentum space , into a non - square region in the polar @xmath325 plane , as shown in fig . [ fig : map ] where @xmath327 is the radial coordinate and @xmath328 is the angular coordinate . in terms of @xmath327 and @xmath328 , we have @xmath329 where @xmath330 at this point , we turn to the brillouin zone in the ( @xmath331 ) plane in fig . [ fig : map ] . due to the symmetry inherent to @xmath332 , @xmath333 is eight times the integral over the region from @xmath334 to @xmath335 in the first quadrant of the polar plane , which is divided into two areas , @xmath336 if @xmath337 and @xmath338 if @xmath339 . thus , @xmath340{c}\text { } 8\int_{\pi/4}^{\pi/2}\left ( \cdots\right ) dt\text { , } 0<c<1,\\ 8\int_{\pi/4}^{\sin^{-1}\frac{1}{c}}\left ( \cdots\right ) dt\text { , } 1<c<\sqrt{2 } , \end{array } \right . \label{it(c ) 0}\ ] ] where @xmath341 is the integrand in eq . ( [ ic(t ) ] ) . making a change of variables from @xmath328 to @xmath342 , we find , with some algebraic manipulations , that @xmath343{c}\text { \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ } k\left ( r^{2}\right ) -e\left ( r^{2}\right ) , \text { } 0<c<1,\\ f\left ( \sin^{-1}\frac{1}{r},r^{2}\right ) -e\left ( \sin^{-1}\frac{1}{r},r^{2}\right ) , \text { } 1<c<\sqrt{2 } , \end{array } \right . \label{it(c)}\ ] ] where @xmath344 and @xmath345 and @xmath346 are the incomplete elliptic integrals of the first and second kind , respectively . note that when @xmath339 , @xmath347 according to eq . ( [ r ] ) is greater than 1 so that @xmath348 and @xmath349 in eq . ( [ it(c ) ] ) are ill defined . this can be circumvented by making use of the identities @xmath343{c}\text { \ \ \ } k\left ( r^{2}\right ) -e\left ( r^{2}\right ) , \text { } 0<c<1,\\ r\left [ k\left ( r^{-2}\right ) -e\left ( r^{-2}\right ) \right ] , \text { } 1<c<\sqrt{2}. \end{array } \right.\end{aligned}\ ] ] finally , we integrate eq . ( [ d0 1 ] ) involving the piecewise function @xmath332 with respect to the radial coordinate @xmath352 @xmath353 and change it into eq . ( [ dd ] ) in the main text after substituting @xmath354 and further algebraic manipulations . de gennes , superconductivity of metals and alloys , " redwood city , california : addison - wesley ( 1989 ) . l. landau and e. lifshitz , statistical physics , " new york : pergamon ( 1979 ) . a. m. black - schaffer and c. honerkamp , j. phys . : condens . matter * 26 * , 423201 ( 2014 ) .	we consider a mixture of a two - component fermi gas and a single - component dipolar bose gas in a square optical lattice and reduce it into an effective fermi system where the fermi - fermi interaction includes the attractive interaction induced by the phonons of a uniform dipolar bose - einstein condensate . focusing on this effective fermi system in the parameter regime that preserves the symmetry of @xmath0 , the point group of a square , we explore , within the hartree - fock - bogoliubov mean - field theory , the phase competition among density wave orderings and superfluid pairings . we construct the matrix representation of the linearized gap equation in the irreducible representations of @xmath0 . we show that in the weak coupling regime , each matrix element , which is a four - dimensional ( 4d ) integral in momentum space , can be put in a separable form involving a 1d integral , which is only a function of temperature and the chemical potential , and a pairing - specific effective interaction , which is an analytical function of the parameters that characterize the fermi - fermi interactions in our system . we analyze the critical temperatures of various competing orders as functions of different system parameters in both the absence and presence of the dipolar interaction . we find that close to half filling , the @xmath1-wave pairing with a critical temperature in the order of a fraction of fermi energy ( at half filling ) may dominate all other phases , and at a higher filling factor , the @xmath2-wave pairing with a critical temperature in the order of a hundredth of fermi energy may emerge as a winner . we find that tuning a dipolar interaction can dramatically enhance the pairings with @xmath3- and @xmath4-wave symmetries but not enough for them to dominate other competing phases .
the asymmetric simple exclusion process ( asep ) is a simple hard core hopping particle model . it consists of a line segment with @xmath1 sites . particles are allowed to hop on to site 1 if it is empty , with rate @xmath2 . any particles on sites @xmath3 to @xmath4 hop on to a site to their right if it is empty , with rate @xmath3 . a particle on site @xmath1 hops off with rate @xmath5 as illustrated in figure [ fig : asep ] . the state of the system at any time is defined by the set of indicator variables @xmath6 , where @xmath7 and the probability , @xmath8 of the system being in state @xmath9 , at time @xmath10 given some initial state at @xmath11 , satisfies a master equation ( see @xcite for details ) . for a review of the extensive literature on this model see derrida @xcite . a significant step forward in the understanding of the mathematical aspects of the model was made with the realization , by derrida et al @xcite , that a stationary state solution , @xmath12 , of the master equation could be determined by the matrix product ansatz @xmath13 with normalisation factor @xmath14 given by @xmath15 provided that @xmath16 and @xmath17 satisfy the dehp algebra [ algebra ] @xmath18 these equations are sufficient to determine @xmath19 but derrida et al @xcite also gave several interesting matrix representations of @xmath16 and @xmath17 and the vectors @xmath20 and @xmath21 , any one of which may be used to determine @xmath19 . the primary result of this paper is to show that each of the three matrix representations of the dehp algebra can be interpreted as a transfer matrix for a different weighted lattice path problem . this then allows the normalisation , correlation function and other properties of the asep model to be interpreted combinatorially as certain weighted lattice path configuration sums see section 2 . one of the path connections is similar to that discussed in derrida et . @xcite the lattice path interpretation has two primary consequences : the first is that it provides a starting point for a new method ( the `` constant term '' or ct method ) for calculating the normalisation and correlation functions see section 3 . this reproduces several existing results ( but by a new method ) and also provides several new results . one of note , the `` @xmath0-expansion '' , arises from a rearrangement of the constant term expression which leads to form of the normalisation in terms of the variables @xmath22 and @xmath23 . the coefficients in this expansion are catalan numbers , the asymptotic form of which enables a uniform approach to computing the asymptotic behaviour of @xmath14 as @xmath24 in the various phases of the asep model see section 5 . the results agree with those found in @xcite , by steepest descent methods . a bonus following from the lattice path interpretation of the algebra representations is that one of the lattice path interpretations ( a slight variation of representation 3 ) has a natural interpretation as a polymer chain having a two parameter @xmath25 interaction with a surface . in this context @xmath14 is a partition function for the `` two - contact '' polymer model see section 4 . we also obtain recurrence relations ( on the length variable ) for the partition function of this polymer model and hence also for the asep normalisation . the second primary consequence of the lattice path interpretation follows from the ct method itself , as the ct method has very natural combinatorial interpretations . for example , the normalisation can be written in several different polynomial forms depending on which variables you use : @xmath26 where all the polynomial coefficients , @xmath27 , are integers . ( all the variables in these polynomials are related by simple equations eg . @xmath28 , @xmath29 etc . see above and in later sections ) . since , in each of these cases , the normalisation arises from a weighted lattice configuration sum , all the above coefficients have a direct combinatorial interpretation as enumerating a particular subset of the paths ( eg . those with exactly @xmath30 steps with the first weight and exactly @xmath31 steps with the second weight ) . however , we show that each of the above polynomial coefficients has an alternative combinatorial interpretation which corresponds to enumerating a different , _ unweighted _ , set of lattice paths eg . in @xcite the coefficient , @xmath32 was given as @xmath33 which , with a simple rearrangement , can be seen to be a particular `` ballot number '' @xmath34 where @xmath35 enumerates ballot paths ( see section 3.1 for details ) of length @xmath10 and height @xmath36 . thus @xmath37 which enumerates a special set of paths with @xmath30 weights of type @xmath38 and @xmath39 weights of type @xmath40 is seen to be determined in terms of the much simpler combinatorial problem of enumerating _ unweighted _ ballot paths of length @xmath41 and height @xmath42 . this correspondence between between the two combinatorial problems ( one pair for each coefficient ) arises as a bijection between the two path problems . this result may be turned around : if a bijection between a particular ballot path problem and the weighted path problem can be proved then it provides an alternative derivation of the normalisation polynomial . finally , in section 6 , the recurrence relations for the asep correlation functions derived in @xcite are shown to follow from the lattice path interpretations . the coefficients of the terms in the recurrence relation are also seen to be various ballot numbers . derrida et al @xcite , provided three different matrix representations for the asep algebra . representation one , @xmath43 @xmath44 where @xmath45 representation two , @xmath46 @xmath47 where @xmath48 and representation three @xmath49 @xmath50 where @xmath51 since these matrices and vectors satisfy the algebraic relations , the normalisation factor for the asep model , can be evaluated using any of the three formulae @xmath52 and @xmath17 matrices on the square lattice , b ) the labelling of the matrix elements.__,title="fig : " ] each of these three representations can be interpreted as the transfer matrix of a particular weighted lattice path problem . if the rows of the @xmath53 matrix and the columns of the @xmath54 matrix are labelled with odd integers @xmath55 and the columns of @xmath53 and rows of @xmath54 are labelled with even integers @xmath56 , then @xmath57 is the weight of a step from an odd height @xmath58 to even height @xmath59 and @xmath60 is the weight of a step from an even height @xmath61 to an odd height @xmath59 . since the rows and columns are labelled with non - negative integers the steps are only in the upper half of @xmath62 . similarly , the elements of @xmath63 and @xmath64 are labelled by @xmath65 and are the weights attached to the initial and final vertices of the paths . the matrices @xmath53 and @xmath54 act successively to the left on the initial vector @xmath63 and @xmath66 is the weighted sum over all paths of length @xmath67 which begin and end at odd height above the @xmath68axis . an example path for each of the three representations is shown in figure [ fig : repex ] . notice that for the first representation the paths with non - zero weight begin at any odd height and end at unit height , for the second they both begin and end at any odd height and for the third they begin and end at unit height . all these paths are defined explicitly below . + derrida et . al . gave another expression for the normalisation factor ( see equation ( 39 ) of @xcite ) . this form does not arise directly from any of the above three representations , however , we will show ( corollary [ alphabeta ] ) that it is the partition function @xmath69 , defined in , corresponding to a `` canonical '' path representation ( see figure [ fig : canp ] for an example ) . in @xcite we provide a combinatorial derivation of the equivalence of the above three and a number of other path representations of the normalisation factor to the `` canonical '' representation . we consider paths whose steps are between the vertices of the half plane square lattice @xmath70 , where @xmath71 ( resp . @xmath72 ) is the set of integers ( resp . non - negative integers ) . a * lattice path * , @xmath73 , of length @xmath74 is a sequence of vertices @xmath75 where @xmath76 , and for @xmath77 , @xmath78 where @xmath79 is the set of allowed steps . for a particular path , @xmath73 , denote the corresponding sequence of steps by @xmath80 , @xmath81 . a * subpath * of length @xmath58 of a lattice path , @xmath73 , is a path defined by a subsequence of adjacent vertices , @xmath82 , of @xmath73 . the set of length @xmath67 * dyck * paths , @xmath83 ; length @xmath84 and height @xmath36 * ballot * paths @xmath85 , and length @xmath84 and heights @xmath86 and @xmath87 * cross * paths @xmath88 all have step set @xmath89 . dyck paths have @xmath90 and @xmath91 , ballot paths have @xmath90 and @xmath92 and cross paths have @xmath93 and @xmath94 , @xmath95 . denote , @xmath96 [ defn : bub ] an * elevated dyck path * or * bubble * , is a subpath , @xmath97 , @xmath98 , for which @xmath99 and @xmath100 for all @xmath101 , @xmath102 . if @xmath103 then the elevated dyck path is a single vertex . the subset @xmath104 of * anchored cross paths * is defined as all the paths in @xmath105 with at least one vertex in common with the line @xmath106 . denote , @xmath107 the set of * one up * paths , @xmath108 is the set of lattice paths of length @xmath67 which have step set @xmath89 , @xmath109 and @xmath110 , i.e. @xmath111 . [ def : js ] the set of * jump step * paths of length @xmath84 , @xmath112 , is the set of lattice paths which have step set @xmath113 with @xmath114 , @xmath115 , @xmath116 . the `` height '' , @xmath117 , of a step @xmath118 is defined as @xmath119 . odd steps with @xmath120 will be called `` jump '' steps . if @xmath121 ( resp . @xmath122 ) the step is called an `` up '' ( resp . `` down '' ) step . denote , @xmath123 _ note : _ jump step paths never visit the @xmath124-axis since such a visit can only occur on an odd step and return to @xmath106 is impossible since all even steps are down . [ def : mh ] * hovering * paths , @xmath125 are one up paths with no vertex on the @xmath124-axis . * separated hovering * paths , @xmath126 , @xmath127 , are the subset of the hovering paths which have @xmath128 . the vertex @xmath129 is marked ( with an empty circle see figure [ fig : allpaths]f ) and known as the * separating vertex. marked separated hovering * paths , @xmath130 , are obtained from @xmath126 by marking ( with an solid circle see figure [ fig : allpaths]i ) subsets of the steps ( or vertices ) which return to @xmath106 . _ note : _ for all paths considered @xmath131 is either odd for all @xmath132 or even for all @xmath132 , i.e. the paths are confined either to the odd sublattice or the even sublattice . [ defn : ret ] a vertex of a ballot or dyck path in common with the @xmath124-axis is called a * contact. all contacts except the initial one are called returns. the polynomial @xmath133 , where @xmath134 is the number of returns for the path @xmath73 , is called the return polynomial * for ballot paths . for each of the three different representations the following lemma converts the matrix formula for the normalization factor into a sum over one of the path sets defined above , where the summand is a product of the step weights @xmath135 , an initial vertex weight @xmath136 , and a final vertex weight @xmath137 . the normalisation factor , @xmath138 for each of the three matrix representations , @xmath139 , can be written as @xmath140w^{f}_{j}(v_{2r})\label{eq:449}\end{aligned}\ ] ] where @xmath141 , @xmath142 , and @xmath143 , and the weight @xmath144 of a particular path @xmath73 with step sequence , @xmath145 is defined , for each of the three cases , as follows . for @xmath146 [ jweights ] @xmath147 for @xmath148 @xmath149 and for @xmath150 [ eqs : w2 ] @xmath151 the above lemma is a direct consequence of the lattice path interpretation of the @xmath53 and @xmath54 matrices as transfer matrices . the equivalence of the different expressions is a consequence of the invariance of the normalisation factor under similarity transformations relating the different matrix representations of @xmath16 and @xmath17 . in order to compute @xmath152 , it turns out to be a little more convenient to rearrange the weights associated with representation three . if we do so we obtain the following corollary . an equivalent set of weights for @xmath152 , is @xmath153 _ note : _ this rearrangement of the weights is only valid for computing @xmath152 . for correlation functions one has less freedom in the weight rearrangement . _ , title="fig : " ] later in the paper ( corollary [ alphabeta ] ) we will show that the normalisation factor can also be expressed in terms of separated hovering paths . @xmath154 where @xmath155 is defined by with [ eqs : w5 ] @xmath156 thus , for any particular path , @xmath157 , all the @xmath158 weighted steps ( if any ) occur to the left of vertex @xmath159 and all the @xmath160 weighted steps ( if any ) occur to the right of @xmath159 . we call this combination of paths and weights the `` canonical '' path representation of the normalisation factor . an example is shown in figure [ fig : canp ] . let @xmath161 be the set of marked separated hovering paths obtained from @xmath126 by marking subsets of the steps which return to @xmath106 then @xmath162 where the weight @xmath163 has a factor @xmath164 for each marked return step which occurs to the the left of @xmath129 and a factor @xmath165 for the other marked return steps . the lemma is proved in @xcite in two stages . first an involution on @xmath166 for which @xmath167 is the fixed set and then a bijection between @xmath167 and @xmath168 . [ z2z5 ] @xmath169 substituting @xmath170 in the weight attached to a path @xmath171 and expanding leads to a sum of terms obtained by weighting each return step to the right of @xmath129 ( if any ) with either @xmath3 or @xmath165 and the remaining returns ( if any ) by @xmath3 or @xmath164 . marking the subsets of steps weighted @xmath164 or @xmath165 determines a set of paths belonging to @xmath161 . in this section we briefly review methods to be used and results obtained previously @xcite as they will be required in the next section . one additional new lemma is stated . the constant term operation , @xmath172 $ ] , is defined by @xmath173 = \text { the constant term in the laurent expansion of $ f(z)$ about $ z=0$.}\ ] ] the number of @xmath174step lattice paths with step set @xmath175 which begin at @xmath176 and end at @xmath177 with no further constraint ( i.e. replacing the constraint @xmath178 in the definition of @xmath179 by @xmath180 ) is the binomial coefficient @xmath181 for which the constant term formula is @xmath182.\ ] ] by the reflection principle @xcite , the number of @xmath174step ballot paths of height @xmath36 is obtained by subtracting the number of unrestricted paths which begin at @xmath183 from those beginning at @xmath176 , both ending at @xmath184 . @xmath185 - ct[(z + 1/z)^t z^{h+2 } ] = ct[\la^t z^h ( 1-z^2)]\ ] ] where @xmath186 . differencing the binomial coefficients expresses @xmath187 in terms of factorials ; for @xmath188 even @xmath189 dyck paths are ballot paths ending at @xmath190 so the number dyck paths with @xmath67 steps is @xmath191 , @xmath192 all of which are solved by @xmath193 $ ] . historically , ballot numbers arise in the combinatorial problem of a two candidate election . if you ask how many ways can @xmath84 votes be caste such that the first candidate ends @xmath36 votes ahead of the second candidate and at any stage of the voting never has fewer votes that the second candidate . previously @xcite we proved the following proposition concerning the return polynomial ( see definition [ defn : ret ] ) . the return polynomial for ballot paths of length @xmath84 and height @xmath36 is given by @xmath194\ ] ] for @xmath195 , @xmath196 is determined by the recurrence relations @xmath197 the first and last equations are the same as for the unweighted ballot paths and are satisfied by @xmath198 $ ] provided that on expansion @xmath199 has no negative powers and @xmath200 ( which will be the case ) . we have introduced the factor @xmath199 to allow the second equation to be satisfied . but @xmath201 and choosing @xmath202 @xmath203&= & ct[\la^{t-1}(z+1/z)(1-z^2)g(z)]\\ & = & { \kappa}ct[\la^{t-1 } z ( 1-z^2 ) g(z ) ] + ct[\la^{t-1}(1/z -z)].\end{aligned}\ ] ] the result follows since the last term may be evaluated to give zero . @xmath204 the first equality follows by expanding in powers of @xmath205 and using . the second is obtained by rewriting the constant term formula as @xmath206\ ] ] and then expanding in powers of @xmath207 . and height @xmath208 . _ , title="fig : " ] this result suggests a bijection between ballot paths of length @xmath84 with @xmath30 returns and ballot paths of length @xmath209 and height @xmath210 and hence suggests a combinatorial proof . such a bijection was given in @xcite . a ballot path can be represented schematically as shown in figure [ fig : bal ] . the bubbles ( see definition [ defn : bub ] ) represent a , possibly empty , arbitrary elevated dyck path . using this schematic representation , proving is then straightforward . for simplicity we show the bijection in the case @xmath211 in figure [ fig : bp ] . b ) all return steps deleted c ) first step deleted produces a ballot path of height @xmath42 and length @xmath41.__,title="fig : " ] the expansion in powers of @xmath212 was also obtained by bijection in @xcite . in this case the bijection is between ballot paths with at least @xmath30 returns , @xmath30 of which are marked , and ballot paths of the same length but of height @xmath213 . with @xmath214 marked returns and b ) its schematic representation in terms of bubbles . c ) each marked return rotated through 90 degrees anti - clockwise ( or equivalently replaced by an up step ) producing a ballot path of height @xmath215 and length @xmath84.__,title="fig : " ] the above methodology is typical of that used for the more complicated two parameter case in the next section . a constant term formula will be derived and then rewritten in four different ways each giving rise to an expansion in pairs of different variables the coefficients of which are ballot numbers and are shown , by bijection , to enumerate various types of lattice path . in the next section we will also need the following lemma . [ lem : ret]let @xmath216 be the set of separated hovering paths ( definition [ def : mh ] ) then @xmath217 since we can shift any hovering path down on to the @xmath124-axis to give a dyck path with vertex @xmath218 marked , the lemma says the number of dyck paths with one contact marked is equal to the number of ( unmarked ) dyck paths two steps longer . a sketch of the bijective combinatorial proof is shown in figure [ fig : mc ] . the construction is a bijection , since , given any length @xmath219 dyck path a unique length @xmath220 marked contact dyck path is determined by deleting the rightmost step and the rightmost step from @xmath221 to @xmath106 ( and marking the left vertex of the latter step ) . ) and c ) adding a final down step produces a unique dyck path of length @xmath219__,title="fig : " ] we begin by computing @xmath152 using the @xmath222 weights we will refer to this as the two contact model . we will generalise the model by allowing arbitrary starting and ending heights for the paths . in particular , let the paths be of length @xmath84 , start at @xmath223 with @xmath224 and terminate at @xmath225 , @xmath226 , i.e. cross paths in @xmath227 . this model can also be thought of as a polymer model with two different surface interactions , or `` contacts '' . the path interacts with the `` thickened '' surface @xmath228 via two parameters , @xmath229 and @xmath230 and the partition function is defined by @xmath231 where the sum is over cross paths of length @xmath84 with given initial and final heights @xmath232 and @xmath233 and the weight @xmath234 is defined by and . this is a generalisation of the asep partition function @xmath152 thus @xmath235 this generalisation allows this partition function to be determined by recurrence relations similar to those for the return polynomial of ballot paths . by considering paths of length @xmath236 which can reach the point @xmath177 by adding one more step the partition function @xmath237 may be seen to satisfy the equations , @xmath238 these partial difference equations can be solved for @xmath237 by using the constant term method . [ prop1 ] with @xmath239 , @xmath240 and @xmath241 , for @xmath242 and @xmath243 @xmath244 + ( \delta_{y^i,1}+\k_2(1-\delta_{y^i,1}))ct[z^{y+y^i-2}\la^t g(z)]\ ] ] where @xmath245 _ note : _ for @xmath246 and @xmath243 @xmath247 substituting into the partial difference equations and noting that @xmath248 may be expanded in even powers of @xmath249 with no inverse powers shows that , and are satisfied . may be taken as the definition of @xmath250 and may then be transformed into @xmath251 to verify that this equation is also satisfied we note that @xmath252 in the case @xmath253 this gives @xmath254 & = ct[\la^{t-1}(1/z - z ) ] + \k_1{z}_{t-2 } ( 1\|y^i;\k_1,\k_2)+\k_2{z}_{t-1 } ( 2\|y^i;\k_1,\k_2)\end{aligned}\ ] ] and follows since the first term is zero . otherwise @xmath255 in which case @xmath256 + \k_2 ct[z^{y^i}\la^{t-1 } g(z)]\ ] ] and using this together with and the fact that the first term in vanishes when @xmath106 @xmath257\\ & = \k_2 ct[z^{y^i-1}(1/z - z)\la^{t-1}]+\k_1{z}_{t-2 } ( 1\|y^i;\k_1,\k_2 ) + \k_2 ^ 2ct[z^{y^i}\la^{t-1}g(z)]\\ & = \k_2 ct[(z^{y^i-2}-(1/z)^{y^i-2})\la^{t-1}]+\k_1{z}_{t-2 } ( 1\|y^i;\k_1,\k_2 ) + \k_2{z}_{t-1 } ( 2\|y^i;\k_1,\k_2)\end{aligned}\ ] ] and again follows since the first term evaluates to zero . an alternative constructive proof of this proposition is given in appendix b. [ crosspaths ] the number of @xmath174step cross paths which begin at @xmath258 and end at @xmath259 is given by @xmath260\\ \intertext{is the partition function restricted to anchored cross paths except that for $ y^i>1 $ the step leading to the first visit to $ y=1 $ has weight $ 1$. } \intertext{in particular } \label{z11 } { z}_t(y\|1;\k_1,\k_2 ) & = \za_t(y\|1;\k_1,\k_2 ) = ct\left[\la^t g(z)z^{y-1}\right].\end{aligned}\ ] ] _ notes : _ * it follows from the constant term formula that @xmath261 * setting @xmath262 gives the number of unweighted paths starting at height @xmath3 . @xmath263 = b_{t+1,y}\ ] ] as expected since the paths biject to ballot paths by adding an initial up step . with @xmath264 , @xmath265 and substituting in gives @xmath266\ ] ] which yields . @xmath267 is the number of cross paths which avoid @xmath106 . this follows since such paths are in simple bijection with the paths starting at height @xmath268 and ending at height @xmath269 , eg . just push the whole path down ( or up ) two units . the second term in is therefore the partition function for anchored cross paths . in the case @xmath270 anchored cross paths always have a first visit to @xmath106 and the step leading to this visit has a factor @xmath230 . removing this factor leaves @xmath271 which is therefore the partition function for anchored cross paths except that for @xmath272 the step leading to the first visit to @xmath106 has weight @xmath3 . setting @xmath273 in gives since the first term vanishes . [ cor : bp3]@xmath274\\ & \notag\\ & = \sum_{j=0}^{(t - y - y^i)/2 + 1}\quad\sum_{k=0}^{(t - y - y^i)/2-j+1}\k_1^j \k_2^k\ , \binom{j+k}{k}\ , b_{t-2j - k-1,y+y^i+k-3}\label{k12}\end{aligned}\ ] ] _ note:_when @xmath275 and @xmath276 the coefficient @xmath277 is indeterminate but setting @xmath278 gives the correct answer . this corresponds to the path which alternates between @xmath221 and @xmath106 . notice that replacing the factorials in the definition of @xmath187 by gamma functions gives @xmath279 for all @xmath280 . rewrite @xmath248 in the form @xmath281 expand in powers of @xmath229 and @xmath230 and use the ct formula for ballot numbers . equation may also be proved by a combinatorial argument . to simplify the proof we only consider the asep case @xmath282 . the extension to general @xmath283 , @xmath232 is straightforward . recall that , from the weight definition , , the returns to @xmath106 are weighted with @xmath284 or @xmath285 depending on whether the return is from below or from above @xmath106 . the binomial coefficient corresponds to choosing a particular sequence of @xmath286 and @xmath287 weighted returns . for each particular sequence of returns we need to show there are @xmath288 possible path configurations . we first represent a particular sequence schematically and then show any path corresponding to the schematic can be bijected to a ballot path with the correct height and length . schematically an example of a particular sequence of @xmath286 and @xmath287 returns is shown in figure [ fig : k1k2seq ] . . _ , title="fig : " ] we now perform three operations to biject a given sequence into a ballot path . * first delete all @xmath287 return steps , see figure [ fig : k1k2bij]a ) . this produces a path of length @xmath289 and height @xmath290 . * next , delete the first up step above @xmath106 ( if any which is the case if @xmath102 ) , see figure [ fig : k1k2bij]b ) . this produces a path of length @xmath291 and height @xmath58 . * finally , delete all @xmath292 steps originally below @xmath106 , see figure [ fig : k1k2bij]c ) . this produces a ballot path of length @xmath293 and height @xmath294 as required . given the sequence of @xmath286 and @xmath287 the reverse direction for the bijection is obtained by simply reversing the forward mapping see figure [ fig : k1k2bij ] . to a ballot path : a ) delete all @xmath287 return steps , then b ) delete first up step above @xmath106 , then c ) delete all steps originally below @xmath106 , to give d ) the final ballot path . _ , title="fig : " ] [ cor : bp2 ] @xmath295 rewrite @xmath248 in the form @xmath296 expand in powers of @xmath297 and @xmath298 and use the ct formula for ballot numbers . as with corollary [ cor : bp3 ] , the result , , may also be proved combinatorially as follows . again , in order to simplify the proof we consider only the asep case @xmath275 . the substitution @xmath299 means that a return which was weighted with @xmath300 is now weighted with either @xmath301 or @xmath3 . the factor @xmath302 is the number of paths with a subset of exactly @xmath61 of the returns from below marked and exactly @xmath58 of the returns from above marked corresponding to the choice of weight @xmath301 or @xmath3 . the binomial coefficient in is the number of ways of choosing a particular sequence of @xmath303 and @xmath304 weighted marks ( reading the path from left to right ) , whilst , for a given sequence , the ballot number represents the number of paths corresponding to the sequence . marked . c ) schematic one up path with only the last step @xmath304 marked . d ) an example showing how the schematic frying pan represents a , possibly empty , one up path . e ) an example showing a frying pan followed by a down step and an up step ( which forms a @xmath303 marked return ) . f ) an example showing a frying pan followed by an up step , then a bubble then a final down step ( which forms a @xmath304 marked return ) . _ , title="fig : " ] the most general path corresponding to a given sequence can be represented schematically by concatenating the corresponding schematic sub - paths shown in figure [ fig : fps]a ) and [ fig : fps]b ) with a final `` frying pan '' shown in figure [ fig : fps]c ) . examples of sub - paths corresponding to the three types of schematics are illustrated in figures [ fig : fps]d ) [ fig : fps]f ) . note , the shaded regions of the schematics represent any number ( possibly zero ) of steps . examples of two possible sequences are illustrated in figure [ fig : kbs ] . thus , for a given return sequence we need to show that there are @xmath305 return marked paths . without loss of generality we choose a typical sequence and represent it schematically as shown in figure [ fig : kbs ] . @xmath304 marked hovering path . a ) for the sequence @xmath306 and b ) for the sequence @xmath307.__,title="fig : " ] thus to prove the ballot number factor in we need to biject any schematic marked return sequence to a ballot path of length @xmath308 and height @xmath309 . we do this by bijecting each schematic in the sequence to a sub - ballot path ( plus , possibly , an extra step ) and then concatenate them all together . * thus , the last frying pan , of length say , @xmath310 , bijects to a height @xmath3 , length @xmath311 ballot path see figure [ fig : dfg ] . + * a marked @xmath304 schematic of length @xmath312 bijects to a height @xmath313 , ballot path with an additional final up step ( see figure [ fig : kbo ] ) . the final length is @xmath314 as an extra step has to be added . + schematic to a height two ballot path , with a final up step . the final path is one step longer than the original.__,title="fig : " ] * finally , a marked @xmath303 schematic of length @xmath315 bijects to a height @xmath3 , ballot path with an additional final up step ( see figure [ fig : kbt ] ) . the final length is unchanged . + schematic to a height one ballot path , with a final up step . _ , title="fig : " ] putting these moves all together is illustrated in figure [ fig : kbokbt ] , which show clearly a ballot path of length @xmath308 and height @xmath316 is obtained . and @xmath317 steps and a ballot path . a ) the marked one up path . b ) after the application of move illustrated in figure [ fig : kbt ] . c ) after the application of move illustrated in figure [ fig : kbo ] . e ) final ballot path after rotating the remaining marks . _ , title="fig : " ] an explicit evaluation of the partition function of the canonical path representation , defined by , is the case @xmath318 of the following result . [ alphabeta ] with @xmath319 and @xmath320 @xmath321 _ notes : _ * for the asep model ( @xmath322 ) with @xmath323 , reduces to the result of derrida et al @xcite equation ( 39 ) @xmath324 which they obtained directly from the algebra ( @xcite appendix a1 , equation ( a12 ) ) without the use of a matrix representation . * the result for the asep model may be written in terms of the return polynomial for ballot paths , thus @xmath325 this formula may also be derived using a path representation based on recurrence relations of derrida , domany and mukamel @xcite . by definition of @xmath158 and @xmath160 @xmath326[1-({\bar{\beta}}-1)z^2]=z^2(\la-{\bar{\alpha}}z)(\la -{\bar{\beta}}z)\ ] ] and from @xmath327 the result follows by expanding in powers of @xmath158 and @xmath160 , substituting in and using the ct formula for ballot numbers . in the asep case @xmath322 the coefficient @xmath328 in is equal to the number of dyck paths with @xmath30 returns to @xmath221 ( see ) . the equality of @xmath329 with @xmath69 , defined by , follows by raising the dyck paths so that they become hovering paths @xmath330 with returns to @xmath106 . the factor @xmath331 in corresponds to weighting the first @xmath61 returns of @xmath73 with @xmath158 and the remainder with @xmath160 . the sum over @xmath10 in is obtained by partitioning the weighted paths according to the position @xmath332 of the @xmath333 return ( ie . the separation vertex ) . the equality of @xmath152 and @xmath69 is also shown directly in @xcite by involution . let @xmath334 and @xmath335 then @xmath336 , @xmath337 and @xmath338 _ note : _ equation is not given in @xcite but ( 34 ) and ( 35 ) of @xcite together give the related formula @xmath339 this expression involves an infinite series whereas our expression is finite . by corollary [ crosspaths ] the coefficient in is the number of cross paths of length @xmath67 with @xmath340 and @xmath341 and the double sum extends over @xmath342 . the factor @xmath343 restricts the sum to anchored cross paths . the equivalence of and is shown in @xcite by constructing an involution on @xmath342 having @xmath344 as is its fixed point set . from @xmath345 and substitution in gives @xmath346 the result follows from using the expansion @xmath347 expanding in powers of @xmath249 and using the ct formula for ballot numbers .. again a combinatorial proof is possible which for simplicity we only give in the asep case . the equality with @xmath348 follows from corollary [ z2z5 ] . to obtain the ballot number formula we obtain a bijection between @xmath349 and the set of ballot paths of length @xmath350 and height @xmath351 . an example of the schematic representation of a path @xmath352 with a given number of @xmath164 and @xmath165 marked return steps and a separating vertex @xmath353 is shown in figure [ fig : cdseq ] . as there are no steps below @xmath106 we have pushed the hovering path down by unit height and consider it as a dyck path . and @xmath165 marked returns and a marked separating vertex @xmath353 . _ , title="fig : " ] as in the proof of lemma [ lem : ret ] , inserting an up step in each path at the position of @xmath353 and taking the union over all possible positions of @xmath353 starting at the last @xmath164 return and up to but not including the first @xmath165 return ( or from the beginning if there are no @xmath164 returns and to the end if there are no @xmath165 returns ) replaces the pair of bubbles on either side of the separating vertex by the set ballot paths of height one and one step longer , see figure [ fig : cdbij]a , b ) . replacing each marked return step with an up step ( and hence increasing the height of the path by 2 each time ) then produces a ballot path of length @xmath350 and height @xmath354 as required , see figure [ fig : cdbij]c ) . and @xmath165 return by an up step produces a ballot path of length @xmath350 and height @xmath354 as in c ) . _ , title="fig : " ] [ cor6 ] @xmath355\ ] ] note : this converts to the integral formula of derrida et al @xcite ( b10 ) with @xmath356 and and using a contour integral to pick out the constant term . it is related to the @xmath73 expansion ( see later ) . ( b10 ) was obtained by finding the eigenvectors of @xmath357 and is therefore an evaluation of @xmath358 . with @xmath359 in and using @xmath360\ ] ] now symmetrize the denominator by multiplying numerator and denominator by @xmath361 . the result follows using @xmath362 because the rest of the expression is now symmetric the contribution from the last term vanishes by replacing @xmath363 by @xmath364 and both @xmath365 and @xmath366 can be replaced by @xmath367 . with @xmath369 corollary [ cor6 ] may be written in the form @xmath370.\ ] ] which may be expanded to give @xmath371 where @xmath372.\ ] ] the asymptotic form of @xmath373 as @xmath24 was obtained in @xcite and will now be used to study the phase diagram for the asep model . first we outline the method by which the asymptotic form was obtained . notice that expanding the factor @xmath374 in in powers of @xmath0 gives an infinite series which is only valid for @xmath375 which is the point at which @xmath0 as a function of @xmath164 passes through its maximum value @xmath376 . instead we use to obtain a recurrence relation . thus noting that @xmath377 and substituting in gives @xmath378 where @xmath379 is the catalan number , . solving with @xmath380 gives @xmath381 which on substituting for @xmath382 in terms of @xmath164 must give a polynomial in @xmath164 . for the asep model @xmath383 and @xmath384 so @xmath385\\ & = { \omega_c}^{-(r+1)}(1 - 2\alpha)\,\,\theta(1 - 2\alpha ) + \sum_{j = r}^\infty c_j{\omega_c}^{j - r}\end{aligned}\ ] ] where @xmath386 is the unit step function . this is @xcite equation ( 3.61 ) . the asymptotic form for @xmath387 was obtained using @xmath388 and approximating the sum by an integral with the result @xmath389 and hence @xmath390 as @xmath391 , @xmath392 and these results agree with @xcite equations ( 48)-(50 ) . in the phase diagram there are therefore * three special regions @xmath393 , @xmath394 , @xmath395 + the partition function @xmath396 is obtained from that in @xmath397 by interchanging @xmath398 and @xmath399 . * three special lines @xmath400 , @xmath401 , @xmath402 the partition function @xmath403 is obtained from that in @xmath404 by interchanging @xmath398 and @xmath399 . * a special point where the lines meet @xmath405 table [ tableo ] shows the asymptotic form of @xmath329 for the above cases [ tableo ] .asymptotic forms of @xmath329 in the regions defined in the phase diagram in figure [ fig : pd ] . [ cols="^,^,^ " , ] the various formulae for @xmath248 when substituted in yield recurrence relations for @xmath406 . for example , using the identity @xmath407 substitution in and using the ct formula for catalan numbers leads to , for @xmath408 @xmath409 substitution in terms of @xmath229 and @xmath230 yields , for @xmath410 @xmath411 which may be initialised by @xmath412 and @xmath413 . the following identity @xmath414 when substituted in gives , using the @xmath415 formula for ballot numbers , for @xmath416 and odd @xmath417 @xmath418 which relates partition functions on lines of constant @xmath419 . the partition function for @xmath106 is determined by and the following proposition then the determines @xmath420 which provides the initial condition for . for @xmath421 @xmath422 from corollary [ alphabeta ] , for @xmath421 @xmath423 and the result follows from corollary [ alphabeta ] with @xmath424 . finally , substituting the identity @xmath425 in and using leads to the recurrence , for @xmath426 , and odd @xmath427 @xmath428 which relates partition functions along lines of constant @xmath84 and may be initialised using the above relations to find the partition functions for @xmath429 and @xmath430 . the probability of finding particles at positions @xmath431 is , using , @xmath432 where the un - normalised @xmath31-point correlation function @xmath433 is given by @xmath434 where @xmath435 . this expression may be thought of as replacing @xmath436 by @xmath16 in @xmath437 at each of the positions @xmath438 which is equivalent to replacing the @xmath54 matrix in a @xmath439 product by a unit matrix . thus in the path representations @xmath433 is obtained by modifying the allowed step definition such that for @xmath440 , the step @xmath441 ( beginning at @xmath442 and ending at @xmath443 ) is always an up step and has weight @xmath3 . we will say that @xmath444 is a _ forced up step_. this is illustrated in figure [ fig : corrpath ] . -point correlation path interpretation ( @xmath445 is shown above ) : each step starting at @xmath442 , @xmath440 must be an up step with weight one.__,title="fig : " ] . in the case @xmath446 , or @xmath262 , it is shown in @xcite that @xmath447 a catalan number . this may also be seen in terms of the third path representation ( corresponding to the third matrix representation of section [ matrepsec ] ) . @xmath448 is just the total number of one up paths of length @xmath67 and these biject to dyck paths of length @xmath219 by adding an up step at the beginning and a down step at the end of each path ( the dotted steps in figure [ fig : corrpath ] ) . the result follows since number of dyck paths of length @xmath449 is well known to be the catalan number @xmath450 . the result also follows from by setting @xmath451 and noting that @xmath452 which is the @xmath453 term . it is also shown in @xcite , equation ( 88 ) that @xmath454 this may be derived combinatorially as follows . again we use the third path representation and to avoid special cases we imagine that the paths are extended to @xmath221 by a further down step . for each path which contributes to @xmath433 we determine a subpath @xmath455 which starts with the last forced up step , @xmath456 , and ends when the path returns to the same height for the first time . this subpath contains a bubble which we suppose has length @xmath457 so that the subpath has length @xmath458 see figure [ bubble1]a . immediate return corresponds to @xmath459 and the maximum value of @xmath460 is determined by the condition , @xmath461 , that there are no further steps beyond @xmath455 . , showing the bubble following the last forced up step . b ) removing the bubble leaves a shorter dyck path _ , title="fig : " ] we can now define a new path obtained by deleting @xmath455 and joining the two ( possibly empty ) resulting sub - paths which remain . this path contributes to @xmath462 . the result follows by partitioning the paths contributing to @xmath463 according to the value of @xmath460 . for a given value of @xmath460 the number of configurations is therefore the product @xmath464 and the number of configurations of @xmath455 which is equal to the number , @xmath465 , of dyck paths of length @xmath457 . for @xmath466 , let @xmath467 then as pointed out in @xcite , equation may be iterated or , combinatorially , @xmath31 bubbles may be removed , to give the explicit formula @xmath468 where the upper limits are @xmath469 with @xmath470 . this formula was previously conjectured by derrida and evans @xcite on the basis of computer calculations up to @xmath471 . equation ( 45 ) of @xcite is the case @xmath472 of the following proposition which we now prove using a lattice path representation . an algebraic proof was given in @xcite . for @xmath473 @xmath474 _ notes : _ * when @xmath475 this reduces to since using @xmath476 the second sum of is just the missing term @xmath477 of the first sum . * in constructing a proof it was found that the first representation in terms of jump - paths was simpler to use than the third which we used in the special case of the previous section . it was explained after definition [ def : js ] , that jump - step paths ( representing @xmath14 ) never intersect @xmath221 . however this is not the case in calculating @xmath433 since the paths are modified by the forced up steps . this allows @xmath221 to be visited and then a forced up step returns the path to @xmath106 . notice that the down step leading to @xmath221 has a weight @xmath160 . _ proof : _ partition the modified jump - step configurations according as the last forced up step , @xmath456 , which starts at height @xmath478 ( case a ) or @xmath221 ( case b ) see figure [ bubble2 ] . . b ) case b : last forced up step starts at @xmath221 . note , the `` bubbles '' represent jump step subpaths . _ , title="fig : " ] [ [ case - a ] ] case a : + + + + + + + a subpath @xmath455 may be identified in exactly the same way as in the case @xmath479 but in this case the bubble of length @xmath449 obtained by deleting the first and last steps of @xmath455 is not an elevated dyck path since it may contain jump steps . however the number of configurations of @xmath455 is still equal to @xmath480 ( see below ) for any value of @xmath58 and since @xmath455 avoids @xmath106 its steps are unweighted . thus on partitioning the paths according to the length @xmath481 of @xmath455 a factor @xmath480 may be removed from the sum over paths having the same value of @xmath482 . when @xmath455 is deleted the remaining steps form a weighted path of length @xmath483 which has only @xmath484 forced up steps . summing over configurations of this path for given @xmath58 and then summing over @xmath485 gives @xmath486 . the first term of the proposition formula is thus derived provided that we can show that the number of configurations of @xmath455 is @xmath480 . now the paths @xmath455 of length @xmath487 biject to @xmath488 by vertical translation through distance @xmath489 and from and @xmath490 the last equality follows since when @xmath491 the one up paths which visit @xmath221 have zero weight and the remaining paths , which have weight @xmath3 when @xmath492 , biject to dyck paths by vertical translation through unit distance . this result is also proved combinatorially in @xcite . [ [ case - b ] ] case b : + + + + + + + the weighted sum over paths may be factorised into three parts . * a factor which arises from the subpath consisting of the first @xmath493 steps . the subpath ends at @xmath106 and has only @xmath484 forced up steps , therefore the sum over these subpaths yields @xmath494 . * a factor @xmath160 which arises from the next two steps which visit @xmath221 and return to @xmath106 . the second of these is the last forced up step @xmath456 having weight @xmath3 . * a factor arising from the subpath consisting of the remaining @xmath495 steps which is a jump step path beginning and ending at @xmath106 , avoiding @xmath221 . the weighted sum over these paths is obtained by setting @xmath496 in the normalising factor for paths of length @xmath495 , thus using corollary [ alphabeta ] @xmath497 the product of these three factors yields the second term of the formula . we have shown that the normalisation of the asep can be interpreted as various lattice path problems . the lattice path problems can then be solved using the constant term method ( ctm ) . the combinatorial nature of the ctm enables us to interpret the coefficients of the normalisation polynomials as various un - weighted lattice path problems usually as ballot paths . one particular form has a natural interpretation as an equilibrium polymer chain adsorption model . the `` @xmath0 '' form of the normalisation is particularly suited to finding the asymptotic expansion of normalisation and hence the phase diagram . we also formulate a combinatorial interpretation of the correlation functions . the lattice path interpretations enable us to make connections with many other models . in particular , because of the strong combinatorial nature of the ctm we are able to find a new `` canonical '' lattice path representation . in a further paper @xcite we show that this representation leads to an understanding of a non - equilibrium model ( the asep model ) in terms of a related equilibrium polymer model . also , having extended the polymer chain model so that the endpoints have arbitrary displacements from the surface the method of gessel and viennot @xcite , @xcite ( see also @xcite ) may be used to express the partition function of a network of non - intersecting paths as a determinant . in particular , the case of two paths gives the partition function for a vesicle model with a two parameter interaction with a surface . a bijection between these vesicles and compact percolation clusters @xcite then enables an analysis of the properties of the clusters attached to damp wall to be made . these applications will also be the subject of a subsequent publication . it is also of combinatorial interest to understand how the various path problems might be related . clearly they are related algebraically as they are all just representations of the same algebra ( and thus related by different similarity transformations ) . in a subsequent paper , @xcite , we show combinatorially ( using bijections and involutions ) how the various path representations are combinatorially equivalent . financial support from the australian research council is gratefully acknowledged . jwe is also grateful for the kind hospitality provided by the university of melbourne and rb is also grateful for the kind hospitality provided by royal holloway college , london university . we would like to thank andrew rechnitzer and aleks owczarek for reading the manuscript and providing useful comments . * appendix * the formula @xmath499\ ] ] clearly satisfies the general equation and the initial condition provided that @xmath500=0.\ ] ] in order to satisfy the boundary condition @xmath501 = { \kappa}_1ct[\la^{t-2}z(\bar z^{y^i } + u(z ) z^{y^i})]+ { \kappa}_2 ct[\la^{t-1}z^2 ( z^{y^i } + u(z ) z^{y^i})]\ ] ] or @xmath502= -ct[\la^{t-2}v(z ) z^{y^i-1}]\ ] ] where @xmath503 replacing @xmath249 by @xmath504 everywhere under the ct operation leaves the value unchanged so @xmath505 and is satisfied provided that @xmath506 or @xmath272 since @xmath243 . hence in this case @xmath272 @xmath507 when @xmath273 the boundary condition is satisfied if @xmath508 \equiv ct[\la^{t-2}(\bar z^2 - 2+z^2 -{\kappa}_1-{\kappa}_2 ( 1+z^2))]= -ct[\la^{t-2}v(z)]\ ] ] and replacing @xmath509 by @xmath510 in the second expression gives @xmath511 which satisfies and hence @xmath512	we show that the known matrix representations of the stationary state algebra of the asymmetric simple exclusion process ( asep ) can be interpreted combinatorially as various weighted lattice paths . this interpretation enables us to use the constant term method ( ctm ) and bijective combinatorial methods to express many forms of the asep normalisation factor in terms of ballot numbers . one particular lattice path representation shows that the coefficients in the recurrence relation for the asep correlation functions are also ballot numbers . additionally , the ctm has a strong combinatorial connection which leads to a new `` canonical '' lattice path representation and to the `` @xmath0-expansion '' which provides a uniform approach to computing the asymptotic behaviour in the various phases of the asep . the path representations enable the asep normalisation factor to be seen as the partition function of a more general polymer chain model having a two parameter interaction with a surface . * pacs numbers : * 05.50.+q , 05.70.fh , 61.41.+e * key words : asymmetric simple exclusion process , lattice paths , constant term method , enumerative combinatorics *
the initial baseline design for the advanced ligo gravitational wave detectors @xcite employs gaussian beams in the arm cavities . the leading noise source in the most sensitive frequency band of the instruments ( @xmath9 hz ) is the thermal noise in the substrate and reflective coating of the mirror test masses . lowering thermal noise is therefore of paramount importance for achieving a higher event rate in ligo . there are a number of other detectors that are being built or upgraded to similar specifications . while we will choose to study advanced ligo for definiteness , our general conclusions should be more widely applicable to any interferometeric detector that needs to limit thermal noise . some of the important parameters that we use are summarized in table [ advligoparams ] . ligo is a fabry - perot interferometer with four mirrored test masses . the resonant beams in the cavity measure the position of the test masses , averaging over the mirrored surface , with the average weighted by the power distribution of the beam . thus , the highly illuminated central area is weighted more than the mirror boundary that is left nearly dark . one way of decreasing the thermal noise is to flatten the beam so that a larger fraction of the mirror is in use . motivated by this intuitive reasoning , oshaughnessy et . @xcite proposed the flat topped mesa beams , which were subsequently explored in detail by them and others @xcite . these beams would lower the thermal noise by a factor of approximately 2.5 compared to the baseline design . the original mesa beam supported by nearly - flat mexican hat mirrors was found to be susceptible to a tilt instability @xcite . this triggered the proposal of a mesa beam supported by nearly - concentric mirrors @xcite . in the same paper , a family of hyperboloidal beams that include all mesa and gaussian beams previously considered was introduced . mesa is currently the leading alternative beam design for advanced ligo , and is being studied experimentally @xcite . in this paper we first discuss the formulation of hyperboloidal beams . the `` nearly - flat '' mesa is created by superposing minimal gaussians with generators uniformly distributed inside a cylinder , and the `` nearly - concentric '' by generators falling inside a cone , and passing through the center of the cavity . these two choices have the same intensity distribution on the mirrors , but the second has a much smaller susceptibility to tilt instability . the hyperboloidal beams smoothly interpolate between these two cases by twisting the generators of the minimal gaussians by an angle @xmath0 . after discussing some geometric properties of the beams , we present a proof that gaussian beams are a special case of the hyperboloidal beams , confirming a conjecture in @xcite . we then compute the three types of mirror thermal noise for a variety of hyperboloidal beam shapes , using a set of simple scaling laws developed in @xcite that simplify previous work @xcite . the first is substrate brownian noise , occuring due to mechanical dissipation in the material ; this is the least significant source of thermal noise . the substrate thermoelastic noise is caused by random thermal expansion . the coating also has both brownian and thermoelastic noise , but these follow the same scaling laws so we consider them as a single source . the coating noise is the most severe of the three types for the fused - silica substrates currently planned for advanced ligo . the substrate thermoelastic noise would dominate in a material like sapphire which has a higher thermal expansion coefficient . we show that the noises decrease with increasing width of the beam , as expected , and that the hyperboloidal beams have larger noise than the relevant mesa beams . the constraint on our ability to lower the noise comes from the need to keep the diffraction loss small . gravitational - wave interferometers must keep a large circulating power in the cavity , and so can not allow significant amounts of light to escape past the edge of the mirrors . the current design constraint used in the most recent papers @xcite is a diffraction loss of 1 part per million ( ppm ) per bounce for 17 cm fused - silica mirrors . the clipping approximation eq . indicates that the desired mesa width is approximately @xmath10 cm . previous work @xcite have shown that the clipping approximation is not accurate for gaussian beams of finite mirrors , and have used fast fourier transform simulations for accurate calculations . we calculate the diffraction losses accurately from eigenvalues of the cavity propagator using an exponentially convergent code that enables us to study the structure of mesa and other hyperboloidal beams as a function of @xmath1 , mirror radius , and twist angle @xmath0 in detail . we find that the diffraction loss is not a monotonic function of @xmath1 , but due to finite mirror effects has anomalous local minima where the loss is significantly below what is expected from the clipping approximation . these minima are observed to become more shallow and eventually disappear as the radius of the mirror is increased . however , for the mirror radii and beam widths relevant for advanced ligo the finite mirror effects are important . we show that they can be used to increase the width of the beam , lowering the noise even further than previous work . finally , we develop an iteration scheme to redesign the mirror , explicitly accounting for finite mirror effects . the iterated mirror is altered to match the phasefront of the primary eigenbeam of the finite mirror cavity , reducing the diffraction loss of this mode , thus allowing even larger beam widths to satisfy the diffraction loss constraint . the mathematical construction of the hyperboloidal beams are discussed in sec . ii , while the asymptotic limit of the hyperboloidal beams are derived in sec . iii . the thermal noise scaling laws are described in sec . the cavity propagator construction and eigenmode decomposition are presented in sec . v , with the results , including finite mirror effects , discussed in sec vi . we then summarize our work in sec . .advanced ligo parameters [ cols="^ , < , < " , ] [ beamintensity ] the beams we study are supported by two identical mirrors facing each other , forming a cavity of length @xmath11 . the mirrors are cylindrically symmetric around the optical axis , which runs along the length of the cavity and will be called the @xmath12 axis . the center of the cavity , equidistant between the mirrors , is @xmath13 ; the mirrors are located at @xmath14 and @xmath15 , where @xmath16 . the transverse distance from the @xmath12 axis will be denoted by @xmath17 , and the angular coordinate by @xmath18 . the cavity is fed with laser light with wavelength @xmath19 , and the distance between the mirrors is fine - tuned so that the cavity resonates in its fundamental mode , with a field amplitude @xmath20 and intensity @xmath21 . in this paper we focus on axisymmetric modes with no @xmath18 dependence in the beams . non - axisymmetric modes are important for studies of the tilt and parametric instabilities @xcite , but are not discussed in detail in this work . the narrowest possible gaussian mode that can exist in a cavity of given length is called the minimal gaussian , which has the intensity distribution \|u(r , z)\|^2 = ( - ) , where @xmath22 . a hyperboloidal beam is the superposition of minimal gaussians chosen such that the symmetry axis of the individual minimal gaussians are generators of a set of coaxial hyperboloids . the beam family has two parameters : @xmath0 , the twist angle one would have to rotate the two basis of a set of coaxial cylinders with respect to each other to obtain the hyperboloids and @xmath1 , the radius of a section perpendicular to the optic axis of the outermost hyperboloid at the end of the cavity . in the case @xmath2 , the propagation axes are parallel and fill a cylinder of radius @xmath1 . this is the mesa beam supported by nearly - flat mirrors . for @xmath3 the lines all cross at @xmath13 forming two cones . this configuration also generates a mesa beam , but one supported by nearly - concentric mirrors . varying @xmath0 smoothly deforms the beam and the mirror shape between the two configurations . some examples of the beam shapes are displayed in fig . [ beamintensity]a . for the cavity to support the desired beams , the phase of the electric field of the beam should be constant on the mirror surface . we will focus our attention on the mirror on the positive @xmath12 side of the cavity . the wavefront can be approximated by the `` fiducial spheroid '' , z = s_(r ) = . for @xmath2 this is the mirror plane @xmath15 ; for @xmath3 , the fiducial spheroid is a sphere centered on @xmath13 , and clearly the lines which generate the hyperboloid are all radii of the sphere . there are two equivalent expressions for the field amplitude evaluated on the fiducial spheroid . the first is the integral expression @xcite u_(r , s _ ) & = & ^r_0_0 ^2_0 d_0 r_0 , where @xmath23 is a complex constant . the second is the method that we use in this paper . the beam is constructed as shown in @xcite by an expansion in gausse - laguerre eigenbeams of spherical mirrors . they are closely related to the gauss - laguerre basis functions given by _ m ( ) = ( -^2/2 ) l_m(^2 ) , where @xmath24 is the @xmath25 laguerre polynomial . then the gauss - laguerre eigenbeams are _ m(r , z ) = _ m + where w(z ) = w_0 , r(z)=z+z_r^2/z , + ( z ) = ( z / z_r ) and @xmath26 . the expansion is written as u_(r , z ) = _ m=0^ a_m^ ( ) _ m ( r , z ) . the expansion coefficients that result in hyperboloidal beams are a_m^ ( ) = ( - ) ^m p ( m+1 , ) . @xmath27 is the incomplete gamma function p(a , x ) & = & _ 0^x e^-t t^a-1 dt + & = & . the mirror shape that supports a hyperboloidal beam is not exactly the fiducial spheroid . we make a correction @xmath28 so that the surface of the mirror is given by @xmath29 . the correction is chosen so that the mirror is located at a phasefront of the beam . we find the mirror surface from @xmath30 by h(r ) = k_0 ^ -1 ( u_(r , s _ ) - u_(0,s _ ) ) . the shape of @xmath28 , or the mirror itself , is generally referred to as a `` mexican hat '' , and some examples are displayed in fig . [ beamintensity]b . as expected , the beams for @xmath0 and @xmath31 are identical , as they correspond to hyperboloids that are simply twisted in opposite directions . there is a duality between @xmath0 and @xmath32 . this was first mentioned in @xcite for all @xmath0 s , then studied in more depth for @xmath33 and @xmath3 in @xcite and , finally , analytically understood in @xcite . this duality extends to several quantities . the beam intensity profiles are identical . the corrections to the mirror shape are opposite , @xmath34 . there are also dualities in the eigenvalues of the propagator @xcite that we will not discuss in this paper . it was conjectured by bondarescu and thorne @xcite that the beam becomes a gaussian in the limit @xmath4 . we will prove this analytically for the intensity profile of the beam , evaluated on the plane @xmath15 which would be the surface of a perfectly flat mirror . the intensity varies slowly enough with @xmath12 that this will also be the intensity profile on the mirror to a good approximation . our proof uses the expression for the beam amplitude in terms of a summation of gauss - laguerre functions . the essential ingredient is the realization that the expansion coefficients take the form @xmath35 as @xmath4 , where the constant depends only on @xmath0 . this allows us to analytically perform the summation to obtain the beam profile . in the limit @xmath36 , the incomplete gamma function @xmath37 , giving @xmath38 . the approximation @xmath15 yields w(z_r ) = w_0 , r(z_r ) = 2 z_r , ( z_r ) = / 4 , and the gauss - laguerre propagators become _ m(\|r , z_r ) = _ m(\|r ) e^- i m e^i ( \|r ) , where @xmath39 and the r - dependent part of the phase has been absorbed into @xmath40 . since in the end we will only be interested in the intensity profile , the exact form of @xmath40 is unimportant . the expansion for @xmath41 then becomes u_(\|r , z_r ) & = & _ m=0^(- ) ^m ( e^- ) ^m e^- l_m(\|r^2 ) e^i ( \|r ) + & = & ( ) ^2 e^- e^i ( \|r ) _ m=0^ ( i ) ^m l_m(\|r^2 ) [ amplitudeanalyticproof ] . we now use the generating function for the laguerre polynomials @xcite _ m=0^ l_m ( x ) t^m = ( - x ) to evaluate the sum in , with @xmath42 . the final result is an intensity profile \| u_(r ) \|^2 = , = . [ analyticgw ] the minimal gaussian @xmath43 is seen to have @xmath44 . the width is symmetric under @xmath45 , as expected from the duality relation @xcite , and goes to infinity at @xmath2 or @xmath46 . this includes every gaussian beam capable of resonating in a cavity of the given length . there are a number of noise sources limiting the sensitivity of ground - based gravitational - wave interferometers . seismic noise causes an effective cutoff in the lowest frequencies that can be measured . fundamental problems such as shot noise and radiation pressure noise , as well as technical issues , are important limitations on the sensitivity throughout the frequency band . however , the major contribution in the most sensitive frequencies of ligo is the thermal noise in the mirrors . reducing the thermal noise is the goal of this paper . the mirror consists of a substrate with a coating , and we must consider noises due to fluctuations of both . the substrate and coating have both thermoelastic and brownian contributions to the noise . thermoelastic noise is caused by expansions in the material caused by random heat flow . brownian noise is due to the coupling of normal modes of vibration by imperfections in the material . as a practical matter , both types of noises in the coating have the same scaling law so they do not have to be considered seperately . in fact , for the fused silica mirrors now under consideration for advanced ligo , the coating noises are the dominant contribution . however , use of a material like sapphire , with a higher coefficient of thermal expansion , would cause the substrate thermoelastic noise to dominate . we will calculate all three types of noise in this paper . a set of simple scaling laws were derived by lovelace @xcite in parallel with oshaughnessy @xcite that are applicable to beams of arbitrary shape . the noise is proportional to an integral depending on the intensity and an overall constant which is independent of the shape of the beam . the noises are given by s_n = c_n _ 0^i(k)^2 k^n dk , [ glnoise1 ] i(k ) = _ 0^r j_0(k r ) \|u(r)\|^2 r dr , where @xmath47 specifies the type of noise under consideration , and @xmath48 is the 2d axisymmetric fourier transform of the beam intensity with @xmath49 the radial wavenumber . the substrate brownian noise has @xmath50 , coating brownian and coating thermoelastic noises have @xmath51 , and substrate thermoelastic noise has @xmath52 . we are interested in comparing noises given by different beam shapes , so the overall constants @xmath53 are not important . the resulting amplitude sensitivity is the square root of the noise , and has units of @xmath54 . these scaling laws were derived for half - infinite mirrors , meaning that effects of the finite radius and thickness of the mirror are ignored . for the specific mirrors under consideration for advanced ligo , the width and thickness of the mirror are large enough compared to the beam width @xcite that this should be a good approximation . the results of @xcite suggest that we can expect corrections of not much more than ten percent to the half - infinite scaling law expressions that we are using . the idealized picture of a locked cavity is that the mirrors are perfectly aligned and a precise distance from each other . the beam should leave one mirror , reflect off the other , and when it returns it should be the same shape and exactly in phase . the beam will have lost some intensity due to diffraction and the finite extent of the mirror . in order to build up a very intense beam with a relatively weak laser , the beam must reflect very many times . the loss per half - trip ( from one mirror to the other ) must be below approximately 60 ppm . the majority of the loss will be due to absorption in the mirror and other factors , not diffraction . the commonly agreed upon budget for losses due to diffraction is 1 ppm . in previous work , the clipping approximation is often used to estimate the diffraction loss by calculating the fraction of the intensity of the beam which falls outside the mirror . it is given by 2 _ r^\|u_(r)\|^2 r dr . [ clippingapproximation ] our numerical code is accurate and fast enough to compute the diffraction loss directly from the eigenvalues of the propagator , which is more accurate . the propagator also allows us to estimate the difficulty of locking the interferometer . the finesse of advanced ligo is about 1200 @xcite . this sets the width of the resonance for the cavity @xcite to be about @xmath55 radians . if any other modes with a small diffraction loss have an argument within this distance of the desired mode , there will be severe problems with locking the cavity . for axisymmetric modes , we find that this is not the case and the arguments of the eigenvalues are well - separated enough that locking with mesa or any of the other beams that we study should be no more difficult than locking the currently proposed mesa . the propagator from a single point @xmath56 on one mirror to a point @xmath57 on the other ( see for instance @xcite ) is ( r,,r, ) = ( 1 + ) e^- i k_0 , where @xmath58 is the path length between the two points , @xmath59 the angle between the cavity axis and the path , and @xmath17 and @xmath18 are the standard cylindrical radial coordinate and azimuthal angle . the cavity is very long compared to the radius of the mirrors so we can immediately make the paraxial approximation @xmath60 . the path length can be approximated as & = & ( ( s_(r ) + s_(r ) - h(r ) - h(r ) ) ^2 + & + & ( r + r - 2 r r ( -))^2 ) ^1/2 + & & l + ( + ) - h(r ) - h(r ) + & - & ( - ) since @xmath61 and @xmath62 . this expression is the path length used to find the phase of the propagator . in the prefactor of the propagator we only need the much simpler @xmath63 in order to compute the amplitude to an accuracy of @xmath64 . we assume that the cavity is axisymmetric , which means that the eigenmodes of the propagator can also be written as eigenstates of rotation . thus we write the complex beam amplitude as @xmath65 . to determine the modes of the cavity , we have an eigenvalue problem given by the integral ^(i)_m v^(i)_m(r , ) = _ 0^r r dr d ( r,,r, ) v^(i)_m(r, ) , where @xmath66 is the associated eigenvalue , and @xmath67 is the mirror radius . integrating over @xmath68 produces ^(i)_m v^(i)_m(r ) = _ 0^r r dr ^r_m(r , r ) v^(i)_m(r ) , [ axieigenvector ] ^r_m(r , r ) & & j_m ( ) , + & & where for convenience we have defined a `` radial kernel '' @xmath69 . @xmath70 is the @xmath71 order bessel function of the first kind . in what follows , we specialize to axisymmetric modes ( @xmath72 ) , as we are not focusing on tilt or parametric instabilities which involve modes with @xmath73 . the above integral must be converted to a discrete sum to be suitable for numerical computation . we choose to do this using a chebyshev quadrature @xcite , which is appropriate for the finite range of @xmath74 . we use @xmath75 collocation points @xmath76 with associated integration weights @xmath77 . the integral is converted to ^(i ) v^(i)_0(r_j ) = _ k=1^n ^r_0(r_j , r_k ) r_k w_k v^(i)_0(r_k ) . this is a matrix eigenvalue problem which is easy to solve numerically . we will order the eigenstates by the number of radial nodes , the number of times that @xmath78 goes to zero . the fundamental mode ( @xmath79 ) has zero nodes and as @xmath80 it limits to the hyperboloidal beam that the cavity is designed to support . we will show that finite mirror effects cause the actual eigenstate to be slightly different . we performed a convergence test where we varied the number of collocation points , @xmath75 , and observed exponential convergence . the diffraction loss for the lowest eigenvalue for the @xmath3 mesa beam with @xmath5 cm , @xmath81 cm changes in the 5th digit when varying @xmath75 between @xmath82 and @xmath83 . we are using @xmath84 or @xmath85 for all of the calculations in this paper . on a typical single processor laptop computer , our code takes @xmath86 seconds for @xmath85 and @xmath87 minute for @xmath84 , to calculate the eigenmodes of any given cavity . the eigenvectors of the propagator are the field amplitudes of the cavity s resonant modes . fine - tuning the length @xmath11 of the cavity to @xmath88 changes the eigenvalue to @xmath89 . this tuning is used to select the desired mode ; when the argument of the eigenvalue is zero it will resonate in the cavity . the magnitude of the eigenvalue is the fraction by which the amplitude changes during a half - trip . we must have @xmath90 for finite mirrors , because some light will always be lost to diffraction . we define the diffraction loss in parts per million ( ppm ) over one half trip through the cavity as = 10 ^ 6 ( 1 - half trip of about one ppm . in the next section , we will discuss the results of our analysis , where we study the noise characteristics of the hyperboloidal beams given the constraint on the diffraction loss . [ dlossvsr ] [ dlossvsd ] [ dlossanddmin ] our goal of reducing the noises in ligo is constrained by the need to keep the diffraction loss at nearly 1 ppm . the hyperboloidal beams have two parameters , @xmath1 and @xmath0 . @xmath1 is roughly the width of the beam , and @xmath0 is the shape . the duality relation reduces the range of @xmath0 that we need to consider . we focus our attention on @xmath91 since the intensity profiles are identical to those in the range from @xmath92 to @xmath93 but the mirrors are nearly concentric as needed to decrease the tilt instability . the mesa profile obtained for @xmath3 has the flat - top shape required to decrease the thermal noises . as @xmath0 goes toward @xmath94 the beam becomes more rounded , losing the flat top and sharp decay near the edge . we have shown that @xmath4 the beam becomes a gaussian whose width is @xmath95 . for @xmath0 near @xmath46 , this gaussian is nearly infinitely wide , and at @xmath43 the beam becomes the minimal gaussian of width @xmath96 . from the clipping approximation we can estimate that at @xmath97 and @xmath98 , @xmath99 the diffraction loss is about 1 ppm . we do not have to consider any values of @xmath0 between these since the widest beam consistent with the diffraction loss constraint would be the @xmath99 gaussian . cm + diffraction loss & phase + 1.0 ppm & 0.0 + 121.7 ppm & -1.5104 + 182.3 ppm & -0.5835 + 334.8 ppm & -2.6677 + 7941.6 ppm & 2.2904 + 45401.8 ppm & 0.8325 + @xmath100 & @xmath101 cm + diffraction loss & phase + 1.0 ppm & 0.0 + 136.2 ppm & -1.5079 + 195.3 ppm & -0.5802 + 302.3 ppm & -2.6630 + 7225.3 ppm & 2.2976 + 44104.7 ppm & 0.8422 + @xmath102 & @xmath103 cm + diffraction loss & phase + 1.0 ppm & 0.0 + 18.6 ppm & -0.4797 + 951.8 ppm & 1.3048 + 3400.8 ppm & 2.7653 + 5870.7 ppm & -2.3322 + 61706.3 ppm & 1.4493 + if the mirrors were infinite in extent , no light would propagate off of the mirror and there would be no diffraction loss , giving eigenvalues of unit magnitude . the clipping approximation assumes that the beam is the one supported by infinite mirrors . this is not the case as diffraction also causes the beam profile to change . the propagator is a more accurate calculation because it finds the precise beam profile supported by the mirrors . the clipping approximation is typically an underestimate of the diffraction loss @xcite . in fig . [ dlossvsr]a , we show that this is indeed usually the case . however , for some ranges of beam width @xmath1 there is an anomalously low diffraction loss below the clipping approximation . to study this effect , we varied the mirror radius and computed the diffraction loss as a function of @xmath1 . the mirrors that we study in this paper have radius @xmath81 cm , so we compared with @xmath104 cm and @xmath105 cm . the local minimum becomes narrower and shallower for increasing mirror radii . this is suggestive of a finite mirror effect that will disappear when the mirror radius is significantly larger than the radius of the beam . [ dlossvsr]b shows the deviation of the beam from the infinite mirror mesa beam . a concentric mesa ( @xmath3 ) beam with @xmath5 cm is chosen . this is the location of a local minimum of the diffraction loss with respect to @xmath1 . the beam intensity is computed directly from the eigenvector of the propagator . the intensity is normalized by integrating over the mirror rather than over the entire mirror plane as with the infinite mirror case . however , this only causes a fractional error in the normalization on the order of @xmath106 . as shown in the figure , the finite mirror causes deviations from the infinite mirror beam ; the plateau of the beam is less flat than expected . when the radius of the mirror is increased , the deviations retain their shape but decrease in size . for @xmath107 cm , the beam is very close to the infinite mirror expectation . there is still a difference near the outer edge of the beam , causing the intensity to decay more quickly with radius than in the infinite mirror case . our numerical results suggest that the anomalous diffraction loss is related to the deviation of the beam from the ideal mesa shape . as @xmath1 increases , the clipping approximation predicts a smooth increase in diffraction loss due to the widening beam . finite mirror effects increase with the ratio @xmath108 , so they alone do not explain this unexpected behavior . for the values of @xmath1 that yield an anomalous diffraction loss , the variations around the plateau ( fig . [ dlossvsr]b ) have an organized shape with an approximate wavelength of @xmath96 . the variations in these cases have a shape such that they alter the falloff of the beam at the edge , i.e. in the last two centimeters of the mirror . this has an obvious beneficial effect on the diffraction loss . we expect that the fundamental mode of the cavity ( the hyperboloidal shape , with no nodes ) , should have the lowest diffraction loss . however , this is not the case for all choices of @xmath0 and @xmath1 . surprisingly , the diffraction loss of the first excited axisymmetric mode can decrease below that of the fundamental mode . this occurs , for @xmath3 , for a small range of @xmath1 around 10.5 cm , with diffraction losses around 4 ppm . the arguments of two eigenvalues remain well separated . [ crossing ] shows the crossing of the two diffraction loss curves . [ crossing ] to investigate the cause of this crossing , we choose the specific value @xmath109 cm and increase the mirror radius to 18 cm . the diffraction loss of the fundamental mode decreases from 3.7 ppm to 0.4 ppm , while the first excited mode only decreases from 3.6 ppm to 2.3 ppm so the diffraction loss of the fundamental mode is now lowest . further increasing the mirror radius to 20 cm causes the losses of the second and third excited modes to cross . despite changes of the mirror radius , the arguments of the eigenvalues change by less than a percent ( we are only considering eigenvalues with losses less than 10,000 ppm because otherwise they would dissipate too quickly to be of interest ) . this dependence on mirror radius tends to confirm that this is a finite mirror effect . having demonstrated that this effect can have substantial and beneficial effects on the diffraction loss , we now turn our attention to studying the parameter space of hyperboloidal beams in more detail . further studies of the precise cause of the anomalous losses may want to focus on the deviation of the beam from its theoretical expectation , as well as the behavior with changing mirror radius . the width of the beam increases with increasing @xmath1 , which averages the fluctuations over more of the mirror surface and therefore decreases the noise . it is less clear how the noise will behave when @xmath0 is changed . we begin by fixing @xmath10 cm and ignoring the diffraction loss constraint . [ noisefixd ] shows that all three types of noise increase as @xmath0 moves away from @xmath92 and @xmath46 . substrate thermoelastic noise is most affected by changing @xmath0 , followed by the coating noises ( recall that both types of coating noise follow the same scaling law ) . [ noisefixd ] as @xmath0 is decreased from @xmath46 , the noises increase if @xmath1 is kept fixed . at the same time , the diffraction loss decreases . if we keep the diffraction loss fixed , the @xmath1 can be increased as @xmath0 decreases toward @xmath94 . widening the beam tends to decrease the noise , which partially offsets the increase from changing @xmath0 . we fixed the loss at 1.4 ppm and found that the noise still increases for beams other than mesa . we expect that for larger diffraction losses this result will still hold . however , at 1 ppm the anomalous behavior of the diffraction loss due to finite mirror effects changes this conclusion . as @xmath0 decreases from @xmath46 toward @xmath94 , the beam loses its flat top and sharp falloff , and approaches the minimal gaussian . also , the @xmath1 corresponding to the local minimum in diffraction loss increases , and the local minimum becomes deeper and wider . [ dlossvsd ] shows the diffraction loss versus @xmath1 for three values of @xmath0 , while fig . [ dlossanddmin ] gives @xmath1 and diffraction loss at the local minimum for a range of @xmath0 . the local minimum of the diffraction loss for a mesa beam ( @xmath3 ) is at @xmath5 cm , and has @xmath6 ppm . as discussed above , the mesa is has the lowest noise of the hyperboloidal family for this diffraction loss . if a diffraction loss of strictly @xmath110 ppm is required , the beam must be reduced to a width of @xmath111 cm in the mesa configuration . alternatively , we can consider other values of @xmath0 . the local minima are displayed in fig . [ dlossanddmin ] , which gives the values of @xmath1 as well as the corresponding diffraction losses . note that for @xmath112 , the diffraction loss at the local minimum is now below 1 ppm . [ oneppmnoises]a shows the maximum @xmath1 that yields a 1 ppm loss , as @xmath0 is varied . there is a discontinuity between @xmath7 and @xmath113 because below @xmath113 the diffraction loss at the local minimum is below 1 ppm . the noises therefore drop substantially when @xmath7 as in fig . [ oneppmnoises]b . the coating noise decreases by 12% and the substrate thermoelastic by 19% , relative to the 1 ppm mesa beam . a strict requirement of 1 ppm diffraction loss therefore combines with the finite mirror effects to make the @xmath7 , @xmath114 cm configuration the best choice . [ oneppmnoises ] restructuring the mirror to specifically account for finite mirror effects allows us to increase @xmath1 in hyperboloidal beams ( thereby reducing thermal noise ) , while keeping within given diffraction loss constraints . the restructured beams can reduce the diffraction loss by a factor of 30 to 100 , allowing a wider beam . for the mesa case this allows for a net noise reduction of 30% for the beam satisfying the 1ppm diffraction loss constraint . as noted above , the original mesa beam used to construct the mirror is infinite in extent . the mirror is designed to be a phasefront of the theoretical beam . since the mirror is actually finite , for @xmath115 there can be substantial effects due to missing light that was incident on the mirror plane outside the mirror radius . to account for these finite mirror effects we reconstruct the mirror , with the goal of making the phase of the first eigenbeam constant at the mirror surface , rather than the phase of the idealized infinite beam . the propagator formulation allows us to explicitly calculate the phase of the eigenbeams , as a function of @xmath17 . as the mirror deviation from the fiducial spheroid @xmath28 enters into the calculation of phase through the propagator , we use an iteration scheme to adjust the mirror to match the eigenbeam phasefront motivated by the argument of the propagator : & & /2 + k_0 ( h(r ) + h(r ) - l ) [ argprop ] + & & - k_0 ( + ) .we see in eq . that if the phase of the eigenbeam is too large at some radius @xmath17 , reducing the value of @xmath28 should act to roughly reduce the phase of the new eigenbeam . with this motivation , we apply the simple iteration scheme : h_i+1(r ) = h_i(r ) - c , where the @xmath116 denotes the @xmath117 iteration , @xmath118 is the first eigenbeam for the mirror with deviation @xmath119 , and @xmath120 is an arbitrary constant less than unity , chosen to prevent overshoot . [ phase1steig ] [ iterations ] cm ( uniterated ) + diffraction loss & phase + 46.5 ppm & 0.0 + 128.5 ppm & -0.4313 + 341.7 ppm & -1.1895 + 10530.8 ppm & -2.1470 + 38445.0 ppm & 3.0277 + @xmath3 & @xmath121 cm ( iterated ) + diffraction loss & phase + 1.0 ppm & 0.0 + 320.6 ppm & -0.4319 + 1100.1 ppm & -1.1920 + 26167.4 ppm & -2.1593 + 66808.7ppm & 2.9873 + @xmath112 & @xmath122 cm ( uniterated ) + diffraction loss & phase + 43.5 ppm & 0.0 + 205.0 ppm & -0.4173 + 371.2 ppm & -1.1665 + 10626.6 ppm & -2.1082 + 50723.6 ppm & 3.0878 + @xmath112 & @xmath122 cm ( iterated ) + diffraction loss & phase + 1.0 ppm & 0.0 + 366.6 ppm & -0.4185 + 2013.4 ppm & -1.1701 + 32628.3 ppm & -2.1267 + 87359.8 ppm & 3.0335 + & @xmath1 [ cm ] & coating noise & substrate brownian noise & substrate thermoelastic noise + @xmath46 & @xmath123 & 0.72 & 0.84 & 0.63 + @xmath124 & @xmath125 & 0.80 & 0.90 & 0.69 + this iteration scheme successfully reduces the relative phase of the eigenbeam , as shown in fig [ phase1steig ] . optimizing the mirror surface to match the phasefront of the primary eigenbeam also acts to reduce the diffraction loss for that mode in general , with the iteration scheme providing convergence towards an apparent lower bound for the diffraction loss , while increasing the diffraction loss for other higher - order eigenbeams . this lower bound increases with @xmath1 ( fig [ davedifflist ] ) . this diffraction loss is plotted against the iteration number for the mesa ( @xmath3 ) case with @xmath126 cm in fig [ iterations ] . the iteration scheme is shown to lower the diffraction loss for this @xmath1 from 46.5 ppm to a @xmath87 ppm lower bound , satisfying the required design constraint . the diffraction losses of higher order modes are more than doubled in the iterated case as illustrated in table [ tablephasesdave ] . [ davedifflist ] [ daveintensity ] the beam for the iterated mirror with @xmath126 cm is close to the original mesa , but with variations in the central plateau of relative amplitude @xmath127 and variations of radial wavelength @xmath128 ( fig [ daveintensity ] ) . this seems to be an unavoidable consequence of a finite @xmath67 , as even the original eigenbeam has roughly similar features . despite changing the variations in the plateau of the beam intensity the process of iteration does not significantly affect the noises computed using eq . . similarly the iterated mirror has variations of the central mirror shape of similar radial scale , with amplitude on the order of 2 nm , shown in fig [ davemirror ] . the most significant feature of the iterated mirror is the inward tilting of the outer edge of the mirror , though preliminary studies show that the primary contribution to reducing the diffraction loss is due to the variations near the center of the mirror . [ davemirror ] reformulating the mirror to account for finite mirror effects allows us to increase the @xmath1 parameter of the beam from @xmath129 cm to @xmath8 cm for a concentric mesa beam while still maintaining a 1ppm diffraction loss . though this design may introduce more complications in the construction of the mirror itself , it allows a significant reduction in noise by broadening the beam . this iteration scheme can also be used for other values of @xmath0 , as shown in table iii , where the iterated mirror for @xmath112 has a diffraction loss lower bound of 1 ppm for @xmath130 cm . however , we find that @xmath3 is optimal for noise reduction . ( see table [ tablenoise ] . ) in this paper , we studied thermal noise and diffraction loss for the hyperboloidal family of light beams and mirror shapes in detail for the first time . this family had been initially proposed to unify the concentric ( @xmath131 ) and nearly - flat ( @xmath2 ) mesa configurations through variations of the twist angle @xmath0 . in this paper we also presented an analytic proof that gaussian beams are a limiting case of the hyperboloidal beam as @xmath4 . this was previously conjectured in ref . we developed a pseudo - spectral code both fast and accurate enough to calculate the diffraction loss directly from the beam propagator . we find that the finite radius of the mirror causes beam shapes to deviate significantly from the infinite - mirror theoretical expectations . this causes a previously unnoticed local minima in the diffraction loss that can be exploited to find a natural beam width @xmath1 for the current diffraction loss constraints of about 1 ppm . for an @xmath3 mesa beam a local minimum occurs at @xmath5 cm giving 1.4 ppm diffraction loss for a mirror of radius @xmath81 cm . if one requires a strict enforcement of the 1 ppm diffraction loss we show that a hyperboloidal beam with @xmath7 and @xmath114 cm has lower noise than that of the @xmath3 mesa with 1 ppm diffraction loss . the coating noise changes by about 12% and the substrate thermoelastic noise and substrate brownian noise change by 5% and 19% , respectively . we also propose new mirror and beam shape configurations that explicitly account for finite mirror effects by reformulating the mirror surface to coincide with the phasefront of the primary eigenbeam . these beams reduce the diffraction loss by more than an order of magnitude for the range of @xmath1 considered here ( between @xmath132 cm and @xmath133 cm ) . this allows the use of wider beams for the same diffraction loss constraints on the primary eigenmode , while the diffraction losses of higher order modes ( both axisymmetric and non - axisymmetric ) are increased . we are able to widen the @xmath3 mesa beam for a @xmath81 cm mirror while keeping the diffraction loss fixed at 1 ppm from a width of @xmath111 cm to @xmath126 cm . this lowers the coating thermal noise by about 30% ( compared to the smaller @xmath1 mesa ) and the other noises ( substrate brownian and substrate thermoelastic noise ) by comparable factors . however , feasibility of the construction of the mirrors must also be taken into account . the non - iterated beams taking advantage of the local minimum in diffraction loss discussed above are supported by strict hyperboloidal or mesa mirrors , which would be no harder to make than the current mesa designs , and would still lower the coating thermal noise by 12% . if one is to consider the more ambitious goal of lowering the coating thermal noise by 28% while using a beam that is very similar to mesa through the iteration scheme described , the limitations on mirror manufacturing errors are likely to be more stringent , but still less than the currently considered conical beams @xcite . in addition , the methods developed here for reducing the diffraction loss of the mesa beam may be applied in the case of the conical beams previously considered @xcite . the phase fronts of conical beams considered there have not been optimized to match the finite mirror surface . recently , parametric instability @xcite was found to be a serious problem in advanced ligo . choosing @xmath1 at the minimum of the diffraction loss curve of the hyperboloidal beams increases the diffraction loss of the higher eigenmodes , thus in principle somewhat improving the parametric instability . the effect is most pronounced in the case of the iterated mirrors which also have the most drastic thermal noise reduction . we thank yanbei chen for guidance , useful discussion and for foreseeing many of the issues presented in this paper . we are grateful to geoffrey lovelace for useful discussions regarding the noise calculations . our calculations make use of the gnu scientific library @xcite , it++ @xcite , and the amd core math library @xcite . we thank our advisors james york , anna flanagan , ira wasserman , saul teukolsky , dong lai and barry barish for support and encouragement . this work was partially supported by funds from the following sources : sofja kovalevskaja programme from the alexander von humboldt foundation , nsf grants ast-0707628 , ast-0606710 and phy-0652952 . apl and rb were supported to attend the gr18 / amaldi7 conference , which was helpful to this work , by a combination of cornell , nsf and iupap / isgrg travel grants .	thermal noise is expected to be the dominant source of noise in the most sensitive frequency band of second generation ground based gravitational wave detectors . reshaping the beam to a flatter wider profile which probes more of the mirror surface reduces this noise . the `` mesa '' beam shape has been proposed for this purpose and was subsequently generalized to a family of hyperboloidal beams with two parameters : twist angle @xmath0 and beam width @xmath1 . varying @xmath0 allows a continuous transition from the nearly - flat ( @xmath2 ) to the nearly - concentric ( @xmath3 ) mesa beam configurations . we analytically prove that in the limit @xmath4 hyperboloidal beams become gaussians . the ideal beam choice for reducing thermal noise is the widest possible beam that satisfies the advanced ligo diffraction loss design constraint of 1 part per million ( ppm ) per bounce for a mirror radius of 17 cm . in the past the diffraction loss has often been calculated using the clipping approximation that , in general , underestimates the diffraction loss . we develop a code using pseudo - spectral methods to compute the diffraction loss directly from the propagator . we find that the diffraction loss is not a strictly monotonic function of beam width , but has local minima that occur due to finite mirror effects and leads to natural choices of @xmath1 . for an @xmath3 mesa beam a local minimum occurs at @xmath5 cm and leads to a diffraction loss of @xmath6 ppm . we then compute the thermal noise for the entire hyperboloidal family . we find that if one requires a diffraction loss of strictly 1 ppm , the @xmath7 hyperboloidal beam is optimal , leading to the coating thermal noise ( the dominant source of noise for fused - silica mirrors ) being lower by about 10% than for a mesa beam while other types of thermal noise decrease as well . we then develop an iterative process that reconstructs the mirror to specifically account for finite mirror effects . this allows us to increase the @xmath1 parameter to @xmath8 cm for a nearly - concentric mesa beam and lower the coating noise by about 30% compared to the original mesa configuration .
since the suggested association between a gamma - ray burst ( grb ) and supernova 1998bw ( galama et al . 1998 ) , the possible general connection between grbs and sne has been discussed ( cen 1998 , wang et al . in contrast , kippen et al . ( 1998 ) found no evidence for a grb - sne connection , and grazianni , lamb & marion ( 1998 ) even conclude that not all sn ib , c / ii can be associated with grbs , although a fraction might be . on the other hand , wang et al . ( 1998 ) found seven candidate sn - grb correlations while bloom et al . ( 1998 ) estimated these sn - grb isotropic explosion energy range from @xmath11 erg / s , and suggested that a subclass of grbs is associated with supernova ib , c / ii . jets have been introduced to solve two problems raised in the grb - sn united picture since 1 ) the great difference between the batse - inferred grb rate with a narrow luminosity function ( @xmath12 h@xmath13 mpc@xmath14 yr@xmath15 ) ( kommers et al . 1999 ; schmidit 1999 ) and the supernova rate ( @xmath16 h@xmath13 mpc@xmath14 yr@xmath15 ) , and 2 ) the great difference between the explosion energy of grbs and supernovae . with a jet the total energy necessary to produce the observational fluxes will be reduced by a factor @xmath17 ( @xmath18 is the lorentz factor of a relativistic jet ) , and the total rate will be greater by a factor @xmath19 ( cen 1998 ; wang et al . 1998 ) . were a wide grb luminosity function or jet to operate , batse - detected grbs would only be a small fraction of all grbs that occur . in this letter , we compare grb rates derived from supernova rates and density evolution to that of the measured rates from the fourth batse catalog of grbs to investigate possible constraints on the grb luminosity function and jet parameters . we also discuss two other related interesting questions : 1 ) can the special relativistic effects of a jet provide by itself an acceptable luminosity function for grbs ? 2 ) can the large amount of undetected bursts contribute to the observational gamma - ray background near 100 kev for an acceptable grb luminosity function ? the former provides an alternative way to test if the gamma - ray burst emission is isotropic or random in the jet s comoving frame . if supernovae are the mysterious sources of grbs , then the grb `` equivalent isotropic '' explosion energy could be within a relatively large range : from @xmath20 erg / s , deduced for grb 990123 ( kulkarni et al 1999 ) to @xmath21 erg / s estimated from sn ib / c - grbs associations ( wang & wheeler 1998 ; bloom et al . we assume the unknown intrinsic luminosity distribution follows a power - law : @xmath22 where @xmath23 is the normalization factor . we fix @xmath24 erg / s . we find that varying @xmath25 from @xmath26 to @xmath27 erg / s affected our results only slightly ; @xmath28 was allowed to vary in the range @xmath29 - @xmath25 . observed spectra of grbs can be approximated as a power - law : @xmath30 with spectral index @xmath31 ( mallozzi et al . for source density , we adopt the broken power - law function we have used before as a rough description for sn ib / c evolution ( che et al . 1999 ) : @xmath32 where @xmath33 is the comoving grb density at redshift @xmath34 per mpc@xmath35 yr@xmath15 . we set @xmath36 , @xmath37 ( lilly 1996 ) , and @xmath38 ( madau 1998 , sadat 1998 ) . we will assume a universe where the calculations are particularly simple : a flat friedmann universe with cosmological constant @xmath39 . we do not expect the results would differ significantly in universes with a significant @xmath40 . in our asssumed universe , the number of observed gamma - ray bursts peak flux brighter than p is : @xmath41 the observed peak flux of a grb at redshift @xmath42 with intrinsic luminosity @xmath43 is : @xmath44 where we fixed the spectral index to be @xmath45 ( mallozzi et al . 1996 ) , and @xmath46 erg @xmath47 s@xmath15 @xmath48 photon @xmath47 s@xmath15 . @xmath49 where @xmath50 is measured in kev and @xmath51 is proper motion distance . the total number of bursts we observe every year is @xmath52 as derived from eqn . we selected the 939 bursts with peak flux greater than @xmath53 from the fourth batse grb catalog ( batse 4b ) to get the measured log @xmath5 - log @xmath6 relation that will be used for comparison . we compared this log @xmath5 - log @xmath6 to grbs following the evolving supernova rate , finding the best fits to the luminosity function range parameter @xmath28 and slope parameter @xmath3 . the rate was normalized by @xmath54 as given in eq . we searched the acceptable range of @xmath28 and @xmath3 with the k - s confidence level @xmath55 . the luminosity function width was allowed to vary from @xmath56 , while the luminosity function slope was allowed to vary from @xmath57 . the search result is shown in figure 1 : the acceptable range of luminosity function widths were @xmath58 , while the acceptable rage for the luminosity function slope was @xmath59 . setting the @xmath60 yr@xmath15 , which is the batse yearly rate corrected for the duty cycle ( meegan et al . 1992 ) , and constraining log @xmath4 and @xmath3 from equation ( 5 ) , we can find lines of constant rate ( @xmath61 ) in the @xmath3 - log @xmath4 frame . if we further suppose that grbs are connected with supernovae , then for the fireball model , the grb event rate should follow the supernovae rate @xmath16 h@xmath13 mpc@xmath14 yr@xmath15 . for a jet with a lorentz factor @xmath62 , however , the grb rate will decrease to @xmath12 h@xmath13 mpc@xmath14 yr@xmath15 . two lines of constant grb rate are plotted in figure 1 and labeled as lines ii and iii , representing isotropically emitting fireball grbs and grbs beamed into a @xmath63 jets respectively . to the left of these two lines , grbs will have a higher event rate than detected by batse . for the fireball model , line ii traverses the k - s test acceptable region at around @xmath64 erg / s , which is enough to produce a gamma - ray burst by shock , but is quite close to the hypothesized theoretical minimum ( meszaros 1999 ) . therefore , grbs occurring at the supernova rate are not in conflict with the isotropic fireball model . line iii shows that the minimum isotropic explosion energy of grbs from a jet is @xmath65 erg / s . from lines ii and iii , it is not possible to discern whether a jet is preferred from isotropic emission , since both cross the allowed k - s fit region . most grbs in the jet model , however , expend more energy than those in an isotropic fireball model , possibly indicating that jet models are more likely to be related to massive stars ( woosley 1998 , sari 1999 ) or hypernovae ( paczynski , 1998 ; iwamoto et al . 1998 ; woosley et al . 1998 ; wheeler et al . 1999 ) . as the luminosity function width @xmath8 increases , the fraction of undetected bursts become larger , which leads to more grb photons contributing to the diffuse gamma - ray background near 100 kev . quantitatively , this can be written as : @xmath66 where @xmath67 is the average luminosity of individual sources , and @xmath68 represents the observational gamma - ray background intensity . we can deduce the required @xmath33 from this equation and estimate the required minimum grb rate from the observation of cosmic ray background ( watanabe et al 1998 , comastri 1998 ) . this result is shown by line i in figure 1 . line i lies outside of k - s test region which indicates that grbs can not dominate the diffuse cosmic gamma - ray background . a relativistic jet or fireball has been considered as a way to solve the enormous energy release of grbs ( paczynski & xu 1994 ; rees & meszaros 1994 ; sari & piran 1997 ; pilla & loeb 1998 ) . though the models of shocks ( internal or external ) for producing a gamma - ray burst usually assume a magnetic field and relativistic particle acceleration ( meszaros et al 1994 ; panaitescu et al 1999 ) , the gamma - ray burst radiation should be confined within the beam . the simple assumption that the grb radiation direction is random in the rest frame of the shocks can not be excluded absolutely . in this section , we test this assumption by checking if the grb luminosity function can be induced by a purely special relativistic effect . we assume that the angle between the line of sight and beaming axis is @xmath69 , the velocity of beaming is @xmath70 and @xmath71 is the intrinsic luminosity of grb in the beaming comoving frame , then the observational flux come from a source at redshift z is ( blandford & konigl 1979 ) : @xmath72 @xmath73 where @xmath74 , the lorentz factor . we assume the beaming - induced @xmath75 , here @xmath76 corresponds to the biggest viewing angle @xmath77 we can see a burst . @xmath78 , seen by the observer within the beam , @xmath79 . the induced luminosity function @xmath80 is : @xmath81 @xmath82 with @xmath83 , the induced luminosity function index is constrained to be @xmath84 , far below the required index for log @xmath5 - log @xmath6 . we indicate this allowed region in figure 1 . we note that this allowed region does not intersect the allowed region created by the necessity of fit to the measured batse log @xmath5 - log @xmath6 . this contradiction implies that the grb luminosity function is not created by purely special relativistic bulk motion . perhaps the real luminosity function of grbs has only a component due to beaming . if so , the beaming induced luminosity function can not dominate but modulate the intrinsically wide luminosity function . for this to be true , the intrinsic dynamic range induced by beaming must be quite modest . from inspection of figure 1 , we see that it if beaming creates only @xmath85 , it will not dominate the wider intrinsic luminosity function . from eqn . 7 , we see that if the gamma - ray burst is beamed into an angle of @xmath86 , then the extra amount of beaming generated by relativistic motion will generate an @xmath85 for @xmath87 , and so could be hidden by the intrinsic luminosity function of grbs . this result can be tested by detecting the off - axis x - ray emissions from beamed grbs , caused by purely special relativistic effects , proposed by woods and loeb ( 1999 ) . up to now , the sources for grbs remain mysterious . though some observations suggest that grbs are associated with supernovae type ib , c / ii , we need more information to confirm this kind of connection . however , the nature of the intrinsic luminosity function is a more general problem . in this letter , we have studied the luminosity function from various aspects and given a clearer sketch on how the luminosity function dynamic range is constrained by the gamma ray burst rate , and what contradictions exist between the observationally favored luminosity and a beaming - induced luminosity function . our results are based on the assumption of connection between grbs and sne . our results show 1 ) if a model for grbs is considered for the consistence with grb rate , we can not ignore the possible intrinsic dynamic range of luminosity function to the rate . 2 ) if we consider the source density evolution of grbs , the required width of luminosity function , log @xmath4 can not be less than 2 . 3 ) the required minimum explosion energy for grbs from the relative grb rate , the limits on the width of the grb luminosity function , and the exclusion of special relativistic motion as the cause of the luminosity function may turn out to be helpful to those studying the grb radiation and progenitor mechanisms . this research was supported by grants from nasa and the nsf . i would like to thank robert nemiroff for helpful discussions and a critical reading of the manuscript . i also thank peter meszaros for his valurable discussion and comments . blandford , r. d. & konigl , a. 1979 , , 232 , 34 bloom , j.s . et al . 1998 . astro - ph/9807050 cen , r. 1998 , 507 , l131 che , h. , yang , y & nemiroff , r. j. , , 1999 , 516 , 559 comastri , a. 1998 , astro - ph/9812307 galama , t.j . 1998 , nature , 395 , 670 graziani , c. , lamb , d.q . & marion , g. h. , 1998 , astro - ph/9810374 iwamoto k. et al . , 1998 , nature , 395 , 672 kippen , r. m. et al . , 506 , l27 kommers , j. m. et al . 1999 , astro - ph/9809300 kulkarni , s. et al . , 1999 , astro - ph/9806016 lilly , s. j. et al . 1996 , , 460 , l1 madau , p. valle , m. d. & panagia , n. 1998 , mnras , 297 , l17 mallozzi , r.s . 1996 , , 471 , 636 meegan , c. a. , et al . 1992 , nature , 353 , 143 meszaros , p. 1999 , private communication meszaros , p. , rees , m. j. & papathanassiou , h. 1994 , , 432 , 181 paczynski , b. 1998 , , 494 , l45 paczynski , b. & xu , g. 1994 , , 427 , 708 panaitescu , a , spada , m. & meszaros , p. , astro - ph/9905026 pilla , r. p. & loeb , a. 1998 , , 494 , l167 piran , t. 1992 , , 389 , l45 rees , m. j. & meszaros , p. , 1994 , , 430 , l93 sadat , r. et al . 1998 , a & a , 331 , l69 sari , r. 1999 , , 519 , l17 sari , r. & piran , t. 1997 , mnras , 287 , 110 schmidt , m. 1999 , astro - ph/9908206 wang , l , wheeler j.c . 1998 , , 504 , l87 watanabe , k. et al . 1998 , astro - ph/9809197 wheeler , j. c. , hoflich , wang , l. 1999 , astro - ph/9904047 woosley , s. e. , eastman r. , schmidt b , 1999 , , 516 , 788 woods , e & loeb , a. astro - ph/9903377	if gamma - ray bursts ( grbs ) are generally associated with supernovae like 1998bw , a relatively wide intrinsic luminosity function is implied , which indicates the existence of a large amount of undetected dim bursts , and a much higher event - rate than is often assumed . if it is assumed that the intrinsic luminosity function of grbs is a power - law : @xmath0 ( @xmath1 , @xmath2 ) , data from the batse 4b catalog can be used to constrain slope index @xmath3 and the dynamic range width log @xmath4 . using a k - s test comparison with the observational log @xmath5 - log @xmath6 , we find constraints on the grb fireball model , grb jets , and the possible grb contribution to cosmic gamma - ray background . we find the acceptable dynamic range for @xmath7@[email protected] results show that jet model is more likely to be related more highly energetic explosion than fireball model.our studies also show that the luminosity function provided by a purely special relativistic effect on a jet is outside of the k - s test acceptable range . were intrinsic beaming to confine the jet to @xmath10 however , the effects of relativistic beaming would not dominate .
srruo@xmath2 is one of the rare 4@xmath3 transition - metal oxides that shows a ferromagnetic order and has attracted considerable interest due to its fascinating electronic and magnetic properties . this compound has a pseudocubic perovskite structure ( a gdfeo@xmath2 type orthorhombic structure ) . its curie temperature @xmath6 and ordered ferromagnetic moment are 160 k and approximately 1@xmath7/f.u . , respectively.@xcite the fundamental and important question concerning this compound is how the electron correlation effect affects its electronic properties . at first glance , because ru 4@xmath3 electrons are generally expected to have a delocalized nature in the crystal , it is natural to consider that the electron correlation effect is not significant and that the ferromagnetic order can be described well in terms of the stoner theory in an itinerant electron picture . however , this compound shows so - called bad metal behavior , and its resistivity exceeds the ioffe - regel limit at high temperatures , indicating a breakdown of the quasi - particle description.@xcite optical spectroscopy has shown that the charge dynamics deviate from fermi liquid behavior.@xcite moreover , the rhodes - wohlfarth ratio @xmath8/@xmath9 , where @xmath8 and @xmath9 represent the paramagnetic dipole moment estimated from the curie constant and the ordered moment at zero temperature , respectively , is approximately 1.3.@xcite this value indicates that srruo@xmath2 is close to the localized - moment limit . these results imply that the correlation effect of the ru 4@xmath3 electrons plays an important role and that the electronic state is not conventional . photoemission ( pes ) spectroscopy provides a direct probe into the electronic band structure . a number of pes experiments have been carried out to clarify the electronic state of srruo@xmath2 . earlier pes studies reported that although the overall structure of the valence spectra consisting of the ru 4@xmath3 and o 2@xmath10 states was roughly reproduced by calculations based on the density functional theory ( dft ) , the ru 4@xmath3 derived peak near the fermi level ( @xmath11 ) was strongly suppressed compared to that derived from dft calculations.@xcite however , subsequent pes studies have revealed , by using bulk sensitive soft x - rays and/or @xmath12 @xmath13 grown thin film samples , that pes spectra are very sensitive to surface conditions , and the valence spectra of the earlier pes studies were largely influenced by the extrinsic surface electronic states.@xcite these subsequent pes studies demonstrated that their valence spectra exhibit a pronounced peak originating from the ru 4@xmath3 state just below @xmath11 as predicted by dft calculations , showing that dft calculations based on an itinerant electron picture can be a good starting point to describe the electronic state of srruo@xmath2 . however , following experimental observations in pes investigations suggest that the ru 4@xmath3 electrons also have localized characteristics . the pes spectra show an incoherent feature , which is not reproduced in dft calculations . this incoherent feature is possibly originating from the electron correlation effect and has been reproduced in a recent calculation based on the combination of the dft and the dynamical mean - field theory ( dft+dmft ) , which can incorporate the electron correlation effect beyond the conventional dft.@xcite in addition , recent angle - resolved pes and optical conductivity experiments have shown that the temperature dependence of ferromagnetic exchange splitting is markedly weaker than that predicted in the stoner theory.@xcite it has been suggested in these studies that exchange splitting persists above @xmath6 but that the long - range ferromagnetic order is destroyed by the spatial and temporal fluctuations of the exchange splitting in a local region above @xmath6 . this fluctuating exchange splitting in a local region above @xmath6 is reminiscent of localized magnetic moments . therefore , it is considered that the ru 4@xmath3 electrons in srruo@xmath2 have both local and itinerant characteristics . in fact , the ru 4@xmath3 electrons in srruo@xmath2 have an instability toward electron localization due to the correlation effect . substituting mn for ru suppresses the ferromagnetic order and induces an antiferromagnetic insulating phase above @xmath1435% of mn concentration.@xcite similar insulating phases have been found for several other ru site substituted systems.@xcite the substitution of la for sr also reduces the ferromagnetic order.@xcite we have recently reported that the ferromagnetic order in sr@xmath0la@xmath1ruo@xmath2 changes into a cluster - glass state for @xmath4 @xmath15 0.3 , whereas the metallic character is retained up to at least @xmath4 = 0.5.@xcite the ac susceptibility depends markedly on the frequency near the cluster - glass ordering temperature @xmath16 , and the frequency dependence of @xmath16 follows the vogel - fulcher law . furthermore , @xmath17sr investigations for @xmath4 @xmath15 0.3 demonstrated that magnetic clusters start developing well above @xmath16 and that the magnetic ordering process of the cluster - glass state is strikingly different from that expected for a conventional long - range ferromagnetic order.@xcite note that the emergence of a cluster - glass state is an indication of the localized character of the ru 4@xmath3 states because it reflects the presence of spatial inhomogeneities in the magnetism . we have also revealed that the electronic specific - heat coefficient increases significantly with la doping , indicating that la doping may enhance the effective mass of the conduction electrons due to the electron correlation effect.@xcite in this study , we performed pes measurements using soft x - rays to study how the electronic state of sr@xmath0la@xmath1ruo@xmath2 evolves with la doping and to observe the tendencies of the electron localization indicated by the emergence of the cluster - glass state . polycrystalline samples of sr@xmath0la@xmath1ruo@xmath2 for @xmath4 @xmath18 0.5 were prepared by the conventional solid - state reaction method as reported previously.@xcite pes experiments were carried out at the soft x - ray beamline bl23su in spring-8.@xcite the base pressure in the sample chamber was maintained below 2@xmath1910@xmath20 pa during the measurements . to obtain clean surfaces , the samples were fractured @xmath12 @xmath13 just before the measurements under ultrahigh vacuum conditions . the pes spectra were measured using a vg - scienta ses2002 analyzer . the energy resolution was approximately 120 mev at a photon energy of @xmath21 = 700 ev and increased up to 300 mev at @xmath21 = 1100 ev . the position of @xmath11 was calibrated by measuring the fermi edge of an evaporated gold film . the sample temperature was controlled by a he - flow cryostat , and the pes spectra were collected over a temperature range from 25 to 200 k. measurements of the x - ray absorption spectrum ( xas ) were performed in the total - electron - yield mode . figure 1 shows the pes spectra of sr@xmath0la@xmath1ruo@xmath2 . the sample temperature was maintained at 25 k during data collection . at this temperature , the samples with @xmath4 @xmath18 0.2 were in ferromagnetic ordered states , and samples with @xmath4 @xmath15 0.3 , which exhibit cluster - glass states at low temperatures , were in paramagnetic states.@xcite however , it should be mentioned here that the valence spectra hardy depend on the temperature in the temperature and @xmath4 range investigated here , as shown later in this section . the incident photon energy was 700 ev . at this photon energy , the ru 4@xmath3 , o 2@xmath10 , and la 5@xmath3 states have relatively large pes cross - sections , whereas the contributions from other valence electrons , such as ru 5@xmath22 , sr 5@xmath22 , and la 6@xmath22 , are essentially negligible because their contributions are approximately two orders of magnitude smaller than that of ru 4@xmath3.@xcite the spectrum for @xmath4 = 0 is very similar to those of previous pes studies using soft x - rays.@xcite the peak just below @xmath11 extending to approximately @xmath23 ev originates from the ru 4@xmath3 states , and the intensity distributed from @xmath23 to @xmath24 ev is mainly attributed to the o 2@xmath10 states.@xcite the line shape of the intensity of the o 2@xmath10 states does not exhibit any noticeable change with la doping . we normalize these spectra by the spectral area from -2.5 to -10 ev , which roughly corresponds to the total intensity of the o 2@xmath10 states . although the la 5@xmath3 states , whose pes cross - sections are comparable that of o 2@xmath10 at @xmath21 = 700 ev,@xcite also contribute to this energy range with increasing @xmath4 as shown later in this section , the la 5@xmath3 contribution compared to the total intensity of the o 2@xmath10 states is less than few percent and essentially negligible by the following reason . the total number of o 2@xmath10 electrons is roughly 18/f.u . assuming that the o 2@xmath10 bands are nearly fully occupied as predicted by dft calculations,@xcite whereas the total number of la 5@xmath3 electrons is considered to be less than 0.5 at most for @xmath4 @xmath18 0.5 . the shape of the ru 4@xmath3 derived peak shows remarkable change ; the intensity just below @xmath11 is suppressed with increasing @xmath4 , and the intensity develops near @xmath25 ev ( indicated by the arrow in fig . it has generally been recognized in previous pes studies that the peak just below @xmath11 and the broad spectral weight near @xmath25 ev in srruo@xmath2 correspond to the coherent and incoherent components of the ru 4@xmath3 state , respectively.@xcite the presence of the incoherent component near @xmath25 ev is supported by a recent dft+dmft calculation for srruo@xmath2,@xcite indicating that it is originating from the electron correlation effect . in this context , the observed spectral change suggests that a spectral weight transfer of the ru 4@xmath3 electrons from the coherent to incoherent components occurs upon doping la . we also display the @xmath4 variation of the spectral area from 0.5 to -2.5 ev , which roughly corresponds to the total number of ru 4@xmath3 electrons contributing to the peak just below @xmath11 , in the inset of fig . 1 . within the experimental accuracy , the spectral area from 0.5 to -2.5 ev seems unchanged , suggesting an absence of a large change in the valence state of ru with la doping . however , we can not discuss the detailed behavior of the valence state evolution of ru by la doping due to the presence of the data scatter of about 10% . since the doped la atoms may provide additional la 5@xmath3 electrons to the valence bands , one may consider that the change of the peak near @xmath11 is caused not by the spectral weight transfer of the ru 4@xmath3 state from the coherent to incoherent components but by the development of the la 5@xmath3 component . to check this possibility , we performed la 3@xmath3-edge resonance pes experiments to identify the la 5@xmath3 contributions in the valence band spectra.@xcite in the la 3@xmath3-edge resonance pes , the following two processes occur simultaneously : @xmath26 these are known as the ( 1 ) direct and ( 2 ) auger processes . because these two processes have the same initial and final states , the la 5@xmath3 spectral weight is expected to be resonantly enhanced when the photon energy corresponds to the la 3@xmath3-edge . the inset of fig . 2 shows the la 3@xmath5 xas spectrum for @xmath4 = 0.3 . a peak is clearly seen at a photon energy of @xmath21 = 834.6 ev . the incident photon energy was chosen to this peak energy for the on - resonance measurements and @xmath21 = 830 ev for the off - resonance measurements . the on- and off - resonance spectra measured at 25 k for each sample are displayed in fig . it is clear that the intensity of the peak at around @xmath27 ev is markedly enhanced in the on - resonance spectra . we normalized these on- and off - resonance spectra by the intensity around @xmath11 and found that the on- and off - resonance spectra overlap almost perfectly around @xmath11 . this reflects an absence of the la 5@xmath3 component in this energy range , where the ru 4@xmath3 component is dominant . here , we note that this normalization method is almost equivalent to that by the incident photon intensity since the difference in the pes cross - section of ru 4@xmath3 between the on- and off - resonance spectra is only about 1%.@xcite because the la 5@xmath3 contribution is limited to the energy range between @xmath23 and @xmath28 ev for all the samples presently investigated , it is considered that the la 5@xmath3 states mainly hybridize with the o 2@xmath10 states and do not influence the change of the spectra near @xmath11 . it should be also noted here that the presence of la 5@xmath3 component contradicts with the simple ionic picture , since the la atom is generally considered to be trivalent in this picture , and thus , the la 5@xmath3 state should be empty.@xcite the presence of la 5@xmath3 state is consistent with our recent magnetic susceptibility measurements , which showed that the @xmath4 variation of the effective moment can not be explained in terms of the simple ionic picture.@xcite since sr@xmath29 is substituted by la@xmath30 , the same fraction of ru@xmath31 is expected to be substituted by ru@xmath30 through electron transfer in the simple ionic picture . the ru@xmath31 and ru@xmath30 ions have effective moments of 2.83 and 1.73@xmath7 , respectively,@xcite and therefore , the effective moment is expected to decrease monotonously with increasing @xmath4 in the simple ionic picture . however , the experimentally obtained effective moment estimated from the curie constant does not shows a monotonous decrease with @xmath4 . both the present la 3@xmath3-edge resonance pes and our recent magnetic susceptibility measurements imply that the ru 4@xmath3 state in sr@xmath0la@xmath1ruo@xmath2 can not be understood in terms of the simple ionic configurations of ru@xmath31 and ru@xmath30 possibly because of its delocalized nature . since the pes experiments were carried out in the soft x - ray range , the pes spectra mainly reflect the bulk electronic states . however , the influence from the surface region of sr@xmath0la@xmath1ruo@xmath2 may not be negligible . in fact , the electronic state in the surface region of srruo@xmath2 seems to be markedly different from that of the bulk state , since the pes spectra obtained for polycrystal samples with scraped or fractured surfaces largely depend on the photon energy , namely , on the probing depth.@xcite here , we subtract the surface components from the pes spectra according to a procedure described in ref . 14 to investigate the intrinsic bulk electronic properties . the pes spectra can be decomposed into surface and bulk components using the following equation : @xmath32 where @xmath33 and @xmath34 represent the intensities of the surface and bulk components , respectively . @xmath3 is the thickness of the surface layer , and @xmath35 is the photoelectron mean free path . since @xmath35 depends on the photon energy , we can estimate the surface and bulk components from two spectra measured at different photon energies . in this analysis , the o 2@xmath10 spectral weights at higher binding energies are fitted by multiple gaussians as shown in fig . 1 , and then we subtract them from the pes spectra to focus on the behavior of the ru 4@xmath3 derived peak . the ru 4@xmath3 derived peaks in the pes spectra measured with @xmath21 = 500 and 1100 ev at 25 k are shown in figs . 3(a ) and 3(b ) , and these spectra are normalized by the area of each spectrum . the spectra measured at @xmath21 = 500 ev are broadened using a gaussian to decrease the energy resolution to that of @xmath21 = 1100 ev so that we can compare the spectra . we estimated @xmath35 to be approximately 10 and 19 @xmath36 for @xmath21 = 500 and 1100 ev , respectively , using semiempirical expressions given in ref . the parameter @xmath3 is assumed to be same order in magnitude with lattice constants,@xcite and we estimated the surface and bulk components by tentatively using @xmath3 values of 6 and 12 @xmath36 [ figs . 3(c)-3(f ) ] . figures 3(c ) and 3(e ) clearly show that the spectral shape of the bulk component does not exhibit any noticeable change with the doubling of @xmath3 , suggesting that the bulk spectra estimated in this procedure are robust against estimation errors in @xmath3 . therefore , the estimated bulk spectra would reflect the intrinsic bulk electronic states , even though the accurate value of @xmath3 is unknown . although the incoherent component ( @xmath141.2 ev ) is stronger in the surface spectra , one can see the spectral weight transfer from the coherent to incoherent component upon la doping in the bulk spectra . the strong enhancement of the incoherent component in the surface spectra can naturally be understood by considering the fact that the surface ru 4@xmath3 state is expected to be less hybridized than the bulk ru 4@xmath3 state . this is because the surface atoms have fewer nearest neighbors . hence , this enhancement further supports that the incoherent component at @xmath141.2 ev originates from the electron correlation effect . we measured the temperature dependence of the pes spectra to study how magnetic ordering affects the valence electronic state of sr@xmath0la@xmath1ruo@xmath2 . figure 4 shows the pes spectra for each sample measured at 25 and 200 k. these spectra are normalized by the area of each spectrum . the ferromagnetic ordering temperatures for @xmath4 = 0 ( @xmath6 = 160 k ) and 0.1 ( @xmath6 = 120 k ) are between 25 and 200 k and well separated from these temperatures . although the cluster - glass ordering temperatures for @xmath4 @xmath15 0.3 are lower than 25 k , the magnetic clusters already start developing at this temperature according to our recent @xmath17sr study.@xcite the pes spectra exhibit no discernible dependence on the temperate over the investigated @xmath4 range . the present study reveals that the substitution of la for sr generates a clear weight transfer of the ru 4@xmath3 state from the coherent to incoherent component . since the incoherent feature is reproduced by a recent dft+dmft calculation for srruo@xmath2 , which can take account of the man- body effect beyond the conventional dft calculations , @xcite the development of the incoherent component seems to be caused by the enhancement of electron correlation by la doping . the enhancement of electron correlation is consistent with the emergence of cluster - glass states for @xmath4 @xmath15 0.3 because both are indicative of the localized character of the ru 4@xmath3 electrons . on the other hand , it is important to note that the large pes intensity at @xmath11 remains up to @xmath4 = 0.5 despite the development of the incoherent component . this suggests that the ru 4@xmath3 state still involves itinerant characteristics and is consistent with the metallic character of sr@xmath0la@xmath1ruo@xmath2 revealed by resistivity and specific - heat measurements.@xcite next , we argue the relationship between this variation in the pes spectra with la doping and the thermodynamic quantities . we previously demonstrated that low temperature magnetization is markedly suppressed with la doping.@xcite this indicates that exchange splitting at low temperatures is reduced with increasing @xmath4 . a dft calculation for srruo@xmath2 predicts that the reduction of the exchange splitting accompanies the enhancement of the density of states near @xmath11 and therefore makes the ru 4@xmath3 derived peak more pronounced.@xcite however , the present pes results show an opposite behavior with increasing @xmath4 . hence , it is considered that the effect of the reduction in exchange splitting on the pes spectra is overwhelmed by the suppression of the pes intensities near @xmath11 due to the renormalization factor @xmath37 = @xmath38 , where @xmath39 is the self - energy originating from the electron correlation effect . we previously reported that the electronic specific - heat coefficient @xmath40 exhibits a monotonic increase with increasing @xmath4 up to @xmath4 = 0.3 , which is accompanied by a significant suppression in the magnetic ordering temperature and the ordered moment.@xcite the increase rate in @xmath40 with la doping becomes small for @xmath4 @xmath15 0.3 , where the low - temperature ordered moment is well suppressed . combined with the features of the pes spectra , we consider that the enhancement in @xmath40 up to @xmath4 = 0.3 is caused by a combination of the suppression of exchange splitting and the development of the electron correlation effect . in contrast , the weak increase in @xmath40 for @xmath4 @xmath15 0.3 likely originates solely from the latter effect . here , we discuss the origin of the enhancement of electron correlation and the suppression of the ferromagnetic order with la doping . according to ref . 28 , the la doping increases the lattice constants and also enhances the gdfeo@xmath2 type distortion by tilting the ruo@xmath41 octahedra . the former change in the crystal structure would enhance the density of states at @xmath11 by weakening the hybridization effect . in contrast , the latter change may diminish the density of states at @xmath11 by lifting the degeneracy of the ru 4@xmath3 derived bands near @xmath11 , which have @xmath42 symmetry.@xcite at first glance , the enhancement of electron correlation with la doping revealed by this study seems to suggest that the evolution of the electronic state caused by the former change is dominant in sr@xmath0la@xmath1ruo@xmath2 . however , in this explanation , the origin of the suppression of ferromagnetic would be puzzling . this is because the presence of the ferromagnetic order is determined by the value of the product of the density of states at @xmath11 and an interaction constant in the stoner theory.@xcite therefore , in this sense , the suppression of the ferromagnetic order can be considered as an indication of the suppression of the density of states at @xmath11 and the broadening of the ru 4@xmath3 bands near @xmath11 . this is clearly inconsistent with the above explanation for the origin of the enhancement of electron correlation . further experimental and theoretical investigations are clearly needed to understand this puzzling coexistence of the suppression of the ferromagnetic order and the enhancement of electron correlation with la doping . it is remarkable that the pes spectra exhibit no significant temperature variations in contrast to their clear @xmath4 dependence . x - ray magnetic circular dichroism spectroscopy studies have revealed that orbital moments are almost quenched in the ferromagnetic state of srruo@xmath2.@xcite therefore , the ferromagnetic order originates from spin moments , and a considerable amount of exchange splitting between the majority and minority spin bands is indispensable to reproduce the ordered moment of approximately 1@xmath7.@xcite in fact , the energy scale of the exchange splitting for @xmath4 = 0 is estimated to be approximately 0.5 ev by dft calculations.@xcite this value is much larger than the energy resolution of present pes study ( 120 mev at @xmath21 = 700 ev ) . therefore , the temperature dependence of the pes spectra should be measurable , if the conventional stoner theory is applicable to the ferromagnetic order in this compound . this is because pes spectra are the simple sum of the majority and minority spin components , and the exchange splitting between these components should decrease with increasing temperature and disappear above @xmath6 in the stoner theory . actually , stoner - like temperature dependence has been reported in pes experiments for ferromagnetic gd metal,@xcite although subsequent detailed spin - resolved pes experiments revealed that the temperature dependence is much more complex than the prediction of the stoner theory.@xcite our pes data are consistent with recent angle - resolved pes and optical spectroscopy studies for @xmath4 = 0,@xcite which indicate that exchange splitting does not disappear at @xmath6 and remains finite well above @xmath6 . in these studies , it has been proposed that such a residual exchange splitting above @xmath6 can be explained based on fluctuating local band theory.@xcite in this theory , the band structure in a local region still has exchange splitting above @xmath6 ; however , the long - range magnetic order is destroyed by spatial and temporal fluctuations . the recent dft+dmft calculation also reproduced residual exchange splitting above @xmath6.@xcite note that the pes spectra of sr@xmath0la@xmath1ruo@xmath2 do not show a discernible temperature dependence for the entire investigated @xmath4 range , whereas the ferromagnetic transition temperature varies widely with @xmath4 . @xmath6 and the ordered moment in the ferromagnetic state for @xmath4 = 0.1 are approximately 120 k and 0.75@xmath7 , respectively , which are approximately 75% of those for srruo@xmath2.@xcite the present pes investigations imply that the exchange splitting for @xmath4 = 0.1 also remains nearly unchanged up to 200 k , which is well above @xmath6 . in this context , it is natural to speculate that exchange splitting for @xmath4 @xmath15 0.3 also remains finite above the cluster - glass ordering temperatures , and it is closely related to the origin of the development of magnetic clusters well above the cluster - glass ordering temperatures observed in the @xmath17sr study.@xcite even though the absence of temperate dependence in the pes spectra for @xmath4 @xmath15 0.3 seem to be consistent with this speculation , this is not conclusive because the low - temperature magnetization for @xmath4 @xmath15 0.3 is significantly smaller than that for srruo@xmath2 , and it may yield too small exchanging splitting to be detected in the present pes experiment.@xcite the electronic states of sr@xmath0la@xmath1ruo@xmath2 were studied by pes experiments using soft x - rays . the ru 4@xmath3 derived peak near @xmath11 is significantly influenced by the substitution of la for sr ; doping la into srruo@xmath2 reduces the pes intensity just below @xmath11 , accompanying the evolution of an incoherent component near @xmath25 ev . the resonant pes experiments at the la 3@xmath5 edge confirmed that the change in the spectral shape of the ru 4@xmath3 derived peak does not originate from the additional contribution of the la 5@xmath3 electrons . we decomposed the pes spectra into the bulk and surface components using the dependence of the photoelectron mean - free path on the photon energy and found that the spectral weight transfer of the ru 4@xmath3 state from the coherent to incoherent components upon la doping occurs in the bulk component . the development of the incoherent component is considered to reflect the enhancement of the electron correlation effect with la doping . this is consistent with the emergence of cluster - glass states for @xmath4 @xmath15 0.3 , which indicates the localized nature of the ru 4@xmath3 electrons . on the other hand , a large portion of the pes intensity at @xmath11 still remains even for @xmath4 = 0.5 , showing that the ru 4@xmath3 state still involves itinerant characteristics . the temperature dependence of the pes spectra is negligibly small and does not follow the prediction of the stoner theory . these results suggest that the electronic and magnetic properties of sr@xmath0la@xmath1ruo@xmath2 are influenced by the electron correlation effect and that the ru 4@xmath3 electrons have both local and itinerant natures . the pes experiments were carried out at the jaea beamline bl23su in spring-8 ( proposal no . 2015b3884 ) . this work was performed under the shared use program of the jaea facilities ( proposal no . 2015b - e25 ) with the approval of the nanotechnology platform project supported by the ministry of education , culture , sports , science , and technology ( proposal no . a-15-ae-0042 ) . a. callaghan , c. w. moeller , and r. ward , inorg . chem . * 5 * , 1572 ( 1966 ) . a. kanbayasi , j. phys . jpn . * 41 * , 1876 ( 1976 ) . p. b. allen , h. berger , o. chauvet , l. forro , t. jarlborg , a. junod , b. revaz , and g. santi , phys . b * 53 * , 4393 ( 1996 ) . o. gunnarsson , m. calandra , and j. e. han , rev . phys . * 75 * , 1085 ( 2003 ) . p. kostic , y. okada , n. c. collins , z. schlesinger , j. w. reiner , l. klein , a. kapitulnik , t. h. geballe , and m. r. beasley , phys . * 81 * , 2498 ( 1998 ) . j. s. dodge , c. p. weber , j. corson , j. orenstein , z. schlesinger , j. w. reiner , and m. r. beasley , phys . rev . lett . * 85 * , 4932 ( 2000 ) . f. fukunaga and n. tsuda , j. phys . . jpn . * 63 * , 3798 ( 1994 ) . k. fujioka , j. okamoto , t. mizokawa , a. fujimori , i. hase , m. abbate , h. j. lin , c. t. chen , y. takeda , and m. takano , phys . b * 56 * , 6380 ( 1997 ) . j. okamoto , t. mizokawa , a. fujimori , i. hase , m. nohara , h. takagi , y. takeda , and m. takano , phys . b * 60 * , 2281 ( 1999 ) . j. park , s .- oh , j .- h . park , d.m . kim , and c .- b . eom , phys . rev . b * 69 * , 085108 ( 2004 ) . w. siemons , g. koster , a. vailionis , h. yamamoto , d. h. a. blank , and m. r. beasley , phys . b * 76 * , 075126 ( 2007 ) . j. kim , j. chung , and s .- j . oh , phys . b * 71 * , 121406 ( 2005 ) . k. horiba , h. kawanaka , y. aiura , t. saitoh , c. satoh , y. kikuchi , m. yokoyama , y. nishihara , r. eguchi , y. senba , h. ohashi , y. kitajima , and s. shin , phys . b * 81 * , 245127 ( 2010 ) . m. takizawa , d. toyota , h. wadati , a. chikamatsu , h. kumigashira , a. fujimori , m. oshima , z. fang , m. lippmaa , m. kawasaki , and h. koinuma , phys . b * 72 * , 060404(r ) ( 2005 ) . h. kumigashira , m. minohara , m. takizawa , a. fujimori , d. toyota , i. ohkubo , m. oshima , m. lippmaa , and m. kawasaki , appl . phys . lett . * 92 * , 122105 ( 2008 ) . d. toyota , i. ohkubo , h. kumigashira , m. oshima , t. ohnishi , m. lippmaa , m. takizawa , a. fujimori , k. ono , m. kawasaki , and h. koinuma , appl . . lett . * 87 * , 162508 ( 2005 ) . s. grebinskij , @xmath43 . masys , s. mickevi@xmath44ius , v. lisauskas , and v. jonauskas , phys . b * 87 * , 035106 ( 2013 ) . m. kim and b. i. min , phys . rev . b * 91 * , 205116 ( 2015 ) . d. e. shai , c. adamo , d. w. shen , c. m. brooks , j. w. harter , e. j. monkman , b. burganov , d. g. schlom , and k. m. shen , phys . lett . * 110 * , 087004 ( 2013 ) . d. w. jeong , h. c. choi , c. h. kim , s. h. chang , c. h. sohn , h. j. park , t. d. kang , d .- y . cho , s. h. baek , c. b. eom , j. h. shim , j. yu , k. w. kim , s. j. moon , and t. w. noh , phys . * 110 * , 247202 ( 2013 ) . g. cao , s. chikara , x. n. lin , e. elhami , v. durairaj , and p. schlottmann , phys . b * 71 * , 035104 ( 2005 ) . r. k. sahu , z. hu , m. l. rao , s. s. manoharan , t. schmidt , b. richter , m. knupfer , m. golden , j. fink , and c. m. schneider , phys . b * 66 * , 144415 ( 2002 ) . m. yokoyama , c. satoh , a. saitoh , h. kawanaka , h. bando , k. ohoyama , and y. nishihara , j. phys . * 74 * , 1706 ( 2005 ) . s. kolesnik , b. dabrowski , and o. chmaissem , phys . b * 78 * , 214425 ( 2008 ) . k. w. kim , j. s. lee , t. w. noh , s. r. lee , and k. char , phys . b * 71 * , 125104 ( 2005 ) . a. j. williams , a. gillies , j. p. attfield , g. heymann , h. huppertz , m. j. mart@xmath45nez - lope , and j. a. alonso , phys . b * 73 * , 104409 ( 2006 ) . r. j. bouchard and j. f. weiher , j. solid state chem . * 4 * , 80 ( 1972 ) . h. nakatsugawa , e. iguchi , and y. oohara , j. phys . : condens . matter * 14 * , 415 ( 2002 ) . i. kawasaki , m. yokoyama , s. nakano , k. fujimura , n. netsu , h. kawanaka , and k. tenya , j. phys . jpn . * 83 * , 064712 ( 2014 ) . i. kawasaki , k. fujimura , i. watanabe , m. avdeev , k. tenya , and m. yokoyama , j. phys . 85 * , 054701 ( 2016 ) . y. saitoh , y. fukuda , y. takeda , h. yamagami , s. takahashi , y. asano , t. hara , k. shirasawa , m. takeuchi , t. tanaka , and h. kitamura , j. synchrotron rad . * 19 * , 388 ( 2012 ) . j. j. yeh and i. lindau , at . data nucl . data tables * 32 * , 1 ( 1985 ) . k. maiti , phys . b * 73 * , 235110 ( 2006 ) . m. y. kimura , k. fukushima , h. takeuchi , s. ikeda , h. sugiyama , y. tomida , g. kuwahara , h. fujiwara , t. kiss , a. yasui , i. kawasaki , h. yamagami , y. saitoh , t. muro , t. ebihara , and a sekiyama , j. phys . : conf . ser . * 592 * , 012003 ( 2015 ) . a. tanuma , c. j. powell , and d. r. penn , surf . sci . * 192 * , l849 ( 1987 ) . h. t. dang , j. mravlje , a. georges , and a. j. millis , phys . b * 91 * , 195149 ( 2015 ) . e. c. stoner , philos . mag . , ser . 7 * 3 * , 336 ( 1927 ) . s. agrestini , z. hu , c .- y . kuo , m. w. haverkort , k .- ko , n. hollmann , q. liu , e. pellegrin , m. valvidares , j. herrero - martin , p. gargiani , p. gegenwart , m. schneider , s. esser , a. tanaka , a. c. komarek , and l. h. tjeng , phys . b * 91 * , 075127 ( 2015 ) . j. okamoto , t. okane , y. saitoh , k. terai , s .- fujimori , y. muramatsu , k. yoshii , k. mamiya , t. koide , a. fujimori , z. fang , y. takeda , and m. takano , phys . b * 76 * , 184441 ( 2007 ) . b. kim , a. b. andrews , j. l. erskine , k. j. kim , and b. n. harmon , phys . * 68 * , 1931 ( 1992 ) . k. maiti , m. c. malagoli , a. dallmeyer , and c. carbone , phys . lett . * 88 * , 167205 ( 2002 ) . v. korenman , j. l. murray , and r. e. prange , phys . b * 16 * , 4032 ( 1977 ) .	to clarify how the electronic state of sr@xmath0la@xmath1ruo@xmath2 evolves with la doping , we conducted photoemission ( pes ) experiments using soft x - rays . the spectral shape of the ru 4@xmath3 derived peak near the fermi level changes significantly with increasing @xmath4 . this variation indicates that a spectral weight transfer from the coherent to incoherent component occurs due to an enhancement of the electron correlation effect . resonant pes experiments at the la 3@xmath5 edge have confirmed that there is no significant contribution of the la 5@xmath3 state in the energy range where the spectral weight transfer is observed . using the dependence of the photoelectron mean free path on the photon energy , we subtracted the surface components from the pes spectra and confirmed that the enhancement of the electron correlation effect with la doping is an intrinsic bulk phenomenon . on the other hand , a large portion of the coherent component remains at the fermi level up to @xmath4 = 0.5 , reflecting that the ru 4@xmath3 state still has itinerant characteristics . moreover , we found that the pes spectra hardly depend on the temperature and do not exhibit a discernible change with magnetic ordering , suggesting that the temperature variation of the exchange splitting does not follow the prediction of the stoner theory . the presently obtained experimental results indicate that the electron correlation effect plays an important role in sr@xmath0la@xmath1ruo@xmath2 and that the ru 4@xmath3 electrons possess both local and itinerant characteristics .
the studies of the quantum systems of three particles of different nature , which are performed within various theoretical approaches , remain to be actual for a long period ( see survey @xcite ) . this is related to both the presence of nontrivial effects , which appear in this simplest many - particle system , such as the famous efimov effect @xcite , and the importance of the theoretical consideration of real three - particle and three - cluster systems of different nature , e.g. , nuclei with three nucleons or with three clusters , hypothetical systems of the type of trineutrons , molecular trimers , etc . important and insufficiently studied are the fine near - threshold effects in the systems of three particles . the present work is devoted to the analysis of the general properties of these effects and the thresholds of stability for three - particle systems . the study of properties of the three - particle thresholds is executed qualitatively on the basis of the analysis of the asymptotic estimates , and also using the high - precision variational calculations with the use of optimized gaussian bases . this approach allows us to investigate , with a high controlled accuracy , even such fine effects as the efimov effect , as well as the structure functions of these near - threshold levels @xcite that are characterized by a very small energy . in the present work , we consider a system of three particles , among which two particles are identical ( we set their masses @xmath0 without loss of generality ) , and the third one can differ from them by the mass and the pairwise interaction potential . we write the hamiltonian of such a system in the case of pairwise interactions in the form ( in the system of units with @xmath1 ) @xmath2 let the intensities of two - particle interaction potentials be defined by the dimensionless interaction constants @xmath3 ( for the pair of identical particles @xmath4 ) and @xmath5 ( for the pairs of particles @xmath6 and @xmath7 ) on the given form of the interaction , where @xmath8 and @xmath9 . we consider the potential functions @xmath10 and @xmath11 mainly with positive values . negative values of @xmath3 and @xmath5 correspond to the attraction , and positive ones to the repulsion . we will study the properties of the energy thresholds of stability for three - particle systems , i.e. the regions of such values of the constants @xmath3 and @xmath5 , at which the _ n_-th energy level of the three - particle system appears below the two - particle thresholds or below zero if the two - particle subsystems are not bound , @xmath12 @xmath13 we will study the energy thresholds of stability for three - particle systems with the zero total angular momentum , @xmath14 , for various ratios of the masses of particles , the form of pairwise interaction potentials , and the symmetry of the wave function relative to the permutations of identical particles ( for the symmetric @xmath15 and antisymmetric @xmath16 states ) . on the plane ( @xmath17 ) , we will present the results of studies in the form of the diagrams of the thresholds of energy levels , namely the thresholds of stability of the energy levels of three particles with the zero angular momentum ( fig.[fig1 ] ) . the systematization of the significant amount of calculations and the qualitative analysis allow us to represent the thresholds of stability in the universal form of the diagrams of thresholds . in fig.[fig1 ] , we give , for the sake of specificity , the example of the diagrams of stability corresponding mainly to potentials @xmath10 and @xmath18 which have the same gaussian form and correspond to almost equal masses . we emphasize that the characteristic properties of a diagram are preserved for nonsingular short - range potentials of more general forms . in fig.[fig1 ] , we show the main peculiarities of such diagrams schematically without holding the scale , but with the preservation of all main regularities . each line of the threshold of the @xmath19-th level ( for the states @xmath20 symmetric relative to the permutation of the identical first and second particles , and for the antisymmetric states @xmath21 ) separates the region of the existence of a bound state ( the region of stability ) with the corresponding @xmath19-th level of three particles from the region , where this level does not exist ( this side of the curve is shaded ) . because the bound states of three particles can appear only under the attraction between different particles ( the potential @xmath22 with the negative constant @xmath23 while the interaction between identical particles ( the potential @xmath24 ) can be both attractive ( @xmath25 ) and repulsive ( @xmath26 ) , the three - particle thresholds are positioned to the left from the axis of ordinates . on the axes , we marked the points where two - particle ground bound states appear : the @xmath27-state with the orbital moment @xmath28 ( @xmath29 , @xmath30 ) and the @xmath31-state with the orbital moment @xmath32 ( @xmath33 , @xmath34 ) . we also show the corresponding two - particle thresholds on the axes by lines with different thicknesses . the inclined dashed lines separate the regions , where the lowest threshold from two two - particle ones is the threshold for two identical particles ( due to the potential @xmath35 , below the dashed line ) or for two different particles ( due to the potential @xmath36 , above the dashed line ) . the three - particle levels exist below the lowest two - particle threshold . the dash - dotted vertical lines in the lower part of fig.[fig1 ] are the asymptotes of three - particle thresholds as @xmath37 . in this case , the lower part of the figure is correlated with the rest parts so as it follows from direct calculations . in the diagram of the thresholds of stability , the bound states of the system of three particles exist , by starting from the lines of thresholds , as a rule , towards the increase in the intensities of the attraction ( i.e. to the left and down on the diagram ) . on the energy diagrams of stability , we can distinguish eight different regions by characteristic peculiarities of the ground and excited three - particle levels : region i the asymptotic region , where @xmath38 , region ii the asymptotes of thresholds as @xmath39 , @xmath40 , region iii the region of the infinite series of efimov levels near the values of the two - particle interaction constants critical as for the appearance of the bound @xmath27 states @xmath41 , @xmath42 ; on the diagram of thresholds , the nonmonotonous and closed curves correspond to the efimov states ( they are marked by the words `` efimov s effect '' ) , region iv the axis @xmath5 in the case of the absence of the interaction between identical particles @xmath43 , when @xmath44 and the conditions of the thomas theorem are satisfied , region v the region where the effect of a `` tube '' is manifested for @xmath45 and arbitrary positive @xmath3 , region vi we indicate the characteristic behavior of thresholds for the symmetric three - particle states in the region @xmath46 with the appearance of an acute `` wedge '' on the line of equality of the threshold binding energies of different pairs of particles , @xmath47 , region vii the region of a `` rearrangement '' of the thresholds of energy levels , region viii the region of the nonmonotonicity of the curves for the thresholds of antisymmetric levels . the characteristic peculiarities of the behavior of the energy thresholds of stability of the three - particle system in different regions on the plane @xmath48 , the general established phenomenon of the nonmonotonicity of thresholds , and the effect of `` traps '' are studied analytically and with the use of numerical calculations within the galerkin variation method with a gaussian basis and various high - precision optimization schemes . consider region i for the ground and excited symmetric states on the diagram of thresholds , where @xmath49 and simultaneously @xmath50 ( the left upper part of the diagram of thresholds in fig.[fig1 ] ) . we rewrite hamiltonian ( [ e1 ] ) in the center - of - mass system as @xmath51 where @xmath52 , @xmath53 jacobi relative coordinates . let a short - range repulsive potential @xmath35 have a maximum at zero and monotonically decrease , and let the potential @xmath36 have a minimum value @xmath54 ( @xmath55 the attraction , and this occurs not obligatorily at @xmath56 ) . in the limit of strong coupling , the main contribution to the ground - state energy is determined by the minimum of the full three - particle potential well , which is positioned at @xmath57 and at some @xmath58 , minimizing the effective total potential energy @xmath59 . for nonsingular short - range potentials @xmath60 and @xmath61 , the threshold line is determined by the formula @xmath62 which determines the value of @xmath58 , where the minimum of the effective attractive potential energy of three particles @xmath63 is attained , and establishes the connection between the constants @xmath3 and @xmath64 in the considered region of the diagram of thresholds . in the simple case of the interaction potentials @xmath60 and @xmath61 in the gaussian form with unit radii , which is thoroughly studied numerically , the condition for the threshold has the form @xmath65 from the condition of the minimum of the left - hand side of ( [ e5 ] ) , we get @xmath66 and , with regard for ( [ e5 ] ) , @xmath67 then , in the limit @xmath68 , the asymptotic formula with regard for the next correction for the lines of the thresholds of the ground and excited levels has finally the following form : @xmath69 here , @xmath70 we note that the quantum numbers @xmath71 in ( [ e8 ] ) and ( [ e9 ] ) correspond to the ground state . the series of states in @xmath31 is due to one - dimensional small oscillations along the coordinate @xmath72 , whereas the series of states in @xmath73 corresponds to three - dimensional oscillations along the coordinate @xmath74 near the minimum of the potential well in the three - particle system . indeed , if we expand the potential energy ( in the case of the gaussian potentials @xmath60 and @xmath61 ) in hamiltonian ( [ e3 ] ) averaged over angles near the minimum in @xmath75 and in the squared deviation @xmath76 , we get , instead of ( [ e3 ] ) , the approximate oscillatory hamiltonian as a function of both coordinates @xmath77 the energy states of the oscillatory hamiltonian ( [ e10 ] ) generate the lines of thresholds ( [ e8 ] ) and ( [ e9 ] ) . the motion of the system of three particles in symmetric states in approximation ( [ e10 ] ) is oscillations relative to the following configuration . at the center of the system , a particle with mass @xmath78 is located . on the same diameter with it , particles 1 and 2 ( @xmath79 ) are positioned , but the wave function is spherically symmetric in angles ( the sphere diameter @xmath80 ) . the distance between particles 1 and 2 does not depend in the main approximation on the mass of the third particle . that is , such a configuration of the system ( the third particle is at the center between the two identical particles ) occurs even in the case where the mass @xmath78 of the third particle is small . the more detailed consideration of the threshold in region i shows that asymptote ( [ e8 ] ) for the ground state is defined as @xmath81 and the coefficient @xmath82 is positive at very small masses . moreover , @xmath83 as @xmath84 , is negative at greater masses and reaches a minimum value at @xmath85 ( @xmath86 ) . at @xmath87 , it approaches the constant , @xmath88 . thus , the linear asymptote of the threshold for the ground state in region i of the threshold diagram in fig.[fig1 ] is reached from the top for very small masses ( @xmath89 ) and is reached from the bottom for the greater mass of the third particle . we also note that , in region i of the diagram of thresholds , the result which is asymptotic in the coupling constant , @xmath90 , is easily generalized to other potentials . in particular , let the interaction potential @xmath61 between different particles be chosen in the gaussian form with another interaction radius @xmath91 . then we get the same configuration of three particles ( the third particle is located between two identical ones ) . moreover , instead of ( [ e7 ] ) , we obtain ( for @xmath92 , when @xmath93 ) @xmath94 in the main approximation , we get @xmath95 where @xmath96 the least value of the coefficient , @xmath97 , is reached in ( [ e14 ] ) at @xmath98 , when @xmath99 and all three particles approach one another at small distances . as @xmath100 , the coefficient @xmath101 grows indefinitely . for @xmath102 , all three particles are , all the more , at small distances , where the small oscillations of the particles relative to the equilibrium position occur . the asymptotics of a threshold remains invariable : @xmath103 . hence , in the region of the diagram of thresholds where @xmath104 and @xmath105 , the three - particle system can possess different configurations even in the case of the simplest interactions . qualitatively , analogous results are obtained for a wider class of the pairs of potentials @xmath60 and @xmath61 . for example , for potentials in the form of exponentials , @xmath106 and @xmath107 , we have , for @xmath108 , @xmath109 in this case , the least value of the coefficient @xmath110 is also attained at @xmath111 , when @xmath112 , and both @xmath58 and @xmath113 grow with increase in @xmath114 . moreover , for an arbitrary pair of the monotonous repulsive potential @xmath24 between the identical particles and the attractive potential @xmath22 between different particles in region i , a linear dependence between the intensities of the potentials takes place , which separates the region of stability of three particles from the region , where the coupling is absent . if the repulsive potential @xmath60 is nonmonotonous and has a sufficient decrease at zero , and if the attractive potential @xmath61 has a minimum at finite distances @xmath115 , a two - cluster configuration of the system of three particles is also possible : in this case , the identical particles are positioned near each other , and the third one is located at a distance @xmath116 from them . all the main qualitative and analytic results concerning the asymptotics of the thresholds of stability in region i are confirmed by the high - precision systematic calculations with two - particle potentials of the gaussian form for a great variety of masses within the variational approach with the use of gaussian bases . the separate calculations were executed for other potentials , and they also support the general schematic fig . [ fig1 ] . we now consider the asymptotic limit of the strong coupling between the identical first and second particles ( region ii , where @xmath117 in the lower part of the diagram of thresholds ( fig.[fig1 ] ) ) . in this limit , the size of subsystem @xmath4 tends to zero , and the variables in ( [ e1 ] ) are separated in the cluster approach . then the three - particle wave functions for hamiltonian ( [ e3 ] ) can be chosen as @xmath118 where @xmath119 is the ground state oscillatory wave function of the relative coordinate @xmath120 of pair @xmath4 , and the wave function of the third particle , being in the effective field of two other particles , depends on the coordinate @xmath121 and is the eigenfunction of the effective hamiltonian reckoned from the two - particle threshold , @xmath122 with the reduced mass @xmath123 and the effective averaged potential @xmath124 consider the motion of the third particle relative to the pair of particles @xmath4 in potential ( [ e19 ] ) which looks , in particular for the gaussian form , as @xmath125 then , for the critical constants @xmath126 , at which the @xmath19-th three - particle near - threshold level in the three - particle system appears , we get the asymptotics as @xmath127 : @xmath128 here , @xmath129 are the critical constants of the appearance of the @xmath19-th level in the system of two particles with the zero orbital moment @xmath28 ( for the gaussian potential with unit radius in the case of the unit masses of particles , the critical constants of the @xmath27-states are as follows : @xmath130 ; @xmath131 ; @xmath132 ; @xmath133 , etc . ) . the expressions analogous to ( [ e21 ] ) can be also obtained in the case where the subsystem of two particles in the state with the wave function ( [ e17 ] ) has a nonzero orbital moment . it follows from ( [ e21 ] ) that the threshold of each three - particle level ( in the states symmetric in the permutations of the pair of particles @xmath4 ) is positioned always to the left from the corresponding vertical asymptote for @xmath134 ( they are marked by dash - dotted vertical lines in the lower part of fig.[fig1 ] ) . we can show analogously that , for the states antisymmetric relative to the permutations of the pair of particles @xmath43 , the thresholds approach , on the contrary , the corresponding vertical asymptotes from the right side . the high - precision calculations for different masses and potentials in the form of a superposition of gaussian functions confirm completely the above - presented conclusions about the asymptotics of thresholds for different levels as @xmath127 . consider region iii in the vicinity of the critical constants of the appearance of the two - particle ground bound @xmath27-states : @xmath135 and @xmath136 in fig.[fig1 ] . this is the region where the efimov effect @xcite is manifested , and the infinite series of efimov weakly bound levels for the system of three ( generally saying , different ) particles is observed . in the diagram of thresholds , it is seen as the infinite collection of closed self - similar ( with the universal scale for highly excited states ) convex curves are accumulated at the critical point ( @xmath137 ) for three pairs of particles interacting in a resonance way . the lines of thresholds for the series of efimov energy levels are considerably stretched along the dashed inclined line ( which starts from the region of the efimov effect ) , where the two - particle thresholds of different pairs of particles in the @xmath27-state coincide , and are somewhat elongated vertically upward along the @xmath3 axis , when only two pairs of particles interact in a resonance way . the fundamental efimov effect is described in detail by v. efimov himself and many other researchers . here , we should like to emphasize only separate points . firstly , the infinite series of symmetric three - particle efimov levels with the zero angular momentum is realized only in the limit @xmath138 , @xmath139 , i.e. in the case where all three pairs of particles in singlet states are at a resonance . outside this region , the number of levels is always bounded , though these levels have the certain specific properties of properly efimov levels especially in the case of two resonating pairs of particles @xmath6 and @xmath7 ( @xmath140 ) even under a sufficiently large decrease in the attraction between the particles of the third pair , @xmath4 . secondly , by the example of the series of efimov levels , we have the ideal demonstration of the energy `` traps '' ( see section 5 for more details ) , when the strengthening of the attraction on the diagram of thresholds leads to a nonmonotonous variation of the number of energy levels : the three - particle levels at first separate from the two - particle threshold and then disappear on it . the efimov levels exist only in a certain interval of the coupling constants of all three pairs of particles interacting in a resonance way . thirdly , for three - particle states antisymmetric in the pair of the identical particles @xmath4 , the singlet point ( @xmath141 ) is nonsingular ( see fig.[fig1 ] , region viii ) . at the same time , the limiting region near the point ( @xmath142 ) , where one pair @xmath4 of the identical particles resonates in the state with the angular momentum @xmath143 , is , as if , `` attractive '' for the corresponding states . in this region of fig.[fig1 ] , there exists a certain anomaly : the curves for the thresholds of antisymmetric states are significantly stretched towards small coupling constants along the line where the two - particle thresholds of two different pairs of particles in the singlet and triplet states coincide . finally , we note that we do not discuss the three - particle antisymmetric states with nonzero angular momentum , where there exists the possibility for a collapse to arise @xcite , which requires a separate consideration . consider the case where the interaction of one pair of particles , @xmath4 is absent ( when @xmath44 ) . this corresponds to region iv on the @xmath64 axis in fig.[fig1 ] , where the conditions of the thomas theorem @xcite are satisfied . then , there always exists the bound ground state of three particles in a wider region of coupling constants @xmath64 of two pairs of particles as compared with an isolated pair of two particles . this means that the three - particle threshold of the ground state for @xmath144 is always positioned to the right from the two - particle critical point in the constant @xmath64 . first of all , we note that the important common point for the thomas and efimov effects is the presence of pairs of particles interacting in a resonance way . moreover , the regions of these effects are positioned near each other in the diagram of thresholds ( fig . [ fig1 ] ) . at the same time , the thomas effect is referred only to the ground state of three particles , whereas the efimov effect concerns the infinite series of near - threshold weakly bound states . secondly , the universality of the thomas theorem consists in that it holds for hamiltonian ( [ e3 ] ) with @xmath44 for an arbitrary attractive potential @xmath36 between two pairs of particles and an arbitrary finite mass @xmath78 of the third particle . moreover , the less the mass @xmath78 , the wider is the region of the manifestation of the thomas effect relative to the interaction constant . we note that , for small masses , the number of three - particle levels becomes great , when the two - particle subsystems are not bound yet . thirdly , there occurs the effect of a `` trap '' for the first excited state at @xmath44 for particles with close masses ( under certain conditions , this concerns higher excited states too ) , as distinct from the ground state . in this case , the existence of an excited level depends nonmonotonically on strengthening the attraction by variation of @xmath145 . in addition , the intersection points of the lines of the thresholds of excited levels with the abscissa axis ( fig.[fig1 ] ) from the side of great values of @xmath145 are located to the right from the corresponding two - particle critical values of the coupling constants , where the binding energies of two particles are equal to zero . we emphasize once more that the schematic figure 1 represents all main regularities so as they follow from the asymptotic estimates and the high - precision calculations . a special attention should be paid to the effect of a `` tube '' for symmetric states at the diagram of thresholds ( region v in the upper part of fig.[fig1 ] ) , where the bounded region near @xmath146 contains at least one bound state of three particles even under the unlimited increase of the repulsion between the pair of identical particles ( @xmath3 is arbitrary ) . the effect of a `` tube '' is seen more clearly in fig.[fig2 ] , where , on the real scale , we show the results of high - precision calculations for the threshold of the ground state and the isolines of the energy of three particles , which is reckoned from the two - particle threshold , for the gaussian potentials with unit interaction radii for the mass @xmath147 . the form of the energy surface in the region @xmath148 and near @xmath149 , indeed , reminds of a `` tube '' positioned vertically . it is seen that the isoenergetic lines @xmath150 for the ground state of three particles with small values of @xmath151 are pulled into the `` tube '' in the upper part of the figure for @xmath152 and great repulsive constants @xmath3 . the isolines do not form a `` tube '' already at @xmath153 and greater values , but reveal the clear asymptotic behavior characteristic of the isolines with great values of @xmath154 , showing a sharp change of the modes near the abscissa axis ( on the given scale ) . it is important that , on the increase of the attraction between particles @xmath6 and @xmath7 ( the motion along the horizontal axis in fig.[fig2 ] ) under a significant repulsion between the pair of particles @xmath4 , the nonmonotonicity in the number of bound states is always observed ( we call this as the effect of a `` trap '' related to the presence of the `` tube '' ) . the phenomenon of a `` tube '' can be explained on a qualitative level . we will show firstly that the number of levels in a `` tube '' is always bounded , but it grows infinitely as @xmath84 . we use the fact @xcite that the pair correlation functions @xmath155 in efimov states @xmath156 for @xmath157 for the adjacent levels satisfy the relation @xmath158 where @xmath159 is the ratio of the energies of adjacent efimov levels in the limit @xmath160 ( @xmath161 in the case of three identical particles , and the danilov minlos faddeev efimov constant @xmath162 determines the index of singularity of the limiting equation of skornyakov ter - martirosyan in the case of the zero interaction radius ) . relation ( [ e22 ] ) takes place due to the fact that , as @xmath160 , the spatial region , where the pair correlation function behaves itself as @xmath163 @xcite , is spreading . it is true from distances of the order of the radius of forces to distances of the order of the size of the system @xmath164 ( here , @xmath165 is the characteristic size of the system in the ground state , @xmath19 is the level number , and @xmath166 is the ratio of sizes of the system for adjacent efimov levels ) . therefore , the pair correlation function for sufficiently great @xmath19 has the following normalizing factor of @xmath167 : @xmath168 at small distances , the pair correlation function is of the same order as that at @xmath169 , therefore we obtain relation ( [ e22 ] ) . in order to approach the region of the `` tube '' from the region , where the efimov effect manifests itself , we consider some additional short - range repulsion @xmath170 between the pair of particles @xmath4 which is taken as a perturbation of the zero hamiltonian , for which the efimov effect is observed . the greater the number of a level , the greater is the size of the three - particle system , which reminds of a spherical concentric `` halo '' @xcite with the ratio of the radii of the adjacent spheres @xmath171 . the greater the size of a system of particles with short - range interaction , the better are satisfied the conditions for the `` gas '' approach @xcite taking into account a contribution of the short - range repulsion @xmath172 to the energy . in the principal approximation in the `` gas '' parameter , a shift of the @xmath19-th level is determined by the two - particle @xmath173-matrix : @xmath174 here , @xmath175 is the two - particle @xmath173-matrix defined for the potential @xmath170 , @xmath176 is the scattering length by the potential @xmath170 ( for the repulsive potentials , @xmath177 ) . these quantities are finite for short - range repulsive potentials of the given form in the limit of the infinitely great repulsion , @xmath104 . in this case , the effective radius and other low - energy parameters which are present in the next terms of the expansion of the energy in the `` gas '' parameter are also finite . these conclusions allow us to consider the most wide region in the constant @xmath3 in the diagram of thresholds from the region of the efimov effect to the region of the `` tube '' with @xmath104 , if the summary potential @xmath178 is replaced by @xmath179 . hence , if the intensity of the repulsive potential increases , @xmath104 , the energy shift @xmath180 obtained due to the repulsion has a finite limit . moreover , the greater the number @xmath19 of an excited level , the more exact is the relation @xmath181 . let @xmath182 be the quantity of the order of the binding energy of the ground state of three particles for the critical two - particle constant . with regard for the energy @xmath183 of the @xmath19-th level of the `` zero '' hamiltonian near the critical constants ( near the efimov region ) , the total energy of the @xmath19-th level in the region of the `` tube '' looks as @xmath184 and @xmath185 . since @xmath186 ( for @xmath187 , the energy @xmath188 becomes positive with increase in @xmath19 for an arbitrary small repulsion which is characterized by the scattering length @xmath176 . this clarifies the above - mentioned conclusion about the finiteness of the number of levels under a deviation from the accumulation point of efimov levels by the constant @xmath3 . that is , for the infinite number of levels to exist , it is necessary that the interaction of all three pairs of particles be resonant , @xmath189 and @xmath190 simultaneously . all the more , the number of levels becomes finite , when the repulsion @xmath191 reaches a significant intensity ( in this case , the @xmath173-matrix is bounded , and the length @xmath176 tends to a finite limit of the order of the radius of forces ) . at the same time , the arbitrariness of the repulsion intensity for the pair @xmath4 , the finiteness of the @xmath192-matrix , and the conditions for the resonance and hence the appearance of the effective long - range interaction for two pairs of particles do not contradict the possibility for the bound states of three particles to exist . the presence of at least one bound state of three particles in the region under study is confirmed by the numerical calculations for different forms of potentials and for different masses . we note that the `` tube '' ( see figs.[fig1 ] and [ fig2 ] ) is asymmetrically positioned relative to the vertical asymptote @xmath193 ( the resonance region for two pairs of particles ) . to a great degree , it is positioned to the left from the asymptote , where the attractive constant @xmath64 is greater by modulus . but , with the further increase in the attractive constant @xmath194 by modulus , we come away from the resonance region . and at sufficiently large repulsive constants @xmath3 , the three - particle level can disappear , arising again only in the limit of a strong coupling ( see figs . [ fig1 ] and [ fig2 ] ) . we emphasize once more that all main qualitative conclusions of this section are confirmed by the numerical calculations . the interesting universal phenomenon of a `` wedge '' ( region vi in fig.[fig1 ] and , respectively , in figs . [ fig3 ] and [ fig4 ] ) is observed on the line where the binding energies of two different pairs of particles are equal , @xmath195 ( on the dashed inclined lines ) . for the sake of specificity , we consider fig.[fig3 ] . as above , we have correctly preserved all the regularities , which are obtained in the calculations , though the scale of the figure is rather arbitrary . the effect of a `` wedge '' consists in the following . if we move along the dashed line ( from the left to the right ) from the region of a strong coupling , @xmath196 , to the side where the attraction decreases , then the threshold line of the @xmath19-th symmetric state of three particles ( in fig.[fig3 ] , they are the lines of the 6-th and 5-th excited states ) is finished by a sharp cusp in the form of a `` wedge '' oriented from the left to the right . the lines of thresholds approach the dashed inclined line from the top and from the bottom with different slopes , and the derivative of the threshold line undergoes a break here . the `` wedge '' arises not due to the nonanalyticity , but as a consequence of the competition of thresholds . on the further motion to the right ( within fig.[fig3 ] ) , firstly the 6-th excited state and then the 5-th one stop to exist . but with the decrease of the attraction along the dashed inclined line , where the energies of the two - particle thresholds are equal , the 5-th excited state appears again , while we approach the efimov region . its threshold line has the form of a `` wedge '' oriented from the right to the left . for higher excited states , the effect of a `` wedge '' holds analogously . inside the closed region of the thresholds of efimov states , the bulbous curves of thresholds have a cusp in the form of a `` wedge '' on the dashed inclined line . we note that , for antisymmetric states , no similar regularities are observed . firstly , we consider the case of great values of the mass @xmath78 of the third particle . if the mass is infinite ( the three - particle model passes into a model with two particles in the short - range field of a fixed center ) , then it is obvious from ( [ e1 ] ) that , in the absence of the potential @xmath197 , the problem of three particles is equivalent to that of two independent particles in the field of the attractive center @xmath198 . the ground state of such a system arises under the same conditions as those for each of the particles in the field of a center : @xmath199 in the general case of arbitrary masses , @xmath200 ) , and , for the gaussian potential with unit radius , we get @xmath201 . we emphasize that , for the infinite mass of the third particle in the case where @xmath44 , the thomas theorem becomes trivial , and the three - particle state exists , by beginning exactly from the two - particle critical constant . for the excited states in the case of the infinite mass @xmath202 , the couplings for three and two particles also appear at the same points on the @xmath64 axis . in addition , the energies monotonically increase with @xmath203 . generally , for @xmath202 , the thresholds of stability are transformed only slightly on the whole plane @xmath48 as compared with the case where @xmath204 , though , in the region of the `` tube '' , there remains only the ground state which differs cardinally from the excited states for this reason . as @xmath202 , the whole `` tube '' is positioned to the left from the asymptote @xmath205 . moreover , the right edge of the threshold of a three - particle state crosses the abscissa axis at the resonance point of two pairs of particles @xmath206 and rises sharply ( almost vertically ) upward . in the resonance region of all three pairs of particles ( @xmath207 ) , the conditions for the appearance of the infinite efimov spectrum , which possesses the universal properties by beginning already from the second excited level , are realized in the standard way . in other regions in the diagram of the thresholds of stability in the case @xmath202 , the general peculiarities seen in fig.[fig1 ] are preserved as well . we now consider the other limiting case of very small mass @xmath208 ( a two - center model of three particles ) . in the region where @xmath209 , heavy particles @xmath4 are not coupled in the absence of the third light particle , and @xmath210 , as well as in the case where there exists the coupling in the pair of heavy and light particles ( pair @xmath6 or @xmath7 ) , we can adiabatically separate the variables . if we make averaging over the fast movement of the light particle ( in the state @xmath211 ) , then we get the equation for the wave function of the coordinate of the relative motion of particles @xmath4 @xmath212 ( in the case of the gaussian potentials with unit radii ) with the additional potential @xmath213 @xmath214 where @xmath215 , and all quantities of the dimension of energy are multiplied by @xmath78 . then , as @xmath84 and for small deviations from the position of equilibrium , we have the oscillatory potential ( @xmath216 ) . since the oscillatory frequency @xmath217 , the spectrum becomes denser for smaller masses . hence , for constants @xmath218 and small masses ( @xmath84 ) , we have the growing number of bound levels of the equidistant spectrum . respectively , the thresholds for small masses will represent the equidistant spectrum in this region . in the diagrams of the thresholds of stability ( see fig.[fig3 ] ) , the indicated regularities are manifested , as @xmath219 , in the enlargement of `` islands '' ( `` traps '' ) of the infinite series of efimov levels , the separation ( due to the merging with the earlier isolated thresholds in the places of a `` wedge '' ) of the increasing number of the new lines of thresholds from the islands , the `` pulling '' of them into the region of the `` tube '' , and the gradual filling of the whole region @xmath220 , @xmath221 . it is easy to trace the change of the general pattern in fig.[fig3 ] with decrease in the mass of the third particle or , conversely , an increase in this mass . we can qualitatively explain the growth of the number of thresholds in the `` tube '' by the following reasoning . because @xmath159 depends on the mass of particles @xcite and tends to 1 with decrease in the mass of the third particle , energy ( [ e24 ] ) depends more and more weaker on the level number @xmath19 . therefore , the number of levels , for which energy ( [ e24 ] ) remains negative , increases . that is , the less the mass , the greater is the number of thresholds which correspond to excited levels and can be `` pulled '' into the `` tube '' . this is completely confirmed by the high - precision calculations for small masses . in particular , the first excited level begins to move into the `` tube '' on the transition from great masses to @xmath222 ( fig.[fig1 ] ) . in the limit @xmath84 , the number of levels in the `` tube '' tends to infinity , but , for any finite mass , the number of levels is finite ( fig.[fig3 ] ) . in the general aspect , the less the mass of the third particle , the more the symmetric states become similar to the first excited one ( for @xmath223 ) . this implies that , in a certain sense , the effects of efimov and thomas and the effect of a `` tube '' have many common features . on the contrary , we may expect that , in the case of the model of two particles in the field of a fixed center and under a variation of the forms of interaction potentials , all three effects can be significantly impoverished , so that they will have few common features . the nonmonotonous change of the number of levels of the three - particle system with increase in the interaction constants in the region of attraction , where the levels appear and then disappear on a two - particle threshold with increase in the attraction ( the effect of a `` trap '' ) is characteristic ( see the diagrams of thresholds in figs.[fig1 ] and [ fig3 ] ) of a wide region of the interaction constants which can be positioned even sufficiently far from the efimov region of the resonance interaction . if the general constant of attraction in the three - particle system grows so that @xmath3 and @xmath5 are linearly related to each other , we can easily reveal a nonmonotonous character of the behavior ( appearance and disappearance ) of levels . that is , the effect of a `` trap '' has a sufficiently universal character . in fig.[fig4 ] , the lines with arrows show some directions of the coordinated increase in the intensities of the interaction potentials , where the three - particle levels appear and then disappear with increase in the interaction constants ( the continuous and dotted lines correspond , respectively , to the absence and presence of the given bound state ) . in this figure , we present ( schematically ) the results of high - precision calculations with the gaussian potentials of unit radii and @xmath224 ( but , for the convenience of a perception , we choose the scale to be arbitrary ) . the figure demonstrates the presence of a `` trap '' even in a very narrow interval of the constant @xmath64 near a critical value of the resonance interaction constant @xmath225 , though this region is small . for the sake of convenience , we schematically present ( to the right in fig.[fig4 ] ) the dependence of the binding energies on the coupling constants with strengthening the attraction along the mentioned directions . in this case , we see the appearance of `` traps '' for the energy levels in both the ground and excited states near two - particle thresholds . in the general aspect , the effect of a `` trap '' for three - particle levels , which demonstrates the essentially different mode of the dependence of the ground - state energy of two particles and that of the three - particle levels on the interaction constants , is a consequence of the two - particle structure of the full interaction potential in the three - particle system . quite nontrivial is the possibility to vary the modes of behavior of the thresholds ( and of the energy levels ) with the help of the interaction potentials which involve at least two modes of attraction with essentially different radii . in this case , we can nontrivially realize a certain generalization of the well - known zeldovich two - particle effect of a `` rearrangement '' of the energy spectra in the system of three particles . for the pairwise interaction potentials with the components of different radii , for example , of the type @xmath226 ( where @xmath227 and @xmath228 ) , one observes a `` rearrangement '' ( fig.[fig5 ] ) of the energy spectrum in the three - particle system ( a three - particle analog of the zeldovich effect ) . due to a great value of the ratio of the interaction radii of the two - component attractive potential ( @xmath229 ) , the spectrum of three particles is close to the superposition of the spectra of two problems involving three particles with the separately taken components of the interaction potential , and the quasidegeneration of the levels occurs . but , at the points of the expected intersection of levels of these two spectra , we observe the effects related to the `` rearrangement '' of levels . for example , the first excited level replaces the ground - state level with increase in the total attraction ( with increase in @xmath3 ) . in its turn , the ground - state level sharply changes the mode of the dependence on the interaction constant . in fig.[fig5 ] , on the real scale , we present the results of calculations of a three - particle completely symmetric energy spectrum ( the binding energies ) as a function of the interaction constant @xmath3 for a specific example of potential ( [ e28 ] ) ( @xmath230 @xmath231 @xmath232 ) . at the same interaction constant @xmath233 , the `` rearrangement '' of the spectrum of four levels occurs successively : @xmath19-th level replaces @xmath234-th level . the ground state level is the only one which , after the `` rearrangement '' , corresponds to the binding energy with a quite different mode of the dependence on @xmath3 . the effect of a `` rearrangement '' of the three - particle spectrum is much more pronounced than that for the two - particle threshold and occurs for a less attraction . by changing the class of the corresponding interaction potentials and their parameters , we can vary the number of levels in the region of a `` rearrangement '' and can also create a `` trap '' for other levels ( see results in @xcite for nonlocal separable potentials ) . the close conditions for a realization of the effects of `` rearrangement '' and `` traps '' for the energy states are known in solid state physics for a long time in the case where , for example , the conditions for a `` quasidegeneration '' of the phonon and optical branches of oscillations are realized , or the crystals with admixtures are considered . even for simple potentials in the gaussian form , the performed high - precision studies of the energy thresholds for three particles reveal one more effect of a sharp change of the modes in the diagrams of the thresholds of stability which can be also named the effect of a `` rearrangement '' . this effect has no analog on the two - particle level . there exist the regions of parameters ( in the @xmath64 axis , these are the regions close to the vertical asymptotes in fig.[fig3 ] ) , where the `` rearrangement '' of the adjacent energy thresholds occurs in the digram of thresholds . that is , on the monotonous change of the interaction intensities , the threshold of the @xmath235-th level sharply approaches that of the @xmath19-th one and occupies its place . this effect is especially clearly manifested in the case of small masses of the third particle ( see fig.[fig6 ] , where the lower part of the diagram of the thresholds of three first excited levels calculated for the mass @xmath147 is shown ) . we note that , in fig.[fig3 ] , this effect of the rearrangement of thresholds is not marked in order to avoid the complication of a perception of the general scheme of three - particle thresholds . the presence of the effect of a `` rearrangement '' is probably related to the following . in addition to the considerable short - range attractive interaction between particles @xmath4 present in hamiltonian ( [ e1 ] ) , there appears an additional effective oscillatory interaction having a great radius at small masses of the third particle . this additional interaction between particles @xmath43 is generated by the motion of the third light particle . then , as @xmath84 , the radius of the oscillatory well grows . as known@xcite , in the presence of two potentials with essentially different radii , there occurs the `` rearrangement '' of the energy spectrum on increasing the intensities of the potentials . in our case where the intensity of the short - range attraction becomes close to the critical constant of the appearance of the @xmath19-th level in this well , the lowest level of the other well falls into the well with less radius . on its previous place , the next level falls , etc . in this case , the spectrum for the well with a greater radius is practically renewed and is preserved , until the increase of the intensity of the short - range attraction induces the appearance of a new , @xmath236-th level , and the `` rearrangement '' repeats . as seen from fig.[fig6 ] , this region of the rearrangement of thresholds contains also `` traps '' , because the lines of thresholds depending on @xmath3 are not monotonous . finally , we note that , for the three - particle problem , the nonmonotonicity of the thresholds of stability is a general regularity , which is rather a rule than the exception . one can assume that the nonmonotonicity of thresholds in a three - particle system is a consequence of the difference of the modes of dependence of the ground - state energy of two particles and three - particle energies for the chosen , even simple , interaction potentials . indeed , for simple attractive potentials in a two - particle system , one mode is always realized ; we will say that this is one configuration . at the same time , for the system of three particles with pairwise interaction potentials and greater number of coordinates , various spatial configurations of three particles can be realized . this can induce a change of the modes of the dependence of three - particle energies and thresholds and , therefore , the nonmonotonous behavior of the number of levels of three particles in various regions of parameters of the interaction potentials . we may expect a more complicated behavior of the thresholds of stability in the problems with four and greater number of particles with pairwise interaction potentials . in conclusion , we emphasize that we have systematically studied the energy thresholds of stability for the general quantum problem of three particles interacting by short - range potentials . analytically and by the high - precision calculations of the energy spectra of three particles , we have established the general nonmonotonic dependence of the thresholds of stability on the strengthening ( or the weakening ) of the attraction between particles . as a result , we have revealed a number of new nontrivial universal effects in the behavior of the three - particle thresholds of stability . in the first turn , we mark the effect of a `` trap '' , when the number of three - particle bound states nonmonotonically depends on the strengthening of the interaction between particles . secondly , we have discovered the effect of a `` rearrangement '' of the energy levels and the thresholds of stability , which has no analog in the two - particle systems . at the same time , we have constructed a three - particle generalization of the zeldovich effect of a `` rearrangement '' of the energy levels for the attractive interaction potentials with the modes of attraction which essentially differ by their interaction radii . thirdly , we have revealed both the effect of a `` tube '' for the energy states in the region of a significant repulsion between a pair of identical particles and the effect of a `` wedge '' on the boundary , where the binding energies of two different pairs of particles coincide . we have made a sufficiently full and clear analysis of the dependences of the thresholds of stability on the masses of particles and the form of a short - range interaction , and have also considered different symmetries of states . the revealed regularities are important for the comprehension of the general properties of three - particle systems of different nature . in addition , the possibilities to describe the characteristic peculiarities of thresholds in three - particle systems with the coulomb interaction , as well as those of the states with nonzero angular momenta , are open . these problems will be analyzed in further publications . challenging is also the necessity to study the thresholds of stability for one- and two - dimensional systems , where one can expect significant differences in properties from those in three - dimensional problems . e. nielsen , d.v . fedorov , a.s . jensen , e. garrido , phys . rep . * 347 * , n 5 , 373459 ( 2001 ) . efimov , yad . fiz . * 12 * , iss . 5 , 10801091 ( 1970 ) . grinyuk , m.v . kuzmenko , i.v . simenog , ukr . zh . * 48 * , n 10 , 10141023 ( 2003 ) , lanl e - print nucl - th/0308003 . shermatov , teor . mat . * 136 * , n 2 , 257270 ( 2003 ) . thomas , phys . rev . * 47 * , n 12 , 903909 ( 1935 ) . grinyuk , i.v . simenog , in : _ physics of multiparticle systems _ ( naukova dumka , kyiv , 1985 ( in russian ) ) , iss . 8 , 6695 . simenog , a.i . sitnichenko , ukr . * 28 * , n 1 , 16 ( 1983 ) . symmetric , @xmath21 antisymmetric states relative to the permutations of particles @xmath4 . the asterisks indicate the intersections of the lines of the thresholds of states with different symmetries , the vertical dash - dotted lines in the lower part of the figure are the asymptotes for the lines of thresholds , and the dashed inclined lines correspond to the equality of the binding energies @xmath237 of different pairs of particles.,width=480 ] . to the right , the dependence of the energies of three particles on the strengthening of the attraction is shown : @xmath238 corresponds to the dependence of the ground state energy on the constant @xmath5 ( at @xmath239 ) with the available `` trap '' in the resonance region , @xmath240 corresponds to the dependence of the energies of the ground and first excited levels on the constant @xmath241 ( at @xmath242 ) with the available `` trap '' for the excited state , @xmath243 demonstrates the presence of a `` trap '' for the ground state ( at @xmath244 ) . ]	we have studied the general properties of the energy thresholds of stability for a three - particle system with short - range interaction . a wide region of the interaction constants and various ratios of the masses of particles are considered . the specific effects characteristic of the near - threshold stationary energy levels of three particles are revealed . the asymptotic estimates are obtained for the thresholds at some limiting cases , and the high - precision variational calculations of the thresholds for various values of the interaction constants and the masses of particles are carried out .
quantum chromodynamics , qcd , took foothold with the discovery of @xmath3 , the spin triplet s wave state @xmath4 of charmonium . despite 25 years of extensive work in charmonium spectroscopy by slac and desy in @xmath5 annihilation and fermilab in @xmath6 annihilation , several nagging problems remained unsolved . among the most important were problems related to spin singlet states and two photon widths of @xmath7 states . neither the @xmath5 or @xmath6 experiments were able to identify the s- and p - wave singlet states @xmath0 and @xmath1 . these states are obviously extremely important for understanding the spin spin hyperfine interaction of @xmath8 . the two photon widths of @xmath7 states @xmath9 are important for understanding relativistic and radiative effects in charmonia , because in the lowest order they are pure qed widths , akin to those of positronium levels . unfortunately , results from different measuring techniques have remained very divergent for about twenty years . in this talk , i am going to address both of these problems . after 25 years of failed efforts to identify it , @xmath0 was finally identified by belle , babar , and cleo . cleo @xcite measured @xmath11 mev , which determines @xmath12 mev . this is an unexpected and extremely interesting result , considering that @xmath13 mev , and most theoretical predictions had @xmath14 mev . the unexpectedly small hyperfine splitting observed for 2s states makes it more interesting than ever to look deeper into the hyperfine splitting of other charmonia , in particular the 1p states . if the confinement potential is lorentz scalar , there is no long range spin spin interaction in @xmath8 . it follows that for all other waves ( @xmath15 hyperfine splitting is zero , so that @xmath16 . to test this prediction it is necessary to identify @xmath1 and measure @xmath17 with precision . in 1992 fermilab e760 studied the reaction @xmath18 and claimed the observation of a signal for @xmath19 . however , higher luminosity runs by fermilab e835 @xcite in 1996 and 2000 failed to confirm this observation . fermilab e835 has also searched for @xmath19 in their 1996/2000 data in the reaction @xmath20 they report @xmath21 , with 13 observed events and a significance of the @xmath19 signal at @xmath22 level . at cleo @xcite we have now firmly identified @xmath19 with a significance of @xmath23 . at cleo data were taken at @xmath24 , with 3.08 million @xmath24 . these data have been analyzed for @xcite @xmath25 both inclusive and exclusive analyses were done , and an accurate determination of @xmath19 mass was made in recoils against @xmath26 s whose energy could be measured with precision . * inclusive analyses : * two independent analyses were made , one in which the photon energy @xmath27 was constrained , and the other in which @xmath28 was constrained . completely consistent results were obtained . * exclusive analysis : * in this analysis , instead of constraining @xmath27 or @xmath28 , seven known decay channels of @xmath29 were measured . once again , consistent results were obtained . the overall result is @xmath30 or @xmath31 two conclusions follow from these results : ( a ) the simple pqcd expectation is not strongly violated , ( b ) the magnitude and sign of @xmath32 is not yet well determined . , in ( left ) inclusive analysis ( right ) exclusive analysis @xcite.,width=211 ] there are two different ways of measuring the two photon width of @xmath34 . in @xmath5 collisions , the @xmath34 state is formed in two photon fusion , @xmath35 , and a subsequent decay of @xmath34 ( usually @xmath36 ) is measured . these measurements , including the latest one from belle @xcite , yield @xmath37 ev . in @xmath6 annihilation , the @xmath34 state is directly formed and its decay into two photons is measured . in these fermilab measurements @xcite , @xmath38 ev is determined . it is this persistent and long - standing large discrepancy which motivated us to make the present measurement at cleo @xcite using 14.4 fb@xmath39 of @xmath5 data taken in the upsilon region to study the two photon fusion reaction @xmath40 the spectrum of photon mass , @xmath41 , in fig . 3 shows clear peaks corresponding to the e1 photon from @xmath36 . fits to these peaks leads to @xmath42 , @xmath43 , @xmath44 ev for @xmath5 , @xmath45 , and ( @xmath46 ) , respectively , using @xmath47 as measured recently by cleo @xcite . in making a comparison of the present results with other recent results , it was noted that the differences originated from using an old ( pdg1990 ) value for @xmath48 which was nearly 40% smaller than the result of the new cleo measurement @xcite . when the results of the belle and e835 measurements are reevaluated using the cleo value of @xmath48 , it is found that all results become completely consistent . since the two gluon decay width of @xmath34 , @xmath49 mev , our measurement of @xmath2 allows us to estimate the strong coupling constant @xmath50 . according to pqcd , @xmath51 . this leads to @xmath52 . if first order radiative corrections are included , we obtain @xmath53 .	at cleo , the charmonium singlet states @xmath0 and @xmath1 have been firmly identified and a long standing discrepancy for @xmath2 has been resolved . address = department of physics and astronomy , northwestern university , evanston , il , 60208 , usa
measurement of branching fractions and transition form factors of conversion decays provides an important test of vector dominance model @xcite and an accurate background estimation in searches for quark - gluon plasma involving a lepton pair @xcite . the expected branching fractions for the @xmath6 and @xmath7 conversion decays are @xmath8 , @xmath9 and an order of magnitude smaller for decays with muons and/or @xmath10 meson in the final state @xcite . various conversion decays of the @xmath11 meson were studied at cmd-2 @xcite and snd @xcite . for the @xmath6 and @xmath7 mesons experimental information is rather scarce : only the branching fractions for the decays @xmath12 and @xmath13 were measured in experiments at nd @xcite and lepton - g @xcite with an accuracy @xmath14 . conversion decays of a vector meson @xmath15 into a pseudoscalar meson @xmath16 and lepton pair @xmath17 ( @xmath18 ) are closely related to the corresponding radiative meson decays @xmath19 @xcite and provide a possibility to study a transition @xmath20 form factor , @xmath21 , as a function of squared mass of virtual photon , @xmath22 . in this work we present the result of the studies of the @xmath6 and @xmath7 conversion decays into a @xmath23 pair and pseudoscalar meson ( @xmath24 or @xmath10 ) performed with the cmd-2 detector at the vepp-2 m collider @xcite . the analysis is based on a data sample collected at 19 energy points in the 720 840 mev c.m . energy range and corresponding to 3.3 pb@xmath0 of integrated luminosity . this statistics contains about 3.3@xmath2510@xmath26 @xmath6 and 1.8@xmath2510@xmath26 @xmath7 decays . more detailed description of this analysis can be found in @xcite . the general purpose detector cmd-2 has been described in detail elsewhere @xcite . the tracking system consists of the cylindrical drift chamber ( dc ) with 250 @xmath27 resolution transverse to the beam plane and double - layer multiwire proportional z - chamber , both also used for the trigger . the tracking system is placed inside a thin ( 0.38 x@xmath28 ) superconducting solenoid with a field of 1 t. the barrel csi calorimeter with a thickness of 8.1 x@xmath28 placed outside the solenoid has energy resolution for photons of about 9% in the energy range from 100 to 700 mev . the angular resolution is of the order of 0.02 radians . the end - cap bgo calorimeter with a thickness of 13.4 x@xmath28 placed inside the solenoid has energy and angular resolution varying from 9% to 4% and from 0.03 to 0.02 radians , respectively , for the photon energy in the range 100 to 700 mev . the barrel and end - cap calorimeter systems cover a solid angle of @xmath29 radians . the decay @xmath12 has been studied using the @xmath30 dominant decay mode @xmath31 . it corresponds to a final state with two opposite charge particles and two photons . one of the significant resonant backgrounds comes from the @xmath32 decay which has the same topology of the final state and more than three orders of magnitude larger probability . another source of resonant background is the @xmath33 decay followed by the dalitz decay of the @xmath30 or @xmath34-quantum conversion in the material in front of the drift chamber . most conversions occur in the beam pipe , at the distance of 1.8 cm from the interaction point . since the dc spatial resolution is not sufficient to separate events with conversions in the beam pipe from those where it occurs at the interaction point , the contribution of this background was subtracted based on monte carlo simulation ( mc ) . the non - resonant background includes contributions from the following qed processes with the same final state topology : @xmath35 , @xmath36 followed by @xmath34-quantum conversions , @xmath37 with one background photon as well as two - quantum annihilation followed by a @xmath34-quantum conversion and one background photon in calorimeters . the data analysis consists of two parts : a measurement of the branching fraction and a study of the transition form factor . the selection criteria used for the measurement of the branching fractions were tuned to select a `` pure '' set of events under study and thus reject events with a large invariant mass of @xmath23 which contain a lot of background . however , this very range is of main interest for a study of the transition form factor . therefore , for the latter a special set of selection criteria including a technique of e/@xmath24 separation @xcite was applied to suppress the main background from 3@xmath24 events . at the first stage of selection , the following criteria were applied to events with two tracks and at least two photons to enrich a data sample with events of the studied decays : * the photon energy @xmath39 mev and its polar angle @xmath40 to suppress background photons in the calorimeters ; * the impact parameter of the tracks @xmath41 cm and z - coordinate of the vertex @xmath42 cm to reject cosmic rays and beam background events ; * the total momentum of the tracks @xmath43 does not strongly differ from the photon momentum @xmath44 in the @xmath33 decay at a given energy @xmath45 mev / c to suppress @xmath32 events as well as @xmath33 events followed by the dalitz decay of the @xmath30 . this condition is illustrated with horizontal lines in fig . [ f : mompar ] where events satisfying previous requirements are shown ; * the angle between the total momentum of the tracks and each photon is more than 1.7 to suppress qed events ; * the invariant mass of the electron - positron pair and most energetic photon @xmath46 is less than @xmath47 to suppress @xmath48 events followed by a conversion of one @xmath34-quantum ; * the opening angle of the tracks @xmath49 to suppress events of the @xmath32 decay , this condition is shown in fig . [ f : mompar ] with a vertical line . using these criteria 390 events were selected . at the next stage , a fit of the @xmath50 invariant mass distribution ( @xmath51 ) was performed at each energy point to determine the number of events with @xmath30 in the final state denoted hereafter as `` @xmath52 '' ( @xmath12 , @xmath33 followed by a @xmath34-conversion and @xmath32 ) . the parameterization included a logarithmic gaussian for the signal and a second order polynomial for the background . the shape of these functions was fixed , so the floating parameters were the numbers of events of each type only . the shape of background was taken from a study of the @xmath51 distributions from the @xmath7 decays into @xmath53 and @xmath54 specially selected in the whole energy range . the shape of the signal was obtained from the fit of the @xmath51 distribution over all selected events with floating parameters of the function describing the signal , see fig . [ f : invm ] . the total number of selected @xmath52 events is 316 . at this stage of analysis the expected number of background events was evaluated using the @xmath55 distribution . its difference for events of the process under study and those of background was employed in the analysis : the former are strongly peaked near zero while events coming from the @xmath32 decay have a smooth wide distribution . events of the @xmath33 decay with conversion , which are also strongly peaked near zero , predominantly populate the angular range @xmath56 . the @xmath55 distribution for the data after imposing all selection conditions but that on @xmath55 is shown by crosses in fig . [ f : dpsi ] . using the modified requirement @xmath57 we reject most of the events coming from the @xmath33 decay with conversion and keep events under study as well as background events from the @xmath32 and qed . the mc @xmath55 distribution for the main background of @xmath32 shown with a hatched histogram in fig . [ f : dpsi ] was approximated with a function @xmath58 with @xmath59 in the @xmath60 angular range . the energy dependence of the number of events selected at the first stage ( with modified requirements on @xmath55 ) was used to determine the number of @xmath32 events . to this end , the number of selected events at each point was written as @xmath61 where @xmath62 , @xmath63 is the cross section of the @xmath7 meson production at the squared c.m . energy @xmath64 , @xmath65 denotes the integrated luminosity at the @xmath66-th energy point , and @xmath67 is a radiative correction . the floating parameters are @xmath68 and @xmath69 cross section values at the resonance peak . the resulting number of resonance events was calculated as a sum over all energy points : @xmath70 . after subtraction of the @xmath12 contribution , the number of @xmath54 events @xmath71 in the @xmath72 range . the above description of the @xmath55 distribution was used to calculate the number of @xmath32 events in the @xmath49 range : @xmath73 the systematic error comes from the uncertainty of the @xmath55 distribution approximations and that of the subtraction of the @xmath12 contribution . since events of @xmath33 decays could not be distinguished from those under study , the detection efficiency for the former was determined as @xmath74 , where @xmath75 denotes the detection efficiency for @xmath76 events and @xmath77 is a probability for a monoenergetic photon from the @xmath33 decay to convert in the material in front of dc . the @xmath77 value was obtained from simulation : @xmath78 the error of the @xmath77 value above comes from the uncertainties in a thickness and composition of materials in front of dc . at the last stage of analysis , the energy dependence of the number of @xmath52 events is fitted with a sum of the contributions from the @xmath79 , @xmath54 , @xmath53 decays and possibly remaining qed events : @xmath80 the number of @xmath76 events at the @xmath66-th energy point was described by the expression ( [ m : nee ] ) : @xmath81 where @xmath65 is an integrated luminosity , @xmath67 is a radiative correction , @xmath82 , @xmath83 , @xmath84 denote the detection efficiency , efficiency of the reconstruction of close tracks and trigger efficiency at the @xmath66-th energy point , respectively . the energy dependence of the born cross section @xmath85 was written using the relativistic breit - wigner approach with the @xmath6 and @xmath7 meson contributions : @xmath86 here @xmath87 is a phase space factor : @xmath88 @xmath89 is an amplitude of the vector meson @xmath15 : @xmath90 and the additional constant @xmath91 in the amplitude describes a possible contribution of higher resonances . the quantities @xmath92 and @xmath93 are the vector meson mass and the cross section at its peak , respectively , with @xmath93 calculated without taking into account other contributions , and @xmath94 is a phase factor . to describe the @xmath6@xmath7 interference , the model with energy - independent interference phases was used with @xmath95 set to zero and @xmath96 @xcite . the @xmath97 is a vector meson propagator described by the expression : @xmath98 the quantities @xmath99 and @xmath100 are vector meson width at the squared c.m . energy @xmath64 and at the vector meson mass , respectively . the number of @xmath53 events was written as @xmath101 with @xmath102 taken from @xcite and a photon conversion probability @xmath103 taken from ( [ m : pconv ] ) . to describe the energy dependence of @xmath54 events , the relativistic breit - wigner with the @xmath7 meson contribution was used : @xmath104 since the total number of @xmath54 events was previously found , see expression ( [ m : n3pi ] ) , the normalization factor @xmath105 was determined from the sum @xmath106 . a possible contribution of qed events which can , e.g. , appear because of the incorrect background subtraction was evaluated from the following expression : @xmath107 the detection efficiencies @xmath82 were determined using the monte - carlo simulation taking into account initial and final state radiation . the calculation of radiative corrections @xmath108 followed ref . @xcite . the events under study can be triggered by two independent triggers : charged and neutral ones . the overall trigger efficiency is estimated to be 99% . the estimation of this value has been performed by analysis of the trigger signals in selected events . since monte - carlo simulation does not completely describe the experiment , a correction @xmath109 for a difference between the efficiencies of close track reconstruction in simulation and experiment was included to describe the data . its value was obtained using events of @xmath32 decays followed by the conversion decay @xmath110 with a similar @xmath55 distribution . the integrated luminosity was determined using events of large angle bhabha scattering with radiative corrections taken into account according to @xcite . table [ t : data ] shows the detailed information about experiment including efficiencies of detection , trigger and close track reconstruction together with the number of @xmath52 events and visible cross section calculated from the following expression : @xmath111 .the energy , integrated luminosity , radiative correction , detection efficiency , the reconstruction efficiency of close tracks , trigger efficiency , number of @xmath52 events and visible cross section of @xmath52 events . [ cols="^,^,^,^,^,^,^,^",options="header " , ] [ t : etamodes ] the result of the analysis is shown in table [ t : etamodes ] . the total number of selected events for each @xmath10 decay mode was expressed as : @xmath112 the cross section parameterization is similar to formula ( [ m : bw ] ) and includes relativistic breit - wigner contributions of the @xmath6 and @xmath7 mesons , but without the @xmath91 term . the interference model with a constant phase @xmath113 is used . we used the results of our study of the process @xmath114 @xcite to fix the values of the amplitude : @xmath115 , from which the following relation between the branching fractions can be obtained : @xmath116 . since the detection efficiency strongly depends on @xmath64 , its value was calculated at each energy point . the radiative corrections @xmath117 , trigger efficiency @xmath118 as well as the reconstruction efficiency of close tracks @xmath109 were calculated by the same method as for the @xmath76 events . the average number of background events has been evaluated by mc simulation . the simulation of corresponding events included generation of `` fake '' photons @xcite . the results of all analyzed @xmath10 decay modes are taken into account to calculate the upper limits on the branching fractions of the @xmath6 and @xmath7 meson decays . the total number of selected events has been described by summation over expressions ( [ m : ncalc ] ) applied to each @xmath10 decay mode . to calculate upper limits , the feldman - cousins approach @xcite was applied with 3 events observed and 5.1 background events expected . the final values of the upper limits were increased by 13% in accordance with the systematic error . the obtained upper limits exceed theoretical predictions by a factor of 2 @xcite : @xmath119 the electromagnetic transition form factor @xmath121 provides information on the electromagnetic structure of interacting particles . its dependence on squared invariant mass of virtual photon @xmath122 can be approximated in a small @xmath122 range as @xmath123 where the parameter @xmath124 is a slope of the transition form factor . in experiment the information about photon virtuality in events of @xmath12 decay could be taken from the invariant mass of the lepton pair @xmath125 . since the previously used selection criteria leave events with small @xmath122 only and the detection efficiency strongly depends on @xmath126 , the cuts on total momentum , angle between the total momentum of tracks and photons , @xmath46 were omitted and the cut on @xmath55 was increased up to 2.5 radians . to suppress background mainly originating from @xmath32 , the kinematic fit requiring energy - momentum conservation and the procedure of @xmath127 separation based on the energy deposition of charged particles in calorimeters @xcite were additionally used . the selected events were divided into groups of @xmath126 . the number of @xmath12 events in each group was determined from the analysis of the @xmath50 invariant mass distribution . after that the @xmath122 dependence of the obtained number of events was fitted to the following function from @xcite @xmath128^{3/2}\cdot \label{q2_m:5}\end{aligned}\ ] ] where @xmath129 is the fine structure constant . the detector resolution on invariant mass @xmath130 mev / c@xmath131 was taken into account . the floating parameters are the normalization factor @xmath132 and the slope of the transition form factor @xmath124 . the result of the fit is shown in fig . [ f : ff ] , the slope value @xmath133 gev@xmath134 . the total systematic error is estimated to be 10% . the main contributions to it are the @xmath127 separation procedure and selection criteria . this error is smaller than a statistical one for all invariant mass groups except for the group with @xmath135 mev / c@xmath131 . the obtained value agrees with the vdm prediction @xmath136 gev@xmath134 within statistical errors . it is worth noting that modes with @xmath137 in the final state are more convenient for studies of the transition form factor @xmath138 @xcite . unfortunately , their study at cmd-2 is complicated because of the large background coming from the dominant decay mode of the @xmath7 meson : @xmath139 and insufficiently powerful @xmath140 separation at low momenta . using a data sample of 3.3 pb@xmath0 in the 720 840 mev c.m . energy range , the branching fraction for the @xmath12 decay has been obtained : @xmath141 and the following 90% c.l . upper limits were set : @xmath142 the authors are grateful to the staff of vepp-2 m for excellent performance of the collider , and to all engineers and technicians who participated in the design , commissioning and operation of cmd-2 . this work is supported in part by grants doe defg0291er40646 , nsf phy-0100468 , pst.clg.980342 , rfbr-03 - 02 - 16280-a , rfbr-03 - 02 - 16477 , rfbr-03 - 02 - 16843 , rfbr-04 - 02 - 16217 , rfbr-04 - 02 - 16233-a , and rfbr-04 - 02 - 16434 .	using 3.3 pb@xmath0 of data collected with the cmd-2 detector in the 720 840 mev c.m . energy range , the branching fraction of the conversion decay @xmath1 has been measured : @xmath2 . the upper limits for the branching fractions of the following conversion decays have been obtained at the 90% confidence level : @xmath3 , @xmath4 and @xmath5 . , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , cmd-2 , light vector mesons , conversion decays 13.40.g , 13.40.h , 14.40.e
silicon photonics is proving to be a very promising platform for quantum information processing . microring resonators are becoming a key component of such systems as they have been shown to be effective as photon - pair sources by means of spontaneous four wave mixing ( sfwm)@xcite . often , it is desirable to have precisely one photon . while sfwm sources generate pairs of photons , single photons can be achieved through heralding . heralding is a technique in which the detection of a single photon from a pair is used to determine the existence of the other . one of the fundamental issues with ring resonators is their inherent @xmath0 loss when critically coupled , regardless of operation in a single bus or double bus configuration@xcite . for single bus resonators , half of the generated photons are lost to scattering within the cavity . double bus resonators are slightly different as the photons are free to leave the ring through either port - resulting in an effective loss of @xmath0 . all of this assumes that the ring resonator is critically coupled to straight waveguides . by straying from this standard coupling scheme , it is possible to dramatically decrease the effective loss of the source . presented here , are experimental results for two different coupling schemes that achieve greatly reduced photon - pair loss when compared to the critically coupled ring resonator . one method for increasing the coincidence efficiency of a microring resonator is to slightly decouple the input waveguide from the ring . a diagram of an asymmetrically coupled double bus ring resonator is shown in fig . [ fig : ring ] . assuming the ring itself to be lossless , the probability of a photon generated inside the ring to couple out the drop port is the ratio between the input and output couplers given as @xmath4 therefore , photons generated within a symmetric double bus ring resonator ( i.e. @xmath5 ) will exit both ports with equal probability . we can quantify the effect of loss through undesired ports of the ring through the relation @xmath6 evident by the above equation describing the coincidence ratio , an increase in the gap between the input waveguide and the ring ( an increase in @xmath7 ) raises the percentage of drop port coincidences from @xmath8 ( symmetric coupling ) to near unity . additionally , heralding efficiency will be proportional to this trend . the drawback of decreasing through port coupling is that the pump will under couple into the ring , resulting in an increase in required pump power to achieve the same level of ring excitation . this is an acceptable trade off as it is much easier to increase the pump power than it is to increase the brightness of the microring source . recently , vernon et . al worked through the complete theory for a single bus ring , reaching a similar conclusion @xcite . in this experiment , illustrated in figure [ fig : setup ] , a tunable continuous wave pump laser was set to the resonant wavelength ( of the microring cavity ) closest to @xmath9 . the linearly polarized pump passed through a series of bandpass filters to minimize the excess laser noise injected into the system . coupling efficiency was increased by utilizing an inverse taper on the silicon waveguide and fusion splicing a short section of high index fiber ( nufern uhna-7 ) to the optical fiber ( smf-28 ) which was buttcoupled to the chip . once on the chip , the pump coupled into the ring resonator ( @xmath10 , @xmath11 , and @xmath12 ) where non - degenerate photon pairs were produced at resonances spaced symmetrically about the pumped resonance . the generated photons then left the resonator through either port before coupling back into fiber . a series of notch filters ( pump rejection ) were used ( for both outputs of the chip ) to remove any remaining pump photons . a subsequent set of arrayed waveguide gratings ( awgs ) spectrally separated the non - degenerate photon pairs into different fibers that each led to a free running single photon detector ( i d quantique id230 ) . time correlation measurements were then collected with a time to digital converter ( i d quantique id800 ) receiving signals from each of the four detectors . coincidence measurements were made for multiple devices with varying gaps ( input gaps ranging from @xmath13 to @xmath14 with a constant drop port gap of @xmath13 ) . for each of these , coincidences were integrated for @xmath15 with a timing resolution of @xmath16 for pump powers ranging from @xmath17 to @xmath18 . the variation in pump power enabled determination of the difference in the required pump power between devices to achieve the same level of ring excitation . as the waveguide - resonator gap on the input side of the ring was incrementally increased , the measured cross coincidences ( through - drop ) rapidly decreased , matching the expectation that the photons would only exit out the drop port ( fig . [ fig : gap_coincidences ] ) . however , the coincidences from the through port remained high . the cause of this was determined to be the generation of photons in the input waveguide leading up to the ring resonator , which is a broadband four wave mixing process . to prove this , the pump laser was tuned to be off - resonance ( eliminating the possibility of photon generation within the ring ) and coincidences were collected . it was immediately obvious that the through port coincidence counts between the two cases ( on - resonance and off - resonance ) were very comparable to each other while the drop and split coincidences were approximately equal to zero in the off - resonance case . therefore , to remove the effect of photon generation from the input waveguide , it was necessary to determine the coincidence ratio of the source using the drop port coincidences ( @xmath19 ) and the split coincidences ( @xmath20 and @xmath21 ) only using the following equation : @xmath22 input gap and a pump power of @xmath18 . ( b ) coincidence plots for a device with a @xmath14 input gap and a pump power of @xmath18 . the inset shows a zoomed in view of the coincidence plots for the cases where the photon pairs are split . , width=302 ] in order to get accurate values for the coincidence ratio , corrections needed to be made to the raw data from each device . first , the variation in fiber - chip coupling efficiency between the two outputs of the silicon chip was addressed by analyzing the spectrum of the ring resonator from both the through and drop ports . the other source of error to correct for was the variation in the loss between the four different optical paths leading from the chip to the detectors . this was addressed by collecting additional sets of data for each device after swapping the paths that the through and drop port photons took after they exited the chip . the resulting coincidence plots ( shown in fig . [ fig : gap_coincidences ] ) clearly show that the drop port coincidences remain significant ( the large drop in total coincidences is due to the decreased coupling of the pump into the ring ) while the split coincidences ( thru - drop and drop - thru ) decreased to approximately zero between a device with symmetric @xmath13 gaps and one with a @xmath14 input gap . comparing the ring spectra of the various devices allowed the determination of how efficiently the pump was coupling into the ring relative to the critically coupled device . this allowed proper comparison of all of the devices ( fig . [ fig : gap_results ] ) . the symmetric device was found to have a relative coincidence ratio ( [ eq : pcoin ] ) of @xmath23 . while this is significantly higher than the theoretical value of @xmath24 , a very small imperfection in the device fabrication can easily result in that ring - gap variation as it is at the most sensitive point on the theory curve . increasing the input gap to @xmath25 resulted in a device with a largely improved coincidence ratio of @xmath26 ( @xmath27 improvement ) and requiring a factor of @xmath28 increased pump power . further improvements were made with input gaps of @xmath29 and @xmath14 having coincidence ratios of @xmath30 ( @xmath31 improvement ) and @xmath32 ( @xmath33 improvement ) respectively while requiring increased pumping by factors of @xmath34 and @xmath35 respectively . as can be seen in figure [ fig : gap_results ] , these results are in agreement with the theory . while the increase in required pump power is an acceptable trade - off for the improved coincidence ratio , the additional pumping requirement can be removed by taking a different approach to the problem . in 1995 , barbarossa _ et al . _ found that resonant wavelengths of a microring cavity could theoretically be suppressed by coupling the input waveguide to the ring at two points @xcite . this design essentially makes a mach - zehnder interferometer ( mzi ) out of the input waveguide and the ring and has since been demonstrated experimentally @xcite . to understand how such a device will operate , it is useful to first think of the individual components . being a cavity , the ring will only support specific wavelengths of light ( where the resonance condition is satisfied ) separated by the free spectral range ( fsr ) . the spectrum of an unbalanced mzi is sinusoidal with the difference in optical path length between the two paths determining where in the spectrum the constructive and destructive interference will occur . for both the ring and the mzi , this is known as phase - matching . for the case of the ring this is phase - matching between consecutive round - trips while in the mzi it is phase - matching between the two different paths . the points of constructive interference in the spectra of these devices can be tuned by adjusting the relative phase between the different paths . in a fabricated device this can be accomplished by heaters or electro - optic phase shifters . the combination of these two elements results in a phase - matching condition that relies on both the resonance condition of the ring and the interference pattern of the mzi . if the spectral width between two wavelengths of constructive interference in the mzi is twice the fsr of the ring , it is possible to suppress every second resonance of the ring . a schematic of such a device is shown in figure [ fig : dmzr_schematic]a . for the case of a photon - pair source , one side of the ring can be used as the input for the pump photons and the drop side as the output for the generated photon - pairs . the mzi ( mzi1 ) on the input side can be tuned to suppress every other resonance , while mzi2 on the output of the ring can be tuned to suppress the resonances allowed by mzi1 ( i.e. they are perfectly out of phase with each other ) . this configuration will ensure the pump laser is critically coupled into the ring while not allowing it to exit out the drop port , and ensures that any photons that are generated at the resonances allowed by the drop port will only exit the over - coupled drop port ( because mzi1 is tuned to not be phased matched with those photons ) . this makes the device function as though it is two independent single bus ring resonators , one for the input side and one for the output side . the input side ring is characterized by the transmission from the input port to the through port while the output side ring is characterized by the transmission from the add port to the drop port . the theoretical spectral response for both the input and output sides are shown in figure [ fig : dmzr_theory ] . this configuration has three key features : ( i ) the pump is critically coupled so the photon generation rate will be maximized ; ( ii ) the pump is filtered from the photons that exit the drop port minimizing noise and reducing the amount of off - chip filtering required ; ( iii ) the photon pairs will always leave out the same over - coupled drop port , yielding @xmath36 coincidence ratio , maximizing heralding efficiency . .device design dimensions [ cols=">,<",options="header " , ] an image of a fabricated device is shown in figure [ fig : dmzr_schematic]b . the design dimensions for the device are shown in table [ table : dmzr_dimensions ] . initial transmission spectra from each side of the device without any thermal tuning are shown in figure [ fig : dmzr_spectrum](a , b ) . in this configuration , all of the resonances are supported by both couplers . upon performing a full 3-d sweep of voltages ranging from @xmath37 to @xmath38 across the three heaters , an optimum heater configuration ( ring heater = @xmath39 , mzi1 heater = @xmath40 , mzi2 heater = @xmath41 ) was found that exhibited the desired spectral dependence [ fig . [ fig : dmzr_spectrum](c , d ) ] . a test setup similar to that shown in figure [ fig : setup](c ) was used for time correlation measurements . coincidences were collected for @xmath42 with a timing resolution of @xmath43 . the coincidence plots for the case where the heaters were not tuned are shown in figure [ fig : dmzr_coincidences](a ) . due to the difference in the coupling gaps between the input side and the output side , the device already had an enhanced coincidence ratio of @xmath44 but the pump was not critically coupled to the ring [ evident by the @xmath45 extinction in figure [ fig : dmzr_spectrum](a ) ] . the coincidence ratio was further improved in this configuration , resulting in a value of @xmath46 . even more notable is the enhancement to the pumping efficiency when compared to the untuned configuration . this is evident from both the @xmath47 extinction of the through port transmission and @xmath48 increase in the drop port coincidence counts [ shown in fig . [ fig : dmzr_spectrum](c ) and fig . [ fig : dmzr_coincidences](b ) respectively ] . in conclusion , this research shows that the coincidence ratio of a silicon ring resonator photon - pair source can be dramatically increased by engineering the coupling of a ring resonator . two approaches were taken , the first decreased the input coupling to the ring but the pump power had to be correspondingly increased . however , this trade - off can be completely overcome with the addition of mach - zehnder legs attached to the resonator allowing for critical coupling of the pump into the resonator , and over coupling of the signal / idler photons through the drop port . this design enables photon - pair sources with inferred near maximal heralding efficiency out of the ring thanks to a measured coincidence ratio of @xmath49 . the heralding efficiency is a very important characteristic of a photon - pair source , especially when studying the quantum mechanical properties of the photons . this is an essential and necessary step toward high performance and scalable quantum computers and communication systems . this work was performed in part at the cornell nanoscale facility , a member of the national nanotechnology coordinated infrastructure ( nnci ) , which is supported by the national science foundation ( grant eccs-1542081 ) . we acknowledge support for this work from the air force research lab ( afrl ) . this material is based upon work partially supported by the national science foundation under award no . any opinions , findings and conclusions or recommendations expressed in this material are those of the author(s ) and do not necessarily reflect the views of afrl or the national science foundation .	silicon ring resonators are used as photon pair sources by taking advantage of silicon s large third order nonlinearity with a process known as spontaneous four wave mixing . these sources are capable of producing pairs of indistinguishable photons but typically suffer from an effective @xmath0 loss . by slightly decoupling the input waveguide from the ring , the drop port coincidence ratio can be significantly increased with the trade - off being that the pump is less efficiently coupled into the ring . ring resonators with this design have been demonstrated having coincidence ratios of @xmath1 but requiring a factor of @xmath2 increase in the pump power . through the modification of the coupling design that relies on additional spectral dependence , it is possible to achieve similar coincidence ratios without the increased pumping requirement . this can be achieved by coupling the input waveguide to the ring multiple times , thus creating a mach - zehnder interferometer . this coupler design can be used on both sides of the ring resonator so that resonances supported by one of the couplers are suppressed by the other . this is the ideal configuration for a photon - pair source as it can only support the pump photons at the input side while only allowing the generated photons to leave through the output side . recently , this device has been realized with preliminary results exhibiting the desired spectral dependence and with a coincidence ratio as high as @xmath3 while allowing the pump to be nearly critically coupled to the ring . the demonstrated near unity coincidence ratio infers a near maximal heralding efficiency from the fabricated device . this device has the potential to greatly improve the scalability and performance of quantum computing and communication systems .
our universe is organized in a cosmic web on megaparsec ( mpc ) scales , with filaments , voids and massive over - densities of matter . these most massive peaks in the large scale matter density are traced by galaxy clusters , the largest gravitationally bound structures in the universe ( e.g. , @xcite ) . large samples of clusters together with subsequent mass and redshift estimates , allow us to constrain the cluster mass function and thus place improved constraints on important cosmological parameters such as @xmath4 , and @xmath5 ( see allen et al . 2011 for a recent review ) as well as studying cluster evolution across cosmic time . several methods are employed to discover galaxy clusters . optical identification produces the largest cluster catalogs by far , and different algorithms focus on different aspects of optical properties , such as spatial galaxy over - densities ( e.g. , @xcite ) , the red sequence ( e.g. , @xcite ) , or the brightest cluster galaxy ( bcg ) ( e.g. , @xcite ) . optical identification schemes usually suffer from projection effects created by observing a three - dimensional object in a two - dimensional plane . however , optical surveys play a crucial role in measuring cluster redshifts ( e.g. * ? ? ? more recent methods include the sunyaev - zeldovich ( sz ) effect and gravitational lensing . the sz effect is caused by cosmic microwave background ( cmb ) photons inverse compton scattering off the high energy electrons of the intracluster medium ( icm ) ( e.g. , @xcite ) . this can be seen as a distortion in the shape of the cmb spectrum and is used to follow - up known clusters as well as for new cluster surveys ( e.g. , @xcite ) . gravitational lensing provides a direct measure of the mass of a cluster and is observable through the deflection , shearing , and magnification of background sources ( e.g. , @xcite ) . lensing and cosmic shear surveys ( e.g. , @xcite ) trace the large scale structure ( lss ) , provide independent cluster mass estimates , map the dark matter within , and place independent constraints with improved calibration on cosmological parameters ( e.g. @xcite ) and potentially could be a useful detection method in future surveys ( e.g. , @xcite ) , as demonstrated by the deep lens survey @xcite . galaxy clusters also appear as extended sources in the x - ray sky . hot electrons in the icm interact with protons and atomic nuclei to cause plasma emission in the x - ray regime ( e.g. , @xcite ) . x - ray selected cluster surveys have been performed with many different combinations of survey depth and area , such as noras ( northern rosat all - sky survey , @xcite ) , 400sd ( 400 square degree survey , @xcite ) , and reflex ( rosat - eso flux - limited x - ray survey , @xcite ) . these studies and more are included in the mcxc catalog @xcite , a compilation of x - ray selected galaxy clusters and their characteristics . the mcxc catalog includes catalogs based on publicly available rosat all sky survey and serendipitous observations . recent examples of _ xmm - newton _ and/or _ chandra _ selected cluster and group surveys include the champ ( _ chandra _ multiwavelength project ) serendipitous galaxy cluster survey @xcite , galaxy groups in the extended _ chandra _ deep field south @xcite , the _ xmm _ cluster survey @xcite , and the _ xmm - newton _ wide - field survey in the cosmos field @xcite . x - ray identification has several advantages over optical . first , x - ray observations of clusters suffer less from the projection effects that limit most optical surveys . because the x - ray emissivity is proportional to the square of the electron density , it provides good contrast over any background ( e.g. , @xcite ) . furthermore , there are several scaling relations based on x - ray data that galaxy clusters are known to follow . for example , the x - ray luminosity versus mass ( @xmath6@xmath7 ) relation is much tighter than the optical richness to mass relation and is a more accurate mass determination method ( e.g. , @xcite ) . one disadvantage to the x - ray identification of galaxy clusters is that it is difficult to detect high redshift clusters because of the @xmath8 dependence of surface brightness on redshift . thus , shallow x - ray surveys only detect the core region of high redshift clusters , which increasingly resemble point - like sources . also , low redshift galaxies can be extended x - ray sources and may appear as false positives in x - ray cluster surveys ( e.g. , @xcite ) . unfortunately , it is also difficult to get accurate measurements of redshifts from x - ray data alone . therefore using x - ray and optical data in a combined program , like the one presented in this paper , is ideal for determining cluster characteristics . since the mass of a galaxy cluster is dominated by dark matter , their total masses are difficult to measure directly . thus , mass estimates are obtained by correlating cluster mass with easily observable quantities that include x - ray luminosity , richness ( number of member galaxies ) , temperature and velocity dispersions ( e.g. , @xcite ) . proving good correlations between independent mass estimates and further constraining them is imperative so that reliable measurements of the cluster mass function can be obtained . this allows us to investigate its evolution in time and thus better study the formation and evolution of structures and improve constraints on cosmological parameters ( e.g. , @xcite ) . for example , in this combined program we correlate the x - ray bolometric luminosity and the optical richness ( @xmath9 ) , comparing our results with other studies ( see section [ subsec : xlumvsopt ] ) . the _ swift _ gamma - ray burst ( grb ) fields provide 125 deg@xmath10 of serendipitous soft x - ray observations . several other groups are working with this dataset but with different focuses than this study @xcite , and we made comparisons to most of these works in @xcite . in this paper , we compare our results to the more recent _ swift _ xrt cluster survey ( swxcs ) @xcite . in @xcite , we detected 442 extended sources in these grb fields . of these , 209 lie in the footprint of the sdss dr8 @xcite . we can use this sdss data to look for over - densities in the photometric redshift distribution of the galaxies near the x - ray source . using the galaxy over - density detection method described in this paper , we confirm 104 of these candidates as galaxy clusters . our own optical observational data , extending to deeper magnitudes , will be presented in future works . the structure of this paper is as follows . sections [ sec : swift ] and [ sec : sdss ] introduce the _ swift _ and sdss data , respectively . section [ sec : method ] describes the method employed to detect clusters . special cases are discussed in detail in section [ sec : cases ] . we discuss the properties of the cluster candidates with confirmed sdss galaxy over - densities in section [ sec : properties ] , including the overall redshift distribution ( [ subsec : overall ] ) , a comparison of the x - ray luminosity to the optical properties ( [ subsec : xlum ] [ subsec : xlumvsopt ] ) , and the red sequence feature common amongst galaxy clusters ( [ subsec : redseq ] ) . we also compare our results to the literature . in section [ subsec : match ] , we match our catalog to other cluster surveys with large footprints on the sky . we briefly discuss cases found in a previous iteration of our x - ray source selection algorithm that are not in our current catalog ( [ sec : notcatalog ] ) . in section [ sec : discussion ] , we conclude with a summary and discussion of our results . we assume a cosmological model with @xmath11 kms@xmath12mpc@xmath12 , @xmath13 , and @xmath14 . the _ swift _ x - ray telescope ( xrt ) observations of grb fields provide a serendipitous x - ray survey . in @xcite , we describe in detail the data reduction and methods used in producing the x - ray cluster catalog . here , we briefly describe the reduction process and source extraction procedures . the _ swift _ xrt has a relatively large field of view ( [email protected] arcmin@xmath10 ) and is sensitive in the energy range of 0.210 kev . these grb observations are of medium - depth , and randomly distributed on the sky for a total area of @xmath16 deg@xmath10 with a median flux limit of 4@xmath1510@xmath17 . these data are ideal for finding galaxy clusters . we downloaded all xrt `` grb '' observations before 2013 - 07 - 27 , and reprocessed the data as described in @xcite . from this , we made images and corresponding exposure maps in different energy ranges : total ( 0.210 kev ) , soft ( 0.52 kev ) , and hard ( 210 kev ) . sources were detected in the images using the ciao tool ` wavdetect ` using a significance of detection threshold of 10@xmath18 . we excluded grbs by matching the known grb positions . to distinguish clusters from agns , we modeled the surface brightness profiles and compared to @xmath19-models representing a range of cluster masses and redshifts . we defined an extended source to be a cluster candidate if it had a s / n ratio @xmath20 , a minimum net photon count of 20 , and a size that is at least 3@xmath21 above the mean size of corresponding point sources at the same off axis angle . since our cluster detection criteria require a minimum of 20 photons , all the cluster candidates should be real astrophysical sources . false positives should arise only from confusing point agn with extended sources . the catalog of these 442 cluster candidates is given in table 4 of @xcite . sdss dr8 provides the optical data for this study . the publicly available data covers 14,555 deg@xmath10 and has magnitude limits of @xmath22 22.0 , 22.2 , 22.2 , 21.3 and 20.5 mag for the @xmath23 , @xmath24 , @xmath25 , @xmath26 , and @xmath27 bands respectively @xcite . of the 442 cluster candidates , 209 are nominally in the sdss dr8 footprint . however , six candidates lie too close to the survey edges and were not considered further ( see figure [ fig : incsource ] ) , to leave us with 203 candidates in the sample . we downloaded catalogs of the positions , magnitudes , photometric redshifts , and available spectroscopic redshifts for all galaxies within a 40 radius of the cluster candidate centers to provide both source and background regions . a small fraction of these regions are incomplete in coverage due to either being on the edge of the sdss sky coverage , bright object masks or cosmic ray masks ( we discuss these in detail in section [ subsec : incomplete ] ) . we searched for galaxy over - densities in three - dimensional space using the galaxy positions and photometric redshifts provided by sdss dr8 . where available , spectroscopic redshifts are used in place of the photometric redshifts . for each cluster candidate , we considered source regions with radii of 2 and 3 centered on the x - ray centroid . we then selected the galaxies within the prescribed source regions and separated them into redshift bins of width @xmath28 . the sdss photometric redshifts include uncertainties , which have an average of @xmath29 @xcite , larger than this binsize ; however , the accuracy of the cluster redshift is improved by @xmath30 and is typically smaller than @xmath28 , where @xmath31 is the number of galaxies above the background in the 2 or 3 source region and within the redshift bin . we search for over - densities by comparing this redshift distribution to that of the local background . we measured the local background in an annulus extending from 25 to 40 and centered on the _ swift _ sources , where the inner radius was chosen to be sufficiently distant from the cluster regions to avoid contamination from the cluster . local backgrounds are needed in our analysis due to cosmic variance and any non - uniformity in the exposure depth . an example of this variability can be seen in figure [ fig : incsource ] where there is a distinct difference in the galaxy density between swcl j215423.1 ( top left ) and swcl j015021.7 ( top right ) . from the background annulus of each source , we chose 100 random regions equal in size to the source region ( so with radii of 2 or 3 ) and calculated the mean and standard deviation of the galaxy count in each redshift bin . this monte - carlo approach is more general because it does not assume background distributions simply described by poisson fluctuations . we compare the average background count and its standard deviation per redshift bin to the redshift distribution of the source region . we consider an over - density peak significant if the number of galaxies in a redshift bin is 3@xmath21 above the averaged background and if the galaxy count is at least seven . setting a minimum galaxy count is important to reduce the false positives for higher redshift clusters ( @xmath32 ) , where the number of detected galaxies is low so that the standard deviation is rather small , and the 3@xmath21 cut can be affected more by systematic uncertainties , such as the incompleteness of galaxies in sdss for higher redshifts . for potential clusters with multiple over - density peaks , we select the most significant redshift bin for that candidate . if the original distribution did not pick up a significant over - density peak , we shift all of the distribution bins to the left by 0.025 and search for over - density peaks as before . using this method , we search at most 40 redshift bins of width @xmath28 for each cluster candidate . since all of our optically confirmed clusters are below @xmath33 , we essentially have at most 32 redshift bins for detection . for the 203 cluster candidates , the expected number of @xmath34 false positives is @xmath35 or approximately 9% of the identifications . the cluster redshift is determined by averaging the galaxy redshifts in the over - density bin . however , this includes foreground and background galaxies in the bin so the cluster redshifts determined favor the center of the over - density bin . to measure a more accurate cluster redshift , we shift the center of the bin @xmath36 using increments of 0.001 . for each case , we perform the same method as before . we keep the most significant over - density as the best redshift estimate and report cluster details in table [ tab : confz ] . this maximizes the likelihood of our method measuring accurate redshifts while maintaining a low number of tests to minimize the number of false positives . our results are listed in table [ tab : confz ] , where ... indicates cases where no 3@xmath21 peak was detected for either the 2 or 3 source region sizes . we confirm sdss over - density peaks for 104 of the 203 cluster candidates in the sdss footprint . this does not mean that the remaining 105 are not clusters because the sdss data are only deep enough to confirm @xmath37 clusters consistently . examples of our sdss cluster confirmation for two _ swift _ xrt images of grb060204b and grb061110a are shown in figures [ fig : sw1 ] and [ fig : sw2 ] , respectively . the blue circles mark the positions of the extended x - ray sources , and the galaxy redshift distributions for the _ swift _ cluster candidates ( red dashed lines ) are shown in separate panels together with the local mean redshift distribution and the @xmath34 level above the local mean . for each candidate with a confirmed over - density peak , we measure the distance between the bcg and the x - ray source center , and calculate the distribution of the physical offsets ( in mpc ) between the bcg and the x - ray source center ( figure [ fig : bcg ] ) . we select the bcg as the galaxy with the brightest sdss @xmath38band magnitude in the source region and the @xmath28 redshift bin . in table [ tab : confz ] , we list the sdss @xmath38band apparent and absolute magnitudes of the bcgs and their physical offsets to their corresponding x - ray centers . it is often assumed that both bcgs and x - ray centroids define the center of the cluster potential wells and thus , their locations should overlap . however , in studies of optically selected galaxy groups it has been found that the brightest halo galaxies can be satellite galaxies instead of central galaxies ( e.g. , @xcite ) . @xcite also found that the number of brightest halo galaxies that were not central increased when studying a higher group mass bin . we show our calculated mass ( details in section [ subsec : xlum ] ) versus bcg - to - x - ray offset in figure [ fig : massvsbcg ] and find the opposite to be true for our sample , that higher mass clusters have low bcg offsets . the bcg - to - x - ray centroid distance distribution exhibits a broad tail extending from a compact core ( e.g. , @xcite ) . we compare our offset distribution with that of 2mass clusters @xcite , which is normalized to match our sample count , and find that the distributions are similar . of the 104 confirmed clusters , there are 73 cases where our method found sdss galaxy over - densities for both the 2 and 3 regions . 70% of these cases agree within the redshift range @xmath39 , where we assigned the more significant peak as the redshift of the cluster . the twenty - four cases where the 2 and 3 regions both detect an over - density peak but the peaks are located in different redshift bins are listed in table [ tab : cases ] . we compared both the significance of the over - density peak and the bcg - to - x - ray center offset ( in mpc ) for both possibilities and assign a score as @xmath40mpc for each cluster in table [ tab : cases ] . if the score is positive ( negative ) then we assign the redshift to that measured by the 2 ( 3 ) region . the final assigned redshifts are shown in bold in table [ tab : cases ] . in this section , we discuss the situation where multiple _ swift _ cluster candidates are closely located in angular position and redshift space and check for complications in the sdss confirmation method . we also discuss how we treat clusters with incomplete sdss coverage of either the source or background areas . we found four cases with _ swift _ cluster candidate pairs within 6 of each other and with similar redshifts ( @xmath41 ) . with the possibility of overlapping source regions , these could result in false positive detections or redshift mis - assignments and thus require a closer examination . for each case , we compare the following for the cluster properties in question : the significance of the over - density peak , confirmation using the other source region size ( so either 2 or 3 ) , the bcg@xmath42to@xmath42x - ray offset , any other redshift peaks over @xmath34 ( i.e. multiple peaks ) , and both x - ray images and galaxy positions . we list the conclusions of our analysis in table [ tab : close ] , and show two examples in figures [ fig : nearbycont ] and [ fig : nearbycont2 ] , where we also include a discussion of our conclusions in the figure captions . swcl j215423.1 and swcl j215413.2 are separated by a distance of 2@xmath437 and both have a high significance peak ( @xmath44 ) around @xmath45 , and swcl j215413.2 has a second galaxy over - density peak at @xmath46 ( with significance @xmath47 ) . it is likely that the @xmath45 peak for swcl j215413.2 is contaminated by the cluster members of swcl j215423.1 . we assign @xmath48 for swcl j215423.1 and @xmath46 for swcl j215413.2 . swcl j232717.2 and swcl j232725.6 are separated by a distance of 4@xmath434 and both have a high significance peak ( @xmath44 ) at @xmath49 , and swcl j232725.6 has a second galaxy over - density peak at @xmath50 ( with significance @xmath51 ) . it is likely that the @xmath49 peak for swcl j232725.6 is contaminated by the cluster members of swcl j232717.2 . we assign @xmath52 for swcl j232717.2 and @xmath50 for swcl j232725.6 ( figure [ fig : nearbycont ] ) . in the remaining two cases , there are no additional over - density peaks in either of the close pairs . for the swcl j092730.1 and swcl j092719.6 pair , we detected over density peaks at @xmath53 ( @xmath54 ) using 2 ( 3 ) source size regions for swcl j092719.6 , but only a peak at @xmath55 using 2 region for swcl j092730.1 . further examining the bcg offsets , significance of the peaks and the x - ray images , we determine that both are clusters , possibly merging . we reach a similar conclusion for the swcl j085552.1 and swcl j085619.7 pair ( figure [ fig : nearbycont2 ] ) . if a cluster candidate is close to the edge of the sdss footprint , then the 40 region surrounding the x - ray source s center will appear to be artificially devoid of galaxies . in addition , there are masking regions associated with bright sources and cosmic rays in the images . to check for coverage completeness for each candidate cluster , the galaxy distribution within 40 of x - ray source center was visually inspected . eight of the candidates have incomplete 3 source regions as shown in figure [ fig : incsource ] . each dot represents a galaxy , the blue solid circles represent the 2 regions and the red dashed circles represent the 3 regions , indicating the largest source region size considered in our analysis . two of the eight cases ( c and g of figure [ fig : incsource ] ) have complete 2 source regions and thus , are mostly complete in sdss and are included in this study . the other six incomplete source regions shown in figure [ fig : incsource ] are excluded from this study . we are examining these cases by performing our own follow - up observations . this lead to the exclusion of 6 candidates ( swcl j012302.8 , swcl j020934.7 , swcl j032216.0 , swcl j075036.6 , swcl j173932.6 and swcl j194530.3 ) , as discussed earlier in section [ sec : sdss ] . there are 33 candidates with incomplete background annuli , and figure [ fig : incback ] shows a few examples . other examples of incomplete backgrounds are similar in shape and incompleteness to those that are shown . each dot represents a galaxy , the blue solid circles are 3 regions containing the source , and the red dashed circles denote the background annulus with inner and outer radii 25 and 40 . since we consider 100 random regions to calculate the average local background and variance , the incomplete background causes the average count of galaxies to artificially decrease and the variance significantly increases . for these 33 cluster candidates , we considered 100 random regions that exclude the incomplete areas . the redshifts and cluster details reported in table [ tab : confz ] reflect the results of this augmented method . here we describe various properties of the 104 sdss confirmed clusters , match them to existing catalogs and discuss the implications for cluster science . figure [ fig : tinker ] shows the distribution of our sdss confirmed cluster redshifts , as compared to the @xcite model for the mass function and redshift evolution of dark matter halos with masses ranging from @xmath56 to @xmath57 , assuming @xmath58 , @xmath59 , @xmath60 , and @xmath61 . although the cosmological parameters are slightly different from the ones used in our paper , the model predictions are sufficient for our current qualitative comparison , providing us with a model to test the completeness of our catalog . figure [ fig : tinker ] compares our observed redshift distribution to that expected from the model . for lower redshifts , our survey appears complete up to @xmath62 and is 80% complete up to @xmath63 . however , the observed distribution is significantly lower for higher redshifts . this is presumably due to the shallowness of the sdss dr8 catalog for galaxies at these redshifts where galaxies are fainter . for the higher - z cluster candidates , we are performing our own , deeper optical observations . an advantage of having an x - ray selected catalog with optical follow - up is that we can more easily identify correlations between properties in the two bands . here , we discuss how we estimate the x - ray luminosity and compare it to the optical properties of the 104 candidate clusters with confirmed sdss galaxy over - densities . we used xspec @xcite to convert the _ swift _ x - ray count rates to luminosities , assuming a galactic neutral hydrogen density of 5@xmath64 @xmath65 , a cluster temperature of 5 kev , and an abundance of 0.4 solar . assuming a @xmath19-model with @xmath19 = 0.65 , we corrected the x - ray luminosities to an aperture of radius 1.0 mpc . in this way the aperture would be similar to that used for the optical richness calculations . the uncertainties in @xmath6 , as seen in figure [ fig : xlum ] , are a combination of the uncertainties of the x - ray photon counts and the 25% systematic uncertainties from changing the assumed @xmath19 value by @xmath660.15 . we also estimate @xmath67 , the mass inside of a radius @xmath68 at which the density of the cluster is 500 times the critical density ( @xmath69 ) , using the relation @xmath70 from @xcite , where @xmath71 . these estimates are included in table [ tab : confz ] . we report observed galaxy counts ( @xmath31 ) and optical richness estimates ( @xmath9 ) in table [ tab : confz ] . the first is the excess galaxy count above the average background in the redshift bin of width @xmath72 . @xmath9 is the estimated number of galaxies with magnitude brighter than @xmath73 , assuming a schechter luminosity function ( e.g. , @xcite ) with a break magnitude @xmath73 = @xmath4221.34 mag and slope @xmath74 = @xmath421.07 based on the results of @xcite for sdss@xmath75 band . we assume @xmath73 evolves with redshift using the correction @xmath76 with @xmath77 as described by @xcite . we converted the limiting apparent @xmath38band magnitude @xmath78 into an absolute magnitude limit for each cluster candidate with a confirmed sdss galaxy over - density . in calculating the limiting absolute magnitudes , we take into account galactic dust extinction and also apply @xmath79-corrections . we use the ned online calculator for galactic extinction values based on the x - ray positions of the cluster candidates . we apply @xmath79-corrections using the template models of @xcite for the elliptical galaxies , the expected dominant galaxy population in clusters . we normalize the schechter luminosity function for each cluster candidate using the limiting absolute magnitude and the observed background - subtracted galaxy count . we then use the redshift dependent @xmath73 to estimate the optical richness , @xmath9 , of the confirmed galaxy clusters . since @xmath31 is found using apparent radii on the sky of 2 and 3 , we correct @xmath9 to an aperture radius of 1.0 mpc , assuming an nfw density profile @xcite . we do this by multiplying @xmath9 by @xmath80 , where @xmath81 is the projected radius of 1.0 mpc , @xmath82 is the aperture radius in mpc corresponding to either 2 or 3 , and @xmath83 is the projected fraction inside radius r. we used the @xmath84 and @xmath85 relations to estimate the scale radius @xmath86 @xcite . the uncertainties in @xmath9 are dominated by the statistical uncertainties in the galaxy counts ( @xmath30 ) and are plotted in figure [ fig : xlum ] . since we calculate the excess galaxy count using a redshift bin width @xmath87 , smaller than the typical photometric redshift uncertainty of @xmath29 ( as discussed in section [ sec : method ] ) , we apply a correction factor of 6.01 for all clusters . at @xmath88 , where any sdss galaxies must be significantly brighter than @xmath89 , this procedure produces large corrections that tend to grossly overestimate @xmath9 . these clusters are indicated in figure [ fig : xlum ] with crosses ( @xmath15 ) . in addition to @xmath31 and @xmath9 , we include @xmath25-band absolute magnitude of the bcgs . these are listed in table [ tab : confz ] . figure [ fig : xlum ] shows @xmath90 as a function of @xmath9 for cluster candidates confirmed using the 3 region . both @xmath6 and @xmath9 correlate well with mass , as @xmath6 probes the gas mass and @xmath9 traces the galaxy members ( e.g. , @xcite ) . * the scaling relation @xmath91 we obtain in log space using an orthogonal regression fit with error dependent weights * is shown in figure [ fig : xlum ] * ( solid line ) . without the error dependent weights , we obtain the scaling relation of @xmath92 , also shown in figure [ fig : xlum ] ( dashed line ) . * the orthogonal regression line minimizes the perpendicular distances between the data points and the line ( e.g. , @xcite ) . * our slope of 2.67 is steeper * than the results of @xcite and @xcite , who report slopes of 1.56 and 1.33 , respectively . * our slope of 1.86 agrees better with these two optically selected studies . * @xcite compares the slopes of the @xmath93 from several studies in their figure 17 and these slopes are * in general * steeper than what we measure here , ranging from 1.84 to 5.86 . comparisons are difficult , as each study can vary in fitting methods and in defining and estimating @xmath9 and @xmath6 . for example , @xcite discuss two methods of linear regression solutions , the ordinary least - squares ( ols ) bisector and orthogonal regression ( e.g. , @xcite , where the ols bisector is the line that bisects the ols solution minimized in the y direction ( ols(y\|x ) ) and the ols solution minimized in the x direction ( ols(x\|y ) ) . @xcite found that the orthogonal regression line was better suited for this fit , especially in cases of large scatter . for ease of comparison and based on their results , we choose to use the orthogonal regression method . we find our slope is on the shallow end of those seen in @xcite . here we mention a few potential reasons for this difference . we define optical richness in section [ subsec : opticalprop ] , and @xcite use a definition based on a physical aperture and apparent magnitude of member galaxies . furthermore , we use a bolometric x - ray luminosity and they use @xmath6 in the [ [email protected] kev ] energy band . both @xcite and @xcite use bolometric x - rayluminosities and report slopes of 1.56 and 3.60 , respectively . since clusters are virialized objects , the predicted scaling relation between @xmath6 and mass is @xmath94 , @xmath95 , and @xmath96 ( e.g. , kaiser 1991 ) . however , with the presence of non - gravitational effects , the scaling relations deviate from the above predictions . one of the recent measurement between @xmath6 and total mass and stellar mass yields @xmath97 and @xmath98 ( anderson et al . 2015 ) and our measured slope * @xmath99 agrees with the @xmath6@xmath73 relation but deviates from the @xmath6@xmath100 relation of anderson et al . ( 2015 ) . and our measured slope of @xmath101 , using a fit without error dependent weights , agrees with the @xmath6@xmath100 relation . * since all of these measurements have their own systematic uncertainties , we do not expect an exact match in the slopes at this stage . however , those studies that measured extreme steep slopes , 45 , for the @xmath6@xmath9 relation were subject to more severe systematic uncertainties . we show @xmath6 as a function of bcg @xmath102 in figure [ fig : xlumvsbcgmag ] . here we test the correlation of luminosity and mass of the icm with the luminosity of the bcg . although there is a large scatter , we observe a general positive correlation when we fit the data with the orthogonal regression method ( e.g. , @xcite ) . the data is well correlated , with a spearman s correlation coefficient of @xmath103 and a probability of 0.03 . points shown in red indicate the bcgs with smallest bcg - to - x - ray center offsets . these clusters show less scatter than clusters with greater bcg - to - x - ray offsets , with correlation coefficients of @xmath104 and @xmath105 and probabilities of 0.04 and 0.27 , respectively . we examined the color distributions for each candidate with a confirmed sdss galaxy over - density to search for the red sequence feature seen in galaxy clusters . in general , cluster members and the surrounding background form a bimodal color distribution that is well represented by a two gaussian fit , as seen in figure [ fig : colorplot ] ( e.g. , @xcite ) . the narrow , taller gaussian is the red sequence feature , prominent in galaxy clusters . the distribution of background and/or foreground field galaxies can generally be approximated with a shallower , wider gaussian , because the background galaxies are from different redshifts and the weak color bimodality is smoothed by redshifts . following @xcite , we shifted between the @xmath106 , @xmath107 , and @xmath108 colors to follow the shifting location of the balmer break with redshift . the redshift ranges used for each color are listed in table [ tab : colredrange ] . for each cluster candidate with a confirmed sdss galaxy over - density , we examine the color distribution using galaxies within 3 of the cluster center and within redshifts @xmath109 , where @xmath110 is our measured cluster redshift . we fit the color distribution to a two gaussian model where we allow the normalization , mean , and width of both gaussians to vary and minimize the @xmath111 statistic . for the lower redshift clusters , @xmath112 , we use a single gaussian model with one prior for the expected color . for these cases , we do not observe color contribution from the background galaxies , likely due to the extent of the clusters on the sky . our two gaussian model includes priors that constrain the width of the red sequence compared to that of the background and constrain the mean of the red sequence to favor the colors used by @xcite in their catalog of gaussian mixture brightest cluster galaxy ( gmbcg ) clusters . we use the second prior because we know from @xcite how the red sequence evolves with redshift ( see figure [ fig : meancolorplot ] ) . we calculate the second prior by averaging the color of gmbcg clusters in redshift bins of width @xmath28 . these are shown by green diamonds in figure [ fig : meancolorplot ] and by the vertical lines in figure [ fig : colorplot ] . we find the red sequence feature in @xmath11385% of the confirmed clusters . the cluster galaxies ( red dots in figure [ fig : colorplot ] , right ) , are the galaxies with color within 2@xmath21 of the red sequence mean . two examples are shown in figure [ fig : colorplot ] , along with corresponding color magnitude diagrams ( cmds ) . here we can observe the red sequence feature clearly both in the narrow histogram and in the clustering of galaxies in color - magnitude space . we also show the mean red sequence color as a function of redshift in figure [ fig : meancolorplot ] . although there is a significant amount of scatter , figure [ fig : meancolorplot ] shows a general positive trend for the g @xmath42 r and r @xmath42 i colors as redshift increases , as expected . we include gmbcg cluster colors from @xcite ( dots ) in figure [ fig : meancolorplot ] as well as the color priors used for our two gaussian fit ( green diamonds ) . in figure [ fig : massredseq ] , we plot the mass of clusters with confirmed redshifts , denoting clusters with a detected red sequence with red triangles and clusters without with black crosses ( @xmath15 ) . we see that in general there is no difference in mass dependence . we performed a kolmogorov - smirnov ( k - s ) test to determine whether the two mass distributions are in fact different . we find there is a 0.56 probability that the distributions are drawn from the same distribution , signifying that the mass distribution of clusters with a detected red sequence is not significantly different than those without a detected red sequence . however , we do detect the red sequence in the six most massive clusters , so clusters with mass greater than @xmath114 . also , we find that the redshift distributions for detected red sequences versus undetected are significantly different , with a k - s test probability of 0.03 . the non - detections of the red sequence in our survey has several origins . the principle problem is that many of the galaxies are relatively faint , potentially leading to large color errors compared to the width of the red sequence . galaxies fainter than the magnitude limit of sdss will not be included in the color distributions , so that the distribution may not fit the two gaussian model . this is especially true for more distant clusters where the observed flux is lower . at low redshifts , the x - ray data is also sensitive enough to include lower mass groups , which lack the well - defined red sequences of rich clusters . @xcite had similar issues in their study of low x - ray luminosity galaxy clusters . we compare our catalog to other studies across multiple wavelengths to test for accuracy and completeness . we chose to compare to catalogs with large footprints since our _ swift _ targets are scattered over a large fraction of the sky . first , we compare with optical catalogs , starting with the 132,684 galaxy clusters in the catalog from @xcite . their method uses a friend - of - friend algorithm incorporating sdss iii galaxies to identify clusters and their bcgs . we used a matching angular radius of up to 1 . out of 44 position matches , 41 agree in redshift within @xmath115 with an average redshift difference of 0.026 . these are listed in table [ tab : gmbcg ] . of the 41 redshift matches , @xmath116% have mass greater than the median mass of our sample , showing that in general we are matching our more massive clusters to the optically selected catalog . @xcite claimed their catalog is 95% complete for @xmath117 and for the redshift range @xmath118 , and our sample is complete for @xmath117 and @xmath119 0.3 . the number of clusters per survey area for the overlapping conditions @xmath117 and 0.05 @xmath120 0.3 are equal ( @xmath121 ) and thus , the completeness statements for both samples are in agreement . we also compare our confirmed cluster redshifts to the gmbcg clusters @xcite . using the above matching procedure , there are 18 position matches and 15 that agree in position as well as redshift , 14 of which are in the @xcite catalog . swcl j121628.2 is not in the @xcite catalog , so the right ascension , declination and separation reported in table [ tab : gmbcg ] refer to the gmbcg catalog . the differences in redshift and position between our catalog and these optical catalogs could arise from the fact that our clusters are x - ray selected and thus centered on the x - ray emission from the clusters as opposed to their optically selected clusters . however , some of the matches with large angular match radii ( @xmath122@xmath43@xmath123 ) might be spurious associations . here we compare to catalogs with selections based on x - ray , sz and lensing data . these trace the location of the gas in the cluster ( e.g. , @xcite ) and should align with our sources better than optically selected surveys , so we use a smaller matching radius of 30 . we search for matches to x - ray selected clusters using the meta - catalog of x - ray detected clusters of galaxies ( mcxc , @xcite ) . mcxc comprises of 1743 clusters and combines the rosat all sky survey with cluster surveys based on serendipitous observations . we find only two previously detected clusters , mcxc_j1557.7 + 3530 at redshift @xmath124 and mcxc_j1557.7 + 3530 at @xmath125 . these match to our clusters swcl j155743.3 at @xmath126 and swcl j025630.7 at @xmath127 @xcite . swcl j025630.7 is also detected with the atacama cosmology telescope ( act ) @xcite using the sz effect . this is the only cluster match when compared to the 148 ghz observations by act of 68 galaxy clusters . we also compared our survey to the first release of sz sources observed by planck @xcite and found that swcl j084749.4 at @xmath128 is a match to psz1_g213.43 + 31.78 at @xmath129 . finally , we compared our confirmed galaxy clusters to the deep lensing survey @xcite but found no matches . in @xcite , we compared our catalog of extended sources to the _ swift _ xrt cluster survey ( swxcs ) @xcite , and found 55 of 72 sources agreed in position to within 30 . @xcite expanded the analysis from using only grb fields in @xcite to also including non - grb fields , increasing the number of fields to @xmath130 and finding 263 cluster candidates in a total solid angle of @xmath131 deg@xmath10 . they cross - correlated their catalog with optical , x - ray , and sz catalogs to match known galaxy and galaxy cluster redshifts that have similar positions . of the 442 extended sources in @xcite , 88 agree within 30 and 68 agree within 10 . although the number of swxcs cluster candidates increases from 72 to 263 @xcite , we do not see a comparative increase in matches because , like @xcite , our analysis includes only _ swift _ grb fields . we compared the positions of our sdss confirmed galaxy clusters to their updated catalog and found 37 position matches in the sdss footprint using a matching radius of 10 . these are shown in table [ tab : liu ] . in @xcite , there are clusters with multiple reported redshifts that come from comparing the swxcs to other studies . for these cases , we report the average of those redshifts in table [ tab : liu ] . of those 37 matches , 23 have redshifts in both catalogs , 22 agree within @xmath115 and 19 agree within @xmath132 with an average redshift difference of @xmath133 . furthermore , 17 of the 37 position matches were also matched to the @xcite catalog and thus are also presented in table [ tab : gmbcg ] . we went through various iterations of the x - ray source selection method before using the latest one discussed in @xcite . there are 10 x - ray sources selected using a previous version of our algorithm that were confirmed as galaxy clusters using sdss data and the method described above . these are listed in table [ tab : notcatalog ] . due to slight changes in the x - ray source selection algorithm , the significances of these sources changed to be below the significance threshold and so are not included in the current catalog or in table [ tab : confz ] . these clusters are worth mentioning for researchers interested in these individual clusters in future studies . in this paper , we present sdss identifications ( table [ tab : confz ] ) , including estimated redshifts , x - ray and optical properties , for the x - ray selected galaxy clusters we identified in @xcite . we confirmed 104 of the 203 cluster candidates in the sdss footprint and estimate that the catalog is 80% complete up to @xmath134 . most of the remaining cluster candidates are expected to be at higher redshifts where the member galaxies are too faint to use sdss for confirmations . this sample significantly increases the number of x - ray selected clusters with confirmed redshifts and it is one of the largest uniformly - selected cluster samples , covering a total sky area of 125 deg@xmath10 . we observe clear red sequences and clustering of galaxies in color - magnitude space in @xmath11385% of the clusters . x - ray source centers and bcg locations show good agreement , with small offsets and distributions similar to other studies . we find clear matches with previously observed clusters . our x - ray luminosities correlate well with optical properties and our @xmath93 slope agrees with other estimates . thus , it is clear the _ swift _ cluster detection technique presented in @xcite is successfully identifying extended x - ray sources that are in fact galaxy clusters . in future studies , we will look at changing the significance threshold of the x - ray source detection @xcite as it may be too stringent currently . for example , we list 10 clusters in table [ tab : notcatalog ] that were confirmed by the sdss data but removed from the current x - ray catalog based on the final choice for the significance threshold . our method combines x - ray and optical techniques and properties to confirm galaxy cluster candidates and can be used for similar studies . in this way we find 63 new galaxy clusters in the sdss footprint that were not detected using the optical cluster finding methods of @xcite and @xcite . the next step is to look deeper in magnitude and beyond the footprint of sdss . although sdss was a good start , the data is too shallow for higher redshift clusters and has incomplete sky coverage . there are @xmath135 undetected _ swift _ extended sources in the sdss footprint , and still @xmath136 outside , for a total of @xmath137 . we have been performing our own followup observations with observing programs taking photometric data at the mdm 2.4 m , kitt peak 4 m , and ctio ( cerro tololo inter - american observatory ) 4 m . we have also observed candidates with redshifts of @xmath138 at the magellan 6.5 m and mdm 2.4 m using multi - slit spectroscopic masks . once we derive galaxy redshifts from our data we will use the method outlined in this paper to confirm additional clusters at higher redshifts . furthermore , our sample will be more statistically significant when it is more complete , which will lead to better constraints on the cluster mass function . funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the us department of energy , the japanese monbukagakusho , and the max planck society . abell , g. o. , corwin , h. g. , jr . , & olowin , r. p. 1989 , , 70 , 1a adami , c. , mazure , a. , pierre , m. , et al . 2011 , a&a , 526 , a18 adelman - mccarthy , j. k. , ageros , m. a. , allam , s. s. , et al . 2007 , apjs , 172 , 634 aihara , h. , allende prieto , c. , an , d. , et al . 2011 , apjs , 193 , 29 allen , s. w. , evrard , a. e. , & mantz , a. b. 2011 , ara&a , 49 , 409 anderson , m. e. , gaspari , m. , white , s. d. m. , wang , w. , & dai , x. 2015 , mnras , 449 , 3806 arnaud , k. a. 1996 , astronomical data analysis software and systems v , 101 , 17 ascaso , b. , wittman , d. , & dawson , w. 2014 , mnras , 439 , 1980 assef , r. j. , kochanek , c. s. , brodwin , m. , et al . 2010 , , 713 , 970 bahcall , n. a. 1988 , ara&a , 26 , 631b barkhouse , w. a. , green , p. j. , vikhlinin , a. , et al . 2006 , , 645 , 955 bell , e. f. , mcintosh , d. h. , katz , n. , & weinberg , m. d. 2003 , apjs , 149 , 289b bell , e. f. , wolf , c. , meisenheimer , k. , et al . 2004 , , 608 , 752 blackburne , j. a. , & kochanek , c. s. 2012 , , 744 , 76 bhringer , h. , voges , w. , huchra , j. p. , et al . 2000 , apjs , 129 , 435 burenin , r. a. , vikhlinin , a. , hornstrup , a. , et al . 2007 , apjs , 172 , 561 carlstrom , j. e. , holder , g. p. , & reese , e. d. 2002 , ara&a , 40 , 643 delia , v. , perri , m. , puccetti , s. , et al . 2013 , a&a , 551 , a142 dai , x. , kochanek , c. s. , & morgan , n. d. 2007 , , 658 , 917 dai , x. , assef , r. j. , kochanek , c. s. , et al . 2009 , , 697 , 506 dai , x. , bregman , j. n. , kochanek , c. s. , & rasia , e. 2010 , , 719 , 119 dai , x. , griffin , r. d. , kochanek , c. s. , nugent , j. m. , & bregman , j. n. 2015 , apjs , 218 , 8 donahue , m. , mack , j. , scharf , c. , et al . 2001 , apjl , 552 , l93 evans , p. a. , osborne , j. p. , beardmore , a. p. , et al . 2014 , apjs , 210 , 8 finoguenov , a. , guzzo , l. , hasinger , g. , et al . 2007 , apjs , 172 , 182 finoguenov , a. , tanaka , m. , cooper , m. , et al . 2015 , a&a , 576 , a130 gal , r. r. , de carvalho , r. r. , lopes , p. a. a. , et al . 2003 , aj , 125 , 2064 girardi , m. , giuricin , g. , mardirossian , f. , mezzetti , m. , & boschin , w. 1998 , , 505 , 74 gladders , m. d. & yee , h. k. c. 2005 , apjs , 157 , 1 guzzo , l. , schuecker , p. , bhringer , h. , et al . 2009 , a&a , 499 , 357 hao , j. , mckay , t. a. , koester , b. p. , et al . 2010 , apjs , 191 , 254 hasselfield , m. , hilton , m. , marriage , t. a. , et al . 2013 , jcap , 7 , 8 hoekstra , h. 2007 , mnras , 379 , 317h hoekstra , h. , & jain , b. 2008 , annual review of nuclear and particle science , 58 , 99 isobe , t. , feigelson , e. d. , akritas , m. g. , & babu , g. j. 1990 , , 364 , 104 kaiser , n. 1991 , , 383 , 104 kochanek , c. s. , white , m. , huchra , j. , et al . 2003 , , 585 , 161 koester , b. p. , mckay , t. a. , annis , j. , et al . 2007 , , 660 , 239 kravtsov , a. v. & borgani , s. 2012 , ara&a , 50 , 353k lin , y. & mohr , j. j. 2004 , , 617 , 879 liu , t. , tozzi , p. , tundo , e. , et al . 2015 , apjs , 216 , 28 lopes , p. a. a. , de carvalho , r. r. , & capelato , h. v. 2006 , , 648 , 209l marriage , t. a. , acquaviva , v. , ade , p. a. r. , et al . 2011 , , 737 , 61 mehrtens , n. , romer , a. k. , hilton , m. , et al . 2012 , mnras , 423 , 1024 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1995 , mnras , 275 , 720 nilo castelln , j. l. , alonso , m. v. , garca lambas , d. , et al . 2014 , mnras , 437 , 2607 oguri , m. , takada , m. , okabe , n. , & smith , g. p. 2010 , mnras , 405 , 2215 oguri , m. , bayliss , m. b. , dahle , h. , et al . 2012 , mnras , 420 , 3213 pasquali , a. , van den bosch , f. c. , mo , h. j. , yang , x. , & somerville , r. 2009 , mnras , 394 , 38 planck collaboration , ade , p. a. r. , aghanim , n. , et al . 2011 , a&a , 536 , a8 planck collaboration , ade , p. a. r. , aghanim , n. , et al . 2014 , a&a , 571 , aa29 piffaretti , r. , arnaud , m. , pratt , g. w. , pointecouteau , e. , & melin , j .- b . 2011 , a&a , 534 , a109 popesso , p. , bhringer , h. , brinkmann , j. , voges , w. , & york , d. g. 2004 , a&a , 423 , 449 puccetti , s. , capalbi , m. , giommi , p. , et al . 2011 , a&a , 528 , a122 refregier , a. 2003 , ara&a , 41 , 645 reichardt , c. l. , stalder , b. , bleem , l. e. , et al . 2013 , , 763 , 127 richard , j. , smith , g. p. , kneib , j .- , et al . 2010 , mnras , 404 , 325 shan , h. , kneib , j .- p . , li , r. , et al . 2015 , arxiv:1502.00313 skibba , r. a. , van den bosch , f. c. , yang , x. , et al . 2011 , mnras , 410 , 417 sunyaev , r. a. , & zeldovich , y. b. 1972 , comments on astrophysics and space physics , 4 , 173 tinker , j. , kravtsov , a. v. , klypin , a. , et al . 2008 , , 688 , 709 tundo , e. , moretti , a. , tozzi , p. , et al . 2012 , a&a , 547 , a57 umetsu , k. , medezinski , e. , nonino , m. , et al . 2014 , , 795 , 163 valentinuzzi , t. , poggianti , b. m. , fasano , g. , et al . 2011 , a&a , 536 , a34 van den bosch , f. c. , aquino , d. , yang , x. , et al . 2008 , mnras , 387 , 79 vikhlinin , a. , burenin , r. a. , ebeling , h. , et al . 2009 , , 692 , 1033 voit , g. m. 2005 , reviews of modern physics , 77 , 207 von der linden , a. , best , p. , kauffmann , g. , & white , s. d. m. 2007 , mnras , 379 , 867 weinberg , d. h. , mortonson , m. j. , eisenstein , d. j. , et al . 2013 , phr , 530 , 87 weinmann , s. m. , van den bosch , f. c. , yang , x. , & mo , h. j. 2006 , mnras , 366 , 2 wen , z. l. , han , j. l. , & liu , f. s. 2012 , apjs , 199 , 34 wittman , d. , dellantonio , i. p. , hughes , j. p. , et al . 2006 , , 643 , 128 swcl j002111.4 + 210438 & 0.674 & & 0.674 & 4.14 & 5.99 & 186.68 & 21.84 & @xmath4223.24 & 0.66 & 3.52e+44 & 4.96e+14 + swcl j002114.5 + 205943 & 0.141 & 0.074 & 0.141 & 6.08 & 6.82 & 13.19 & 16.94 & @xmath4222.02 & 0.06 & 3.75e+43 & 1.75e+14 + swcl j002823.6 + 092705 & 0.195 & 0.215 & 0.195 & 7.09 & 13.06 & 22.54 & 17.82 & @xmath4223.23 & 0.02 & 7.95e+43 & 2.71e+14 + swcl j003317.8 + 193925 & & 0.246 & 0.246 & 3.17 & 3.41 & 7.98 & 19.85 & @xmath4222.20 & 0.61 & 9.62e+43 & 2.95e+14 + swcl j005136.8 + 074351 & & 0.757 & 0.757 & 3.90 & 2.32 & 432.92 & 19.26 & @xmath4226.40 & 0.74 & 8.50e+43 & 1.94e+14 + swcl j012310.5 + 375549 & 0.749 & 0.762 & 0.762 & 4.52 & 2.37 & 521.41 & 22.29 & @xmath4223.54 & 1.11 & 9.45e+43 & 2.06e+14 + swcl j015751.5 + 170139 & & 0.529 & 0.529 & 4.85 & 6.20 & 52.61 & 20.66 & @xmath4223.34 & 0.01 & 1.25e+44 & 2.89e+14 + swcl j015752.9 + 165933 & 0.510 & 0.525 & 0.510 & 5.30 & 10.00 & 50.36 & 20.38 & @xmath4223.15 & 0.29 & 2.26e+44 & 4.22e+14 + swcl j015803.5 + 165005 & 0.223 & 0.233 & 0.223 & 7.46 & 11.91 & 20.55 & 16.95 & @xmath4224.33 & 0.01 & 8.09e+43 & 2.69e+14 + swcl j020003.8 + 084024 & 0.196 & & 0.196 & 4.10 & 6.96 & 12.48 & 16.24 & @xmath4224.90 & 0.02 & 1.93e+44 & 4.70e+14 + swcl j020006.4 + 084454 & 0.419 & & 0.419 & 4.01 & 8.73 & 26.16 & 19.07 & @xmath4223.85 & 0.03 & 1.32e+44 & 3.21e+14 + swcl j025547.9 + 000902 & 0.573 & 0.570 & 0.573 & 4.00 & 7.71 & 78.20 & 20.89 & @xmath4223.39 & 0.63 & 3.36e+44 & 5.17e+14 + swcl j025630.7 + 000601 & 0.374 & 0.391 & 0.374 & 15.49 & 45.33 & 112.34 & 18.22 & @xmath4224.48 & 0.01 & 7.60e+43 & 2.35e+14 + swcl j035259.4@xmath42004338 & 0.413 & 0.301 & 0.301 & 3.77 & 7.32 & 23.86 & 18.59 & @xmath4222.94 & 0.36 & 4.58e+44 & 7.51e+14 + swcl j054716.7 + 641156 & 0.271 & 0.369 & 0.369 & 6.19 & 10.37 & 44.17 & 19.36 & @xmath4226.24 & 0.80 & 8.25e+43 & 2.48e+14 + swcl j075043.2 + 310400 & 0.378 & 0.375 & 0.378 & 4.69 & 14.16 & 33.79 & 19.16 & @xmath4223.59 & 0.43 & 1.09e+44 & 2.94e+14 + swcl j075900.8 + 324449 & 0.579 & 0.751 & 0.579 & 3.70 & 7.53 & 75.21 & 21.99 & @xmath4222.36 & 0.64 & 3.50e+44 & 5.28e+14 + swcl j075908.5 + 324237 & 0.722 & & 0.722 & 3.99 & 5.85 & 481.81 & 21.15 & @xmath4224.13 & 0.58 & 2.45e+44 & 3.83e+14 + swcl j080022.5@xmath42085931 & & 0.273 & 0.273 & 3.98 & 4.25 & 11.45 & 18.94 & @xmath4223.12 & 0.32 & 1.28e+44 & 3.47e+14 + swcl j082113.9 + 320018 & 0.746 & 0.728 & 0.746 & 7.58 & 9.09 & 859.00 & 21.16 & @xmath4224.28 & 0.08 & 9.68e+44 & 8.86e+14 + swcl j083340.9 + 331118 & 0.682 & 0.675 & 0.682 & 8.42 & 18.06 & 673.84 & 20.66 & @xmath4224.35 & 0.22 & 5.10e+44 & 6.21e+14 + swcl j084749.4 + 133142 & 0.391 & 0.391 & 0.391 & 7.40 & 21.23 & 55.50 & 19.41 & @xmath4223.22 & 0.07 & 1.64e+45 & 1.56e+15 + swcl j084959.1 + 521711 & 0.201 & & 0.201 & 3.95 & 8.00 & 13.75 & 18.95 & @xmath4222.06 & 0.15 & 1.20e+44 & 3.48e+14 + swcl j085523.8 + 110202 & 0.317 & & 0.317 & 4.50 & 10.88 & 20.82 & 19.67 & @xmath4222.04 & 0.40 & 1.10e+44 & 3.07e+14 + swcl j085552.1 + 465925 & 0.477 & & 0.477 & 4.01 & 7.77 & 29.24 & 19.67 & @xmath4223.81 & 0.69 & 1.27e+44 & 3.02e+14 + swcl j085619.7 + 470018 & 0.479 & 0.474 & 0.479 & 3.53 & 6.41 & 23.93 & 20.27 & @xmath4223.05 & 0.05 & 6.83e+43 & 2.05e+14 + swcl j090714.8 + 351020 & 0.218 & 0.212 & 0.218 & 5.08 & 10.73 & 17.49 & 17.18 & @xmath4223.57 & 0.36 & 5.01e+43 & 2.00e+14 + swcl j092607.0 + 314854 & 0.268 & & 0.268 & 3.44 & 8.54 & 14.40 & 18.95 & @xmath4222.61 & 0.25 & 7.47e+43 & 2.49e+14 + swcl j092649.8 + 301346 & 0.570 & 0.567 & 0.567 & 5.33 & 7.64 & 84.84 & 18.86 & @xmath4225.48 & 0.78 & 3.58e+44 & 5.40e+14 + swcl j092719.6 + 301348 & 0.396 & 0.331 & 0.331 & 4.25 & 9.07 & 25.43 & 18.31 & @xmath4223.72 & 0.04 & 1.52e+44 & 3.71e+14 + swcl j092730.1 + 301046 & 0.303 & & 0.303 & 4.41 & 10.72 & 20.16 & 19.05 & @xmath4223.22 & 0.23 & 3.67e+44 & 6.54e+14 + swcl j092852.3 + 002137 & 0.535 & 0.091 & 0.091 & 4.09 & 4.32 & 13.27 & 16.91 & @xmath4220.52 & 0.02 & 2.77e+43 & 1.50e+14 + swcl j093041.3 + 170400 & 0.320 & & 0.320 & 4.88 & 10.96 & 20.56 & 20.00 & @xmath4222.23 & 0.46 & 7.25e+43 & 2.36e+14 + swcl j093045.4 + 165930 & 0.180 & 0.177 & 0.180 & 5.84 & 9.58 & 15.81 & 17.07 & @xmath4223.62 & 0.19 & 4.15e+43 & 1.82e+14 + swcl j095257.1 + 102440 & 0.400 & 0.398 & 0.400 & 4.07 & 11.79 & 29.35 & 18.83 & @xmath4223.92 & 0.02 & 1.52e+44 & 3.55e+14 + swcl j095513.4 + 181215 & 0.423 & 0.421 & 0.423 & 4.57 & 11.61 & 33.60 & 19.54 & @xmath4223.33 & 0.57 & 1.34e+44 & 3.23e+14 + swcl j095515.5 + 180357 & 0.745 & 0.747 & 0.745 & 8.29 & 11.16 & 962.85 & 20.37 & @xmath4225.03 & 0.19 & 3.47e+44 & 4.68e+14 + swcl j101341.5 + 430651 & 0.400 & 0.400 & 0.400 & 4.09 & 11.11 & 26.72 & 18.30 & @xmath4224.53 & 0.03 & 1.37e+44 & 3.34e+14 + swcl j111736.0 + 033711 & 0.403 & 0.400 & 0.403 & 4.40 & 12.78 & 35.37 & 19.67 & @xmath4223.46 & 0.26 & 3.71e+44 & 6.17e+14 + swcl j113427.6@xmath42070208 & 0.239 & 0.304 & 0.239 & 5.02 & 10.36 & 17.31 & 17.44 & @xmath4224.04 & 0.01 & 1.21e+44 & 3.41e+14 + swcl j114232.3 + 505623 & 0.254 & & 0.254 & 3.47 & 9.06 & 15.30 & 16.77 & @xmath4224.81 & 0.02 & 8.59e+43 & 2.74e+14 + swcl j114503.1 + 600811 & 0.268 & 0.262 & 0.268 & 5.61 & 14.15 & 23.77 & 18.31 & @xmath4223.34 & 0.01 & 5.52e+43 & 2.06e+14 + swcl j114553.0 + 595320 & & 0.546 & 0.546 & 4.53 & 4.85 & 45.29 & 21.27 & @xmath4222.63 & 0.36 & 1.39e+44 & 3.04e+14 + swcl j115811.3 + 452903 & 0.389 & 0.407 & 0.389 & 6.05 & 17.12 & 39.64 & 18.39 & @xmath4224.48 & 0.22 & 1.04e+44 & 2.83e+14 + swcl j120137.8 + 104936 & 0.277 & 0.745 & 0.277 & 3.37 & 7.66 & 13.36 & 17.82 & @xmath4223.78 & 1.16 & 1.39e+44 & 3.63e+14 + swcl j121628.2 + 353820 & 0.347 & 0.358 & 0.358 & 4.52 & 8.68 & 25.13 & 19.57 & @xmath4223.74 & 0.81 & 1.12e+44 & 3.03e+14 + swcl j121711.8 + 353745 & 0.492 & 0.472 & 0.492 & 3.89 & 8.83 & 37.41 & 20.33 & @xmath4223.00 & 0.25 & 2.30e+44 & 4.32e+14 + swcl j123313.9 + 210217 & 0.516 & 0.516 & 0.516 & 6.51 & 13.68 & 69.95 & 19.74 & @xmath4224.07 & 0.11 & 2.26e+44 & 4.20e+14 + swcl j123612.4 + 290222 & 0.104 & 0.105 & 0.104 & 5.21 & 7.07 & 16.07 & 16.09 & @xmath4223.29 & 0.21 & 7.48e+43 & 2.75e+14 + swcl j124308.5 + 170639 & 0.722 & 0.145 & 0.145 & 6.12 & 7.85 & 17.97 & 15.52 & @xmath4225.00 & 0.30 & 3.71e+43 & 1.74e+14 + swcl j124312.1 + 170454 & 0.132 & 0.136 & 0.136 & 6.84 & 7.23 & 16.93 & 15.52 & @xmath4224.99 & 0.20 & 5.65e+43 & 2.27e+14 + swcl j125957.2 + 155717 & 0.700 & 0.685 & 0.700 & 4.74 & 6.82 & 296.56 & 21.28 & @xmath4223.71 & 0.62 & 2.83e+44 & 4.25e+14 + swcl j130332.1 + 591556 & 0.218 & 0.147 & 0.218 & 4.81 & 9.70 & 15.75 & 17.37 & @xmath4223.56 & 0.42 & 1.06e+44 & 3.20e+14 + swcl j130911.8 + 611521 & 0.253 & & 0.253 & 3.67 & 7.88 & 13.44 & 12.55 & @xmath4229.00 & 0.01 & 8.34e+43 & 2.69e+14 + swcl j130959.1 + 612530 & 0.248 & 0.247 & 0.248 & 6.19 & 12.52 & 20.77 & 17.85 & @xmath4223.44 & 0.04 & 1.05e+44 & 3.12e+14 + swcl j131521.9 + 164155 & 0.342 & 0.242 & 0.242 & 4.79 & 5.97 & 13.81 & 16.35 & @xmath4224.70 & 0.01 & 3.78e+44 & 6.93e+14 + swcl j133051.0 + 420641 & 0.597 & 0.588 & 0.588 & 4.93 & 6.65 & 86.40 & 20.37 & @xmath4223.82 & 0.60 & 1.70e+44 & 3.35e+14 + swcl j133055.8 + 420015 & & 0.113 & 0.113 & 3.92 & 4.31 & 11.24 & 14.57 & @xmath4223.36 & 0.00 & 1.13e+44 & 3.54e+14 + swcl j135914.0 + 470528 & 0.526 & 0.527 & 0.526 & 8.86 & 17.97 & 95.73 & 20.55 & @xmath4223.32 & 0.28 & 1.03e+44 & 2.56e+14 + swcl j140637.3 + 274348 & 0.600 & 0.758 & 0.600 & 5.34 & 10.78 & 125.86 & 20.08 & @xmath4224.41 & 0.17 & 1.35e+45 & 1.20e+15 + swcl j140639.0 + 273546 & 0.232 & 0.233 & 0.232 & 5.95 & 14.44 & 24.57 & 18.37 & @xmath4222.67 & 0.07 & 1.39e+44 & 3.75e+14 + swcl j140726.4 + 274738 & 0.177 & 0.174 & 0.174 & 4.46 & 10.17 & 22.41 & 16.06 & @xmath4224.26 & 0.01 & 7.85e+43 & 2.72e+14 + swcl j140907.4 + 242406 & 0.566 & 0.579 & 0.566 & 4.66 & 9.67 & 72.16 & 20.32 & @xmath4223.58 & 0.44 & 1.24e+44 & 2.80e+14 + swcl j141221.8 + 165216 & 0.571 & 0.438 & 0.571 & 4.32 & 7.80 & 65.16 & 21.09 & @xmath4223.32 & 0.43 & 2.25e+44 & 4.04e+14 + swcl j143211.6 + 362225 & 0.663 & 0.623 & 0.663 & 4.75 & 9.94 & 225.21 & 21.34 & @xmath4223.47 & 0.11 & 1.63e+44 & 3.10e+14 + swcl j144209.2 + 333414 & 0.451 & 0.638 & 0.638 & 6.42 & 6.84 & 151.86 & 22.04 & @xmath4223.64 & 0.85 & 3.31e+44 & 4.90e+14 + swcl j144604.5 + 203334 & 0.275 & & 0.275 & 3.19 & 6.77 & 12.01 & 19.09 & @xmath4222.68 & 0.17 & 8.84e+43 & 2.75e+14 + swcl j151550.9 + 442056 & 0.270 & 0.109 & 0.109 & 5.44 & 7.48 & 19.24 & 16.39 & @xmath4223.00 & 0.02 & 1.04e+44 & 3.36e+14 + swcl j152252.9 + 253527 & 0.546 & 0.626 & 0.546 & 7.20 & 13.06 & 89.91 & 19.59 & @xmath4224.67 & 0.01 & 5.83e+44 & 7.41e+14 + swcl j152316.4 + 254754 & & 0.796 & 0.796 & 7.60 & 3.03 & 1156.89 & 21.59 & @xmath4224.16 & 0.84 & 1.12e+44 & 2.24e+14 + swcl j155117.4 + 445118 & 0.697 & & 0.697 & 5.66 & 9.43 & 328.74 & 21.99 & @xmath4222.95 & 0.58 & 3.76e+44 & 5.09e+14 + swcl j155159.8 + 445748 & 0.213 & 0.207 & 0.213 & 6.23 & 12.02 & 20.09 & 16.60 & @xmath4224.27 & 0.01 & 7.90e+43 & 2.67e+14 + swcl j155555.3 + 410548 & 0.368 & & 0.368 & 5.10 & 14.71 & 32.07 & 18.17 & @xmath4224.29 & 0.01 & 3.32e+44 & 5.89e+14 + swcl j155708.6 + 354100 & 0.429 & 0.440 & 0.429 & 4.20 & 10.80 & 30.22 & 19.66 & @xmath4223.50 & 0.15 & 1.10e+44 & 2.85e+14 + swcl j155743.3 + 353020 & 0.166 & 0.161 & 0.166 & 35.68 & 54.31 & 97.45 & 15.91 & @xmath4224.56 & 0.04 & 8.77e+44 & 1.22e+15 + swcl j160637.5 + 321351 & 0.065 & 0.058 & 0.065 & 5.22 & 7.27 & 25.61 & 16.22 & @xmath4222.02 & 0.14 & 7.15e+43 & 2.73e+14 + swcl j160956.9 + 301052 & 0.114 & 0.276 & 0.114 & 4.64 & 10.02 & 21.14 & 17.66 & @xmath4221.55 & 0.17 & 4.36e+43 & 1.95e+14 + swcl j164637.4 + 363021 & & 0.077 & 0.077 & 3.54 & 2.93 & 9.67 & 17.39 & @xmath4221.02 & 0.17 & 2.12e+43 & 1.28e+14 + swcl j164956.4 + 313021 & 0.734 & & 0.734 & 4.82 & 8.39 & 617.61 & 21.73 & @xmath4223.83 & 0.59 & 4.32e+44 & 5.41e+14 + swcl j170542.2 + 112451 & & 0.409 & 0.409 & 3.69 & 6.25 & 25.12 & 19.83 & @xmath4223.19 & 0.59 & 3.12e+44 & 5.52e+14 + swcl j170716.1 + 235208 & 0.084 & 0.085 & 0.085 & 4.83 & 4.79 & 15.17 & 20.35 & @xmath4222.59 & 0.28 & 3.88e+43 & 1.85e+14 + swcl j170757.1 + 235135 & 0.212 & 0.213 & 0.212 & 6.34 & 10.28 & 17.97 & 17.64 & @xmath4223.43 & 0.01 & 2.47e+44 & 5.42e+14 + swcl j172011.1 + 692315 & 0.470 & 0.270 & 0.270 & 4.90 & 6.77 & 15.61 & 20.80 & @xmath4222.34 & 0.72 & 6.44e+43 & 2.27e+14 + swcl j200031.1 + 085259 & & 0.047 & 0.047 & 3.94 & 3.28 & 2.21 & 18.63 & @xmath4219.87 & 0.15 & 8.29e+41 & 1.73e+13 + swcl j215411.4@xmath42001127 & 0.381 & 0.214 & 0.381 & 6.14 & 14.54 & 41.65 & 20.52 & @xmath4221.98 & 0.20 & 1.69e+44 & 3.84e+14 + swcl j215411.6@xmath42000654 & 0.213 & 0.106 & 0.106 & 3.68 & 2.77 & 7.72 & 18.60 & @xmath4220.85 & 0.10 & 4.34e+43 & 1.96e+14 + swcl j215413.2 + 000413 & 0.227 & 0.166 & 0.227 & 4.60 & 10.58 & 18.72 & 17.67 & @xmath4223.65 & 0.04 & 1.70e+44 & 4.25e+14 + swcl j215423.1 + 000526 & 0.169 & 0.170 & 0.169 & 8.58 & 15.76 & 28.35 & 16.10 & @xmath4224.23 & 0.07 & 1.64e+44 & 4.31e+14 + swcl j215520.0@xmath42001115 & 0.373 & 0.371 & 0.371 & 3.83 & 7.08 & 26.74 & 19.81 & @xmath4223.70 & 0.85 & 1.16e+44 & 3.06e+14 + swcl j222432.9@xmath42021216 & 0.204 & 0.162 & 0.162 & 4.00 & 5.54 & 13.21 & 18.57 & @xmath4221.52 & 0.25 & 4.69e+43 & 1.99e+14 + swcl j222438.0@xmath42022231 & 0.499 & 0.496 & 0.499 & 6.44 & 13.73 & 71.02 & 19.91 & @xmath4223.83 & 0.06 & 1.83e+44 & 3.72e+14 + swcl j222439.0@xmath42021111 & & 0.292 & 0.292 & 5.16 & 8.99 & 24.41 & 18.44 & @xmath4223.89 & 0.56 & 4.75e+43 & 1.85e+14 + swcl j222444.0@xmath42022034 & 0.714 & 0.718 & 0.714 & 3.84 & 6.71 & 546.21 & 22.08 & @xmath4223.29 & 0.44 & 2.81e+44 & 4.20e+14 + swcl j222506.2@xmath42020611 & 0.524 & & 0.524 & 3.45 & 6.35 & 40.70 & 20.50 & @xmath4223.20 & 0.75 & 1.47e+44 & 3.20e+14 + swcl j222516.4@xmath42020825 & 0.467 & 0.585 & 0.467 & 3.56 & 8.61 & 34.29 & 20.42 & @xmath4222.99 & 0.60 & 1.74e+44 & 3.69e+14 + swcl j222954.1 + 194350 & 0.287 & 0.290 & 0.287 & 7.65 & 15.63 & 28.27 & 18.00 & @xmath4223.63 & 0.03 & 1.12e+44 & 3.15e+14 + swcl j224206.9 + 233408 & 0.402 & & 0.402 & 5.38 & 13.98 & 40.30 & 19.40 & @xmath4223.76 & 0.13 & 3.19e+44 & 5.61e+14 + swcl j231257.7 + 182543 & 0.325 & 0.548 & 0.325 & 3.83 & 8.73 & 21.26 & 19.98 & @xmath4222.50 & 0.54 & 1.07e+44 & 2.99e+14 + swcl j231733.7 + 322828 & 0.400 & 0.376 & 0.400 & 4.35 & 13.24 & 37.54 & 19.32 & @xmath4223.25 & 0.07 & 1.61e+44 & 3.67e+14 + swcl j232244.2 + 055601 & 0.235 & & 0.235 & 4.18 & 7.80 & 13.89 & 17.51 & @xmath4224.04 & 0.31 & 5.55e+43 & 2.11e+14 + swcl j232248.4 + 054810 & 0.320 & 0.244 & 0.244 & 5.92 & 9.94 & 23.97 & 16.85 & @xmath4224.56 & 0.01 & 1.72e+44 & 4.24e+14 + swcl j232717.2 + 263108 & 0.223 & 0.223 & 0.223 & 11.50 & 13.29 & 31.74 & 16.85 & @xmath4224.33 & 0.05 & 1.41e+44 & 3.80e+14 + swcl j232725.6 + 263506 & 0.257 & 0.227 & 0.257 & 5.71 & 8.79 & 16.05 & 17.99 & @xmath4223.13 & 0.38 & 1.72e+44 & 4.21e+14 + swcl j233009.3 + 264459 & & 0.522 & 0.522 & 3.68 & 4.00 & 36.17 & 19.82 & @xmath4223.99 & 0.56 & 1.07e+44 & 2.64e+14 + swcl j002114.5 + 205943 & * 0.141 * & 0.074 & 6.075 & 5.425 & 0.056 & 0.108 & 0.269 & + swcl j035259.4@xmath42004338 & 0.413 & * 0.301 * & 4.398 & 3.775 & 0.358 & 0.004 & @xmath420.145 & + swcl j054716.7 + 641156 & 0.271 & * 0.369 * & 3.193 & 6.190 & 0.282 & 0.797 & @xmath420.485 & + swcl j075900.8 + 324449 & * 0.579 * & 0.751 & 3.696 & 3.017 & 0.644 & 0.838 & 0.421 & + swcl j092719.6 + 301348 & 0.396 & * 0.331 * & 3.717 & 4.255 & 0.522 & 0.037 & @xmath420.664 & + swcl j092852.3 + 002137 & 0.535 & * 0.091 * & 3.443 & 4.092 & 0.604 & 0.022 & @xmath420.799 & + swcl j113427.6@xmath42070208 & * 0.239 * & 0.304 & 5.024 & 4.663 & 0.010 & 0.013 & 0.122 & + swcl j120137.8 + 104936 & * 0.277 * & 0.745 & 3.367 & 4.565 & 0.085 & 1.159 & 0.675 & + swcl j124308.5 + 170639 & 0.722 & * 0.145 * & 3.898 & 6.120 & 0.601 & 0.303 & @xmath421.038 & + swcl j130332.1 + 591556 & * 0.218 * & 0.147 & 4.808 & 4.112 & 0.422 & 0.332 & 0.142 & + swcl j131521.9 + 164155 & 0.342 & * 0.242 * & 5.038 & 4.786 & 0.533 & 0.014 & @xmath420.435 & + swcl j140637.3 + 274348 & * 0.600 * & 0.758 & 5.342 & 4.929 & 0.168 & 0.545 & 0.515 & + swcl j141221.8 + 165216 & * 0.571 * & 0.438 & 4.325 & 3.997 & 0.434 & 0.860 & 0.535 & + swcl j144209.2 + 333414 & 0.451 & * 0.638 * & 3.256 & 6.423 & 0.314 & 0.851 & @xmath420.519 & + swcl j151550.9 + 442056 & 0.270 & * 0.109 * & 3.903 & 5.444 & 0.152 & 0.017 & @xmath420.648 & + swcl j152252.9 + 253527 & * 0.546 * & 0.626 & 7.201 & 5.655 & 0.014 & 0.015 & 0.516 & + swcl j160956.9 + 301052 & * 0.114 * & 0.276 & 4.639 & 3.975 & 0.173 & 0.362 & 0.410 & + swcl j172011.1 + 692315 & 0.470 & * 0.270 * & 3.066 & 4.902 & 0.695 & 0.724 & @xmath420.583 & + swcl j215411.4@xmath42001127 & * 0.381 * & 0.214 & 6.142 & 4.248 & 0.202 & 0.446 & 0.876 & + swcl j215411.6@xmath42000654 & 0.213 & * 0.106 * & 3.542 & 3.680 & 0.211 & 0.098 & @xmath420.159 & + swcl j215413.2 + 000413 & 0.227 & * 0.166 * & 4.597 & 8.218 & 0.041 & 0.460 & @xmath420.788 & + swcl j222516.4@xmath42020825 & * 0.467 * & 0.585 & 3.564 & 3.523 & 0.602 & 0.996 & 0.407 & + swcl j231257.7 + 182543 & * 0.325 * & 0.548 & 3.832 & 3.597 & 0.543 & 0.530 & 0.065 & + swcl j232248.4 + 054810 & 0.320 & * 0.244 * & 4.636 & 5.925 & 0.080 & 0.009 & @xmath420.501 & + swcl j085552.1 & swcl j085619.7 & 4.799 & 0.477 & 0.479 & 4.01 & 3.53 & 0.694 & 0.050 & no & 1 : cluster at 0.477 , 2 : @xmath27 = 0.479 + swcl j092730.1 & swcl j092719.6 & 3.795 & 0.303 & 0.331 & 4.41 & 4.25 & 0.254 & 0.036 & no & 1 : cluster at @xmath27 = 0.303 , 2 : @xmath27 = 0.331 + swcl j215423.1 & swcl j215413.2 & 2.745 & 0.169 & 0.166 & 8.58 & 8.22 & 0.063 & 0.449 & yes & 1 : strong cluster at 0.169 , 2 : @xmath27 = 0.227 + swcl j232725.6 & swcl j232717.2 & 4.378 & 0.227 & 0.223 & 6.42 & 11.50 & 0.328 & 0.048 & yes & 1 : cluster at @xmath27 = 0.257 , 2 : strong cluster at @xmath27 = 0.223 + swcl j015752.9 + 165933 & 0.036 & 29.471 & 16.992 & 0.507 & & 0.510 + swcl j015803.5 + 165005 & 0.036 & 29.515 & 16.835 & 0.209 & & 0.223 + swcl j020003.8 + 084024 & 0.101 & 30.014 & 8.674 & 0.215 & & 0.196 + swcl j025630.7 + 000601 & 0.049 & 44.129 & 0.101 & 0.355 & 0.375 & 0.374 + swcl j035259.4@xmath42004338 & 0.017 & 58.247 & @xmath420.727 & 0.325 & 0.314 & 0.301 + swcl j054716.7 + 641156 & 0.932 & 86.793 & 64.209 & 0.363 & & 0.369 + swcl j083340.9 + 331118 & 0.513 & 128.421 & 33.197 & 0.709 & & 0.682 + swcl j084749.4 + 133142 & 0.034 & 131.956 & 13.528 & 0.359 & & 0.391 + swcl j090714.8 + 351020 & 0.566 & 136.802 & 35.167 & 0.210 & & 0.218 + swcl j092730.1 + 301046 & 0.789 & 141.864 & 30.188 & 0.373 & 0.293 & 0.303 + swcl j095257.1 + 102440 & 0.061 & 148.237 & 10.410 & 0.375 & & 0.400 + swcl j095513.4 + 181215 & 0.324 & 148.808 & 18.209 & 0.327 & 0.316 & 0.423 + swcl j095515.5 + 180357 & 0.983 & 148.818 & 18.050 & 0.622 & & 0.745 + swcl j101341.5 + 430651 & 0.097 & 153.424 & 43.116 & 0.423 & & 0.400 + swcl j111736.0 + 033711 & 0.817 & 169.387 & 3.625 & 0.442 & & 0.403 + swcl j113427.6@xmath42070208 & 0.046 & 173.616 & @xmath427.036 & 0.251 & & 0.239 + swcl j114232.3 + 505623 & 0.069 & 175.636 & 50.940 & 0.260 & & 0.254 + swcl j114553.0 + 595320 & 0.010 & 176.471 & 59.889 & 0.182 & & 0.546 + swcl j115811.3 + 452903 & 0.701 & 179.560 & 45.492 & 0.409 & 0.416 & 0.389 + swcl j121628.2 + 353820 & 0.058 * & 184.117 * & 35.640 * & & 0.266 & 0.358 + swcl j121711.8 + 353745 & 0.458 & 184.306 & 35.635 & 0.443 & 0.459 & 0.492 + swcl j123313.9 + 210217 & 0.288 & 188.312 & 21.040 & 0.544 & & 0.516 + swcl j130959.1 + 612530 & 0.166 & 197.492 & 61.423 & 0.243 & 0.254 & 0.248 + swcl j131521.9 + 164155 & 0.062 & 198.840 & 16.699 & 0.223 & & 0.242 + swcl j140637.3 + 274348 & 0.054 & 211.655 & 27.731 & 0.655 & & 0.600 + swcl j140639.0 + 273546 & 0.232 & 211.664 & 27.600 & 0.263 & 0.259 & 0.232 + swcl j140726.4 + 274738 & 0.083 & 211.861 & 27.795 & 0.159 & 0.171 & 0.174 + swcl j143211.6 + 362225 & 0.845 & 218.041 & 36.361 & 0.572 & & 0.663 + swcl j151550.9 + 442056 & 0.145 & 228.963 & 44.347 & 0.116 & & 0.109 + swcl j152252.9 + 253527 & 0.037 & 230.720 & 25.591 & 0.549 & & 0.546 + swcl j155159.8 + 445748 & 0.050 & 238.000 & 44.963 & 0.201 & 0.208 & 0.213 + swcl j155555.3 + 410548 & 0.046 & 238.980 & 41.097 & 0.355 & 0.375 & 0.368 + swcl j155708.6 + 354100 & 0.054 & 239.285 & 35.683 & 0.413 & 0.390 & 0.429 + swcl j155743.3 + 353020 & 0.245 & 239.427 & 35.508 & 0.148 & 0.169 & 0.166 + swcl j170757.1 + 235135 & 0.036 & 256.989 & 23.859 & 0.221 & & 0.212 + swcl j215423.1 + 000526 & 0.381 & 328.598 & 0.084 & 0.175 & 0.156 & 0.169 + swcl j222432.9@xmath42021216 & 0.378 & 336.131 & @xmath422.203 & 0.242 & & 0.162 + swcl j222438.0@xmath42022231 & 0.164 & 336.158 & @xmath422.372 & 0.507 & & 0.499 + swcl j222444.0@xmath42022034 & 0.061 & 336.184 & @xmath422.343 & 0.658 & & 0.714 + swcl j222954.1 + 194350 & 0.132 & 337.473 & 19.731 & 0.272 & & 0.287 + swcl j224206.9 + 233408 & 0.391 & 340.522 & 23.571 & 0.432 & & 0.402 + swcl j231257.7 + 182543 & 0.983 & 348.233 & 18.414 & 0.460 & & 0.325 + swcl j231733.7 + 322828 & 0.216 & 349.394 & 32.476 & 0.405 & & 0.400 + swcl j232248.4 + 054810 & 0.039 & 350.701 & 5.802 & 0.244 & & 0.244 + swcl j232717.2 + 263108 & 0.225 & 351.825 & 26.522 & 0.230 & & 0.223 + swcl j002114.5 + 205943 & 0.051 & 5.3095 & 20.9956 & & 0.141 + swcl j002823.6 + 092705 & 0.045 & 7.0987 & 9.4517 & 0.224 & 0.195 + swcl j012303.8 + 375609 & 0.150 & 20.7625 & 37.9361 & & + swcl j015752.9 + 165933 & 0.018 & 29.4699 & 16.9924 & 0.507 & 0.510 + swcl j020003.8 + 084024 & 0.026 & 30.0158 & 8.6730 & 0.215 & 0.196 + swcl j020745.0 + 002053 & 0.098 & 31.9375 & 0.3498 & & + swcl j035259.4@xmath42004338 & 0.033 & 58.2471 & @xmath420.7274 & 0.328 & 0.301 + swcl j075900.8 + 324449 & 0.065 & 119.7527 & 32.7480 & & 0.579 + swcl j084749.4 + 133142 & 0.096 & 131.9544 & 13.5276 & 0.349 & 0.391 + swcl j092649.8 + 301346 & 0.043 & 141.7085 & 30.2296 & 0.559 & 0.567 + swcl j092719.6 + 301348 & 0.086 & 141.8327 & 30.2309 & 0.302 & 0.331 + swcl j092730.1 + 301046 & 0.058 & 141.8744 & 30.1798 & 0.312 & 0.303 + swcl j095206.7 + 102137 & 0.108 & 148.0280 & 10.3622 & & + swcl j095513.4 + 181215 & 0.052 & 148.8058 & 18.2051 & 0.416 & 0.423 + swcl j101341.5 + 430651 & 0.058 & 153.4235 & 43.1152 & 0.449 & 0.400 + swcl j114232.3 + 505623 & 0.037 & 175.6344 & 50.9393 & 0.260 & 0.254 + swcl j114503.1 + 600811 & 0.048 & 176.2613 & 60.1362 & 0.285 & 0.268 + swcl j124312.1 + 170454 & 0.126 & 190.8026 & 17.0812 & 0.142 & 0.136 + swcl j133051.0 + 420641 & 0.150 & 202.7158 & 42.1111 & & 0.588 + swcl j133055.8 + 420015 & 0.076 & 202.7310 & 42.0048 & & 0.113 + swcl j140637.3 + 274348 & 0.049 & 211.6547 & 27.7302 & 0.655 & 0.600 + swcl j140639.0 + 273546 & 0.109 & 211.6611 & 27.5974 & 0.252 & 0.232 + swcl j143211.6 + 362225 & 0.084 & 218.0467 & 36.3740 & 0.572 & 0.663 + swcl j143223.3 + 361752 & 0.028 & 218.0967 & 36.2980 & & + swcl j152252.9 + 253527 & 0.152 & 230.7187 & 25.5926 & 0.557 & 0.546 + swcl j155117.4 + 445118 & 0.083 & 237.8208 & 44.8551 & & 0.697 + swcl j155743.3 + 353020 & 0.082 & 239.4292 & 35.5065 & 0.158 & 0.166 + swcl j164956.4 + 313021 & 0.082 & 252.4857 & 31.5070 & & 0.734 + swcl j173932.8 + 272051 & 0.038 & 264.8860 & 27.3480 & & + swcl j194004.2 + 782415 & 0.051 & 295.0164 & 78.4033 & & + swcl j215507.7 + 164725 & 0.028 & 328.7821 & 16.7907 & & + swcl j222438.0@xmath42022231 & 0.092 & 336.1594 & @xmath422.3742 & 0.507 & 0.499 + swcl j222444.0@xmath42022034 & 0.006 & 336.1832 & @xmath422.3428 & 0.658 & 0.714 + swcl j222516.4@xmath42020825 & 0.068 & 336.3192 & @xmath422.1411 & & 0.467 + swcl j222954.1 + 194350 & 0.159 & 337.4729 & 19.7292 & 0.272 & 0.287 + swcl j232248.4 + 054810 & 0.053 & 350.7014 & 5.8036 & 0.244 & 0.244 + swcl j232725.6 + 263506 & 0.037 & 351.8565 & 26.5843 & 0.059 & 0.257 + swcl j015803.5 + 165006 & 0.223 & 0.232 & 0.223 & 7.42 & 11.92 & 3.43 & 16.95 & @xmath4224.33 & 0.01 & 8.08e+43 & 2.69e+14 + swcl j095513.4 + 181216 & 0.423 & 0.421 & 0.423 & 4.51 & 11.59 & 5.58 & 19.54 & @xmath4223.34 & 0.57 & 1.29e+44 & 3.15e+14 + swcl j095515.7 + 180359 & 0.745 & 0.747 & 0.745 & 8.35 & 11.14 & 160 & 20.37 & @xmath4225.03 & 0.21 & 3.34e+44 & 4.57e+14 + swcl j083340.9 + 331117 & 0.682 & 0.676 & 0.682 & 8.42 & 18.06 & 112 & 20.66 & @xmath4224.36 & 0.22 & 5.38e+44 & 6.43e+14 + swcl j124313.2 + 170443 & 0.132 & 0.134 & 0.132 & 6.28 & 10.16 & 3.31 & 15.80 & @xmath4224.02 & 0.03 & 6.49e+43 & 2.48e+14 + swcl j224206.2 + 233350 & 0.403 & & 0.403 & 5.69 & 15.04 & 7.18 & 19.40 & @xmath4223.76 & 0.15 & 3.56e+44 & 6.01e+14 + swcl j115811.3 + 452904 & 0.389 & 0.407 & 0.389 & 5.90 & 17.08 & 6.58 & 18.39 & @xmath4224.48 & 0.22 & 1.08e+44 & 2.91e+14 + swcl j160637.7 + 321349 & 0.065 & 0.064 & 0.065 & 5.22 & 7.25 & 4.23 & 16.22 & @xmath4222.02 & 0.14 & 7.45e+43 & 2.80e+14 + swcl j082114.0 + 320018 & 0.746 & 0.732 & 0.746 & 7.39 & 9.08 & 142 & 21.16 & @xmath4224.28 & 0.08 & 9.15e+44 & 8.55e+14 + swcl j232248.3 + 054810 & 0.319 & 0.245 & 0.245 & 5.63 & 9.50 & 22.8 & 16.85 & @xmath4224.56 & 0.01 & 1.78e+44 & 4.33e+14 + swcl j222954.1 + 194349 & 0.286 & 0.291 & 0.286 & 8.18 & 16.66 & 5.03 & 18.00 & @xmath4223.63 & 0.04 & 1.13e+44 & 3.19e+14 + swcl j143211.7 + 362226 & 0.663 & 0.623 & 0.663 & 4.66 & 10.36 & 39.0 & 21.34 & @xmath4223.47 & 0.10 & 1.79e+44 & 3.29e+14 + sdss identification of _ swift _ clusters in the grb060204b field . ( bottom left ) the x - ray image with extended sources a@xmath42d indicated by the blue circles . ( other panels ) photometric redshift distributions ( red dashed line ) of galaxies within 2 of source center are shown for the four labeled sources . also shown are the averaged distribution of random positions of field galaxies ( solid black line ) along with 3@xmath21 above the average ( dotted black line ) . the black squares indicate any sdss spectoscopic redshifts for galaxies within the 2 region . the blue vertical lines indicate the confirmed redshifts . in this field , a is confirmed at @xmath27 = 0.174 at a significance of 4.46@xmath21 , b is confirmed at @xmath27 = 0.600 ( 5.34@xmath21 ) , c is confirmed at @xmath27 = 0.232 ( 5.95@xmath21 ) , and d has no sdss confirmation.,scaledwidth=80.0% ] sdss identification of _ swift _ clusters in the grb061110a field . ( bottom left ) the x - ray image with extended sources a@xmath42e indicated by the blue circles . ( other panels ) photometric redshift distributions ( red dashed line ) of galaxies within 2 ( a , d , e ) and within 3 ( b , c ) of source center are shown for the five labeled sources . also shown are the averaged distribution of random positions of field galaxies ( solid black line ) along with 3@xmath21 above the average ( dotted black line ) . the blue vertical lines indicate the confirmed redshifts . in this field , a is confirmed at @xmath27 = 0.467 at a signifcance of 3.56@xmath21 , b is confirmed at @xmath27 = 0.292 ( 5.16@xmath21 ) , c is confirmed at @xmath27 = 0.162 ( 4.00@xmath21 ) , d is confirmed at @xmath27 = 0.714 ( 3.84@xmath21 ) , and e is confirmed at @xmath27 = 0.499 ( 6.44@xmath21).,scaledwidth=100.0% ] offsets between bcgs and the x - ray source centers for the 104 confirmed clusters ( red , solid line ) . for comparison , the black , dashed line is the result from @xcite , normalized to our study.,scaledwidth=50.0% ] an example of nearby contamination . left : the _ swift _ x - ray image . swcl j232725.6 ( upper source , left middle : 2 detection , right middle : 3 detection ) and swcl j232717.2 ( lower source , right panel , 3 detection ) have centers that are 4.38 apart . both source size regions for swcl j232717.2 detected a galaxy cluster at @xmath139 and the 3 source size region for swcl j232725.6 detected a cluster at @xmath140 . swcl j232725.6 also detected a galaxy cluster at 0.257 via 2 detection . we determine the 3 confirmation for swcl j232725.6 is contamination and that swcl j232725.6 is a cluster at @xmath141 . , scaledwidth=100.0% ] an example of nearby contamination . left : the _ swift _ x - ray image . swcl j085619.7 ( middle panel , left source ) and swcl j085552.1 ( right panel , right source ) have centers that are 4.8 apart . the 2 source size region for swcl j085552.1 and the 2 and 3 source size regions for swcl j085619.7 all have significant over - density peaks at @xmath142 0.48 . the distributions from the 2 regions are shown here . we conclude that swcl j085619.7 and swcl j085552.1 are both clusters . , scaledwidth=80.0% ] distributions of sdss galaxies relative to source regions . each dot is a galaxy and the source regions are indicated by the red , dashed lines ( 3 radius ) and blue , solid lines ( 2 radius ) . the top row ( a and b ) shows examples of cluster candidates with complete source regions in sdss dr8 . the remaining cases show various degrees of incompleteness : c and g are complete for the 2 region and thus are included in the survey . the remaining six are too incomplete and are rejected from the survey , so that the number of cluster candidates in this survey is 203.,scaledwidth=90.0% ] incomplete backgrounds of sdss fields . each dot is a galaxy , the blue , solid circle represents the 3 region and the red , dashed circles encompass the background annulus from 25 to 40 . left : an example of background that is @xmath143% complete . middle , right : examples of background regions that are @xmath144% incomplete.,scaledwidth=100.0% ] observed redshift distribution ( red , solid ) as compared to a model prediction ( dashed , @xcite ) . standard error bars of @xmath145 are used . the dashed and dotted histograms show redshift distributions of matches between our clusters and the @xcite and gmbcg @xcite catalogs.,scaledwidth=70.0% ] x - ray bolometric luminosity versus optical richness . the points are the confirmed @xmath146 clusters and the black lines are the best fit orthogonal regression lines for the clusters with @xmath147 . * the equations for the solid and dashed lines are @xmath91 and @xmath92 , respectively . the former represents the best fit that includes error dependent weights and the latter represents the best fit without any weighting . * the @xmath15 s indicate clusters with @xmath88 , most likely with over - estimated richnesses due to the magnitude limits of sdss and the faintness of distant galaxies . for this reason , these are excluded from the linear fit . , scaledwidth=70.0% ] x - ray bolometric luminosity as a function of bcg absolute magnitude . here we observe a roughly positive trend , with spearman s correlation coefficient and probability of @xmath103 and 0.03 . this indicates a positive correlation ( with large spread ) between gas mass ( from @xmath6 ) and bcg luminosity ( from @xmath102 ) . the red points indicate bcg - to - x - ray - center offsets of @xmath148 mpc and show a higher correlation than offsets of @xmath149 mpc , with @xmath150 and @xmath151 and probabilities of 0.04 and 0.27 , respectively.,scaledwidth=70.0% ] color distributions and color magnitude diagrams . left : color distributions showing clear red sequences ( red gaussian fit ) and field galaxies ( blue gaussian fit ) , in g @xmath42 r for swcl j232717.2 ( top ) at @xmath152 and r @xmath42 i for swcl j222438.0 at @xmath153 ( bottom ) . right : the corresponding cmds with red points indicating galaxies within 2@xmath21 of the red sequence gaussian . the line indicates the best fit to these points with mean @xmath154 and gaussian dispersion @xmath21.,scaledwidth=90.0% ] mean red sequence color as a function of redshift for every galaxy cluster ( red stars ) that has an sdss galaxy over - density and convergent gaussian fit of the red sequence . the points are clusters from the gmbcg catalog @xcite . the green diamonds are the color priors from the gmbcg catalog used in the two gaussian fits . , scaledwidth=80.0% ] cluster mass ( in @xmath155 ) versus redshift . here we look for a mass trend in clusters with detected red sequences in sdss ( red triangles ) versus those without ( black @xmath15 ) . we detect the red sequence in the 6 most massive clusters , although in general , we do not observe a mass trend . , scaledwidth=80.0% ]	we study 203 ( of 442 ) _ swift _ agn and cluster survey extended x - ray sources located in the sdss dr8 footprint to search for galaxy over - densities in three dimensional space using sdss galaxy photometric redshifts and positions near the _ swift _ cluster candidates . we find 104 _ swift _ clusters with a @xmath0 galaxy over - density . the remaining targets are potentially located at higher redshifts and require deeper optical follow - up observations for confirmation as galaxy clusters . we present a series of cluster properties including the redshift , bcg magnitude , bcg - to - x - ray center offset , optical richness , and x - ray luminosity . we also detect red sequences in @xmath1 of the 104 confirmed clusters . the x - ray luminosity and optical richness for the sdss confirmed _ swift _ clusters are correlated and follow previously established relations . the distribution of the separations between the x - ray centroids and the most likely bcg is also consistent with expectation . we compare the observed redshift distribution of the sample with a theoretical model , and find that our sample is complete for @xmath2 and is still 80% complete up to @xmath3 , consistent with the sdss survey depth . these analysis results suggest that our _ swift _ cluster selection algorithm has yielded a statistically well - defined cluster sample for further studying cluster evolution and cosmology . we also match our sdss confirmed _ swift _ clusters to existing cluster catalogs , and find 42 , 23 and 1 matches in optical , x - ray and sz catalogs , respectively , so the majority of these clusters are new detections .
the spread of contagion ( information diffusion or spread of an infection ) is a universal phenomenon that is extensively studied in the context of physical , biological , and social networks . such cascades can have one or multiple sources ( or _ seeds _ ) and spread from infected nodes to neighbors through the link structure . a motivating application for the study of influence is viral marketing strategies @xcite , in which the influence of a set @xmath1 of people in a social network is the number of adoptions triggered if we give @xmath1 free copies of a product . the problem also has important applications beyond social graphs , such as placing sensors in water distribution networks for detecting contamination @xcite . a popular model for information diffusion is _ independent cascade _ ( ic ) , in which an independent random variable is associated with each ( directed ) edge @xmath2 to model the degree of influence of @xmath3 on @xmath4 . a single propagation instance is obtained by instantiating all edge variables . we then study the distribution of a property of interest , such as the number of infected nodes , over these random instances . the simplest and most studied ic model is _ binary ic _ , in which the range of the edge random variables is binary . a biased coin of probability @xmath5 is flipped for each directed edge @xmath2 . accordingly , the edge can be either _ live _ , meaning that once @xmath3 is infected , @xmath4 is also infected , or _ null_. this model was formalized in a seminal work by kempe et al . @xcite and is based on earlier studies by goldenberg et al . note that each direction of an undirected edge @xmath6 may have its own independent random variable , since influence is not necessarily symmetric . a particular propagation instance is specified by the set of live edges , and a node is infected by a seed set @xmath1 in this instance if and only if it is reachable from a seed node . the _ influence _ of @xmath1 is formally defined as the expectation , over instances , of the number of infected nodes . instead of working directly on this probabilistic ic model , kempe et al . @xcite proposed a simulation - based approach , in which a set @xmath7 of propagation instances ( graphs ) is generated in monte carlo fashion according to the influence model . the average influence of @xmath1 on @xmath7 is an unbiased estimate that converges to the expectation on the probabilistic model . the ability to compute influence with respect to an _ arbitrary _ set of propagation instances has significant advantages , as it is useful for instances generated from traces or by more complex models @xcite , which exhibit correlations between edges that can not be captured by the simplified ic model @xcite . moreover , the average behavior of a probabilistic model on a small set of instances captures its `` typical '' behavior , which is often more relevant than the expected value when the variance is very high . a basic primitive in the study of influence are _ influence queries _ : compute ( or approximate ) the influence of a query set @xmath1 of seed nodes . with binary influence , this amounts to performing graph searches from the seed set in multiple instances . unfortunately , this does not scale well when many queries are posed over graphs with millions of nodes . even more computationally challenging is the fundamental _ influence maximization _ problem , which is finding the most potent seed set of a certain size or cost . the problem was formalized by kempe et al . @xcite and inspired by richardson and domingos @xcite . kempe et al . showed that , even when the influence function is deterministic ( but the number @xmath8 of seeds is a parameter ) , the problem encodes the classic max cover problem and therefore is np - hard @xcite . moreover , an inapproximability result of feige @xcite implies that any algorithm that can guarantee a solution that is at least @xmath9 times the optimum is likely to scale poorly with the number of seeds . chen et al . @xcite showed that computing the exact influence of a single seed in the binary ic model , even when edge probabilities are @xmath10 , is # p hard @xcite . using simulations , the objective studied by kempe et al . @xcite is then to find a set @xmath1 of seeds with maximum average influence over a fixed set of propagation instances . a natural heuristic is to use the set of most influential individuals , say those with high degree or centrality @xcite , as seeds . this approach , however , can not account for the dependence between seeds , missing the fact that two important nodes may `` cover '' essentially the same communities . kempe et al . @xcite proposed a greedy algorithm ( greedy ) instead . it starts with an empty seed set and iteratively adds to @xmath1 the node with maximum _ marginal _ gain in influence ( relative to current seed set ) . since our objective is monotone and submodular , a classical result from nemhauser et al . @xcite implies that the influence of the greedy solution with @xmath8 seeds is at least @xmath11 of the best possible for any seed set of the same size . from feige s inapproximability result , this is the best approximation ratio guarantee we can ( asymptotically and realistically ) hope for . greedy has become the gold standard for influence maximization , in terms of the quality of the results . greedy , however , does not scale to modern real - world social networks . the issue is that evaluating the marginal contribution of each node requires a directed reachability computation in each instance ( of which there can be hundreds ) . several performance improvements to greedy have thus been proposed . leskovec et al . @xcite proposed celf , which are `` lazy '' evaluations of the marginal contribution , performed only when a node is a candidate for the highest marginal contribution . chen et al . @xcite took a different approach , using the reachability sketches of cohen @xcite to speed up the reevaluation of the marginal contribution of all nodes . while effective , even with these and other accelerations @xcite , the best current implementations of greedy do not scale to networks beyond @xmath12 edges @xcite , which are quite small by modern standards . to support massive graphs , several studies proposed algorithms specific to the ic model , which work directly with the edge probabilities instead of with simulations and thus can not be reliably applied to a set of arbitrary instances . borg et al . @xcite recently proposed an algorithm based on reverse reachability searches from sampled nodes , similar in spirit to the approach used for reachability sketching @xcite . their algorithm provides theoretical guarantees on the approximation quality and has good asymptotic performance , but large `` constants . '' very recently , tang et . @xcite developed tim , which engineers the ( mostly theoretical ) algorithm of borgs et al . @xcite to obtain a scalable implementation with guarantees . a significant drawback of this approach is that it only works for a pre - specified seed set size @xmath8 , whereas greedy produces a sequence of nodes , with each prefix having an approximation guarantee with respect to the same - size optimum . in applications we are often interested not in a single point , but in a trade - off curve that allows us to find a sweet spot of influence per cost or characterize the network . tim also scales very poorly with the seed set size @xmath8 , and the evaluation only considered seed sets of up to 50 nodes . the degreediscount @xcite heuristic refines the natural approach of adding the next highest degree node . mia @xcite converts the binary ic sampling probabilities @xmath13 to deterministic edge weights and works essentially with one deterministic instance . irie , by jung et al . @xcite , is a heuristic approximation of greedy addition of seed nodes , and has the best performance we are aware of for an algorithm that produces a sequence of seed nodes . in each step , the probability of each node to be covered by the current seed set @xmath1 is estimated using another algorithm ( or simulations ) . they then use eigenvector computations to approximate marginal contributions of all nodes . of those approaches , the irie heuristic scales much better and is much more accurate than other heuristics . in particular , it performs nearly as well as greedy on many research collaboration graphs @xcite . [ [ contributions . ] ] contributions . + + + + + + + + + + + + + + we design a novel sketch - based approach for influence computation which offers scalability with performance guarantees . our main contribution is skim ( sketch - based influence maximization ) , a highly scalable ( approximate ) implementation of the greedy algorithm for influence maximization . we also introduce _ influence oracles _ : after preprocessing that is almost linear , we can answer _ influence queries _ very efficiently , considering only the sketches of the query seed set . we can apply our design on inputs specified as a fixed set of propagation instances , as in kempe et al . @xcite , with influence defined as the average over them . we also handle inputs specified as an ic model , where influence is defined as the expectation . our model is defined precisely in section [ model : sec ] . we now provide more details on our design . the exact computation of an influence query requires expensive graph searches from the query seed set @xmath1 on each of @xmath14 instances . the exact greedy algorithm for influence maximization requires a similar computation for each marginal contribution . we address this scalability issue by working with sketches . the core of our approach are per - node summary structures which we call _ combined reachability sketches_. the sketch of a node compactly represents its influence `` coverage '' across @xmath14 instances ; we call this its _ combined reachability set_. the combined reachability sketch of a node , precisely defined in section [ sketch : sec ] , is the bottom-@xmath15 min - hash sketch @xcite of the combined reachability set of the node . this generalizes the reachability sketches of cohen @xcite , which are defined for a single instance . the parameter @xmath15 is a small constant that determines the tradeoff between computation and accuracy . bottom-@xmath15 sketches of sets support cardinality estimation , which means that we can estimate the influence ( over all instances ) of a node or of a set of nodes from their combined reachability sketches . the estimate has a small relative error and good concentration @xcite . our use of combination sketches and state - of - the - art optimal estimators is key to obtaining the best balance between sketch size and accuracy . our skim algorithm for influence maximization is presented in section [ binaryim : sec ] . it scales by running the greedy algorithm in `` sketch space , '' always taking a node with the maximum _ estimated _ ( rather than exact ) marginal contribution . skim computes combined reachability sketches , but only until the node with the maximum estimated influence is computed . this node is then added to the seed set . we then update the sketches to be with respect to a _ residual _ problem in which the node that is selected into the seed set and its `` influence '' are no longer present . skim then resumes the sketch computation , starting with the residual sketches , but ( again ) stopping when a node with maximum estimated influence ( in the current , residual , instance ) is found . a new residual problem is then computed . this process is iterated until the seed set reaches the desired size . since the residual problem becomes smaller with iterations , we can compute a very large seed set very efficiently . we also prove that the total overhead of the updates required to maintain the residual sketches is small . in particular , for a set @xmath16 of @xmath14 _ arbitrary _ instances , the algorithm can be run to exhaustion , producing a full permutation of the nodes in @xmath17 } \|g^{(i)}\|+m \epsilon^{-2}\log^2 n)$ ] time , where @xmath18 is the sum over nodes of the maximum indegree ( over instances ) . for all @xmath19 , the first @xmath8 nodes we select have with a very high probability ( at least @xmath20 for a constant @xmath21 ) influence that is at least @xmath22 times the maximum influence of a seed set of the same size @xmath8 . these are worst - case bounds . we propose an adaptive approach that exploits properties of actual networks , in particular a skewed influence distribution , to achieve faster running times with the same guarantees . our use of the residual instances by skim is the key for maintaining the accuracy of the greedy selection through the execution and providing with high probability , approximation ratio guarantees that nearly match those of exact greedy . section [ binaryq : sec ] presents our influence oracles , which preprocess the input to compute combined reachability sketches for all nodes . for instances @xmath7 with @xmath23 nodes and @xmath24 edges , the sketches are built in @xmath25 total time . the influence of a set @xmath26 can then be approximated from the sketches of the nodes in @xmath1 . the oracle applies the union cardinality estimator of cohen and kaplan @xcite to estimate the union of the influence sets of the seed nodes . the query runs in time @xmath27 and unbiasedly with a well - concentrated relative error of @xmath28 . while preprocessing depends on the number of instances , the sketch size and the approximation quality only depend on the sketch parameter @xmath15 . the asymptotic bounds we obtain are novel also from a theoretical perspective , and significantly improve the state of the art , even for influence maximization on a single ( deterministic ) instance ( select a seed set in a directed graph with maximum reachable set ) . section [ experiments : sec ] presents an extensive experimental study . besides demonstrating the scalability of our algorithms on real - world networks , we compare skim with existing approaches , including exact greedy ( when size allows ) , the state - of - the - art irie heuristic , and tim . we obtain ic models from networks by using the well - studied weighted and uniform @xcite probabilities . our algorithms scale up to very large graphs with barely any compromise on quality over exact greedy , with theoretical guarantees . on instances generated by an ic model , we achieve more than an order of magnitude speedup over the best greedy heuristics , which are designed specifically for this model . even for a fixed small seed set size , skim is significantly faster than tim . moreover , our algorithm is efficient and accurate enough to be executed exhaustively , producing a full permutation of the nodes for networks with billions of edges . for the first time , we provide the full ( approximate ) pareto front of influence versus seed set size . these relations showcase a basic property of the network , and the general pattern that a small fraction of nodes influences a large fraction of the network . in contrast , most previous studies we are aware of only considered seed sets with at most 50 nodes , revealing only a very restricted view of this relation . a _ propagation instance _ @xmath29 is specified by the edge set @xmath30 . the _ influence _ of a set of nodes @xmath1 in instance @xmath31 is the number of nodes reachable from @xmath1 using the edges @xmath30 : @xmath32 where the predicate @xmath33 holds if @xmath34 or if there is a forward path from a node in @xmath1 to the node @xmath3 . our input is specified as a set @xmath35 of @xmath36 propagation instances @xmath37 on the same set of nodes . the influence of @xmath1 over all instances @xmath7 is the average single - instance influence : @xmath38 } { \mathop{\sf inf}}({g^{(i)}},s).\ ] ] the set of propagation instances can be derived from cascade traces or generated by a probabilistic model . the input can also be specified as a probabilistic model , such as independent cascade ( ic ) @xcite , which defines a distribution @xmath39 over instances @xmath40 that share a set @xmath41 of nodes . in this case , the influence of @xmath39 is defined as the expectation @xmath42 we are interested in _ influence oracles _ and in _ influence maximization_. influence queries are specified by a seed set @xmath43 and the goal is to compute ( or estimate ) the influence @xmath44 . influence oracles , after efficient preprocessing of the input , allow us to support very fast queries . influence maximization is the problem of finding a seed set @xmath45 with maximum influence , where @xmath46 is given . we are interested in efficiently computing a seed set whose influence is close to the maximum one , as well as in computing a sequence of seeds so that each prefix has influence that is close to maximum for its size . at the heart of our approach are _ combined reachability sketches _ , which are summary structures @xmath47 that we associate with each node @xmath3 . the combined sketches can be defined with respect either to a set @xmath35 of @xmath36 instances or to a probabilistic model @xmath39 . we first consider as input a set of @xmath36 instances . we define the _ reachability set _ of a node @xmath3 in instance @xmath31 as @xmath48 where @xmath49 means that @xmath4 is reachable from @xmath3 in @xmath31 . considering all instances , the _ combined reachability set _ is a set of node - instance pairs : @xmath50 the influence of a set of nodes @xmath1 on instances @xmath7 can thus be expressed as @xmath51 } \bigl\| \bigcup_{u\in s } r(g^{(i)},u ) \bigr\| = \frac{1}{\ell } \bigl\| \bigcup_{u\in s } r_u \bigr\|.\ ] ] this is the average over the instances @xmath7 ( with @xmath52 $ ] ) of the number of nodes reachable from at least one node in @xmath1 . the combined reachability sketch of a node captures its reachability information _ across _ instances . the sketches we use are the bottom-@xmath15 min - hash sketches @xcite @xmath53 of the combined reachability sets @xmath54 : we associate with each node - instance pair @xmath55 an independent random rank value @xmath56 $ ] , where @xmath57 $ ] is the uniform distribution on @xmath58 $ ] . the _ combined reachability sketch _ of @xmath3 is the set of the @xmath15 smallest rank values amongst @xmath59 : @xmath60 where bottom-@xmath15 of a set is its subset consisting of the @xmath15 smallest values . when there is a single instance ( @xmath61 ) the combined reachability sketches are the same as the reachability sketches of cohen @xcite . we define the _ threshold rank _ @xmath62 of each node @xmath3 as @xmath63 which is the @xmath15th lowest rank value in @xmath64 . ( for a set @xmath65 of cardinality @xmath66 , we define @xmath67 . ) therefore , when @xmath68 we have @xmath69 , and @xmath70 otherwise . the cardinality @xmath71 can be estimated from @xmath47 using a bottom-@xmath15 cardinality estimator . the estimate is @xmath72 if @xmath70 ( i.e. , if @xmath73 ) and is @xmath74 otherwise . this estimate has a coefficient of variation ( cv ) , which is the ratio of the standard deviation to the mean , that is never more than @xmath75 and is well concentrated @xcite . by applying chernoff bounds with @xmath76 , we obtain that using @xmath77 , the probability of having relative error larger than @xmath78 is at most @xmath79 . therefore , we can be correct with high probability on estimating the influence of all nodes . [ structranks : sec ] instead of using ranks drawn from @xmath57 $ ] , we can work with integral permutation ranks with respect to a permutation on the @xmath80 node - instance pairs . we can also structure the permutation so that each sequence in positions @xmath81 to @xmath82 for integral @xmath83 has each node appear in exactly one pair . the associated instance with a node @xmath4 in chunk @xmath84 is randomly selected from instances @xmath85 for which the pair @xmath86 does not have a permutation rank of @xmath87 or less ( independently for each node ) . one can show that this can only improve estimation accuracy @xcite . only the first @xmath88 positions can be included in combined reachability sketches of nodes . when estimating influence , we can convert permutation ranks to random ranks using the exponential distribution @xcite . we can also estimate cardinality of a subset of the @xmath89 elements directly from permutation ranks @xmath90 $ ] , using the unbiased estimator @xmath91 , where the threshold @xmath92 is the @xmath15th smallest permutation rank . this estimator can be interpreted as setting aside the element with permutation rank @xmath92 , and estimating the fraction ( of the other @xmath93 elements ) that is in our set by the fraction of such elements with rank smaller than @xmath92 , which is @xmath94 . we now define sketches with respect to a binary ic model @xmath39 , presented as a graph with probabilities @xmath13 associated with its edges . the influence of a set of nodes @xmath1 is @xmath95 the sketches we define for @xmath39 also contain at most @xmath15 rank values , but provide approximation guarantees with respect to . the sketches can be interpreted as the sketches computed for @xmath14 instances generated according to the model @xmath40 as @xmath96 . when doing so , at the limit , each unique rank value corresponds to a unique instance , so we do not need to explicitly represent `` instances . '' we work with structured permutation ranks ( section [ permranks : sec ] ) . since it suffices to consider the first @xmath97 ranks , this conveniently removes the dependence of the rank representation on @xmath14 . we can similarly apply an estimator to the @xmath15th smallest rank @xmath98 to estimate influence : instead of estimating cardinality ( which goes to infinity with @xmath14 ) and dividing by @xmath14 using the estimator @xmath99 we take the limit as @xmath96 and estimate influence using @xmath100 . in this section we present our sketch - based influence maximization ( skim ) algorithm . we first review greedy , the greedy algorithm for influence maximization ( working with @xmath14 instances ) presented by kempe et al . greedy is applied with respect to the influence objective @xmath44 , as defined in equation . it starts with an empty seed set @xmath101 . in each iteration , it adds to @xmath1 the node @xmath4 with maximum _ marginal gain _ , @xmath102 this is the same as choosing @xmath4 maximizing @xmath103 . skim approximates exact greedy by ensuring that at _ each iteration _ , with sufficiently high probability , or in expectation over iterations , the node we choose to add to the seed set has a marginal gain that is close to the maximum one . to do so , it suffices to compute sketches only to the point that the node with the maximum estimated marginal gain is revealed . to maintain accuracy , we maintain a residual problem and respective sketches . skim constructs ( partial ) combined reachability sketches by adapting a construction of reachability sketches @xcite : it processes node - instance pairs @xmath104 by increasing rank , performing a reverse reachability search in @xmath105 from @xmath3 . the sketch @xmath53 of each visited node @xmath4 is augmented with the rank @xmath106 of the pair . for a given value of @xmath15 , the first node @xmath3 whose sketch reaches size @xmath15 is also the node with maximum estimated influence . this is because the bottom-@xmath15 cardinality estimate of a node depends only on the @xmath15th smallest rank in @xmath47 , @xmath62 ( which is a complete sufficient statistic for cardinality estimation from the sketch @xcite ) ; see equation . for the node @xmath3 , @xmath62 is equal to the rank @xmath106 of the last processed pair @xmath104 . for other nodes @xmath4 with incomplete sketches , we know that @xmath107 , so their estimate is lower . sketch building is suspended once the node @xmath4 with maximum estimated influence is found . skimthen adds @xmath4 to the seed set and generates a _ problem , with @xmath4 and all node - instance pairs it covers removed from the instances @xmath39 . the ( partially computed ) sketches of each remaining node @xmath3 are updated using @xmath108 , which deletes from the sketch the ranks of all covered node - instance pairs . the process of building sketches is then resumed on the residual problem , working with updated partial sketches and instances . we continue processing node - instance pairs in increasing rank order , starting from the first rank that exceeds @xmath109 and skipping pairs that are already covered . @xmath110 hash map of node - instance pairs to nodes @xmath111 @xmath112 shuffle the @xmath80 node - instance pairs @xmath104 we provide pseudocode for skim as algorithm [ greedyssc : alg ] . instead of maintaining the actual partial sketches @xmath53 , the algorithm only keeps their cardinalities @xmath113 $ ] . to support correct and efficient updates of the sketches , we maintain an inverted index @xmath114 $ ] that lists , for each rank value @xmath106 we processed , all nodes @xmath4 such that @xmath115 . the entry for rank @xmath106 is created and populated when we perform a reverse reachability search from pair @xmath104 . the algorithm outputs the list of pairs @xmath116 , where @xmath117 is a permutation of the nodes according to the order they are selected into the seed set , and @xmath118 is the marginal influence of @xmath119 . the surprising property of our construction is that this whole iterative process is very efficient . if we run skim with a fixed @xmath120 , section [ sec : analysis ] will show that we obtain the following worst - case performance guarantees : [ skimtime : thm ] skim runs in time @xmath121 , where @xmath122 . the permutation @xmath117 of nodes has the property that with probability @xmath123 , for all @xmath124 $ ] , the set of seed nodes @xmath125 , has @xmath126 . it is not hard to show that the influence of a node @xmath4 in the residual problem of iteration @xmath84 is equal to its marginal influence with respect to @xmath127 in the original problem . therefore , @xmath118 , which is the influence of @xmath119 in the residual problem of iteration @xmath84 , is the marginal influence of @xmath119 , with respect to @xmath127 in the original problem . thus , by definition , for all @xmath124 $ ] and @xmath125 , @xmath128 } i_i$ ] . we also show that the partial sketches correctly capture a component of the sketches computed for the residual problem : at the end of an iteration selecting @xmath4 , each updated partial sketch @xmath47 is equal to the set of entries of the combined reachability sketch @xmath129 of @xmath3 in the residual problem that have rank value at most @xmath109 . the content of each sketch @xmath47 before computing the residual is clearly a superset of all reachable node - instance pairs @xmath130 with rank @xmath131 in the residual problem . we can then verify that entries are removed from @xmath47 only and for all covered node - instance pairs with @xmath131 . we now analyze the running time of skim . all updates of the residual problem together take time linear in the size of @xmath7 , since nodes and edges that are covered by the current seed set are removed once visited and never considered again . the remaining component of the computation is determined by the number of times ranks are inserted ( and removed ) from sketches . inserting a value to @xmath47 involves a scan of all ( remaining ) incoming edges to @xmath3 in an instance . removals of ranks can be charged to insertions . so we need to bound the total number of rank insertions : the expected total number of rank insertions at a particular node is @xmath132 . consider a sketch @xmath53 . we can show , viewing the sketches as uniform samples of reaching pairs , that each rank value removal corresponds to cardinality and hence influence ( marginal gain)being reduced in expectation by a factor of @xmath133 . the initial influence is at most @xmath23 , so there are at most @xmath134 insertions until the marginal influence is reduced below @xmath135 , at which point we do not need to consider the node . the running time is dominated by the sum over nodes @xmath4 , of the number of times a rank is inserted to the sketch of @xmath4 , times the in - degree of @xmath4 ( the maximum over instances ) . from the lemma , we obtain a bound of @xmath136 on the total number of insertions . thus , we obtain a bound of @xmath137 on the running time of the algorithm . to obtain an approximation that is within @xmath138 with good probability , we can choose a fixed @xmath139 , for some constant @xmath21 . the relative error of each influence estimate of a node in an iteration is at most @xmath78 with probability of at least @xmath20 . since we use polynomially many estimates ( maximize influence among @xmath23 nodes in each of at most @xmath23 iterations ) , all estimates are within a relative error of @xmath78 with probability that is polynomially close to @xmath140 . lastly , we bound the approximation ratio of the `` approximate '' greedy algorithm we work with , which uses seeds with close to maximum instead of maximum marginal gain : [ approxgreedy : lemma ] with any submodular and monotone objective function , approximate greedy , which iteratively chooses a node with marginal gain that is at least @xmath141 of the maximum , has an approximation ratio of at least @xmath142 . the same claim holds in expectation when the selection is well concentrated , that is , its probability of being below @xmath143 times the maximum decreases exponentially with @xmath144 . the argument extends the analysis of exact greedy by nemhauser et al . @xcite . for any @xmath8 , and after selecting any set @xmath145 of seeds , the maximum marginal gain by adding a single node is always at least @xmath146 of the maximum possible gain for @xmath8 nodes . when using the approximation , this is at least @xmath147 of the maximum possible gain . therefore , after approximate greedy selection of @xmath8 nodes , the influence is at least @xmath148 using the first order term of the taylor expansion . this worst - case analysis is too pessimistic , both for the approximation ratio and running time . in our experiments , we tested skim with a fixed @xmath15 , and observed that the computed seed sets had influence that is much closer to the exact greedy selection than indicated by the worst - case bounds . the explanation is that the influence distribution on real inputs is heavy - tailed , with the vast majority of nodes having a much smaller influence than the one of maximum influence . one factor of @xmath149 in the worst - case running time is due to a `` union bound '' ensuring a relative error of @xmath78 for all nodes in all iterations , with high probability . with a heavy tail distribution , we can identify the maximum with a small error if we ensure a small error only on the few nodes that have influence close to the maximum . furthermore , when the maximum influence is separated out from other influence values , our approximate maximum is more likely to be the node with actual maximum influence . moreover , the estimation error over iterations averages out , so as the seed set gets larger we can work with lower accuracy and still guarantee good approximation . we propose incorporating error estimation that is _ adaptive _ rather than worst - case . this facilitates tighter confidence bounds on the estimation quality of our output . it also allows us to adjust the sketch parameter @xmath15 during computation in order to meet pre - specified accuracy and confidence levels . let the _ discrepancy _ in an iteration be the gap between the actual maximum and the marginal influence of the selected seed . we will bound the sum of discrepancies across iterations by maintaining a confidence distribution on this sum . the estimation uses two components . ( i ) the exact marginal influence @xmath150 of the selected node in each iteration , as well as the sum @xmath151 , which is the influence of our seed set . the value @xmath150 is computed when generating the residual problem . ( ii ) noting in each iteration the size of the second largest sketch ( excluding the last processed rank ) . intuitively , if the second largest sketch is much smaller than the first one , it is more likely that the first one is the actual maximum . we bound the discrepancy in a single iteration using chernoff bounds . the probability that the sum of independent bernoulli trials falls below its expectation @xmath152 by more than @xmath153 is @xmath154 < \left ( \frac{\exp(-\nu)}{(1-\nu)^{(1-\nu ) } } \right)^\mu.\ ] ] we use this to bound the probability that the discrepancy exceeds @xmath155 , where @xmath156 is the exact marginal gain of our selected seed node . we consider the second largest sketch size , @xmath157 ( the last rank of @xmath158 is not considered part of the sketch even if included ) . we use @xmath159 , @xmath160 , and @xmath161 in equation to obtain a confidence level . finally , to maintain an upper bound on the confidence - error distribution of the sum of discrepancies , we take a convolution , after each iteration , of the current distribution with the distribution of the current iteration . skim can be adapted for higher concurrency by running the sketch - building phases in batches of ranks . we can also adapt it to process inputs presented as an ic model instead of as a set of instances . this yields a more efficient implementation than when generating a set of instances using simulations and running skim on them . in ic - model skim , the residual problem is a collection of partial models and sketch building is performed on the probabilistic model . we omit details due to space limitations . we now present an accurate and efficient oracle for binary influence , which is based on precomputing a combined reachability sketch ( as defined in section [ sketch : sec ] ) for each node . we preprocess a set of @xmath14 instances @xmath35 using @xmath162 computation and working storage of @xmath163 per node . the preprocessing generates combined reachability sketches @xmath53 of size @xmath163 for each node @xmath164 . [ binaryinforacles : thm ] given a set @xmath165 of combined reachability sketches for @xmath39 with parameter @xmath15 , influence queries @xmath44 for a set @xmath1 of nodes can be estimated in @xmath166 time from the sketches @xmath167 . the estimate is nonnegative and unbiased , has cv at least @xmath75 , and is well concentrated , meaning that the probability that the relative error exceeds @xmath168 decreases exponentially with @xmath144 . we next present the two components of our oracle : estimating the influence of @xmath1 from the sketches of the nodes in @xmath1 and efficiently computing all combined reachability sketches . we show how to use the combined reachability sketches of a set of nodes @xmath1 to estimate the influence of @xmath1 , as given in equation . in graph terms , this means estimating the cardinality of the union @xmath169 from the sketches @xmath47 , with @xmath34 . the influence @xmath44 is the union cardinality divided by the number of instances @xmath14 and , accordingly , is estimated using @xmath170 . our estimators use the threshold rank @xmath62 of each node @xmath3 ; see equation . from the bottom-@xmath15 sketches of each set @xmath64 for @xmath34 we can unbiasedly estimate the cardinality of the union @xmath169 . one way to do this is to compute the bottom-@xmath15 sketch of the union @xcite , which has threshold value @xmath171 and apply the cardinality estimator @xmath172 . this would already conclude the proof of theorem [ binaryinforacles : thm ] . in our implementation , we use a strictly better union cardinality estimator that uses all the ( at most @xmath173 ) values in the set of sketches instead of just the @xmath15th smallest : @xmath174 this estimator , proposed by cohen and kaplan @xcite , can be computed from the @xmath175 sketches in time @xmath166 , by first sorting the @xmath175 sketches by decreasing threshold , and then identifying for each distinct rank value the threshold of the first sketch that contains it . when the sets @xmath64 are all the same , the estimate is the same as applying an estimator to the bottom-@xmath15 sketch on the union , but equation can have up to a factor of @xmath176 lower cv when the sets @xmath64 are sufficiently disjoint . moreover , this estimator is an optimal sum estimator in that it minimizes variance given the information available in the sketches . we can also derive a permutation version of equation . the simplest way is to treat the permutation rank @xmath92 as a uniform rank @xmath177 which is the probability that the rank of another node is smaller than @xmath92 . when there is a single instance @xmath29 , the combined sketches are simply reachability sketches @xcite . reachability sketches @xmath178 for all nodes can be computed very efficiently , using at most @xmath179 edge traversals in total , where @xmath18 is the number of edges @xcite . shuffle the @xmath80 node - instance pairs @xmath104 algorithm [ reachsketch1:alg ] computes combined sketches by applying the pruned searches algorithm of cohen @xcite on each instance @xmath180 , obtaining a sketch @xmath181 for each node , and combining the results . the combined sketch @xmath53 is obtained by taking the bottom-@xmath15 values in the union of the @xmath14 sketches , defined as @xmath182 the algorithm runs in @xmath183 time . rather than storing all sets of sketches , we can compute and merge concurrently or sequentially , but after each step , take the bottom-@xmath15 values in the current bottom-@xmath15 set and the newly computed sketch for instance @xmath105 : @xmath184 . therefore , the additional run time storage requirement for sketches is @xmath185 . this gives us the worst - case bounds on the computation stated in theorem [ binaryinforacles : thm ] . we implemented our algorithms in c++ using visual studio 2013 with full optimization . all experiments were run on a machine with two intel xeon e5 - 2690 cpus and 384gib of ddr3 - 1066 ram , running windows 2008r2 server . each cpu has 8 cores ( 2.90ghz , 8@xmath18664kib l1 , 8 @xmath186256kib , and 20mib l3 cache ) , but all runs are sequential for consistency . we ran our experiments on benchmark networks available as part of the snap @xcite and webgraph @xcite projects . more specifically , we test _ social _ ( , , , , , , , ) , _ collaboration _ ( ) , and web ( , ) networks . is obtained from by reversing all arcs ( influence follows the reverse direction of links ) . kempe et al . @xcite proposed two natural ways of associating probabilities with edges in the binary ic model : the _ uniform _ scheme assigns a constant probability @xmath187 to each directed edge ( they used @xmath188 and @xmath189 ) , whereas in the _ weighted cascade _ ( wc ) scheme the probability is the inverse of the degree of the head node ( making the probability that a node is influenced less dependent on its number of neighbors ) . we consider the wc scheme by default , but we will also experiment with the uniform scheme ( un ) . these two schemes are the most commonly tested in previous studies of scalability @xcite . [ cols= " < , > , > , > , > , > , > , > , > , > , > , > " , ] figure [ fig : oracleerror ] shows in detail how the error of the estimator ( @xmath190 axis ) decreases when the seed set size increases ( @xmath191 axis ) . to better evaluate the performance of estimating the _ union _ of several reachability sets , we use the following _ neighborhood generator _ for queries : for each query , it first picks a node @xmath3 at random with probability proportional to its degree . from @xmath3 it exhaustively grows a bfs of the smallest depth @xmath192 such that the tree contains at least @xmath8 nodes . the nodes for the seed set are then uniformly sampled from this tree . with this generator , we expect the reachability sets of seed nodes to highly overlap . looking at the figure , we observe that the estimation error of our oracle decreases rapidly for increasing @xmath8 . also , running queries from the neighborhood generator ( right ) compared to the uniform one ( left ) , has almost no effect on the estimation error ; for 50 seed nodes it is even better on many instances . , title="fig:"],title="fig : " ] finally , figure [ fig : ell ] ( right ) reports the performance of the oracle for fixed instances on the general ic model . we vary the number @xmath14 of instances generated by simulations when building the oracle , but compute the error on a different set of 8192 instances . since our oracle implementation is optimized for fixed instances , we see a higher error with @xmath193 . we can also see that the error decreases with the number of simulations . we conclude that for an ic model oracle , it is beneficial to construct sketches that have approximation guarantees with respect to the ic model itself ( cf . section [ icsketches : sec ] ) rather than work with simulations . we presented highly scalable algorithms for binary influence computation . skim is a sketch - space implementation of the greedy influence maximization algorithm that scales it by several orders of magnitude , to graphs with billions of edges . skim computes a sequence of nodes such that each prefix has a probabilistic guarantee on approximation quality that is close to that of greedy . we also presented sketch - based influence oracles , which after a near - linear processing of the instances can estimate influence queries in time proportional to the number of seeds . our experimental study focused on instances generated by an ic model , since the fastest algorithms we compared with only apply in this model . our experiments revealed that skim is accurate and faster than other algorithms by one to two order of magnitude . in future work , we plan to develop a skim - like algorithm for _ timed influence _ , where edges have lengths that are interpreted as transition times and we consider both the speed and scope of infection @xcite . we also plan to use sketches to efficiently estimate the jaccard similarity of the influence sets of two nodes , which we believe to be an effective similarity measure @xcite .	propagation of contagion through networks is a fundamental process . it is used to model the spread of information , influence , or a viral infection . diffusion patterns can be specified by a probabilistic model , such as independent cascade ( ic ) , or captured by a set of representative traces . basic computational problems in the study of diffusion are _ influence queries _ ( determining the potency of a specified _ seed set _ of nodes ) and _ influence maximization _ ( identifying the most influential seed set of a given size ) . answering each influence query involves many edge traversals , and does not scale when there are many queries on very large graphs . the gold standard for influence maximization is the greedy algorithm , which iteratively adds to the seed set a node maximizing the marginal gain in influence . greedy has a guaranteed approximation ratio of at least @xmath0 and actually produces a sequence of nodes , with each prefix having approximation guarantee with respect to the same - size optimum . since greedy does not scale well beyond a few million edges , for larger inputs one must currently use either heuristics or alternative algorithms designed for a pre - specified small seed set size . we develop a novel sketch - based design for influence computation . our greedy sketch - based influence maximization ( skim ) algorithm scales to graphs with billions of edges , with one to two orders of magnitude speedup over the best greedy methods . it still has a guaranteed approximation ratio , and in practice its quality nearly matches that of exact greedy . we also present _ influence oracles _ , which use linear - time preprocessing to generate a small sketch for each node , allowing the influence of any seed set to be quickly answered from the sketches of its nodes .
zheng conducted different numerical experiments which led to estimates of exponents @xmath8 , @xmath9 ( where @xmath10 is the so - called fisher - huse exponent ) , and @xmath11 . from all direct measurements of @xmath8 he concluded that @xmath3 . using this value , he found an estimate of @xmath10 in agreement with the ncop value ( @xmath12 whereas @xmath13 ) . thus , the only strong departure from the ncop values is for exponent @xmath8 . therefore , in the following , we focus on growth law of @xmath5 in large systems at late times ( as opposed to the early - time approach favored by zheng ) . in order to reach late times satisfactorily , we need a better control of the conservation of energy than with the simple second - order scheme used by zheng . the following results were obtained with a third - order bilateral symplectic algorithm @xcite with a timestep @xmath14 . the conservation of initial energy is better than @xmath15 in relative value in all runs presented . to investigate phase - ordering , we use the same initial conditions as zheng ( @xmath16 where the sign is random and @xmath17 calculated to yield the desired energy density @xmath18 ) . we present results for two sets of parameter values , @xmath19 ( set a , used by zheng ) , and @xmath20 ( set b , used in @xcite ) . the initial energy density ( @xmath21 for set a , and @xmath22 for set b ) , was chosen very close to its minimum value allowed by the random - sign initial conditions ( @xmath23 for set a , and @xmath24 for set b ) . this ensures that `` thermal '' fluctuations are minimized , since the energy density is then as far as possible from the critical energy density @xmath25 ( @xmath26 for set a , and @xmath27 for set b ) . the typical domain size @xmath5 was determined by the mid - height value of @xmath28 , the normalized two - point correlation function calculated for simplicity along the principal axes of the lattice using the continuous field @xmath29 or the reduced `` spin '' variable @xmath30 : @xmath31 . ( no significant difference was found between the two cases , and only results using @xmath32 are shown below . ) we first checked that dynamical scaling holds by observing the collapse of @xmath33 curves at different times after some transient ( not shown ) @xcite . for both sets of parameters , the expected ncop law ( @xmath34 , i.e. @xmath35 ) is reached at late times , but a rather long transient is observed , especially for set a ( fig . [ f1]ab ) . the same data plotted in logarithmic scales is thus misleading . if the data for set b reach the `` normal '' scaling ( see fig . [ f1]d and its inset ) at late times , the corresponding plot for parameter set a ( fig . [ f1]c ) seems to indicate a value of @xmath4 between @xmath36 and @xmath37 ( typical of the value estimated by zheng ) if one ignores the systematic trend upward of the local exponent ( see inset of fig . [ f1]c ) @xcite . following zheng , we also measured the decay of the autocorrelation function @xmath38 , but , again , for rather large systems and late times , choosing , in particular , initial reference times @xmath39 larger than the `` microscopic '' transient time . in the usual ncop phase ordering framework , we expect @xmath40^{\lambda } \sim \left(\frac{t_0}{t}\right)^{\lambda / z}\ ] ] where the fisher - huse exponent @xmath41 . again , plotting @xmath42 vs @xmath43 for parameter set a may yields an `` effective '' exponent @xmath9 smaller than its ncop value ( and close to the value given by zheng ) , but a closer look reveals a systematic increase of the instantaneous exponent ( fig . [ f2]a and inset ) . however , plotting @xmath44 vs @xmath5 , the expected scaling is observed ( fig . considering now parameter set b , the expected scaling laws are observed rather easily ( fig . [ f2]b , d ) . this confirms further that for the parameter values chosen by zheng , the onset of the asymptotic scaling regime is delayed . for the sake of completeness , and in order to probe the validity of the early - time scaling approach taken by zheng , we also performed numerical simulations to measure the short - time growth of the squared magnetization @xmath45 ( @xmath46 being defined as the spatial average of either @xmath47 or @xmath48 ) for samples with zero initial magnetization . in this case , for ncop scaling , we expect @xmath49 . our data on logarithmic scales barely reaches the expected behavior at @xmath50 ( fig . [ f3]a , b ) . note that the corrections have different sign for the two parameter sets . on linear scales , however , our data reveals the expected proportionality of @xmath51 and @xmath52 ( fig . [ f3]c , d ) . the above results show that domain growth in the microcanonical @xmath0 model eventually falls into the ncop universality class ( @xmath35 , @xmath41 ) after some possibly long transient behavior . in this section , we suggest that there are two main factors at the origin of these corrections to scaling : space discretization and energy conservation . for the sets of parameters studied by zheng ( notably parameter set a ) , domain growth initially appears to be slower than the expected ncop law ( the effective value of @xmath8 measured at short times is larger than two ) . this is similar to earlier observations on coupled map lattices , both for the ncop and cop cases . first measurements of domain growth seemed to indicate slower growth in those discrete - space , discrete - time chaotic models @xcite , but it was shown later that this non - trivial scaling is only transient and that , for late - enough times , normal scaling is recovered @xcite . it was also argued that these long transients gradually disappear in the continuous - space limit which is well - defined in these systems @xcite . here , it is easy to observe that , increasing @xmath53 ( and @xmath54 , since there is only one free parameter ) , the slowing - down of domain growth due to lattice effects diminishes , as suggested by the difference between parameter sets a and b. for parameter set b , the transient actually presents faster growth than asymptotically : corrections to scaling are dominated by another phenomenon , stronger than lattice effects . the ( conserved ) total energy in the system can be partitioned into three components , kinetic , potential , and interaction energy : @xmath55 as domain growth proceeds following random initial conditions , the interaction energy decreases with time as a large part of it is contained in the interfaces separating domains , the density of which decreases like @xmath56 . potential decreases and kinetic energy increases ( fig . [ f4 ] ) . starting from `` ordered '' initial conditions ( all sites in the same phase , e.g. @xmath57 ) , the three components of the total energy are almost constant , except for a slight ( quasi logarithmic in time ) increase of interaction energy and decrease of potential energy due to the nucleation of `` thermal '' droplets and to relaxation further into potential wells . this effect is indeed larger for parameter set a for which the minimal possible energy density @xmath58 is further away from the critical energy density , leading to stronger thermal fluctuations than for parameter set b. kinetic energy can be interpreted as a temperature in the microcanonical context @xcite . we can thus see the observed increase of kinetic energy as an increase of the temperature of the system . for the ( canonical ) ising model , it is well known @xcite that the prefactor of the growth law of @xmath59 decreases to zero as the temperature approaches its critical value . we suggest to see the growth law observed here for the @xmath0 model as including a `` temperature - dependent '' prefactor : @xmath60 quantitatively ( fig . [ f4 ] ) , the kinetic energy seems to reach its asymptotic value like : @xmath61 assuming its analyticity , we can write the prefactor @xmath62 : @xmath63 where @xmath64 is the asymptotic ( @xmath65 ) prefactor of the domain growth law . injecting ( [ deltae ] ) and ( [ expand ] ) into ( [ growth ] ) , we finally expect the following ansatz to hold for the domain growth law : @xmath66 equation ( [ ansatz ] ) provides excellent fits to our data for the growth of @xmath59 . we find @xmath67 , @xmath68 , and @xmath69 for parameter set a ( with @xmath70 and correlation coefficient 0.99996 ) , and @xmath71 , @xmath72 , and @xmath73 ( with @xmath74 and correlation coefficient 0.9997 ) for parameter set b. the corrective terms have opposite signs in both cases : this indicates that eq . ( [ ansatz ] ) is only relevant for parameter set b , because , in analogy with the ising model , only negative values of @xmath75 are allowed ( the prefactor of the growth law decreases with increasing temperature ) . a first conclusion is thus that the main corrections to scaling for parameter set a ( typical of those used by zheng ) are due to lattice effects and are _ not _ consequences of the conservation of global energy . we note that , similarly , a fit of domain growth in coupled map lattices also yields positive values of @xmath75 , indicating that this sign is a signature of lattice effects @xcite . on the other hand , the above framework does provide the relevant explanation for the corrections to scaling observed for parameter set b , which can thus be traced back to the fluxes between the various components of the energy induced by the decrease of interfaces between domains as phase - ordering proceeds . in this comment , we have shown that zheng reached erroneous conclusions when studying phase ordering in the microcanonical lattice @xmath0 model . this system , like other chaotic , deterministic , dynamical systems presenting phase ordering , does display the expected domain growth scaling laws , i.e. those of the non - conserved order parameter case ( model a , @xmath35 , @xmath41 ) . we have shown further that the main influence of the conservation of energy is to introduce corrections to scaling , but that the long transients which plagued zheng s approach are due to lattice effects . zheng offered , as an explanation for his non - conventional results , that the system considered falls into the class of model c , where a conserved density is coupled to the order parameter @xcite . studies of phase - ordering in model c @xcite show that model b - like , but also model a - like behavior can be observed . in this context , the microcanonical @xmath0 lattice can be considered as a model c system quenched into its `` bistable region '' ( where @xmath35 is observed @xcite ) . since @xmath5 is precisely defined as the midheight value of @xmath76 , it is in our opinion not so surprising that we obtain already a good collapse before the scaling regime sets in . in this sense our appoach is different from that of zheng who tried to collapse @xmath77 . of course , plotting the quantity of interest against @xmath78 and looking for a straight line is not the most accurate way of determining @xmath8 ( here we use this only because the log - log plots do not exhibit a constant asymptotic slope ) . nevertheless , the value @xmath79 measured by zheng can safely be rejected since the local slope of @xmath5 against @xmath80 increases with time ( not shown here ) .	zheng [ phys . rev . e * 61 * , 153 ( 2000 ) ] claims that phase ordering dynamics in the microcanonical @xmath0 model displays unusual scaling laws . we show here , performing more careful numerical investigations , that zheng only observed transient dynamics mostly due to the corrections to scaling introduced by lattice effects , and that ising - like ( model a ) phase ordering actually takes place at late times . moreover , we argue that energy conservation manifests itself in different corrections to scaling . 2 the problem of the dynamical foundations of statistical mechanics has received renewed attention recently , when , in the spirit of the famous work of fermi , pasta , and ulam @xcite , researchers directly studied the evolution of isolated , many - degrees - of - freedom hamiltonian systems with the aim of relating their microscopic , deterministic , chaotic motion to their macroscopic statistical properties @xcite . in this context , the two - dimensional lattice @xmath0 model is of special interest because it is known to exhibit , within the canonical ensemble , a second - order phase transition in the ising universality class . recent work @xcite aimed , in particular , at studying the corresponding behavior in the isolated , microcanonical case , whose equations of motion can be written : @xmath1 where the sum is over the four nearest neighbors of site @xmath2 on a square lattice . in the same spirit , zheng @xcite has considered the a priori simpler problem of the phase - ordering process which takes place when the microcanonical @xmath0 lattice is suddenly `` quenched '' below the critical point . universality of domain growth laws is nowadays a fairly well - established topic @xcite . it is well documented , even for less traditional systems such as deterministic , possibly chaotic , spatially - extended dynamical systems @xcite . for a scalar order parameter , two main universality classes can be distinguished depending on whether it is locally conserved or not . the ising model is prototypical of the non - conserved case , and so should be the lattice @xmath0 model , at least in the usual canonical ensemble point of view . however , and this was the interesting point raised by zheng , the presence of the energy conservation in the microcanonical case might have an influence on the dynamical scaling laws associated with phase ordering . in this sense , the question is whether the phase ordering of model c @xcite is in the same universality class as model a. using numerical simulations in the so - called `` early - time '' regime , zheng confirmed that the usual dynamical scaling laws seem to hold , but with exponents at odds with those both the non - conserved order parameter ( ncop ) and the conserved order parameter ( cop ) class @xcite . in particular , he found @xmath3 ( @xmath4 is the exponent governing the algebraic growth of @xmath5 , the typical size of domains ) , between its ncop and cop values ( @xmath6 and @xmath7 ) . zheng `` explains '' this surprising result by the influence on scaling of `` the fixed point corresponding to the minimum energy of random initial states '' ( present when considering domain growth ) . here we show that a more careful numerical investigation actually leads to conclude that the microcanonical lattice @xmath0 model shows normal , ncop phase - ordering . we argue that energy - conservation , but also lattice effects , have sub - leading influences on scaling . we discuss in particular the effect of the increase of the `` bulk temperature '' due to the progressive disappearance of interfaces during the growth transient . we attribute the erroneous results of zheng to the danger of using `` early - time '' methods and naive logarithmic plots in problems with large microscopic times and/or corrections to scaling .
the majority - vote model with noise defined on a regular lattice is a system of spins where each one is allowed to be on two states only @xcite . in this two - state model , each spin assumes the state of the majority of its neighboring spins with probability @xmath6 and the opposite state with probability @xmath7 . the system presents an order - disorder phase transition as the noise parameter @xmath7 reaches a critical value @xmath8 . studies on the regular square lattice found @xmath9 @xcite , and critical exponents equal to those for the equilibrium ising model in accordance with the conjecture by grinstein _ _ @xcite . the two - state majority - vote model ( mv2 ) was also studied on a variety of complex networks @xcite , such as undirected and directed random graphs @xcite , small - world networks @xcite , and barabsi - albert scale - free networks @xcite . on undirected and directed random graphs , it was shown that @xmath8 is an increasing function of the connectivity of the graphs @xcite . on small - world networks , the critical noise is an increasing function of the rewiring probability @xcite . more general , these studies have shown that mv2 models defined on different complex networks belong to different universality classes and the calculated critical exponents depend on the topology of the complex network @xcite . the generalization to a three - state majority vote model ( mv3 ) on a regular square lattice was considered by @xcite , where the authors found @xmath10 . the resulting critical exponents for this non - equilibrium mv3 model are in agreement with the ones for the equilibrium three - state potts model @xcite , again supporting the conjecture of ref . @xcite . in this paper we present an extensive study of the critical behavior of the three - state majority - vote model on erds - rnyi s random graphs @xcite . monte carlo ( mc ) simulations and standard finite - size scaling techniques are used to determine the critical noise parameter @xmath0 , as well as the exponents @xmath2 , @xmath3 and @xmath4 for several values of the mean connectivity @xmath1 of the graph . we also study the globally connected network case . the phase diagram of the system is presented , and compared to our previously obtained diagram for the two - state model @xcite . exact results for quantities of interest are obtained for the limiting cases @xmath11 , and @xmath12 , which agree with the simulation results . this work is organized in the following way : in section [ sec2 ] we describe the non - equilibrium three - state majority - vote model with noise , and introduce the relevant quantities used in our simulations . sections [ sec3 ] and [ sec4 ] contain our results along with a discussion . finally , in the last section we present our conclusions . the three - state majority - vote model with noise is defined by a set of spin variables \{@xmath13 } , where each spin is associated to one vertex of an erds - rnyi s random graph and can have the values @xmath14 and @xmath15 . the connectivity of a vertex is defined as the total number of bonds connected to it , that is @xmath16 , where @xmath17 if there is a link between the sites @xmath18 and @xmath19 and @xmath20 otherwise . a random graph is completely characterized by the mean number of connections per site , _ i.e. _ the average connectivity @xmath1 , and the total number of sites @xmath21 . the system evolves in time according to the following rules : for each spin we determine the state of the majority of its neighboring spins , that is , all the spins that are linked to it . with probability @xmath6 the new state of the spin agrees with the majority state of its neighbors and it disagrees with probability @xmath7 , which is known as the noise parameter . in the case of a tie between the three possible states , each state is chosen with equal probability @xmath22 . in the case of a tie between two majority states , the spin assumes each one of these states with equal probability @xmath23 , and the minority state with probability @xmath7 . finally , in the case of a single majority state , the two minority states occur with equal probability @xmath24 , and the majority state with probability @xmath6 . it is clear that the rules just described present the @xmath25 symmetry with respect to the simultaneous change of all states @xmath26 . let @xmath27 be the number of neighbors of site @xmath18 in state @xmath28 , therefore @xmath29 . according to the above rules we can write the following probabilities for a given spin to assume the state @xmath30 : @xmath31 the probabilities for the other two states are obtained by the symmetry operations of the @xmath25 group . for example , let us consider a neighborhood corresponding to the fourth and fifth rules , where we have a single majority state . in this case , the two minority states ( say , states 2 and 3 ) occur with equal probability @xmath24 , and the majority state with probability @xmath32 . we can write @xmath33 from the fourth rule , and @xmath34 , from the fifth rule . it is worth mentioned that the condition @xmath35 (= probability of choosing the majority state 1 ) @xmath36 (= probability of choosing a minority state , either 2 or 3 ) is valid for @xmath37 , and we conclude that @xmath38 is the limit value for the noise parameter in the present three state mv model . moreover , the probabilities defined by eq.([rules ] ) satisfy @xmath39 . to study the critical behavior of the model we consider the magnetization @xmath40 , the susceptibility @xmath41 , and the binder s fourth - order cumulant @xmath42 . these quantities are defined by @xmath43 @xmath44\ ] ] @xmath45 where n is the number of vertices of the random graph with fixed @xmath1 , @xmath46 denotes time averages taken in the stationary regime , and @xmath47 stands for configurational averages . in eqs . ( [ eq : mag])-([eq : cumu ] ) @xmath48 is defined in analogy to the magnetization in the three - state potts model as the modulus of the magnetization vector , that is @xmath49 , whose components are given by @xmath50,\ ] ] where the sum is over all sites in the graph , @xmath51 is the kronecker delta function , and we introduce the factor @xmath52 in order to normalize the magnetization vector . in the critical region we assume the following finite - size scaling ( fss ) relations @xcite @xmath53 @xmath54 @xmath55 where @xmath56 , and the universal scaling functions @xmath57 , @xmath58 and @xmath59 only depend on the scaled variable @xmath60 . the above fss relations follow from the _ ansatz _ that the correlation length scales with the number of nodes in the graph , that is @xmath61 , which is consistent with an effective dimensionality equal to unity . from the size dependence of @xmath40 and @xmath41 we can obtain the exponents @xmath2 and @xmath3 , respectively . the correlation length exponent @xmath62 is calculated from the size dependence of the derivative of binder s fourth - order cumulant with respect to the noise parameter , @xmath63 . furthermore , we use the hyperscaling relation @xmath64 to estimate the effective dimensionality of the system @xmath65 in order to check the fss prediction for an effective dimensionality equal to unity . first we notice that there are only two independent components of the magnetization vector @xmath66 , since they obey the relation @xmath67 moreover , the norm of the magnetization is invariant with respect to any @xmath68 group symmetry operation . let us consider the limit @xmath11 , where the probability of agreeing with the majority state equals unity . in this situation , after a transient period the system reaches the ordered steady state . without loss of generality we can assume @xmath69 for all sites . in this case @xmath70 and @xmath71 , thus @xmath72 . it is also straightforward to check that @xmath73 , and @xmath74 . in the opposite limit ( @xmath12 ) the system reaches a disordered steady state , where the average of each component of the magnetization vanishes . in fact , in this limit the probability of a given spin agreeing with the majority of its neighbors equals the probability of it agreeing with any of the other two minority states , _ i.e. _ @xmath75 . it is possible to write the probability distribution for the order parameter as a gaussian distribution in the form @xcite @xmath76 where @xmath77 . from this distribution we obtain that @xmath78 , @xmath79 , and @xmath80 . moreover , it follows that @xmath81 , @xmath82 , and @xmath83 . these exact results as well as the predict size @xmath21-dependence of the relevant quantities are in agreement with numerical results from simulations for all networks considered . we begin our simulations generating a random graph of size @xmath21 and mean connectivity @xmath1 in a disordered configuration where the state of each spin is @xmath84 or @xmath15 with the same probability . we used systems of size @xmath85 , and varied @xmath1 from @xmath30 to @xmath86 . to perform the dynamics we choose a site at random and , for a given fixed value of the noise parameter , we update its state in accordance with the dynamics rules given by eq . ( [ rules ] ) . a monte carlo step ( mcs ) is defined as @xmath21 updates . we waited @xmath87 mcs needed for the system to reach the steady state , and the time averages , @xmath46 , were estimated from the next @xmath88 mcs . the values of mcs used vary with @xmath21 , @xmath1 and @xmath7 , typically we used @xmath89 mcs . for all sets of parameters , we have generated at least @xmath90 distinct random networks in order to calculate the configurational averages @xmath91 . [ mag - susc ] shows the magnetization @xmath40 and the susceptibility @xmath41 as functions of the noise parameter . the data were obtained from simulations on random graphs with @xmath92 sites and several values of the average connectivity @xmath1 . in part ( a ) each curve for @xmath40 , for a given value of @xmath1 , clearly indicates that there exists a phase transition from an ordered state to a disordered state where the magnetization vanishes . we also notice that the transition occurs at a value of the critical noise parameter @xmath0 , which is an increasing function of the mean connectivity @xmath1 of the random graph . in part ( b ) we show the corresponding behavior of the susceptibility @xmath41 . the value of @xmath7 where @xmath41 has a maximum is here identified as @xmath93 . we also perform simulations for globally connected networks , that is , for random graphs with @xmath21 nodes and connectivity @xmath94 . in the thermodynamical limit @xmath95 , the magnetization is given by @xmath96 ( the dashed line in fig . 1(a ) ) , from which we obtain the mean - field values @xmath97 and @xmath98 . fig . 1(c ) shows the dependence of the mean - field magnetization at @xmath99 on the system size . the straight line confirms the scaling relation given by @xmath100 . the slope of the resulting straight line yields the exponent @xmath101 . a similar analysis for the susceptibility at @xmath99 yields @xmath102 , that is , @xmath103 . note that the critical behavior for the globally connected network is in agreement with the exact results discussed above in the limit case of @xmath104 . the mean - field results for the critical noise parameter and critical exponents are given in table [ tab1 ] . [ cumul ] we plot binder s fourth - order cumulant @xmath42 for different system sizes @xmath21 and four distinct values of @xmath1 . the critical noise parameter @xmath0 , for a given value of @xmath1 , is estimated as the point where the curves for different values of @xmath21 intercept each other . we also obtained @xmath105 for the critical value of the cumulant at @xmath0 , which is independent of the connectivity @xmath1 of the graph . the dependence of @xmath0 on @xmath1 yields the phase diagram for the mv3 model shown in fig . [ phased ] . the phase diagram of the mv3 model on random graphs shows that for a given graph ( fixed @xmath1 ) the system becomes ordered for @xmath106 , whereas it has zero magnetization for @xmath107 . we notice that the increase of @xmath0 is more pronounced for small values of @xmath1 . the error bars in @xmath0 ( see table [ tab1 ] ) are much smaller than the symbols . in the figure , it is also included the corresponding phase diagram for the mv2 model obtained from monte carlo simulation in our previous work @xcite . for both models the system exhibits an ordered state for all values of the mean connectivity greater than one . this is in agreement with the limiting value of @xmath108 for the existence of a percolating cluster and , therefore , the onset of long - range order in the system . however , when @xmath109 we obtain the upper limits @xmath110 and @xmath111 , for the mv2 and mv3 models respectively . [ nu ] shows the dependence of the derivative of the binder s fourth - order cumulant at @xmath112 on system size . for clarity we have added in each curve the respective value of @xmath1 . the straight lines , obtained from simulations with different values of the mean connectivity @xmath1 , confirm the scaling relation given by eq . ( [ ufss ] ) . for fixed @xmath1 , the slope of the resulting straight line equals the exponent @xmath4 . the results displayed in table [ tab1 ] indicate a weak dependence of the correlation length exponent with @xmath1 . in order to obtain independent estimations for the critical noise parameter @xmath113 , as well as the exponent ratios @xmath2 and @xmath3 , for different values of the mean connectivity @xmath1 , we consider the functions @xmath114 and @xmath115 defined as @xmath116 @xmath117 where @xmath118 . the above functions relate the magnetizations and susceptibilities calculated with two different system sizes , @xmath119 and @xmath120 . in fact , substituting the finite - size relation ( [ mfss ] ) into eq . ( [ phi ] ) , we obtain @xmath121 at the critical value @xmath113 , the last term vanishes and we obtain @xmath122 . analogously , substituting the finite - size relation ( [ xfss ] ) into eq . ( [ psi ] ) , we have @xmath123 . in fig . [ phipsi ] we show the dependence with noise of the functions @xmath114 and @xmath115 , for @xmath124 and several values of @xmath119 and @xmath120 . from the intersection points in fig . [ phipsi](a ) and fig . [ phipsi](b ) we obtain independent estimations for the critical noise parameter @xmath113 . for all networks considered in the simulations , we have obtained a quite satisfactory agreement between the two values of @xmath113 determined in this way and the corresponding ones that follow from the analysis of binder s cumulant ( fig . [ cumul ] ) . moreover , we used the relations @xmath122 and @xmath123 to calculate the exponent ratios for different values of the mean connectivity @xmath1 . table [ tab1 ] shows the results for different values of the mean connectivity @xmath1 . we call the readers attention to the difference between the calculated values of the exponents in the case of @xmath125 and the corresponding results with @xmath126 . this might be an indication that , for @xmath125 , we need to take into account logarithmic corrections to the finite - size scaling relations @xcite . [ mdatacol ] shows the data - collapse plot for @xmath127 , which is a universal function of the combined variable @xmath128 . we have also obtained quite good data - collapse for @xmath129 . the collapsing of curves for five different system sizes corroborates the quoted values for @xmath8 , @xmath2 , @xmath3 and @xmath4 . on fig . [ udatacol ] we present two different ways to obtain the data - collapsing for the universal scaling function @xmath130 . part ( a ) shows the standard data - collapse that follows from simulations with different values of system size @xmath21 , for the case of mean connectivity @xmath124 fixed . in part ( b ) we have fixed @xmath92 and used the data from simulations for varying connectivity . it is worth mentioning that in the last case the collapse for different values of @xmath1 was obtained by using the values for the critical parameter @xmath8 and the exponent @xmath131 for the corresponding value of @xmath1 ( see table [ tab1 ] ) . in general , the universal scaling functions only depend on the scaled variable @xmath60 for a given system . however , the two data - collapsing in fig . [ udatacol ] indicate that @xmath59 does not depend on the specific value of @xmath1 , contrary to what is observed for @xmath57 and @xmath58 . .[tab1 ] the critical noise @xmath8 , the critical exponents , and the effective dimensionality @xmath65 , for the mv3 model on random networks with mean connectivity @xmath1 . the mean - field ( mf ) results are shown in the last line . [ cols=">,<,<,<,<,<",options="header " , ] table [ tab1 ] summarizes the values ( along with errors ) of the critical noise parameter @xmath0 , the critical exponents @xmath2 , @xmath3 and @xmath4 , and the effective dimensionality of the system . for all @xmath1 considered , including the mean - field limit @xmath5 , the value @xmath132 follows from the hyperscaling relation ( eq . ( [ hsc ] ) ) , in agreement with the scaling _ ansatz _ for the magnetization and susceptibility eqs . ( [ mfss ] , [ xfss ] ) . in fact , since our original work @xcite several authors have studied different spin models on varied complex networks , always finding an effective dimensionality equal to one @xcite . even though this result seemed surprising at first , it is a direct consequence of the scaling for the correlation length @xmath133 , with @xmath134 . there are no previous works studying the three - state majority - vote model on erds - rnyi s graphs , to allow a direct comparison of the present results . yet , for completeness , it would be of interest to mention earlier simulations of the majority - vote model on other kinds of networks . the only works on the mv3 model to this date , considered the model on a regular square lattice @xcite . they find the universal value of binder s fourth - order cumulant to be @xmath135 , and the exponents @xmath136 and @xmath137 , all of which are in agreement with the results for the equilibrium three - state potts model @xcite . the present simulations of the mv3 model on globally connected networks yielded @xmath138 , and the following mean - field exponents : @xmath139 , @xmath140 , @xmath98 , @xmath141 , and @xmath142 . from our simulation results we can conclude that the mv3 model defined on a regular square lattice , on erds - rnyi s random graphs , and on the corresponding globally connected network ( the mean - field limit ) belong to different universality classes . comparing the current results with the ones previously obtained for the two - state model on random graphs @xcite , we first notice that the ordered region ( @xmath143 ) in the phase diagram of fig . [ phased ] is larger for mv3 than for mv2 . this is expected since now we might obtain three possible majority states . it should also be clear from the calculated exponents that these models do not belong to the same universality class . in particular , @xmath144 for mv2 , and @xmath145 for mv3 . we have obtained the phase diagram and critical exponents of the three - state majority - vote model with noise on random graphs . the second - order phase transition which occurs in the model with mean connectivity @xmath146 has exponents that show a slight variation along the critical line . nevertheless , our monte carlo simulations provide an effective dimensionality @xmath65 equal to one for all values of @xmath1 . this result , which is in agreement with several previous studies on spin models defined on complex networks , was shown to be a consequence of the _ ansatz _ that the correlation length scales with the number of nodes . future work on the two- and three - state potts model on random graphs would be of interest in order to provide a direct comparison with our results in light of the conjecture by grinstein _ et al . _ , which states that reversible and irreversible models with same symmetry belong to the same universality class . marques , phys . lett . a * 145 * , 379 ( 1990 ) . t. tom , m.j . de oliveira , m.a . santos , j. phys . a : math . gen . * 24 * , 3677 ( 1991 ) . de oliveira , j. stat . phys . * 66 * , 273 ( 1992 ) . santos and s. teixeira , j. stat . phys . * 78 * , 963 ( 1995 ) . mendes and m.a . santos , phys . e * 57 * , 108 ( 1998 ) . w. kwak , j .- s . yang , j .- i . sohn and i .- m kim , phys . e * 75 * , 061110 ( 2007 ) . yang , i .- m . kim and w. kwak , phys . e * 77 * , 051122 ( 2008 ) . g. grinstein , c. jayaprakash and y. he , phys . * 55 * , 2527 ( 1985 ) . campos , v.m . de oliveira and f.g.b . moreira , phys . e * 67 * , 026104 ( 2003 ) . pereira and f.g.b . moreira , phys . e * 71 * , 016123 ( 2005 ) . f.w.s . lima , u.l . fulco and r.n . costa filho , phys . e * 71 * , 036105 ( 2005 ) . f.w.s . lima , int . c * 17 * , 1257 ( 2006 ) . luz and f.w.s . lima , int . c * 18 * , 1251 ( 2007 ) . f.w.s . lima , a.o . sousa and m.a . sumuor , physica a * 387 * , 3503 ( 2008 ) . z .- x . wu and p. holme , phys . e * 81 * , 011133 ( 2010 ) . p. erds and a. rnyi , publ . inst . hung . sci . * 5 * , 17 ( 1960 ) . d.j . watts and s.h . strogatz , nature ( london ) * 393 * , 440 ( 1998 ) . barabsi and r. albert , science * 286 * , 509 ( 1999 ) . c. castellano , s. fortunato and v. loreto , rev . phys * 81 * , 591 ( 2009 ) . r. albert and a .- barabsi , rev . phys * 74 * , 47 ( 2002 ) . dorogovtsev and j.f.f . mendes , _ evolution of networks : from biological nets to the internet and www _ ( oxford university press , oxford , 2003 ) . a. brunstein and t. tom , phys . e * 60 * , 3666 ( 1999 ) . t. tom and a. petri , j. phys . a * 35 * , 5379 ( 2002 ) . mod . phys . * 54 * , 235 ( 1982 ) . h. hong , m. ha and h. park , phys . lett . * 98 * , 258701 ( 2007 ) . s. caracciolo and a. pelissetto , nucl . b * 53 * , 693 ( 1997 ) .	the three - state majority - vote model with noise on erds - rnyi s random graphs has been studied . using monte carlo simulations we obtain the phase diagram , along with the critical exponents . exact results for limiting cases are presented , and shown to be in agreement with numerical values . we find that the critical noise @xmath0 is an increasing function of the mean connectivity @xmath1 of the graph . the critical exponents @xmath2 , @xmath3 and @xmath4 are calculated for several values of connectivity . we also study the globally connected network , which corresponds to the mean - field limit @xmath5 . our numerical results indicate that the correlation length scales with the number of nodes in the graph , which is consistent with an effective dimensionality equal to unity .
planetary astrophysics is undergoing an epoch of explosive growth driven by the discovery of about 850 exoplanets in two decades . continuous improvement in detection techniques will uncover many more planets in the near future which are too small or too far from their host star to be detected at present . observations suggest that planet formation is ubiquitous , with more than 40% of nearby sun - like stars harboring at least one planet less massive than saturn @xcite . these prolific discoveries have revealed the amazing diversity of exoplanetary systems and indicated that planets may form under a wide range of conditions and have considerable mobility . the diversity of exoplanets has been particularly highlighted by the kepler s space mission . to date , kepler has detected over 2300 planet candidates @xcite , from which more than a hundred are confirmed planets ( http://kepler.nasa.gov ) . about one third of kepler s candidates are associated with multiple transiting systems and the vast majority of them are expected to be real multi - planetary systems @xcite . some of these systems show remarkable properties , like the five coplanar , alternating rocky and icy planets orbiting kepler-20 in less than 80 days @xcite . interactions between planets and their parent protoplanetary disk are likely to have played a major role in shaping such compact system . a striking feature of kepler s multi - planetary systems is the large number of planet pairs far from mean - motion resonances ( hereafter mmr , * ? ? ? planet pairs near resonances show some tendency to have period ratios slightly greater than resonant values ( see figure [ fig : kepler ] ; period ratios are defined as the outer planet s orbital period divided by the inner planet s ) . for instance , there are almost twice as many planet pairs with period ratios between 2.0 and 2.1 than between 1.9 to 2.0 . this intriguing feature could naturally arise from tidal orbital circularization of close - in resonant planet pairs @xcite . however , as stressed by the histograms in figure [ fig : kepler ] , the same trend holds for planet pairs where the orbital period of the inner planet exceeds 10 days , most particularly near first - order resonances . for such systems , tidal circularization is unlikely to have caused significant divergent evolution of the planets orbits , as shown in section [ sec : tidal ] of the appendix . an example of a planetary system with a period ratio slightly greater than resonant is the kepler-46 system ( previously known as koi-872 ) . the 0.9 solar - mass star is surrounded by a @xmath0 transiting planet on a 33.6 day period ( kepler-46b ) . large transit time variations ( ttv ) of kepler-46b are caused by the presence of an outer planet , kepler-46c , whose mass ( @xmath1 ) and orbital period ( 57.0 days ) have been determined by @xcite . the same authors have also reported the detection of a @xmath2 planet candidate on a 6.8 day period ( kepler-46d , not yet confirmed ) . the main properties of the kepler-46 planetary system are summarized in table [ tab : tablekoi ] . the low - eccentricity and quasi - coplanar orbits in this system suggest that the planets reached their present location most probably through disk - planet interactions . still , the ratio or orbital periods between kepler-46c and kepler-46b is @xmath3 , which is slightly greater than @xmath4 . capture in the 5:3 mmr is not a natural outcome of planet migration driven by disk - planet interactions . it requires a rate of convergent migration that is large enough to cross the 2:1 mmr , but not too large so that the planets do not reach their 3:2 mmr . capture into a second - order resonance also requires small but not zero eccentricities , which disk - planet interactions tend to damp quite efficiently . should disk - planet interactions have led kepler-46b and kepler-46c to reach their 5:3 mmr , it is unclear how their period ratio would have then increased to their present value . the initial motivation of this work is to investigate under what circumstances disk - planets interactions may account for the orbital period ratio between kepler-46c and kepler-46b . in section [ sec : hydro ] , we present results of hydrodynamical simulations modeling the early evolution of both planets as they were embedded in their parent protoplanetary disk . these saturn - mass planets are expected to open a partial gap around their orbit . a rapid convergent migration causes the planets to merge their gap and to evolve into a common gap . the planets convergent migration is found to stall with period ratios between 1.5 and 1.7 , depending on the disk s density profile inside the common gap . the observed period ratio can be reproduced without the planets being in resonance . furthermore , we show that disk - driven migration of partial gap - opening planets may lead to significant divergent evolution of planet pairs over typical disk lifetimes . such divergent evolution is shown to arise from the interaction between planets and the wakes of their companions . this mechanism is illustrated for super - earth - mass planets in section [ sec : results_nbody ] with both hydrodynamical simulations and customized three - body integrations . divergent evolution of gap - opening planet pairs mediated by wake - planet interactions could partly account for the diversity of period ratios in kepler s multi - planetary systems . concluding remarks are provided in section [ sec : conclusion ] . bottom : ratio of orbital periods of all planet pairs among kepler s candidate multi - planetary systems detected from quarters 1 to 6 ( x - axis ) versus the orbital period of the inner planet in each pair ( in days , y - axis ) . bins are @xmath5 wide . from left to right , vertical dashed lines show the location of the 5:4 , 4:3 , 3:2 , 5:3 and 2:1 mean - motion resonances . top : histograms of the orbital period ratio for all planet pairs ( black ) , and for planet pairs where the inner planet s period is greater than 10 days . data were extracted from http://planetquest.jpl.nasa.gov / kepler/. ] .[tab : tablekoi]kepler-46 planetary system @xcite [ cols="<,<,<,<",options="header " , ] we model in this section the early evolution of the kepler-46 planetary system when the planets were embedded in their parent protoplanetary disk . our study does not address the formation of the planets . we focus instead on the planets orbital evolution due to disk - planet and planet - planet interactions . two - dimensional hydrodynamical simulations of disk - planets interactions were carried out using the code fargo @xcite . the physical model and numerical setup of the hydrodynamical simulations are described in section [ sec : setup ] . results of simulations follow in section [ sec : results_hydro ] . _ disk model _ we adopt a two - dimensional disk model for the protoplanetary disk in which the planets orbiting kepler-46 formed . disk - planets interactions determine the rate of convergent migration of the planets and the damping rate of their eccentricity . both quantities are sensitive to the disk s surface density , temperature , turbulent viscosity and radiative properties . the parameter space is therefore particularly large , and the need to assess the planets orbital evolution over typically a thousand planet orbits has led us to make simplifying assumptions for the disk model and to fix some of the disk parameters . disk self - gravity is discarded given the low disk masses adopted in this work ( the toomre - q parameter associated with the unperturbed density profile does not go below 7 ) . a locally isothermal equation of state is used where the vertically - integrated pressure @xmath6 and density @xmath7 satisfy @xmath8 , with the sound speed @xmath9 being specified as a fixed function of the cylindrical radius @xmath10 . the sound speed is related to the disk s pressure scale height @xmath11 through @xmath12 with @xmath13 the keplerian angular velocity . the ( fixed ) temperature profile is taken proportional to @xmath14 so that the disk s aspect ratio @xmath15 is uniform . we take the standard value @xmath16 . a locally isothermal equation of state is generally not appropriate to model the migration of an embedded low - mass planet , as it underestimates the magnitude of the corotation torque that the planet experiences from the disk ; see @xcite for a recent review on planet migration . this is not an issue in this study , however , since we consider partial gap - opening planets for which the corotation torque has a rather weak effect on migration . this may be quantified through the dimensionless parameter @xmath17 ( @xmath18 denotes the planet - to - star mass ratio ) which is @xmath19 in our simulations . this lower limit implies that the direction and speed of planet migration are weakly sensitive to the choice for the equation of state @xcite . the effects of turbulence are modeled by a constant shear kinematic viscosity @xmath20 . we take @xmath21 in standard code units ( defined below ) which translates into a viscous alpha parameter @xmath22 at the initial location of the inner planet . this value of @xmath23 is typical of the magnetically active regions of protoplanetary disks with magneto - hydrodynamic turbulence driven by the magneto - rotational instability . the present orbital periods of the planets in the kepler-46 system imply that the planets have probably interacted with such active regions in their parent disk . note , however , that the disk s magnetic field is not included in our simulations . we also point out that the planets that we consider are of large enough mass so that small - scale turbulent fluctuations can be safely ignored when modeling the interaction with their parent disk . in other words , the viscous diffusion approximation is expected to be a reasonable approximation for the large planet masses we consider @xcite . the disk is set up in radial equilibrium , with the centrifugal acceleration and the radial acceleration related to the pressure gradient balancing the gravitational acceleration due to the central star . the initial gas surface density is taken proportional to @xmath24 . its value at the initial location of the inner planet is taken as a free parameter , which we denote by @xmath25 . + _ planet parameters _ we model the orbital evolution of the outer two planets in the kepler-46 system : kepler-46b ( hereafter refereed to as the inner planet ) and kepler-46c ( as the outer planet ) . the @xmath26 unconfirmed planet candidate with a period of 6.8 days is assumed to have had a negligible impact on the orbital evolution of kepler-46b and kepler-46c . the planets are assumed to have already formed at the beginning of our simulations , they therefore take their present mass . kepler-46b has a physical radius @xmath27 however , its mass could not be constrained from ttv ; see @xcite . the mass - radius diagram of exoplanets detected by transiting methods indicates that the mass of kepler-46b most probably lies between @xmath28 and @xmath29 . its planet - to - primary mass ratio ( @xmath30 ) is therefore varied in the range @xmath31 $ ] . the ttv signal of kepler-46b constrains the mass of kepler-46c to be @xmath32 @xcite . the planet - to - primary mass ratio of kepler-46c is thus fixed to @xmath33 . the initial orbital radius of kepler-46b , @xmath34 , is taken to be the code s unit of length . that of kepler-46c is arbitrarily set to @xmath35 , so that the ratio of orbital periods @xmath36 initially . for the disk aspect ratio and viscosity taken in our study , the two planets are expected to open a partial gap around their orbit @xcite . + _ numerical setup _ the hydrodynamical equations are solved in a cylindrical coordinate system @xmath37 centered on to the star with @xmath38r_{\rm inner}$ ] and @xmath39 $ ] . the computational grid has @xmath40 zones evenly spaced along the radial direction and @xmath41 azimuthal sectors . the frame rotates with the keplerian frequency at the inner planet s location , and the indirect terms that account for the acceleration of the central star by the disk and the planets are included in the equations of motion . a standard outflow boundary condition is taken at the grid s inner edge , while damping is used in a so - called wave killing - zone extending from @xmath42 to @xmath43 ( disk quantities are damped towards their instantaneous axisymmetric profile ) . to avoid a violent relaxation of the disk due to the sudden introduction of the planets , the mass of the planets is gradually increased over 10 orbital periods . the gravitational potential of the planets is smoothed over a softening length , @xmath44 , equal to @xmath45 with @xmath11 evaluated at the planets location . the calculation of the force exerted by the disk on the planets excludes the content of the planets circumplanetary disk , the size of which being about @xmath46 of the planets hill radius @xcite . the hill radius of each planet is resolved by about 10 grid cells along each direction initially . + _ code units _ results of simulations are expressed in the standard units of disk - planet calculations : the mass unit is the mass of the central star ( denoted by @xmath47 and equal to @xmath48 for the kepler-46 system ) . the length unit is the initial orbital radius of the inner planet ( @xmath34 ) , and the time unit is the initial orbital period of the inner planet ( @xmath49 ) divided by @xmath50 . whenever time is expressed in orbits , it refers to the orbital period at the initial location of the inner planet . the free parameters in our simulations are the mass of the inner planet and the initial surface density of the disk . both quantities determine the rate of convergent migration and the evolution of the orbital eccentricities . we consider three different masses for the inner planet : @xmath51 , @xmath52 and @xmath29 ( recall that the mass of the outer planet is fixed at @xmath53 ) . for each planet mass , a set of 9 simulations was carried out varying @xmath25 , the unperturbed surface density at the initial orbital radius of the inner planet . we took @xmath25 in the range @xmath54 $ ] , which corresponds to an initial disk mass between @xmath55 and @xmath56 . typical outcomes of the simulations are described in section [ sec : illust ] . we find that disk - planets interactions may lead to significant divergent evolution of the planet orbits , which we interpret in section [ sec : divergent ] . to obtain evolution of the planets orbital elements to a steady state , a simple model for disk dispersal is included and its results presented in section [ sec : disp ] . the robustness of our results to varying some physical and numerical parameters is discussed in section [ sec : robust ] of the appendix . figure [ fig : ex ] displays typical outcomes of our hydrodynamical simulations of the kepler-46 system . from left to right in the figure we show the time evolution of the planets semi - major axes , orbital period ratio , eccentricities and resonant angles . here @xmath57 and @xmath58 denote the mean longitude and longitude of periastron , with the subscripts i and o referring to the inner and outer planets , respectively . the top row shows results with @xmath59 and @xmath60 . as the planets progressively clear a partial gap around their orbit , convergent migration does not occur at a constant rate . in particular , the inward migration of the outer planet slightly accelerates at early times due to a stage of type iii migration @xcite before slowing down upon approaching the 2:1 mmr with the inner planet . the planets are locked in the 2:1 mmr from about 300 orbits onwards , as can be seen from the libration of the resonant angles associated with this resonance . the resonance capture increases the inner planet s eccentricity to about 0.035 . the increase in the outer planet s eccentricity is much more modest , reaching about @xmath61 , even though the mass ratio of the planets is near unity . this modest increase in the outer planet s eccentricity arises from the contribution of the indirect term nearly cancelling the contribution from the direct term to the term that is first order in the eccentricities in the development of the disturbing function , which governs the response of the outer planet at a 2:1 mmr ( see , e.g. , * ? ? ? * ; * ? ? ? damping of the eccentricities follows from the gravitational interaction with the background gas disk . interestingly , the ratio of orbital periods rises from 2.02 to 2.15 in only 1500 orbits , while the eccentricities are damped . the physical origin of this divergent evolution is discussed in section [ sec : divergent ] . note that resonant coupling is maintained throughout this divergent evolution , the width of the 2:1 mmr increasing as the eccentricities decrease . disk s surface density at 400 orbits for @xmath62 and @xmath60 ( top ) , and for @xmath63 and @xmath64 ( bottom ) . the outer planet s mass is @xmath65.,title="fig : " ] disk s surface density at 400 orbits for @xmath62 and @xmath60 ( top ) , and for @xmath63 and @xmath64 ( bottom ) . the outer planet s mass is @xmath65.,title="fig : " ] the second row of panels in figure [ fig : ex ] displays results for @xmath66 and @xmath67 . this case is particularly interesting as it shows that convergent migration may stall with an orbital period ratio very close to the observed value ( @xmath68 ) . the accelerating inward migration of the outer planet is particularly visible in the first 100 orbits . this runaway migration is fueled by the dense gas region located between the two planet gaps . this dense material is continuously funneled beyond the outer planet s orbit upon embarking on horseshoe u - turns with respect to the outer planet . the density between the two planet gaps therefore decreases , and so does the migration rate of the outer planet . convergent migration then proceeds at a slower pace and the planets end up merging their gap , thereby forming a common gap in the disk . this is illustrated in figure [ fig : dens ] , which compares the disk s surface density at 400 orbits for the two aforementioned simulations which have @xmath59 and @xmath69 ( top panel ) , and @xmath66 and @xmath67 ( bottom panel ) . while in the former case a narrow ring of gas is left between the two planet gaps , the latter case shows that both planets end up in a common gap . back to the second row of panels in figure [ fig : ex ] , we see that the orbital period ratio remains approximately stationary from about 400 orbits , and so do the eccentricities . the time - averaged eccentricity of the inner planet is close to 0.01 ; that of the outer planet is @xmath70 . these values are consistent , albeit slight larger than those inferred from observations ( see table [ tab : tablekoi ] ) . note from the critical angles that the planets are not locked in the 5:3 mmr . lastly , the third row of panels in figure [ fig : ex ] is for @xmath71 and @xmath72 . here , convergent migration is rapid enough that the planets approximately reach the nominal location of their 3:2 mmr ( the minimum value of the period ratio is @xmath73 at 170 orbits ) . the planet eccentricities increase to about 0.03 ( inner planet ) and 0.015 ( outer planet ) . subsequent damping of the eccentricities occurs along with a slight divergent evolution of the planets . the rate of divergent evolution is much reduced compared to the 2:1 mmr case shown in the top row of panels in figure [ fig : ex ] . we will analyse this result further in section [ sec : divergent ] . we also point out that no steady - state has been reached at the end of the simulation . for this simulation , as for all simulations of the kepler-46 system , results are shown until the inner planet reaches @xmath74 , below which the proximity of the grid s inner edge ( at @xmath75 ) starts affecting the planets orbital evolution . the final period ratio obtained in all our simulations is displayed in figure [ fig : finalopr ] as a function of @xmath25 for the three masses assumed for kepler-46b . by final period ratio , we mean either its value obtained at the end of the simulation ( after between 1500 and 2000 planet orbits ) , or that obtained when the inner planet s orbital radius falls below @xmath76 . figure [ fig : finalopr ] highlights the three major outcomes of our simulations : ( i ) capture in 2:1 mmr generally followed by rapid divergent evolution , ( ii ) capture in 3:2 mmr followed by slow divergent evolution , and ( iii ) convergent migration stalling with a period ratio between 1.6 and 1.7 , depending on the disk s density distribution inside the planets common gap . several simulations can reproduce the observed period ratio of kepler-46b and kepler-46c . the best agreement with observations is obtained for @xmath66 , our upper mass value . this mass would imply a mean density for kepler-46b about 20% larger than jupiter s ( that is , about 1.6 g @xmath77 ) . the results of hydrodynamical simulations in section [ sec : illust ] show that the orbital period ratio of two partial gap - opening planets may increase from a near resonant value . this divergent evolution is reminiscent of the late evolution of short - period planet pairs through star - planet tidal interactions . in that case , tidal circularization causes a slow divergent evolution of the orbits @xcite , a mechanism known as tidally - driven resonant repulsion @xcite . it is based on energy dissipation of the planet pair while its total angular momentum remains constant @xcite . do we have an analogous resonant repulsion mechanism due to disk - planets interactions ? to address this question , it is instructive to first look at the evolution of two embedded planets that do not interact gravitationally . in figure [ fig : fono ] we compare the results of two hydrodynamical simulations : one in which the gravitational interaction between the planets is switched off ( both the direct and the indirect interaction terms are discarded ) and another where the planet - planet interaction is fully accounted for . the inner planet s mass is @xmath78 and the initial disk surface density at @xmath79 is @xmath80 . the other disk and planet parameters are those described in section [ sec : setup ] . when planet - planet interactions are on , capture into the 2:1 mmr is followed by rapid divergent evolution and damping of the eccentricities . this case is similar to that presented in the top row of panels in figure [ fig : ex ] . quite surprisingly , we see that a very similar divergent evolution is also obtained when planet - planet interactions are switched off , while the planets are not resonantly coupled ( the resonant angles do not librate ) and their eccentricities take vanishingly small stationary values . this comparison highlights that , when disk - planet interactions are considered , eccentricity damping is not necessarily responsible for the planets divergent evolution . the divergent evolution of a planet pair may instead be driven by changes to the disk - planet interaction brought about through the proximity of the planets to each other , such as the development of an interaction between each planet and the wake of its companion . we argue below that the damping of the eccentricities can actually be a consequence of divergent evolution driven by wake - planet interaction . before describing the process of wake - planet interaction in more detail , we briefly report that , in some other simulations , not illustrated here , with planet - planet interactions switched off , a tiny increase in the orbital eccentricities is observed when the orbital period ratio takes near resonant values . eccentricities may reach a few @xmath81 before being damped by the disk . although this is a small effect , it suggests that planets could be weakly resonantly coupled through interaction with the wakes and/or the circumplanetary disk of their companions . in our simulations , the mass of the circumplanetary disks does not exceed one percent of the planets mass . + _ wake - planet interactions _ the results of simulations shown in figure [ fig : fono ] make clear that disk - driven divergent evolution of a planet pair is not necessarily driven by eccentricity damping and it may continue to operate even when the eccentricities are very small . we here indicate how the interaction between a planet and the wake of its companion could act as a driving mechanism for divergent evolution . consider for instance the inner planet . its outer wake carries positive fluxes of angular momentum and energy . a fraction of this angular momentum flux can be deposited in the coorbital region of the outer planet through the dissipation of shocks , and then transferred to the outer planet via an effective positive corotation torque ( see * ? ? ? associated with the angular momentum flux @xmath82 deposited in the coorbital region of the outer planet is an energy flux @xmath83 , where @xmath84 denotes the inner planet s mean motion ( or angular velocity ) . from this energy flux , @xmath85 is used to keep the outer planet and its coorbital region nearly circular ( see figure [ fig : fono ] ; here , @xmath86 is the outer planet s mean motion ) . the remaining energy flux , @xmath87 , is dissipated in the outer planet s coorbital region . reciprocally , a fraction of the negative inward flux of angular momentum carried by the inner wake of the outer planet can be deposited in the inner planet s coorbital region and subsequently transmitted to the inner planet . again , this exchange is associated with energy dissipation in the disk at a rate equal to the ( absolute value of ) the product of the flux of angular momentum transmitted to the inner planet by the outer one and the difference in the planets angular velocities . this dissipation occurs , as is usual in disk - planet interactions , through the tidally excited density waves steepening into shocks ( e.g. , * ? ? ? numerically , shocks are dealt with through a combination of numerical diffusion and the application of an artificial viscosity . as a locally isothermal equation of state is adopted in our simulations , this dissipated energy is effectively lost to the system through cooling . while the total orbital energy of the two planets is dissipated due to the wake - planet interactions described above , their total angular momentum remains unaltered . as studied in the context of star - planet tidal interactions , energy dissipation at constant angular momentum leads to the divergent evolution of a planet pair @xcite . we now derive an equation satisfied by the period ratio of the two planets as a direct consequence of considering the evolution of their total orbital energy and angular momentum . we denote by @xmath88 , @xmath89 and @xmath90 the semi - major axis , eccentricity and mass of the inner planet ( with the subscript changed to o denoting the same quantities for the outer planet ) . the total angular momentum of the two planets is @xmath91 which gives , correct to second order in the eccentricities , @xmath92 we are looking for changes in the planets semi - major axes and eccentricities that occur on similar timescales , we may therefore discard the eccentricity terms in the expressions for @xmath93 @xmath94 and @xmath95 . denoting by @xmath96 and @xmath97 the torques exerted by the disk on the inner and outer planets , respectively , conservation of angular momentum reads @xmath98 the total orbital energy is given by @xmath99 and conservation of energy reads @xmath100 where @xmath101 the planets mean motion . here the right hand side of eq . ( [ eq_e ] ) contains three contributions . the first arises from the energy dissipation in the disk resulting from the action of the torques @xmath102 as these are negative this energy is removed from the orbital motion . the second originates from orbital circularization due to disk - planet interactions . the quantities @xmath103 and @xmath104 are the circularization times for the planets that arise from the disk torque and energy dissipation acting on each planet through the effects of its own wake @xcite . the third and final contribution @xmath105 stems from the energy dissipation due to wake - planet interactions . we remark that wake dissipation in the coorbital region of the outer planet contributes @xmath106 to @xmath107 , in addition to a corresponding contribution arising from the coorbital region of the inner planet , as indicated above . note that terms of order @xmath108 are retained above even though the eccentricities are small as it is assumed that @xmath109 are small enough so that dissipation through circularization could be as important as , or even dominate the other effects under some circumstances . combining eqs . ( [ eq_j ] ) and ( [ eq_e ] ) , we find @xmath110 when wake - planet interactions are negligible ( @xmath111 ) , eq . ( [ eq_pr ] ) shows that , far from resonance , where the eccentricities are very small , the planet s period ratio @xmath112 decreases as a consequence of convergent migration , provided that @xmath113 as expected , the latter condition is that the inward migration rate of the outer planet should exceed the inward migration rate of the inner planet . as the eccentricities increase as resonance is approached , a steady state may be reached in which the planets period ratio and eccentricities take stationary values . if the eccentricity damping timescales are sufficiently short , disk - driven circularization of the orbits can reverse convergent migration , that is the planets period ratio can increase . ( [ eq_pr ] ) highlights that wake - planet interactions may also lead to divergent evolution of a planet pair . divergent evolution driven by wake - planet interactions does not require the planets to be resonantly coupled , as already inferred from the simulations shown in figure [ fig : fono ] . similar divergent evolution of a non - resonant planet pair embedded in a disk , comprising a jupiter and a super - earth , was reported by @xcite . when divergent evolution is primarily due to wake - planet interactions , as the period ratio of the planet pair increases away from a resonant value , the distance to resonance increases , and the forced eccentricities therefore decrease ( e.g. , * ? ? ? it implies that orbital circularization is a natural consequence of divergent evolution driven by wake - planet interactions . this is actually a major difference between the divergent evolutions driven by wake - planet interactions and by orbital circularization ( through the disk or stellar tides ) : eccentricity damping is the consequence of the former mechanism , while it is the cause of the latter . note that , with both mechanisms , the divergent evolution of a resonant planet pair maintains the libration of the resonant angles by damping the eccentricities . we call _ disk - driven resonant repulsion _ the divergent evolution of a resonant planet pair embedded in a disk , following the terminology introduced by @xcite for the divergent evolution of a resonant planet pair due to star - planet tidal interactions . we stress that disk - driven resonant repulsion can have two origins : ( i ) disk circularization of the orbits and ( ii ) wake - planet interactions . in other words , disk - driven repulsion comprises ( i ) _ circularization - driven repulsion _ and ( ii ) _ wake - driven repulsion_. the results of hydrodynamical simulations shown above indicate that , for the disk and planet parameters that we have considered , wake - planet repulsion dominates circularization - driven repulsion . this point will be further illustrated in sect . [ sec : nbody_method ] . + _ when is wake - driven repulsion important? _ we have seen above that the period ratio of a planet pair embedded in a disk may increase due to angular momentum transfer between the planets through their wakes . we further discuss in this paragraph under what conditions wake - planet interactions may lead to efficient divergent evolution of a planet pair . we separate the planet pair into a donor and a recipient . the bigger the donor s mass , the stronger its wake , and the easier for the wake to deposit angular momentum in the coorbital region of the recipient planet through the dissipation of shocks . this requires the dimensionless parameter @xmath17 of the donor planet to typically exceed unity ( @xmath18 is the planet - to - star mass ratio and @xmath114 the disk s aspect ratio ) . the donor planet can thus be a partial or a deep gap - opening planet . the angular momentum that is deposited by the donor s wake is ( at least partly ) transferred to the recipient planet through an effective corotation torque @xcite . if the recipient s mass is too big , the density in its coorbital region will be too low for this effective corotation torque to matter . in other words , the recipient could be a partial gap - opening planet or even a type - i migrating planet . these considerations show why divergent evolution is neither expected for two type - i migrating planets ( the donor s wake being too weak ) , nor for two type - ii migrating planets ( the recipient s gap being too deep ) . we have checked this with hydrodynamical simulations , not illustrated here . the above considerations also help understand the different rates of divergent evolution obtained in figure [ fig : ex ] . the fact that both planets open a partial gap is a favorable situation for wake - planet interactions to play a significant role . in the top row of panels , where the planets become locked in the 2:1 mmr , the high - density region between the two gaps ( see upper panel in figure [ fig : dens ] ) causes significant deposition of angular momentum in the planets coorbital region and therefore very efficient divergent evolution . in contrast , in the bottom row of panels , where the planets merge their gap and become locked in the 3:2 mmr , the moderate disk density left between the planets can only trigger slow divergent evolution . in the middle row of panels , where the period ratio stalls near 1.7 , a steady state is reached with background convergent migration balanced by wake - planet divergent evolution . + _ evolution towards a steady state _ when the planets are widely separated , they are expected to interact with the disk independently of each other and they can then enter a convergent migration phase . as they approach each other both planet - planet interactions , and wake - planet interactions where the planets interact with each others wakes can occur . both these processes involve exchanges of orbital energy and angular momentum between the planets . planet - planet interactions can result in resonant trapping associated with the growth of eccentricities , while wake - planet interactions act to produce divergent migration without the growth of eccentricities . from eq . ( [ eq_pr ] ) we see that both circularization- and wake - driven resonant repulsions act together to increase the period ratio . if wake - driven repulsion dominates after the eccentricities have damped to low values , with @xmath115 remaining constant , the period ratio should increase linearly with time , which is roughly what is obtained in the simulation presented in the top row of panels in figure [ fig : ex ] . however , in the longer term @xmath116 should decrease with time as the planets separate , which indicates that a steady state should be ultimately reached with convergent migration driven by background disk torques balancing wake - driven repulsion . we stress that wake - planet interactions cause a non - linear evolution of the disk s density profile near the planets , which may affect the time evolution of the background torques responsible for convergent migration . at least in the simulation without planet - planet interaction of figure [ fig : fono ] , a slow decrease in the background torques could explain why the period ratio decreases then increases , instead of reaching a stationary non - resonant value . another possibility is that convergent migration and disk - driven repulsion behave differently as the planet pair moves closer to the star , if the torques responsible for both mechanisms have different radial dependences . investigation of these aspects is deferred to a future study . the above simulations of the kepler-46 system show that the planets period ratio can take a range of values , depending on the convergent migration rate during the early evolution of the system . the limited duration of the simulations ( up to a few thousand orbits ) and the fact that planets still migrate at the end of the simulations however raise the question of the long - term evolution of the system . even when the inward migration of the planets gets stalled , further evolution of the period ratio can be maintained by disk - driven resonant repulsion . a steady state may _ a priori _ be achieved after depletion of the protoplanetary disk . disk dispersal typically occurs between @xmath117 and @xmath118 years after formation of the central star . photoevaporation driven by extreme ultraviolet radiation from the star opens a gap in the disk at separations of typically a few au for sun - like stars ( e.g. , * ? ? ? * ; * ? ? ? * ) . for planets below that separation the background disk will be depleted on up to a viscous timescale at the orbital separation where photoevaporation sets in . assuming a local viscous alpha parameter of a few @xmath119 , viscous draining of the inner disk could operate in a few @xmath120 years , that is typically a few @xmath121 orbital periods at the present location of kepler-46b and kepler-46c . to assess the impact of disk evaporation on our results , we have restarted two simulations adopting a simple exponential decay of the surface density profile . this is done by solving @xmath122 in addition to the hydrodynamical equations , where @xmath123 is the azimuthally - averaged density profile at restart time ( from which evaporation switches on ) , @xmath124 is an arbitrarily small density profile in a steady state ( taken to be not zero for numerical convenience ) , and @xmath125 is the evaporation timescale . for illustrative purposes , we have considered three short evaporation timescales : 100 , 500 and 2000 orbits , simulations being restarted at 1000 orbits . results are shown in figure [ fig : disp ] . the upper panels are obtained with @xmath126 and @xmath60 , for which a rapid divergent evolution occurs after capture in the 2:1 mmr . we see that the evolution is frozen out at a few evaporation timescales after restart , the planets reaching stationary eccentricities and semi - major axes . similarly , the lower panels in figure [ fig : disp ] show the results for @xmath127 and @xmath67 , for which convergent migration stalls at about the observed period ratio in the kepler-46 system . in that case , including disk evaporation stalls the migration of both planets , as expected , but it does not significantly affect the planets eccentricities and period ratio . the final period ratio remains very close to the observed value . in section [ sec : hydro ] we have presented results of hydrodynamical simulations modeling the early evolution of kepler-46b and kepler-46c as they were embedded in their parent protoplanetary disk . the variety of period ratios obtained in our simulations raises the question of whether disk - planets interactions could partly explain the diversity of period ratios in kepler s multi - planetary systems . we review below some of the processes that might account for such diversity . as highlighted in figure [ fig : kepler ] , kepler s multi - planetary systems show a clear tendency for planet pairs near resonances to feature period ratios slightly greater than strict commensurability . divergent evolution of a short - period resonant planet pair due to star - planet tidal interactions could partly explain this feature @xcite . although the efficiency of tidal dissipation remains largely uncertain , it is thought that efficient tidal circularization requires the inner planet to orbit its star in less than a few days . as can be seen in figure [ fig : kepler ] many of kepler s candidate systems have an inner planet beyond 10 days , for which tidal circularization is unlikely to have caused significant resonant repulsion ( see section [ sec : tidal ] of the appendix ) . for these systems we propose that resonant repulsion driven by disk - planets interactions could have played a prominent role in determining the observed period ratios . the vast majority of the planets in kepler s multi - planetary systems have physical radii between 0.1 and 0.3 jupiter radii , which places them in the super - earth to neptune - mass range . from the mass - radius diagram of kepler candidates followed up by radial velocity , the median mass of a 0.2 jupiter - radius planet is about 10 to 15 earth - masses ( see , e.g. , http://exoplanets.org ) . based on generally considered temperatures and viscosities in protoplanetary disks ( corresponding to @xmath128 and @xmath129 a few @xmath81 ) such planets are expected to experience type i migration . recent n - body experiments by @xcite have shown that convergent migration of type - i migrating planets yields very little departure from strict commensurability in contrast with kepler s data . he showed that the inclusion of stochastic forces in addition to type i migration could reproduce the observed distribution of period ratios quite nicely . such agreement has been obtained assuming that the initial period ratio distribution is the observed one , and taking the same convergent migration timescale and the same amplitude of stochastic forces for all kepler s multiple systems , which remains quite uncertain given the expected diversity of physical properties in protoplanetary disks . still , his results indicate that type i migration alone may not be able to account for kepler s diversity of period ratios . based on our results of hydrodynamical simulations for the kepler-46 system , we propose that some of kepler s planet candidates in multi - planetary systems could have opened partial gaps in their parent disk , thereby escaping the type i migration regime . this could be the case if , for instance , planets formed and/or migrated in regions of low turbulent activity ( dead zones ; see , e.g. , * ? ? ? * ) or if the disk s aspect ratio takes smaller values than commonly adopted . aspect ratios @xmath130 are likely typical of the short orbital separations at which kepler s candidate systems are detected . to address this possibility , we carried out hydrodynamical simulations with two planets in the super - earth mass range . the inner planet s mass @xmath131 , the outer planet s mass @xmath132 and the central star is one solar mass . the planet - to - primary mass ratios are therefore @xmath133 and @xmath134 , that is an order - of - magnitude smaller than those of kepler-46b ( median value ) and kepler-46c . in order to get similar gap depths in the super - earths and kepler-46 simulations , we adopted values of @xmath114 and @xmath23 such that the dimensionless parameters in the gap - opening criterion of @xcite take the same values . for the super - earths simulations , this yields @xmath135 and @xmath136 . all other simulation parameters are otherwise identical , except the grid resolution that we increased to @xmath137 . figure [ fig : m10 ] displays the results of two simulations . in the top panels , the unperturbed disk s surface density is @xmath138 . we see that the planets lock themselves in the 3:2 mmr from about 1300 orbits . divergent evolution proceeds with a decrease in the planets eccentricities . after reaching exactly 1.5 at about 1500 orbits , the period ratio increases to 1.54 at 5000 orbits . as illustrated by the screenshot of the disk surface density , the proximity of the planets allows their wakes to penetrate into each others coorbital region , where they can deposit part of their angular momentum flux . as we have shown in section [ sec : divergent ] , transfer of angular momentum between planets through wake - planet interactions will result in a loss rate of the total orbital energy of the two planets . this has the consequence of increasing the period ratio and decreasing the eccentricities . in the bottom panels of figure [ fig : m10 ] , the unperturbed disk surface density parameter is decreased to @xmath139 . here , convergent migration due to the background disk torques is slow enough to lock the planets in the 2:1 mmr , but this time the resonant capture does not lead to divergent evolution . instead , a steady state is reached in which both the eccentricities and the orbital period ratio take stationary values . this steady state is a consequence of negligible wake - planet interactions in this case . as illustrated in the bottom - right panel of figure [ fig : m10 ] , the planets are sufficiently far from each other that their wakes primarily transfer energy and angular momentum to regions of the disk that are not coorbital with the planets . the results of section [ sec : m10 ] support the idea that disk - driven resonant repulsion of partial gap - opening super - earths may lead to period ratios that substantially differ from nominal resonant values . as shown in section [ sec : divergent ] , this mechanism has two origins : circularization of the orbits and wake - planet interactions . its efficiency depends on the planets masses and the evolution of the disk s surface density . the survey of such a large parameter space is out of reach of hydrodynamical simulations , but three - body simulations with customized prescriptions for migration , eccentricity damping and disk dispersal can help illustrate the diversity of period ratios that can be achieved . this is the strategy that we have adopted . for illustration purposes we have considered fixed planet masses : @xmath140 , @xmath132 and @xmath141 ( as in section [ sec : m10 ] ) . our three - body simulations solve the equations of motion for the two planets and the central star using a standard technique . planets are assumed to be coplanar . disk - planet interactions are incorporated by applying appropriate dissipative forces ( for details , see * ? ? ? inspection at eq . ( [ eq_pr ] ) shows that , in order to model the planets evolution , we need to specify ( i ) the rate of convergent migration before disk - driven repulsion sets in , ( ii ) the damping rate of the eccentricities , ( iii ) the efficiency of wake - driven divergent evolution , and ( iv ) how disk dispersal is modeled . point ( iii ) remains uncertain at this stage , as the term @xmath142 in the right - hand side of eq . ( [ eq_pr ] ) , which models energy dissipation due to wake - planet interactions , should depend sensitively on the planet masses , their mutual separation and distance from the central star , and several disk quantities ( including its temperature and viscosity ) . a systematic study of this dependence goes beyond the scope of this paper , and is therefore left for future work . in order to do simple modeling , and to minimize the number of free parameters , we do not attempt to parameterize the @xmath142 term . instead , we model disk - driven repulsion as circularization - driven repulsion only . in other words , in eq . ( [ eq_pr ] ) we compensate the absence of the @xmath142 term in our three - body integrations by considering circularization timescales @xmath103 and @xmath143 that are shorter than in the hydrodynamical simulations . damping of the planets eccentricity is modeled as an exponential decay with a constant damping timescale to orbital period ratio , which we take as a free parameter ( specified below ) . to model the convergent migration prior to resonant repulsion , we adopt a simple prescription that reproduces the results of hydrodynamical simulations . denoting by @xmath144 the ratio of semi - major axes ( @xmath145 ) , we found good agreement using @xmath146 with @xmath147 \times \frac { ( t / \tau)^3 } { \sqrt{1+(t/\tau)^6 } } , \label{eq : f}\ ] ] where pr and @xmath148 are fitting parameters . the expression in eq . ( [ eq : prescmig ] ) tends to a constant value for @xmath149 , which mimics the fact that convergent migration is found to nearly stall before resonant repulsion sets in . in eq . ( [ eq : f ] ) , pr denotes the period ratio at which convergent migration approximately stalls , and @xmath148 the time at which this occurs . the good agreement between this migration prescription and our hydrodynamical simulations is illustrated in figure [ fig : fit ] . note that in the three - body simulations , disk - driven migration is applied to the outer planet only . we display in figure [ fig : john1 ] the results of a three - body simulation with @xmath150 and @xmath151 orbits , which mimics the convergent migration obtained in the hydrodynamical simulation of section [ sec : m10 ] with @xmath152 ( see right panel of figure [ fig : fit ] ) . in the top - left panel of figure [ fig : john1 ] , only the semi - major axis of the outer planet varies in the first 1500 orbits as our prescription for disk - driven migration is applied to the outer planet . once resonant repulsion sets in near 1500 orbits , the semi - major axis of the inner planet decreases slightly , that of the outer planet increases slightly as well . the eccentricity damping timescale here is 20 planet orbits , and this rather short timescale is found to give very similar divergent evolutions in the three - body and hydrodynamical simulations ( see top - right panel in figure [ fig : john1 ] ) . at 5000 orbits , the period ratio is 1.53 in the three - body run , and 1.54 in the hydrodynamical run . this small difference can be qualitatively explained by the fact that the repulsion occurs at a slightly larger pace in the hydrodynamical simulation than has been modelled in the three - body run . furthermore , we point out that the eccentricity peaks near the 2:1 and 3:2 mmrs are about a factor of 2 to 3 smaller in the three - body simulation ( compare with the eccentricities of the hydrodynamical run shown in the second upper panel of figure [ fig : m10 ] ) . the reason for this difference is most likely a reflection of the fact that the actual circularization time in the hydrodynamical run is longer than that adopted in the three - body integration , consistent with the operation of wake - planet repulsion in addition to circularization - driven repulsion as discussed above . we also incorporated a simple model for disk dispersal , which is taken to occur between @xmath120 and @xmath117 orbits . since planets are assumed to be already formed at the beginning of the simulations , this range of timescales is meant to account for different planet formation timescales . the longest time ( @xmath117 orbits ) corresponds to a case where both planet formation and capture into resonance following convergent migration occur rapidly . the shortest one ( @xmath120 orbits ) stands for the opposite . once disk dispersal switches on , the eccentricity damping timescale is progressively increased as an exponential growth with characteristic timescale @xmath153 orbits . this timescale is shorter than the typical viscous draining timescale of a disk inside the photoevaporation radius . it implies that the efficiency of disk - driven resonant repulsion is nearly entirely determined by the time from the beginning of the simulations at which dispersal switches on . we display in figure [ fig : john2 ] the results of a series of three - body simulations using the model described in section [ sec : nbody_method ] . recall that @xmath154 , @xmath132 and @xmath155 , which can be thought of being representative of kepler s sample of multi - planetary systems ( the planets mass ratio is arbitrary , it takes the same value as in the hydrodynamical simulations in section [ sec : m10 ] ) . planets are initiated with a period ratio of 2.4 . the prescription for convergent migration given at eq . ( [ eq : prescmig ] ) is used with @xmath156 and @xmath157 orbits . this value of pr is meant to explore the diversity of period ratios that can be achieved by disk - driven resonant repulsion away from the 3:2 mmr . the eccentricity damping timescale is varied from 20 to 200 orbits , and the time at which disk dispersal switches on from @xmath120 to @xmath117 orbits ( see section [ sec : nbody_method ] ) . given the duration of disk - driven resonant repulsion , our results are found to have very little dependence on the `` convergent migration timescale '' @xmath148 . simulations were carried out over @xmath158 orbits , and the period ratio time - averaged over the last 1000 orbits of the simulations is indicated by the color bar . for a given eccentricity damping timescale , the later disk dispersal occurs , the longer planets eccentricities are damped , and therefore the larger the final period ratio . at a given time prior to disk dispersal , the shorter the eccentricity damping timescale , the larger the final period ratio , again as expected . for the shortest damping timescales that we have considered , period ratios can reach @xmath159 1.75 . with such short timescales , planets can evolve continuously from the 3:2 mmr to the 2:1 mmr ( that is , the 2:1 resonant angles start to librate when the 3:2 resonant angles no longer librate ) . had we taken even shorter damping timescales or delayed disk dispersal even more , resonant repulsion would have led to period ratios exceeding 2 . the main conclusion that can be drawn from figure [ fig : john2 ] is that partial gap - opening super - earths may experience significant disk - driven resonant repulsion away from the nominal 3:2 mmr . under very favorable circumstances ( particularly efficient repulsion maintained over a particularly long timescale ) period ratios can easily exceed nominal resonant values . although we have considered fixed planet masses for illustration purposes , we believe that disk - driven resonant repulsion is a generic mechanism that may have likely occurred among kepler s multi - planetary systems . results of three - body simulations with prescribed convergent migration and eccentricity damping . the central star has a solar mass . the mass of the inner and outer planets is @xmath160 and @xmath161 , respectively . convergent migration is modeled using the expression in eq . ( [ eq : prescmig ] ) with pr = 1.5 and @xmath162 orbits . the eccentricity damping timescale is varied from 20 to 200 orbits ( y - axis ) , and it remains fixed until disk dispersal is switched on . the timescale at which disk dispersal switches on ranges from @xmath120 to @xmath117 orbits after the beginning of the simulations ( x - axis ) , and the evaporation timescale is set to @xmath163 orbits . the orbital period ratio obtained after @xmath158 orbits , and time - averaged over the last 1000 orbits , is colour - coded.,title="fig : " ] the multi - planetary systems detected by the kepler mission have very diverse architectures , which reflect the wide range of conditions under which planets may form and evolve . the quasi - coplanar orbits in kepler s multiple systems suggest that interactions between planets and their parent protoplanetary disk should play a prominent role in shaping such diverse architectures . in particular , close planet pairs feature a variety of orbital period ratios and many of them are not in mean - motion resonance . convergent migration of multiple planets due to disk - planet interactions may lead to resonant systems in disks where turbulence is low enough not to disrupt resonances @xcite . resonant planet pairs formed by disk - planet interactions may experience divergent evolution of their orbits due to tidal orbital circularization @xcite . this so - called _ tidal resonant repulsion _ may increase the orbital period ratio of close - in planet pairs quite substantially depending on the efficiency of star - planet tidal interactions @xcite . tidal resonant repulsion could help account for the diversity of period ratios in kepler s close - in planet pairs . we have shown in this paper that disk - planets interactions could also form planet pairs with a variety of period ratios . we have first focused on the kepler-46 planetary system , in which the two saturn - mass planets kepler-46b and kepler-46c have a period ratio @xmath164 @xcite . we have presented results of hydrodynamical simulations modeling the early evolution of these planets as they were embedded in their parent disk . for typical disk temperatures and turbulent viscosities , both planets are expected to open a partial gap around their orbit . a rapid convergent migration causes the planets to merge their gaps and to evolve into a common gap . depending on the density s structure inside the common gap , the planets convergent migration is found to stall with a variety of period ratios between 1.5 and 1.7 ( see figure [ fig : finalopr ] ) . the observed period ratio of kepler-46b / c can be reproduced without the planets being in mean - motion resonance . our results also highlight that disk - planets interactions may significantly increase the orbital period ratio of a planet pair from a near - resonant value , a mechanism that we term _ disk - driven resonant repulsion_. it has two origins : ( i ) orbital circularization by the disk , and ( ii ) the interaction between each planet and the wake of its companion . as the wake of a planet turns into a shock at some distance from the planet , it may deposit angular momentum in the coorbital region of another planet through an effective corotation torque . we have shown that this mechanism can dissipate the total orbital energy of a planet pair without changing its total angular momentum , a configuration that increases the planets period ratio and decreases their eccentricity . resonant repulsion caused by wake - planet interactions can be particularly efficient when at least one of the planets opens a partial gap around its orbit ( see section [ sec : divergent ] ) . for the disk and planet parameters adopted in this work , wake - planet interactions are much more efficient at driving the repulsion of a planet pair than disk circularization . however , there may be circumstances for which this balance is different . the strength of wake - planet interactions increases as the radial separation between planets decreases , independently of their eccentricity . if wake - planet interactions are the dominant process for the divergent evolution of a planet pair , then one could expect the period ratio to reach a constant value , which could be different from a resonant ratio . our results of simulations report , however , many cases where the period ratio nearly stalls but then increases . there could be two reasons for this behavior . first , wake - planet interactions cause a ( non - linear ) evolution of the disk s density profile near the planets , which causes the torques driving convergent migration and wake - planet interaction to change over time . furthermore , independently of the previous consideration , preliminary results indicate that the torques responsible for convergent migration and for wake - driven repulsion have different dependences on the stellar distance as the planets migrate . both effects are found to strengthen wake - driven repulsion as compared to convergent migration as planets migrate inwards . this would explain why , in our simulations , the period ratio of inwardly migrating planet pairs increases after reaching a nearly constant value . the above considerations indicate that wake - planet interactions may naturally increase the period ratio of planet pairs away from resonant values . one of our low - mass disk models for kepler-46 , in which the saturn - mass planets are captured in the 2:1 mmr , a period ratio of 2.3 is attained in as short as a few thousand orbits . we note that a significant divergent evolution of sub - jovian planets was obtained in some of the hydrodynamical simulations of @xcite and @xcite . the similarity between their results of simulations and ours ( convergent migration nearly stalled close to resonance , followed by divergent evolution of the planets orbits alongside the damping of their eccentricity ) suggests that the divergent evolution obtained in these simulations is likely to arise from wake - planet interactions . many planets in kepler s multiple systems are in the super - earth to neptune - mass range , and have orbital periods below 100 days . at these short periods the aspect ratio of a disk should be rather small , thus many kepler s planet candidates could have opened partial gaps in their parent disk . in contrast to the saturn - mass planets in the kepler-46 system , super - earths are not expected to merge their gaps and to evolve in a common gap , unless a very fast convergent migration brings them to very short mutual separations . we have argued instead that close - in super - earths are good candidates for efficient resonant repulsion caused by wake - planet interactions . for instance , one of our simulations shows that the orbital period ratio of two @xmath159 15 earth - mass planets increases from 1.5 to 1.55 in few thousand orbits . disk - driven repulsion via wake - planet interactions could explain why many planet pairs in kepler s multiple systems have period ratios slightly greater than resonant . while resonant repulsion driven by stellar tides should only be relevant to planet pairs with orbital periods below a few days , disk - driven repulsion may also apply to planets on longer periods . the ttv signals of kepler s short - period planet pairs indicate that some of these planets have small but not zero free eccentricities @xcite . free eccentricities get quickly damped by tidal orbital circularization . however , the presence of a disk of planetesimals or a non - smooth gas disk dispersal could give planets some free eccentricity . close - in planet pairs that seem to have undergone resonant repulsion , but which have not zero free eccentricities , like kepler-23b / c and kepler-28b / c , could indicate that repulsion was primarily mediated by disk - planets interactions rather than by star - planet tidal interactions . we foresee at least three directions to pursue work on disk - driven resonant repulsion . firstly , the effects of disk turbulence , which we have modeled with a constant viscosity , need to be addressed carefully . the levels of stochastic fluctuations required to disrupt the process as planet masses decrease , and to produce some residual free eccentricity should be explored . secondly , planet mass growth should be taken into account as it may also lead to some divergent evolution @xcite . thirdly , a monte - carlo approach with three - body integrations using simple prescriptions for the effects of disk - planet interactions should be used to compare the distribution of orbital elements among kepler s planet pairs with the synthetic distributions predicted by disk - driven resonant repulsion . cb is supported by a herchel smith postdoctoral fellowship of the university of cambridge . we thank gwenal bou , cathie clarke , jrme guilet , cristobal petrovich and yanqin wu for useful discussions . we also thank the referee for a very helpful report . hydrodynamical simulations were performed on the darwin supercomputer of the university of cambridge high performance computing service using strategic research infrastructure funding from the higher education funding council for england . we show here that star - planet tidal interactions are unlikely to cause significant divergent evolution of two planets in the super - earth mass range when the orbital period of the inner planet exceeds about 10 days . in the constant time lag model , taking tidal dissipation within the planet to be much larger than within the star , the orbital circularization timescale , @xmath165 , is given by equation ( 25 ) of @xcite and can be expressed as follows : @xmath166 where @xmath167 is the tidal quality factor of the inner planet , @xmath168 its mean density , @xmath169 its physical radius , @xmath170 its orbital period and @xmath47 is the star s mass . divergent evolution of two resonant planets may occur as a result of tidal dissipation of orbital energy at approximately constant orbital angular momentum @xcite . the timescale for divergent evolution , which we denote by @xmath171 , satisfies @xmath172 with @xmath88 and @xmath89 the semi - major axis and forced eccentricity of the inner planet , and @xmath173 its change in semi - major axis during divergent evolution ( it can be taken to be the planet s distance from resonance ) . we have @xmath174 , where @xmath175 is the outer planet s mass , @xmath90 the inner planet s mass ( @xmath176 denotes the outer - to - inner planet mass ratio ) , and @xmath177 is a constant of order unity that varies with the resonance ( see , e.g. , equation 38 of * ? ? ? * ) . using eq . ( [ eq_app1 ] ) and taking @xmath178 , we obtain @xmath179 this order of magnitude estimate shows that planet pairs orbiting sun - like stars should experience negligible tidally - driven resonant repulsion over gigayear timescales when the inner planet is a super - earth , with @xmath180 @xmath181 and orbital period greater or equal to 10 days , even if @xmath182 and @xmath183 . moreover , if the inner planet is a jupiter - like planet , then assuming @xmath184 and @xmath185 g@xmath77 , eq . ( [ eq_app2 ] ) shows that divergent evolution of more than a few percent away from resonance is expected over a few gigayears as long as the inner planet s orbital period does not exceed 10 days . in the course of our numerical experiments , we have tested the robustness of our results by varying the assumptions made in the physical model and by varying the grid resolution . _ physical model _ we performed several simulations reducing the disk s aspect ratio to 4% and 3% while keeping the same viscous alpha parameter . although not illustrated here , these simulations show again a variety of outcomes , but none reproducing the observed period ratio between kepler-46c and kepler-46b . decreasing the disk s aspect ratio makes the gaps carved by the planets deeper . when convergent migration is rapid enough for the planets to cross their 2:1 mmr and to evolve in a common gap , we find that the planets generally lock themselves into the 3:2 , 4:3 or even 7:5 mmr with very little departure from exact commensurability . in these simulations , the density in the common gap is too low for wake - planet interactions to cause significant divergent evolution ( see section [ sec : divergent ] ) . these simulations thus suggest that partial gap opening is required to obtain a period ratio consistent with the observed value . furthermore , as the structure of the gaps plays a prominent role in the final evolution of the planets period ratio , we carried out additional simulations where the planets were first held on fixed circular orbits for 500 orbits before migrating . this preliminary stage allowed enough time for each planet to build up a gap with a stationary density structure . the results of the simulations are shown in the left panel of figure [ fig : gapres ] for @xmath78 and three values of @xmath25 . starting with pre - evolved gaps changes the initial rate of convergent migration . notwithstanding quite different initial evolutions , two out of these three models end up following the same evolution with or without pre - evolved gaps , and reach the same final period ratio at the end of the simulations . the model with @xmath186 reaches a slightly smaller period ratio ( decreased from 1.65 to 1.6 when starting with pre - evolved gaps ) . + _ grid resolution _ the effect of resolution was checked by doubling the number of grid cells along each direction for a few simulations . the results of three high - resolution runs ( @xmath187 ) are compared to the nominal - resolution runs ( @xmath188 ) in the right panel of figure [ fig : gapres ] . doubling the grid resolution slightly changes the initial rate of convergent migration , leading to slightly different initial evolution . as for the above series of simulations with pre - evolved gaps , the model with capture into 2:1 mmr followed by resonant repulsion is not affected by doubling the resolution . the two models with common gap evolution seem again more susceptible to a variation in the initial convergent migration , but lead to small differences in the final outcome . interestingly , both high - resolution models with @xmath189 and @xmath190 yield a period ratio @xmath191 , in very good agreement with the observed value .	the kepler mission is dramatically increasing the number of planets known in multi - planetary systems . many adjacent planets have orbital period ratios near resonant values , with a tendency to be larger than required for exact first - order mean - motion resonances . this intriguing feature has been shown to be a natural outcome of orbital circularization of resonant planetary pairs due to star - planet tidal interactions . however , this feature holds in multi - planetary systems with periods longer than ten days , for which tidal circularization is unlikely to provide efficient divergent evolution of the planets orbits . gravitational interactions between planets and their parent protoplanetary disk may instead provide efficient divergent evolution . for a planet pair embedded in a disk , we show that interactions between a planet and the wake of its companion can reverse convergent migration , and significantly increase the period ratio from a near - resonant value . divergent evolution due to wake - planet interactions is particularly efficient when at least one of the planets opens a partial gap around its orbit . this mechanism could help account for the diversity of period ratios in kepler s multiple systems comprising super - earth to sub - jovian planets with periods greater than about ten days . diversity is also expected for pairs of planets massive enough to merge their gap . the efficiency of wake - planet interactions is then much reduced , but convergent migration may stall with a variety of period ratios depending on the density structure in the common gap . this is illustrated for the kepler-46 system , for which we reproduce the period ratio of kepler-46b and c.
materials with potentially useful thermoelectric properties are currently under intense theoretical and experimental investigation , mainly due to the prospect of applications , e.g. , for conversion of waste heat into electricity in thermoelectric generators , for applications to refrigeration , or for on - chip cooling and energy efficiency in microelectronics applications@xcite . apart from possible applications , thermoelectric materials can also serve as an interesting testing ground for theoretical approaches to electrical and thermal transport in solids@xcite . as the scale of the individual components in semiconducting devices is approaching the nano - size , a description of thermal transport through quantum dots is also attracting a lot of experimental and theoretical attention@xcite . in this paper we address the thermoelectric properties of a nanoscale size quantum dot exhibiting the kondo effect , which we describe in terms of a single level anderson impurity model with two conduction electron leads at fixed chemical potentials . the quantum dots that we consider have sizes of @xmath23 and can be tuned from the kondo to the mixed valence and empty orbital regimes by a gate voltage @xcite . short segments of carbon nanotubes @xcite connected to leads exhibit similar physics , so our results could also be of relevance to such systems . very recent experiments on nanoscale quantum dots @xcite are beginning to probe the effect of kondo correlations on the thermopower , although as we shall argue in the conclusions a quantitative comparison with theory is still some way off . the thermoelectric properties of dilute magnetic impurities in non - magnetic metals , such as ce@xmath24la@xmath25al@xmath26 and ce@xmath24la@xmath25b@xmath27 , are closely related to those of quantum dots ( see sec . [ model ] ) so we discuss these here also . understanding the thermoelectric properties of magnetic impurities is also a useful starting point for understanding those of heavy fermions within the dynamical mean field theory approach@xcite although , in these systems , crystal field effects and non - resonant channels play a crucial role for the thermopower , and need to be taken into account for a quantitative comparison to experiment @xcite . the approach that we use in this paper , wilson s numerical renormalization group ( nrg ) method @xcite , gives reliable results for transport properties in all parameter and temperature regimes of interest@xcite . the present calculations were carried out for the anderson model with finite coulomb repulsion , as is appropriate for nanoscale size quantum dots . we implemented recent developments in the calculation of dynamical quantities within the nrg , including the use of the self - energy@xcite and the full density matrix ( fdm ) generalization @xcite ( see also ref . ) of the reduced density matrix approach@xcite within the complete basis set of eliminated states@xcite . in particular , the fdm approach allows calculations of dynamical properties at all excitation energies @xmath28 relative to the temperature @xmath3 , thereby simplifying the calculation of transport properties which require knowledge of excitations , @xmath28 , above and below the temperature @xcite . the outline of the paper is as follows . in sec . [ model]-[transport ] we describe the anderson impurity model for quantum dots and dilute magnetic impurities and we specify the relevant transport quantities that we calculate for these two different physical realizations of the model . the nrg method used in this paper is described in sec . [ method ] together with results for occupancies which we use to define kondo , mixed valence and empty orbital regimes in the strong correlation limit . [ tdependence ] presents the temperature dependent transport properties of quantum dots and sec . [ tdependence - compare ] compares these to the corresponding quantities for dilute magnetic impurities . results for the figure of merit , power factor and lorenz number ratios for quantum dot and magnetic impurity systems are presented in sec . [ zt ] . [ scaling ] investigates the extent to which universal scaling functions apply to the thermopower and thermal conductance of quantum dots in the kondo regime . in sec . [ gate - dependence ] we present our results for the gate voltage ( local level ) dependence of transport quantities for quantum dots ( magnetic impurities ) . conclusions and a discussion of the relevance of our results to recent experiments on nanoscale size quantum dots is presented in sec . [ conclusions ] . appendix [ reduction ] discusses the reduction of the two - lead anderson model to a single channel model , appendix [ extra - results ] contains some additional results for moderately and weakly correlated quantum dots , and appendix [ full - density - matrix ] provides details of the fdm approach @xcite and an alternative detailed derivation of the fdm expression for local green s functions , which we have used to obtain the results in this paper . finally appendix [ transport - derivations ] gives an outline of the derivation of thermopower and thermal conductance for quantum dots . a nanoscale quantum dot is described by the single level anderson impurity model with two conduction electron leads @xmath29 here , @xmath30 is the kinetic energy of conduction electrons with wavenumber @xmath31 and spin @xmath32 in lead @xmath33 , @xmath34 is the local level energy , @xmath5 is the coulomb repulsion on the dot and @xmath35 is the tunnel matrix element of the dot level to conduction electron states in lead @xmath33 . the operators @xmath36 create ( destroy ) conduction electron states @xmath37 and @xmath38 create ( destroy ) local d - level states @xmath39 . we assume a flat density of states of magnitude @xmath40 per spin channel for both leads , where @xmath41 is the half bandwidth of each lead . the single - particle broadening ( half - width at half - maximum ) of the d - level is given by @xmath42 , where @xmath43 are the contributions to the broadening from the left and right leads . in this paper , we follow the convention used in quantum dot work and use as unit of energy not @xmath44 , but the full - width at half - maximum @xmath45 . since the @xmath46-state of the quantum dot in ( [ qdot - two - leads ] ) only couples to the even combination @xmath47 of the lead electron states , one can show ( see appendix [ reduction ] ) that , to a very good approximation , the above model can be reduced to the following single - channel anderson model @xmath48 where the tunneling amplitude @xmath49 is given by @xmath50 , so that the hybridization strength of the dot to the leads is given by @xmath42 . this is also the appropriate model for describing dilute magnetic impurities in non - magnetic metals@xcite . in fact , for both systems ( see below and appendix [ reduction ] ) the calculation of the linear transport properties reduces to the calculation of the equilibrium d - level spectral density of the single - channel model @xmath51,\ ] ] where @xmath52 is the fourier transform of the retarded d - level green function of ( [ siam ] ) . hence , all results in this paper , including those for dilute magnetic impurities , are obtained by solving the single - channel model ( [ siam ] ) using the nrg ( as explained in sec . [ method ] ) to obtain @xmath53 . thermoelectric transport through the quantum dot ( [ qdot - two - leads ] ) is calculated for a steady state situation in which a small external bias voltage , @xmath54 , and a small temperature gradient @xmath55 is applied between the left and right leads . left and right leads are then at different chemical potentials @xmath56 and @xmath57 , and temperatures @xmath58 and @xmath59 , with @xmath60 and @xmath61 . we follow the approach for deriving the electrical conductance , @xmath21 , the thermal conductance , @xmath62 , and thermoelectric power , @xmath18 , through an interacting quantum dot@xcite using the non - equilibrium green s function formalism . for completenes , an outline of this derivation @xcite can be found in appendix [ transport - derivations ] . the final expressions are given by @xmath63 \label{transport - expressions - k}\end{aligned}\ ] ] where @xmath64 are the transport integrals @xmath65 here , @xmath66 denotes the magnitude of the electronic charge and @xmath67 denotes planck s constant . the quantity @xmath68 is related to the spectral density @xmath69 via @xmath70 at @xmath71 , the conductance acquires the value @xmath72 where @xmath73 is the occupancy of the dot and we have used the friedel sum rule , @xmath74 for integer occupation , @xmath75 , and equal coupling to the leads , @xmath76 , the conductance reaches the unitary value @xmath77 , which we henceforth denote by @xmath78 . it is of interest to compare the transport properties of a quantum dot , with the corresponding quantities for electrons scattering from a dilute concentration , @xmath79 , of magnetic impurities in a clean host metal with constant density of states @xmath80 per spin . as for quantum dots , the relevant model for such dilute magnetic impurities is the single channel anderson model ( [ siam ] ) with hybridization strength @xmath44 . in order to obtain the thermopower , @xmath18 , the thermal conductivity , @xmath81 , and resistivity , @xmath22 ( or conductivity @xmath82 ) for such a dilute concentration of magnetic impurities we use the kubo formalism , see appendix a of ref . for the details , and find for these quantities @xmath83.\label{imp3}\end{aligned}\ ] ] the transport integrals @xmath84 , appearing here , are now defined by @xmath85 where @xmath86 is the transport time of electrons , which is given in terms of the impurity spectral density @xmath53 by @xmath87 in order to compare impurity transport properties with those of quantum dots , using the same units , we shall use rescaled quantities , e.g. for quantum dots @xmath88 and @xmath89 and for impurities @xmath90 and @xmath91 where @xmath92 is the unitary resistivity of electrons scattering from a dilute concentration @xmath93 of magnetic impurities . while the physics governing the transport properties of electrons scattering from dilute magnetic impurities , described by ( [ siam ] ) , is expected to be similar to that governing the transport properties of electrons through quantum dots ( also described by ( [ siam ] ) ) , differences are also expected , particularly for the respective thermopowers or the thermal conductance ( conductivity ) , for the following reason : the transport expressions for quantum dots arise from integrals @xmath64 which involve the @xmath94th moments of @xmath53 convoluted with the derivative of a fermi function , whereas those for magnetic impurities arise from nth moments of @xmath95 convoluted with the same derivative . at low temperatures , a sommerfeld expansion for @xmath96 and @xmath97 results in different signs for the thermopower in the two different situations , since derivatives of @xmath98 and @xmath99 have opposite signs . on the other hand , at higher temperatures , moments of @xmath98 and @xmath99 are determining factors for transport . particularly the moments @xmath96 ( @xmath97 ) , entering the thermopower , and @xmath100 ( @xmath101 ) , entering the thermal conductance ( conductivity ) , probe differences in the behavior or @xmath53 and @xmath95 at high temperature . consequently we expect significant quantitative differences for the thermopower and thermal conductance ( conductivity ) of quantum dots and dilute magnetic impurities at high temperatures . we discuss these differences in sec . [ tdependence - compare ] . in the kondo ( solid lines ) , mixed valence ( dashed lines ) and empty orbital ( dotted lines ) regimes . we define these regimes by @xmath102 , @xmath103 and @xmath104 , respectively . ] , of , ( a ) , the occupancy of the quantum dot , @xmath73 , and , ( b ) , the double occupancy , @xmath105 , at several temperatures . ] we calculate the spectral function @xmath53 and the transport properties of quantum dots , by using the nrg approach@xcite . this method is numerically exact and can be used to calculate both static thermodynamic properties as well as finite temperature dynamic and transport properties @xcite . in brief , the nrg procedure @xcite consists of the following steps , ( i ) , a logarithmic mesh of @xmath106 is introduced about the fermi level @xmath107 , and , ( ii ) , a unitary transformation of the @xmath108 in ( [ siam ] ) is performed such that @xmath109 is the first operator in a new basis , @xmath110 , which tridiagonalizes @xmath111 , i.e. @xmath112 , where the hoppings @xmath113 for a flat conduction band@xcite . the hamiltonian ( [ siam ] ) with the above discretized form of the kinetic energy is now iteratively diagonalized by defining a sequence of finite size hamiltonians @xmath114 for @xmath115 up to a maximum chain length @xmath116 . for each @xmath117 , this yields the excitations @xmath118 and many body eigenstates @xmath119 of @xmath120 at a corresponding set of energy scales @xmath121 defined by the smallest scale in @xmath120 , @xmath122 . since the number of states grows as @xmath123 , for @xmath124 only the lowest 600 or so states of @xmath120 are retained . these are used as a basis for constructing @xmath125 . for @xmath124 , both the retained and eliminated ( high energy ) states of @xmath126 , together with the corresponding eigenvalues , are stored . this information is subsequently used to evaluate the spectral function @xmath53 within the fdm approach @xcite described in appendix [ full - density - matrix ] . this evaluation makes use of , ( i ) , the completeness of eliminated states @xcite , allowing a multiple - shell evaluation of green s functions @xcite , avoiding double counting of excitations , and , ( ii ) , the reduced density matrix approach to greens functions , introduced to the nrg by hofstetter @xcite . in addition , we calculate the spectral function via the correlation part of the self - energy @xmath127 following bulla , pruschke and hewson @xcite , via @xmath128,\nonumber\\ \sigma(\omega , t ) & = & u \frac{\langle\langle n_{d,-\sigma}d_{\sigma } ; d_{\sigma}^{\dagger}\rangle\rangle}{\langle\langle d_{\sigma};d_{\sigma}^{\dagger}\rangle\rangle } \equiv u \frac{f_{\sigma}(\omega , t)}{g_{d\sigma}(\omega , t)}\label{correlationgf}\end{aligned}\ ] ] since the fdm entering the definition of the green s functions , see appendix [ full - density - matrix ] , contains the complete spectrum from all nrg iterations , asymptotically high and low temperatures can be investigated more easily than within previous approaches @xcite , which involved at a given temperature @xmath3 , choosing an appropriate energy shell to extract @xmath53 . in addition , the regime @xmath129 , which was problematical in previous approaches , can now be addressed , since contributions from all excitations ( for all energy shells ) are taken into account in the expression for the green s function within the fdm approach . the calculations reported here have been carried out for a discretization parameter @xmath130 , retaining @xmath131 states per nrg iteration and a hybridization strength @xmath132 ( in units of the half - bandwidth @xmath41 ) . the maximum chain length diagonalized was @xmath133 . we use the full width @xmath134 as our energy unit throughout . results for a wide range of temperatures from @xmath135 to @xmath136 were obtained to fully characterize the transport properties of quantum dots and dilute magnetic impurities . we note that , in practice , the regime @xmath136 is probably not accessible in experiment due to other effects which become important at high temperature , and which we do not take into account , e.g. phonons , multiple levels , crystal field states etc . calculations for strong ( @xmath137 ) , moderate ( @xmath138 ) and weak ( @xmath139 ) correlations were carried out for a range of dimensionless gate voltages , @xmath4 , defined by @xmath140 with this definition , the gate voltage for mid - valley occurs at @xmath141 for all @xmath5 . due to particle - hole symmetry , calculations were carried out for @xmath142 , with those for @xmath143 being obtained via a particle - hole transformation . this results in @xmath144 , occupancy @xmath145 , double occupancy @xmath146 , thermopower @xmath147 , with @xmath2 and @xmath0 remaining unchanged . the behavior of @xmath1 , @xmath2 and @xmath0 under @xmath144 , follows from their definition and the behavior of the spectral function , @xmath148 under @xmath144 . [ figure1 ] shows the temperature dependence of the dot level occupancy , @xmath149 , for gate voltages in the kondo , mixed valence and empty orbital regimes , for @xmath150 . for @xmath151 we use the occupancy at @xmath71 to delineate between the different regimes . specifically , the kondo regime is defined by gate voltages around mid - valley ( @xmath152 ) with @xmath153 ( see caption to fig . [ figure1 ] ) . similarly , the mixed valence and empty orbital regimes are defined by gate voltages corresponding to @xmath154 and @xmath155 , respectively ( see fig . [ figure1 ] ) . in the kondo regime , a characteristic low temperature scale , the kondo scale @xmath156 , can be defined via @xcite @xmath157 where @xmath158 . the mid - valley kondo scales for @xmath159 and @xmath160 , are @xmath161 and @xmath162 , respectively . within the fdm approach , the thermodynamic value of the dot occupancy @xmath163 $ ] , where @xmath20 is the fdm defined in appendix [ full - density - matrix ] , and the value obtained from the spectral sum rule @xmath164d\omega$ ] , with @xmath165 defined in ( [ correlationgf ] ) , are identical by construction , as we also verified numerically . [ figure2 ] shows the gate voltage dependence of the dot occupancy ( and for completeness also the double occupancy @xmath105 ) at a number of temperatures for the strong correlation case @xmath166 . before presenting the results , a few words are in order concerning the physical significance of the kondo , mixed valence and empty orbital regimes for strong coulomb correlations ( @xmath151 ) on the dot ( for more detailed information we refer the reader to ref . ) . the kondo regime , @xmath167 , corresponds to the formation of a localized spin on the dot at intermediate temperatures ( @xmath168 ) . in this temperature range , physical properties exhibit logarithmic temperature dependences , the hall - mark of the kondo effect . at @xmath169 the localized spin is quenched by the lead electrons , resulting in a many - body singlet at @xmath71 and a narrow kondo resonance ( of width @xmath156 ) in the dot spectral density at the fermi level . physical properties are characterized by spin fluctuations on scales @xmath170 , charge fluctuations on a scale @xmath171 and renormalized fermi liquid excitations at @xmath172 . the dot spectral density is well understood @xcite : it has a three peaked structure , with single - particle charge excitations at @xmath34 and @xmath173 and a temperature dependent kondo resonance at the fermi level . the mixed valence regime , corresponds to gate voltages such that the level @xmath34 is within @xmath44 of the fermi level . the charge on the dot fluctuates between @xmath174 and @xmath75 resulting in an average charge @xmath175 . the physics is governed by quantum mechanical charge fluctuations on a scale set by @xmath44 . the empty orbital regime , corresponds to @xmath176 and @xmath177 . physical properties are dominated by charge fluctuations , primarily via thermal activation ( with an activation energy @xmath34 ) . even though @xmath178 , the physics in this regime corresponds to that of a non - interacting resonant level model with a resonant level of width @xmath44 at energy @xmath179 . the temperature dependence of transport properties of a quantum dot described by the model ( [ qdot - two - leads ] ) is shown in fig . [ figure3 ] for several values of the gate voltage , ranging from the kondo regime ( fig . [ figure3]a , d , g ) , to the mixed valence ( fig . [ figure3]b , e , h ) and empty orbital ( fig . [ figure3]c , f , i ) regimes and for strong coulomb correlations on the dot ( @xmath180 ) . moderate to weak correlations are described briefly in sec . [ moderate - correl ] and appendix [ extra - results ] . depending on the regime , the transport properties exhibit different characteristic temperature dependences , which we describe in detail below for each transport property in turn . here , and in several other figures in the paper , we use arrows to indicate the evolution , with increasing gate voltage @xmath181 about mid - valley ( @xmath182 ) , of the various transport properties . the general trends in the electrical conductance @xmath21 of kondo correlated quantum dots are well understood@xcite : in short , as @xmath183 , the conductance approaches a maximum value ( see fig . [ figure3]a ) , indicating that the quantum dot appears `` transparent '' to electrons tunneling through it , and a logarithmic behavior around @xmath184 marks the crossover from the weakly coupled regime at @xmath185 to the strongly coupled regime at @xmath172 . an issue , less discussed in the literature , which we point out here , is the appearance of a finite temperature peak in the conductance , @xmath186 , on entering the mixed valence regime ( see fig . [ figure3]b ) . this feature becomes particularly pronounced in the empty orbital regime ( see fig . [ figure3]c ) . this effect has been observed in experiments on lateral quantum dots @xcite and a comparison to theoretical calculations shows good agreement@xcite ( see also the discussion of the resistivity of dilute magnetic impurities in sec . [ tdependence - compare ] ) . the thermopower exhibits a particularly interesting temperature dependence in the kondo regime , fig . [ figure3]d , with two sign changes at @xmath187 and @xmath188 , and , correspondingly , three extrema at @xmath189 , @xmath190 and @xmath191 . the detailed behavior of @xmath192 and @xmath193 as a function of gate voltage will be described below ; here , it suffices to note that neither @xmath192 nor @xmath193 are low energy scales , and @xmath193 is typically on a scale of order @xmath13 ( see fig . [ figure3]d and fig . [ figure4]a below ) . the low temperature `` kondo '' peak in @xmath18 at @xmath189 is found to scale with @xmath156 ( as defined in eq . ( [ haldanetk ] ) ) , as shown in the inset to fig . [ figure3]d . thus , in contrast to @xmath192 and @xmath193 , @xmath194 can be considered a low energy scale in the kondo regime . the central positive peak in @xmath18 first grows with positive magnitude on moving away from the kondo regime ( fig . [ figure3]d ) and then decreases in magnitude in fig . [ figure3]e on entering the mixed valence regime . simultaneously , the `` kondo '' peak in @xmath18 acquires a large negative value while merging with the high energy ( negative ) peak at @xmath191 on entering the mixed valence regime ( fig . [ figure3]e ) . well into the mixed valence regime , the thermopower exhibits a single negative peak on a scale @xmath13 with a distinct shoulder at higher temperatures due to the peak at @xmath191 . this picture continues to hold in the empty orbital regime ( see fig . [ figure3]f ) , with the shoulder at @xmath191 having almost disappeared . the thermopower remains negative for all gate voltages @xmath142 in this regime . the above behavior in the temperature dependence of the thermopower in the kondo regime is explained in terms of the structure of the single - particle excitations in @xmath53 . at low temperatures , a sommerfeld expansion for @xmath18 gives@xcite @xmath195 showing that the sign of the thermopower depends on the slope of the spectral density at the fermi level . for @xmath172 and @xmath142 , the kondo resonance lies above the fermi level , so the slope of the spectral density at the fermi level is positive , resulting in a negative thermopower . this remains true on further increasing the temperature , but as shown in ref . , eventually the kondo resonance is suppressed at @xmath196 resulting in a negative slope of the spectral density at @xmath197 for @xmath142 ( with the opposite being true for @xmath143 ) . consequently , the thermopower changes sign at the temperature @xmath192 which roughly corresponds to the temperature at which the kondo resonance vanishes . at @xmath185 , the determining factor for the sign of the thermopower is no longer the slope of the spectral function at @xmath197 , but the number of states available below or above the fermi level . these determine the overall sign of the transport integral @xmath96 in the expression for the thermopower in eq . ( [ transport - expressions - s ] ) . for @xmath142 , there are @xmath198 states below the fermi level and @xmath199 states above the fermi level . consequently , the integral of @xmath200 for @xmath201 is greater than its counterpart for @xmath202 , so @xmath203 and the thermopower is again negative at @xmath185 . this occurs at @xmath204 , which is found to be of order @xmath13 ( see below ) . due to the factor @xmath205 coming from the derivative of the fermi function in @xmath96 , the negative thermopower at @xmath206 acquires a maximum negative value and then decreases as @xmath205 at @xmath207 , exhibiting no further sign changes , as confirmed also numerically . we note that , away from half - filling ( @xmath152 ) , the modified second order perturbation in @xmath5 approach@xcite gives an incorrect sign for the slope of the spectral density at the fermi level in the kondo regime . this results in a wrong sign for the thermopower at @xmath169 in the kondo regime@xcite compared to our nrg calculations ( which agree with those of ref . ) . approximate approaches using an infinite @xmath5 anderson model @xcite could also not access the low temperature kondo regime . ( filled symbols , solid lines ) and @xmath193 ( open symbols , dashed lines ) at which @xmath18 changes sign as a function of @xmath208 for @xmath209 ( left panel : quantum dot , right panel : magnetic impurity ) . the critical gate voltage , @xmath15 , beyond which no sign change occurs in @xmath18 at finite @xmath4 is indicated in the legend for each case . ] the sign changes of the thermopower of strongly correlated quantum dots at the temperatures @xmath192 and @xmath193 in the kondo regime , are particularly interesting . they provide a `` smoking gun '' signature for kondo behavior in quantum dots , and could be used in future experiments as sensitive probes of strong correlations and kondo physics . it is therefore interesting to give a detailed characterization of the dependence of @xmath192 and @xmath193 on gate voltage and interaction strength @xmath16 . we show in fig . [ figure4]a - b the loci of @xmath192 and @xmath193 as a function of @xmath142 for quantum dots and magnetic impurities for three interaction strengths . although @xmath18 vanishes at @xmath152 , @xmath192 and @xmath193 have finite limiting values there . these are difficult to determine numerically due to the vanishingly small thermopower in this limit , and they are difficult to obtain analytically , since @xmath192 and @xmath193 lie outside the fermi liquid regime where analytic calculations are possible . estimates of these values at the smallest gate voltage are close to the limiting values . they are tabulated in table [ tabletk ] , together with the relevant kondo scales at mid - valley . whereas the limiting values of @xmath192 are comparable for both quantum dots and magnetic impurities , the limiting values of @xmath193 for quantum dots are approximately twice larger than for magnetic impurities . by carrying out additional calculations , using a finer grid of gate voltages , we determined the critical gate voltages @xmath210 , beyond which no sign change occurs ( indicated in fig . [ figure4]a - b ) . for each value of @xmath16 , we find that @xmath15 corresponds to entering the mixed valence regime , i.e. @xmath15 corresponds to a local level position @xmath211 in the single channel anderson model . . kondo temperature @xmath156 ( in units of @xmath13 ) at mid - valley @xmath152 ( symmetric point ) for @xmath212 . also shown are the temperatures @xmath192 and @xmath193 ( in units of @xmath13 ) at which the thermopower , @xmath18 , changes sign at @xmath213 ( the smallest @xmath4 studied ) . the numbers in brackets are the corresponding temperatures for magnetic impurities . [ cols="^,^,^,^",options="header " , ] the electronic contribution to the thermal conductance of a strongly correlated quantum dot , shown in fig . [ figure3]g - i , also exhibits interesting behavior : a crossing point at @xmath171 is found in the kondo regime and for gate voltages approaching the mixed valence regime ( fig . [ figure3]g and inset ) . such ( approximate ) crossing points are typical signatures of strong correlations and are well known in other contexts , including @xmath214 and heavy fermions@xcite , dissipative two - level systems @xcite and doped mott - insulators@xcite . on entering the mixed valence and empty orbital regimes ( fig . [ figure3]h - i ) , two - peaks develop on either side of the crossing point ( the lower peak being at @xmath215 and the upper one at @xmath216 ) . these qualitative features in @xmath62 can be related to @xmath53 , as in the case of @xmath18 ( see also sec . [ compare - thermal - cond ] ) . the effect of reducing correlations to a moderate value , @xmath217 , is shown in fig . [ figure11 ] of appendix [ extra - results ] : the trends are similar to those described above , with a significantly diminished kondo regime . in particular , the evolution with gate voltage of @xmath21 is similar to that in the strongly correlated case ( see fig . [ figure11]a - c ) and the thermopower exhibits two sign changes as a function of temperature in the kondo regime ( fig . [ figure11]d ) , with a rapid evolution to a single negative peak in the mixed valence and empty orbital regimes ( fig . [ figure11]e - f ) . however , the crossing point in @xmath0 in the kondo regime becomes less evident for moderate correlations ( fig . [ figure11]g ) , and , the two - peaked structure for @xmath0 in the mixed valence and empty orbital regimes is replaced by a single peak with a shoulder ( fig . [ figure11]h - i ) . these general trends , for correlated quantum dots , contrast with those for weakly correlated quantum dots , shown in fig . [ figure12]a - c of appendix [ extra - results ] for @xmath139 . these exhibit no sign change in the thermopower for any gate voltage @xmath142 . similarly , the thermal conductance for weakly correlated quantum dots shows no crossing point , exhibiting only a single finite temperature peak . the fdm approach allows us to easily investigate the high temperature asymptotics of transport properties . as we discuss also in the context of dilute magnetic impurities in sec . [ tdependence - compare ] below , earlier transport calculations @xcite could not discern the highest temperature peak in @xmath18 ( occurring at @xmath191 for @xmath166 , see fig . [ figure3]d ) , nor the peak in the thermal conductivity ( see discussion in sec . [ tdependence - compare ] below ) . here , we are able to do so . in addition , the numerical calculations recover the high temperature asymptotics of the transport properties : @xmath218 , @xmath219 and @xmath220 for @xmath207 . note that , for the anderson model , the logarithmic corrections in the kondo regime occur at intermediate temperatures @xmath221 : the corrections at @xmath207 go over to the above power laws . it is interesting to quantify the differences in the transport properties of quantum dots given by ( [ transport - expressions - g]-[transport - expressions - k ] ) with the analogous transport properties of dilute magnetic impurities given by ( [ imp1]-[imp3 ] ) . this is shown in fig . [ figure5 ] for the temperature dependence of transport properties in the kondo , mixed valence and empty orbital regimes for @xmath222 . in the kondo regime , and for temperatures @xmath223 , the conductance of a quantum dot is a universal function of @xmath224 , i.e. @xmath225 ( e.g. see ref . ) . the same holds for the analogous quantity for dilute magnetic impurities , namely the resistivity , i.e. @xmath226 ( e.g. see ref . ) . since @xmath2 and @xmath20 are different physical quantities , the functions @xmath227 and @xmath228 are different and they can not be made to coincide by using a common kondo scale @xmath156 ( e.g. the kondo scale defined in eq . [ haldanetk ] ) . this is seen in fig . [ figure5]a , which shows that the conductance curves for quantum dots are shifted in temperature , on a logarithmic scale , relative to the resistivity curves of magnetic impurities . the two functions @xmath227 and @xmath228 are rigorously identical only in the fermi liquid regime @xmath172 . experimentally , however , the accessible range of temperatures is that around @xmath229 , say one decade below and one decade above @xmath156 . for this region of temperatures , the two functions @xmath227 and @xmath228 can be made to coincide by redefining them as new functions @xmath230 and @xmath231 , respectively , with _ different _ respective kondo scales , @xmath232 and @xmath233 such that @xmath234 , see ref . . in the mixed valence and empty orbital regimes , fig . [ figure5]b - c shows that the conductance of a quantum dot differs significantly from the resistivity of magnetic impurities ( with significant deviations at @xmath235 ) . in particular , the aforementioned finite temperature peak in the conductance of a quantum dot is absent in the resistivity of magnetic impurities . a signature of this peak in @xmath22 is seen at most in the kondo regime at temperatures of order @xmath13 ( see fig . [ figure5]a ) and is absent in the mixed valence and empty orbital regimes . these differences to the quantum dot case , arise , as described in sec . [ transport ] , due to the different way in which the spectral function appears in the respective transport integrals . these differences reflect also the absence of universality outside the kondo regime . in fig . [ figure5]d - f we see that , up to an overall sign change , due to @xmath53 appearing differently in the transport integrals as explained in sec . [ transport ] , the thermopower of magnetic impurities behaves in a qualitatively similar way to that of a quantum dot , with two sign changes at @xmath6 and @xmath236 ( shown in fig . [ figure4]b ) and three extrema . in the kondo regime , the position , @xmath194 , of the kondo enhanced peak in the thermopower of magnetic impurities is found to scale with @xmath156 , just as for the quantum dot case ( see sec . [ thermopower - dot - tdep ] ) . a significant difference between @xmath18 for magnetic impurities and quantum dots is the much larger high temperature peak ( at @xmath206 ) for the former in the kondo regime ( by as much as a factor @xmath237 , see fig . [ figure5]d ) . this difference holds to some extent also in the mixed valence regime ( fig . [ figure5]e ) . in the empty orbital regimes the thermopowers show a single peak at @xmath171 with a similar magnitude for both cases ( fig . [ figure5]f ) . the electronic contribution to the thermal conductivity of magnetic impurities , @xmath19 , shows significant differences to the corresponding thermal conductance of quantum dots , @xmath0 , see fig . [ figure5]g - i . for example , whereas @xmath0 exhibits interesting structure with either one ( in the kondo regime ) or two ( in the mixed valence and empty orbital regimes ) peaks around @xmath171 , @xmath19 only exhibits a single peak in all regimes , and this peak occurs at a much larger temperature @xmath207 . the reason for the latter difference is the following : the main contribution to the thermal conductance and thermal conductivity come from the integrals @xmath100 and @xmath101 in eq . ( [ trans - dot ] ) and eq . ( [ trans - imp]),respectively , which involve integrals of @xmath238 and @xmath239 , respectively . for the former , the peaks in @xmath53 at @xmath34 and @xmath173 result in a peak in the integrand at @xmath240 , whereas in the latter , the dips in @xmath95 at @xmath34 and @xmath173 shift the main contribution to the integral to much higher energies @xmath241 . correspondingly , the temperature of the peaks in @xmath0 in the former are at @xmath242 and for the latter are at @xmath243 , in agreement with the numerical results . the existence of two peaks in @xmath244 in the mixed valence and empty orbital regimes as opposed to a single peak in the kondo regime is also easily explained : the two peaks reflect the sampling of the two incoherent features at @xmath34 and @xmath173 in @xmath245 appearing in the moment @xmath100 for @xmath0 . in the kondo regime , these excitations lie close to each other and only one peak results . similarly , the single peak in @xmath19 for all regimes results from the strong suppression of the above incoherent excitations in @xmath95 appearing in the moment @xmath101 . a measure of the thermoelectric efficiency of a quantum dot device is the dimensionless figure of merit defined by @xmath246 , where @xmath247 is the phonon contribution to the thermal conductance . hence for high efficiency , one requires either large @xmath1 or small total thermal conductance relative to electrical conductance or both conditions simultaneously . a calculation of @xmath248 for quantum dot systems would therefore require knowledge of the material specific phonon contribution to the thermal conductance @xmath247 . similarly , for magnetic impurity systems a calculation of the dimensionless figure of merit , @xmath249 , would require knowledge of the material specific phonon contribution to the thermal conductivity @xmath250 . this is outside the scope of the present paper , so instead we show in fig . [ figure6]a - c results at @xmath166 for the quantity @xmath251 , for quantum dots , and @xmath252 , for magnetic impurities ( with the latter being depicted on the negative axis for clarity ) . in addition , we also show in fig . [ figure6]d - f an appropriate rescaled power factor ( @xmath253 for quantum dots , and @xmath254 for magnetic impurity systems ) . this is another useful measure of an efficient thermoelectric system , by - passing lack of knowledge of the total thermal conductances ( conductivities ) . since , in the low temperature limit , the thermopower of the anderson model vanishes linearly with temperature in all regimes , a significant figure of merit is found only at finite temperature , as seen in fig . [ figure6]a - c . in the kondo regime , fig . [ figure6]a , enhanced regions of @xmath255 are found in four temperature regions , ( i ) , at @xmath256 , due the low temperature kondo enhancement of the thermopower , however the magnitude of @xmath255 is tiny ( see inset to fig . [ figure6]a ) , ( ii ) , at temperatures of order @xmath13 in the region @xmath257 , where @xmath255 can be of order @xmath258 for both quantum dots and magnetic impurities , ( iii ) , at temperatures @xmath206 , with enhancements comparable to those for region ( ii ) , and , ( iv ) , in the asymptotic region @xmath207 , where @xmath255 saturates to a finite value which is larger for quantum dots than for magnetic impurities ( discussed below ) . the behavior of the figure of merit in the mixed valence and empty orbital regimes is complicated , see fig . [ figure6]b - c . in the mixed valence regime , significant enhancements are found , for quantum dots , at temperatures somewhat below @xmath13 , see fig . [ figure6]b , and in the asymptotic regime @xmath207 . similar enhancements are found also for the empty orbital case ( fig . [ figure6]c ) . for magnetic impurities , similar enhancements to quantum dots are found on temperature scale of order @xmath13 , but at @xmath259 the enhancements are much smaller than for quantum dots ( see fig . [ figure6]b - c ) . the latter effect is due to the much larger thermal conductivities ( even at higher temperatures ) of magnetic impurities as compared to those of quantum dots ( see discussion above and fig . [ figure6]g - i ) . the power factor @xmath260 is enhanced in the same regimes ( i)-(iii ) as the figure of merit , see fig . [ figure6]e - f , but vanishes as @xmath261 in the limit @xmath207 ( using the asymptotic behavior of @xmath1 and @xmath2 from sec . [ asymptotics ] ) . in the kondo regime , @xmath260 exhibits a much larger peak above @xmath193 for magnetic impurities as compared to quantum dots ( see fig . [ figure6]d ) . this reflects the observation made above ( sec [ thermopower - compare ] ) that the highest temperature peak in @xmath18 at @xmath206 for magnetic impurities is significantly enhanced as compared to that of quantum dots . for quantum dots , the main enhancement in @xmath260 in the kondo regime is in the range @xmath262 . these trends differ little from those observed in the mixed valence and empty orbital regimes for both quantum dots and magnetic impurities ( see fig . [ figure6]e - f ) . we comment on the enhancement of @xmath255 in the region @xmath207 , which can result in @xmath263 ( e.g. in the mixed valence and empty orbital cases ) . this enhancement reflects a violation of the wiedemann - franz law at @xmath207 . the latter states that the thermal conductance ( conductivity ) is proportional to the electrical conductance ( conductivity ) multiplied by temperature , i.e. that the lorenz number @xmath264 , defined for quantum dots by @xmath265 and for magnetic impurities by @xmath266 is independent of temperature and takes on the universal value @xmath267 . since , @xmath268 , a significant reduction of @xmath269 can result in an enhancement of @xmath255 . in fig . [ figure6]g - i we see that @xmath269 is much suppressed at @xmath207 , thereby allowing for significant enhancements in @xmath255 in this limit . this enhancement is seen for all regimes , especially for the mixed valence and empty orbital regimes . we note , however , that from fig . [ figure6]g - i ( and inset ) , the wiedemann - franz law is , on the whole , reasonably well satisfied at temperatures @xmath270 , and becomes exact in the fermi liquid regime @xcite(for other violations of the wiedemann - franz law see ref ) . for a quantum dot , scaled by its limiting low temperature absolute value @xmath271 versus @xmath272 for @xmath150 and a range of gate voltages in the kondo regime . here , @xmath273 @xcite . ( b ) the thermal conductance @xmath62 , scaled by @xmath274 , versus @xmath275 , for parameters as in ( a ) , with @xmath276 defined in ( [ alpha ] ) and @xmath277 a kondo scale defined in ( [ tk - theta ] ) . the electrical conductance @xmath278 versus @xmath275 at @xmath213 is also shown . the kondo scale defined by @xmath279 is related to @xmath277 by @xmath280 . ] by analogy to the scaling properties of the electrical conductance @xmath225 , where @xmath227 is a universal function of @xmath281 in the kondo regime , it is interesting to establish to what extent such scaling is present in the thermopower , @xmath18 , and the thermal conductance , @xmath62 , of strongly correlated quantum dots . we investigate this here , for @xmath166 and for values of the gate voltage in the kondo regime ( see fig . [ figure7 ] ) . in the fermi liquid regime , @xmath172 , we have @xcite @xmath282 scaling can therefore be expected for @xmath283 , once the occupancy ( and gate voltage ) dependent factor @xmath284 is scaled out . in the above , @xmath273 is a measure of the inverse kondo scale and can be extracted@xcite from the numerical value of @xmath285 using eq . ( [ thermopower - fl ] ) and the calculated values of @xmath73 from fig . [ figure1 ] . we see from fig . [ figure7]a that @xmath286 does indeed scale with @xmath287 for a range of gate voltages in the kondo regime . this scaling extends up to temperatures comparable to @xmath156 with significant deviations setting in above this temperature scale . this is not surprising given the fact that the thermopower is a highly sensitive probe of the particle - hole asymmetry in the spectral density . for the thermal conductance , @xmath62 , we expect from the wiedemann - franz law , @xmath288 , to see a scaling in @xmath289 similar to that in @xmath278 . this is confirmed in fig . [ figure7]b which shows @xmath290 versus @xmath275 for several gate voltages in the kondo regime , where @xmath276 is defined by @xmath291 and @xmath277 is a kondo scale defined by @xmath292 we see that , for @xmath166 , @xmath293 is a universal function of @xmath275 for temperatures extending up to at least @xmath294 , just as @xmath225 is a universal function of @xmath224 for temperatures extending up to at least @xmath295 . increasing @xmath16 , and thereby reducing @xmath296 allows universality to extend to still higher temperatures . suppressing charge fluctuations , e.g. by working within a kondo model , allows these universal scaling functions to be defined for all temperatures . note also , that although these universal functions @xmath227 and @xmath297 have a similar functional dependence on @xmath224 and @xmath275 respectively , they are shifted relative to one another on an absolute temperature scale . the difference between @xmath297 and @xmath227 for temperatures around @xmath298 accounts for the violation in the wiedemann - franz law on this scale , as noted previously ( see inset to fig . [ figure6]g ) . the wiedemann - franz law is only satisfied exactly in the fermi liquid regime @xmath172 . one can collapse @xmath21 onto @xmath289 by scaling the temperature axis of the former by @xmath299 . in the universal regime @xmath270 , small deviations between @xmath300 and @xmath301 arise for @xmath302 and @xmath303 . for dilute magnetic impurities , our conclusions for scaling in the kondo regime are essentially the same as those above for quantum dots ( see also ref . ) . at several characteristic temperatures @xmath304 for , ( a ) , the quantum dot case , and , ( b ) , the impurity case . the vertical dashed lines delineate the kondo regime ( around @xmath182 ) from the surrounding mixed valence and empty orbital regimes . ] the gate voltage dependence of transport through quantum dots and magnetic impurities is shown for several representative temperatures in fig . [ figure8 ] for the electrical and thermal transport and in fig . [ figure9 ] for the thermopower . for magnetic impurities , different @xmath305 should be understood as corresponding to changes in the local level position @xmath34 relative to the fermi level , as invoked by the application of pressure ( either chemical via doping or hydrostatic ) . in some rare earth systems@xcite the application of pressure has been shown to tune the magnetic impurities from the kondo to the mixed valence and empty orbital regimes . throughout this section we denote by @xmath192 and @xmath193 the minimum temperatures of @xmath6 and @xmath7 in the limit @xmath306 ( see fig.[figure4 ] and table [ tabletk ] ) . we show that the different behavior in the gate voltage dependence of the thermopower , at fixed temperature @xmath3 , can be classified in terms of the relative value of @xmath3 to @xmath192 and @xmath193 . the gate voltage dependence of the electrical and thermal conductance of quantum dots is shown in fig . [ figure8]a - c . the former exhibits , for @xmath185 , coulomb blockade peaks at @xmath307 with a suppression of the conductance in the mid - valley region around @xmath152 . on decreasing the temperature , the kondo effect becomes operative resulting in an enhancement of the conductance in the region between the coulomb blockade peaks ( see fig . [ figure8]a ) . this picture is well known . at @xmath185 , the thermal conductance of quantum dots also exhibits coulomb blockade peaks@xcite , not directly evident in the plot of @xmath0 versus gate voltage ( fig . [ figure8]b ) , where only weak signatures of these are discernible . they become clearer in the gate voltage dependence of @xmath308 which by the wiedemann - franz law ( see sec . [ wf - lorenz ] ) is proportional to the electrical conductance @xmath21 , as seen in fig . [ figure8]c . differences between fig . [ figure8]a and fig . [ figure8]c indicate the degree of deviation from the wiedemann - franz law . these deviations are largest at @xmath207 for all gate voltages , as previously observed in fig . [ figure6]g - i . the same observations , using a different terminology , can be made for the case of magnetic impurities in fig . [ figure8]d - f : the valence fluctuation peaks at @xmath207 are now seen in the resistivity of mixed valence impurities , whereas kondo impurities ( @xmath309 ) have a small resistivity at @xmath207 and a large unitary resistivity at @xmath71 . the behavior of the thermal conductivity @xmath19 is similarly understood in terms of the wiedemann - franz law @xmath310 with @xmath311 , as seen by comparing fig . [ figure8]d and fig . [ figure8]f . the gate voltage dependence of the thermopower of quantum dots is shown in fig . [ figure9]a . the particle - hole symmetry about @xmath182 ( see sec . [ model ] ) implies @xmath312 at all temperatures , so we only discuss @xmath142 . we focus mainly on the kondo regime , @xmath313 , and discuss the remaining gate voltages by reference to fig . [ figure3 ] . there are three main types of behavior , characterized by the following temperatures , @xmath3 , relative to @xmath192 and @xmath193 : ( i ) , @xmath314 , as exemplified by @xmath315 , ( ii ) , @xmath257 as exemplified by @xmath316 and @xmath317 , and , ( iii ) , @xmath206 , as exemplified by @xmath318 . in case ( i ) , the kondo resonance is asymmetric about the fermi level @xcite , lying slightly above it for @xmath142 . the slope of the spectral density at @xmath197 is positive , resulting by eq . ( [ sommerfeld - exp ] ) in a negative thermopower , as observed for @xmath315 . the same holds , at still lower temperatures , @xmath172 , where fermi liquid theory @xcite gives the explicit expression ( [ thermopower - fl ] ) . case ( ii ) , @xmath257 , is the most interesting for quantum dots , for several reasons : first , this temperature range is experimentally accessible since for @xmath180 , we have @xmath319 and @xmath320 . second , there is an overall sign change in @xmath321 , relative to case ( i ) , for a finite range of gate voltages ( see fig . [ figure9]a ) . third , a further sign change occurs at finite @xmath181 , and , fourth , the thermopower is large enough for a significant range of gate voltages to enable its measurement . the sign change at a finite gate voltage occurs when @xmath322 as a function of @xmath4 in fig . [ figure3]d reaches the value zero and becomes negative . for gate voltages outside the kondo regime , the thermopower , as a function of gate voltage , @xmath321 , either approaches zero at @xmath323 ( as happens for @xmath324 in fig . [ figure9]a ) or does not saturate for @xmath323 ( e.g. for @xmath317 in fig . [ figure9]a ) . in terms of fig . [ figure3]f ( see the arrows ) , the former occurs for temperatures to the left of the minimum in @xmath1 in fig . [ figure3]f , and the latter occurs for the opposite case . the latter case is half - way to case ( iii ) , @xmath325 , which exhibits a thermopower approximately linear in gate voltage , with no sign change at any @xmath142 . this is similar to the `` sawtooth '' behavior of @xmath321 found for multi - level quantum dots weakly coupled to leads at @xmath207 in ref . ( for related experimental work see ref . ) . the behavior of the thermopower in multi - level open quantum dots has also been investigated @xcite . the same classification ( i)-(iii ) , as for quantum dots , can be used to explain the local level dependence of the thermopower of magnetic impurities shown in fig . [ figure9]b . in this paper we investigated the thermoelectric properties of strongly correlated quantum dots , described by the single level anderson impurity model connected to two conduction electron leads . for this purpose , we used wilson s nrg method and calculated the local green s function and transport properties by using the full density matrix approach @xcite . since this approach builds into the density matrix all excitations obtained in the nrg approach , it is particularly well suited to finite temperature transport calculations , allowing us , for example , to investigate also the high temperature asymptotics of transport properties . for strong correlations and in the kondo regime , the thermopower exhibits two sign changes , at temperatures @xmath6 and @xmath7 with @xmath8 . we found that @xmath9 , where @xmath10 is the position of the kondo induced peak in the thermopower , @xmath11 is the kondo scale , and @xmath12 . the loci of @xmath6 and @xmath7 merge at a critical gate voltage @xmath14 , beyond which no sign change occurs . we determined @xmath15 for different @xmath16 finding that @xmath15 coincides , in each case , with entry into the mixed valence regime . no sign change is found outside the kondo regime or for weak correlations , @xmath17 . thus , a sign change in @xmath18 at _ finite _ @xmath4 is a particularly sensitive signature of strong correlations and kondo physics . this effect could be measurable in quantum dots , as it manifests itself in an overall sign change in @xmath1 for a finite range of gate voltages on increasing temperature @xmath3 from below @xmath326 to values in the range @xmath327 , which is an accessible range since @xmath328 . the results for quantum dots were compared also to the relevant transport coefficients of dilute magnetic impurities in non - magnetic metals : the electronic contribution , @xmath19 , to the thermal conductivity , the thermopower , @xmath1 , and the impurity contribution to the electrical resistivity , @xmath20 . as regards the temperature dependence of the respective transport quantities , we find , in the mixed valence and empty orbital regimes , two peaks in @xmath62 as compared to a single peak in @xmath81 . similarly , @xmath21 exhibits a finite temperature peak on entering the mixed valence regime , whereas such a pronounced peak is absent in @xmath22 , even far into the empty orbital regime . as for quantum dots , we find that the low temperature kondo peak position in the thermopower of magnetic impurities scales with @xmath156 . we compared and contrasted the figure of merit , power factor and the extent of violation of the wiedemann - franz law in quantum dots and dilute magnetic impurities , finding enhanced figures of merit at temperatures where the wiedemann - franz law is strongly violated . finally , we clarified the extent of scaling , as a function of @xmath224 , in the thermopower and thermal conductance of quantum dots in the kondo regime . we comment on a recent experiment in ref . which we believe shows evidence of kondo correlations in the thermopower of a strongly correlated quantum dot . in this experiment , the thermovoltage across a kondo correlated quantum dot is investigated as a function of gate voltage and lattice temperature . this can be compared to our @xmath321 in fig . [ figure9]a . the gate voltage @xmath329 in ref . is related to our dimensionless gate voltage , @xmath4 , via @xmath330 , i.e. @xmath331 . mirror reflecting our results for @xmath321 in fig . [ figure9]a about @xmath152 allows a qualitative comparison with the experimental measurements in ref . . using the experimental estimate @xmath332 from ref . , we can translate the four experimental temperatures @xmath333 and @xmath334 at which the thermopower was measured into our theoretical temperatures in units of @xmath13 . we assume strong coulomb correlations on the dot @xmath166 and show the results for the thermopower in fig . [ figure10 ] . from table [ tabletk ] , the lowest experimental temperature corresponds to @xmath335 , the next lowest temperature ( @xmath336 ) lies close to @xmath337 , where the thermopower changes sign in the kondo regime , and the highest two temperatures lie between @xmath192 and @xmath193 . the lowest temperature measured , @xmath338 , indeed shows a positive thermopower above mid - valley , in agreement with our results for @xmath314 . upon increasing the temperature , the experiment shows a sign change of the thermovoltage for a finite range of gate voltages ( relative to mid - valley ) , which is consistent with our prediction of such a sign change in the kondo regime for @xmath339 . the onset , with increasing temperature , of an additional oscillation in @xmath321 about @xmath152 in the experiments is therefore consistent with our results . the experimental data deviates from our calculated thermopower in the mixed valence and empty orbital range of gate voltages @xmath340 , with the theoretical results showing a much larger thermopower in this region of gate voltages . these deviations are expected at @xmath340 since additional levels present in real quantum dots , but absent in our model , start being populated . this significantly influences transport through the quantum dot . qualitatively , however , we are able to interpret these experiments on the thermopower of kondo correlated quantum dots for gate voltages @xmath341 . for a more quantitative comparison to theory , further investigations are needed . at temperatures @xmath304 corresponding to the experimental ones @xmath58 from ref . using @xmath342 . since the experimental gate voltage @xmath343 we plotted @xmath1 versus @xmath344 to facilitate comparing with the experimental data . ] calculations for single quantum dots and dilute magnetic impurities , are a starting point for dealing with a finite density of quantum dots , such as self - assembled quantum dots , or for a finite concentration of magnetic or mixed valence impurities in bulk ( e.g. for tl impurities in pbte in ref . ) . the transport properties of such systems , modeled by a random distribution of anderson impurities , will be determined by ( [ imp1]-[imp3 ] ) subject to the charge neutrality condition @xmath345 , where @xmath73 is the occupancy of the dot ( impurity ) , @xmath346 is the occupancy of the relevant conduction band and @xmath94 is the total electron filling . coupled with material specific electronic structure information and the effects of phonons , such calculations , will be important for understanding the potential of materials such as self - assembled quantum dots or pbte@xmath25tl@xmath24 systems for thermoelectric applications . financial support from the forschungszentrum jlich ( v.z . ) and supercomputing time from the john von neumann institute for computing ( jlich ) is gratefully acknowledged . we thank j. von delft and a. weichselbaum for useful comments and discussions . the reduction of the single - level two - lead anderson model ( [ qdot - two - leads ] ) for a quantum dot , to a single - channel model is , in general , approximate , but as we show here , the approximation is very good ( or even exact ) . one notices first , that the @xmath46-state of the quantum dot in ( [ qdot - two - leads ] ) only couples to the even combination @xmath347 of the lead electron states . by using the following canonical transformation @xmath348 noting that normalization of even / odd states implies @xmath50 , we can rewrite ( [ qdot - two - leads ] ) in terms of even ( @xmath66 ) and odd ( @xmath349 ) lead states , as follows @xmath350 here , @xmath351 , @xmath352 is the hamiltonian for the odd lead electrons with @xmath353 , and @xmath354 is a potential scattering term between even and odd lead electrons . hence , the odd lead electrons do not couple to the dot directly , but only indirectly via the potential scattering term . the magnitude of this is given by @xmath355 , which is vanishingly small at low energies . moreover , it vanishes identically for degenerate leads @xmath356 . the calculations we report in this work , using the single channel anderson model @xmath357 are therefore a very good approximation , even in general , to those obtained from the two - lead model ( [ qdot - two - leads ] ) and identical to those for the case @xmath356 . since @xmath50 , the hybridization strength @xmath358 of the single channel model is seen to be the relevant single - particle broadening , @xmath359 , of the two - lead model ( [ qdot - two - leads ] ) . in this paper we follow the convention in the quantum dot community of using the full - width at half - maximum , @xmath45 , as the unit of energy . finally , we note , that a reduction to a single - channel model is , in general , not possible for multi - level or double quantum dots attached to two leads @xcite . [ figure11 ] shows results for a moderately correlated quantum dot , @xmath138 exhibiting the same trends as those found for the strongly correlated case @xmath166 ( fig . [ figure3 ] ) . , ( d - f ) , the thermopower , @xmath1 , and , ( g - i ) , the normalized thermal conductance , @xmath89 , multiplied by a factor @xmath360 for clarity of presentation , as function of @xmath304 , in the moderately correlated regime @xmath138 and a range of dimensionless gate voltages , @xmath361 , in the kondo ( first column ) , mixed valence ( second column ) and empty orbital ( third column ) regimes . the range of @xmath4 is indicated in the top panels for each regime and the increment used was @xmath362 . arrows indicate the evolution of the transport quantities with increasing @xmath4 . ] , ( b ) , the thermopower , @xmath1 , and , ( c ) , the normalized thermal conductance , @xmath89 , multiplied by a factor @xmath360 for clarity of presentation , as function of @xmath304 , in the weakly correlated regime @xmath139 and a range of dimensionless gate voltages , @xmath363 . the range of @xmath4 is indicated in the top panel and the increment used was @xmath362 . arrows indicate the evolution of the transport quantities with increasing @xmath4 . ] for completeness , we show an example of transport through a weakly correlated quantum dot with @xmath139 in fig . [ figure12 ] . in this case , the thermopower remains negative for all gate voltages @xmath142 ( fig . [ figure12]b ) . similarly , the thermal conductance exhibits only a single peak for all gate voltages ( fig . [ figure12]c ) . in this appendix we give an alternative derivation of the finite temperature green s function within the fdm approach of weichselbaum and von delft @xcite . a concise derivation , implementing arbitrary abelian symmetries , has also been given in ref . . we consider a general fermionic retarded green s function @xmath364_{+}\rangle\nonumber\\ & = & -i\theta(t ) { \rm tr}\left[\rho(a(t)b + ba(t))\right],\nonumber\end{aligned}\ ] ] where @xmath365 are fermionic operators , e.g. for the d - level green s function of our quantum dot @xmath366 and @xmath367 . the trace is evaluated for an appropriate density matrix @xmath20 by using the complete set of states introduced by anders and schiller @xcite . these consist of the set of states @xmath368 obtained from the eliminated eigenstates , @xmath369 , of @xmath126 , and the degrees of freedom , denoted collectively by @xmath66 , of the sites @xmath370 , where @xmath116 is the longest chain diagonalized . the retained low energy states of @xmath126 are denoted @xmath371 , and @xmath372 extends these to the hilbert space of @xmath373 by the additional environment degrees of freedom @xmath66 of the sites @xmath370 . the eigenstates ( retained and eliminated ) , @xmath374 , and eigenvalues , @xmath375 , of @xmath126 satisfy @xmath376 . completeness of the states @xmath377 is expressed by @xcite @xmath378 where @xmath379 is the last iteration for which all states are retained . for iterations @xmath380 , the set of states @xmath381 consists of both retained ( @xmath31 ) and eliminated ( @xmath382 ) states . the following decomposition of ( [ completeness ] ) is useful @xcite : @xmath383 where the last equation follows from the fact that the hilbert space of retained states at iteration @xmath117 ( supplemented by the degrees of freedom @xmath66 for sites @xmath384 ) spans the same hilbert space as all eliminated states from all subsequent iterations . by using the decomposition of unity ( [ completeness ] ) twice within the trace in the expression for @xmath385 , the following lehmann representation can be found for this green s function@xcite @xmath386\nonumber\\ g_{ab}^{ii } & = & -i\theta(t)\sum_{m = m_{0}+1}^{n-1}\sum_{le , ke'}\left[e^{i(e_{l}^{m}-e_{k}^{m})t } \langle ke'm\|b\rho\|lem\rangle\langle lem\|a\|ke'm\rangle + e^{i(e_{k}^{m}-e_{l}^{m})t } \langle lem\|\rho b\|ke'm\rangle\langle ke'm\|a\|lem\rangle\right]\nonumber\\ g_{ab}^{iii } & = & -i\theta(t)\sum_{m = m_{0}+1}^{n-1}\sum_{lem , ke'}\left[e^{i(e_{k}^{m}-e_{l}^{m})t } \langle lem\|b\rho\|ke'm\rangle\langle ke'm\|a\|lem\rangle + e^{i(e_{l}^{m}-e_{k}^{m})t } \langle ke'm\|\rho b\|lem\rangle\langle lem\|a\|ke'm\rangle\right]\nonumber \end{aligned}\ ] ] where the double sum over @xmath387 ( coming from two applications of ( [ completeness ] ) ) is decomposed into contributions @xmath388 ( first term ) , @xmath389 ( second term ) and @xmath390 ( third term ) . in the last two terms , use has also been made of ( [ useful - identity ] ) . in the time evolution @xmath391 , we have made use of the nrg approximation @xmath392 , so that @xmath393 . peters et al . @xcite evaluated the above expression for the green s function by using an approximate density matrix @xmath394 , defined by the eliminated states of the longest chain diagonalized , i.e. @xmath395 where @xmath396 ( or @xmath397 ) is chosen appropriately @xcite to ensure that @xmath398 is a good approximation to the partition function of the infinite system at temperature @xmath399 . note also , that since @xmath116 is the last iteration , all states in the above expression are considered as eliminated states in order that ( [ completeness ] ) be satisfied . this procedure can be repeated for each chain length @xmath400 , using a density matrix @xmath401 to obtain shell green s functions @xmath402 defined at a corresponding set of temperatures @xmath403 ( or @xmath404 ) ( for clarity we henceforth omit the subscript @xmath405 for @xmath406 ) . since these shell green s functions , @xmath407 , contain only excitations of order the characteristic scale , @xmath121 , of @xmath126 , or larger , the green s function @xmath408 can only be evaluated at frequencies @xmath409 . information at @xmath410 is not available . this restriction is overcome by the fdm approach that we now describe . weichselbaum and von delft @xcite evaluated the above green s function by using the fdm of the system made up of the complete set of eliminated states from all iterations @xmath411 . specifically , the fdm is defined by @xmath412 where @xmath413 is the partition function made up from the complete spectrum , i.e. it contains all eliminated states from all @xmath414 . consequently , evaluating the green s functions by using the above fdm , allows an arbitrary temperature @xmath3 to be used for all frequencies @xmath28 , and , in particular , allows accurate calculations to be carried out at @xmath129 . consider the following density matrix for the mth shell ( defined , however , in the hilbert space of @xmath415 ) : @xmath416 normalization , @xmath417=1 $ ] , implies @xmath418 where @xmath419 . then the fdm can be written as a sum of weighted density matrices for shells @xmath411 @xmath420 the calculation of the weights @xmath421 is outlined in sec . [ fdm - sub1 ] . substituting @xmath422 into the above lehmann representation for @xmath423 and fourier transforming yields @xmath424 . the first term , @xmath425 , is easily evaluated by using the orthonormality of the eliminated states @xmath426 , orthonormality of environment degrees of freedom in @xmath427 , with @xmath428 and the trace over the @xmath429 environment degrees of freedom in @xmath430 to obtain ( for @xmath431 ) @xmath432 the second term , @xmath433 , is also easily evaluated and results for @xmath434 ( the @xmath116th term vanishes , as all states are counted as eliminated states at this iteration ) @xmath435 the third term , @xmath436 , takes the form @xmath437 inserting @xmath438 between @xmath439 and @xmath440 in @xmath441 and between @xmath440 and @xmath439 in @xmath442 gives @xmath443 and @xmath444 in sec . [ fdm - sub2 ] we show that the second terms in the above expressions vanish , i.e. @xmath445 on using @xmath446 from eq.([useful - identity ] ) the terms involving @xmath447 are evaluated as @xmath448 note that these expressions are finite only for @xmath449 . using the definition of the reduced density matrix@xcite @xmath450,\ ] ] we arrive at the following expression for @xmath451 @xmath452 hence , the final expression for @xmath453 is given by @xmath454 where , in the last two terms we rearranged the summations over @xmath455 and @xmath117 and introduced the full reduced density matrix @xmath456 note that the meaning of this quantity is completely analogous to the reduced density matrix introduced by hofstetter in ref . except that one obtains reduced density matrices at iteration @xmath117 by eliminating environment degrees of freedom @xmath457 from the fdm ( [ fdm ] ) instead of the density matrix for iteration @xmath116 . in addition , the former is built from the complete set of eliminated states , as opposed to the retained states of iteration @xmath116 in the approach of ref . . the above expression for @xmath458 is identical to that in ref . . we have checked that the sum rule for the spectral function @xmath459 $ ] @xmath460 is satisfied exactly ( to machine precision ) when using the discrete ( unbroadened ) form of the spectral function as in ref . . the expression for @xmath421 in ( [ weights ] ) involves @xmath461 which contains eigenvalues from all iterations @xmath431 . in evaluating these expressions , one should therefore use the absolute energies for the @xmath462 . since , in practice , the iterative diagonalization of the hamiltonian @xmath120 involves subtraction of groundstate energies and rescaling at each @xmath117 ( see ref . ) , one has to keep track of the subtracted groundstate energies and return to the actual physical energies relative to a common absolute energy reference in evaluating @xmath421 and @xmath461 . we take this absoute energy reference to be the ground state energy of the last wilson iteration @xmath116 . thus , if @xmath463 is the true groundstate energy of @xmath126 , we use @xmath464 , @xmath465 in evaluating @xmath466 and @xmath461 @xmath467 using the expression for @xmath439 we easily find that @xmath468 hence @xmath469 in eq . ( [ rho - b ] ) involves matrix elements of the form @xmath470 for @xmath471 , which vanish , since all retained state at iteration @xmath117 have no overlap with eliminated states at iterations @xmath471 ( i.e. , eliminated states of previous iterations are not used to obtain retained states of later iterations ) . the same arguments can be used to prove eq . ( [ b - rho ] ) . for completeness , we outline here the derivation of thermoelectric transport through a strongly interacting quantum dot@xcite . the electrical , @xmath472 , and heat current , @xmath473 , from the left lead to the quantum dot can be expressed in terms of the particle number @xmath474 and energy @xmath475 of the left lead , via @xmath476\\ i^{q}_{l}&=&\dot{h}_{l}-\mu_{l}\dot{n}_{l}=\frac{1}{i\hbar}[h_{l}-\mu_{l}n_{l},h ] , \end{aligned}\ ] ] where @xmath477 is the hamiltonian ( [ qdot - two - leads ] ) . in terms of the lesser green s function s @xmath478 and @xmath479 , the above currents are given by @xmath480,\\ i^{q}_{l}&=&-\frac{2}{\hbar}re\left[\sum_{k\sigma } t_{l}(\varepsilon_{lk\sigma}-\mu_{l } ) g^{<}_{d\sigma , kl\sigma}(t , t)\right].\end{aligned}\ ] ] the lesser green s function @xmath481 can be expressed via equations of motion solely in terms of green s functions of the dot and the non - interacting green s function for the left lead . after some lengthy algebra @xcite , one finds the following expressions for the currents in terms of the retarded , @xmath482 , advanced , @xmath483 and lesser green s function , @xmath484 of the dot : @xmath485,\\ i^{q}_{l}&=&-\frac{i}{\hbar}\sum_{\sigma}\int d\omega(\omega-\mu_{l } ) \tilde{\gamma}_{l}\left [ g_{d\sigma}^{<}(\omega)\right . \nonumber\\ & + & \left . f_{l}(\omega)(g_{d\sigma}^{r}-g_{d\sigma}^{a})\right],\end{aligned}\ ] ] where @xmath486 is the fermi function of the left lead and @xmath487 is the hybridization strength of the dot to the left lead as defined in sec . [ model ] . by using current conservation @xmath488 , one can eliminate the lesser green s function from the above expressions to arrive at the final expressions used in this paper @xmath489 the quantity @xmath490 acts as a transmission function and is given by @xmath491 the electric and heat currents are expanded to linear order in @xmath492 and @xmath493 @xmath494 defining , thereby , the transport coefficients @xmath495 . in terms of the latter , the transport properties are given by @xmath496 finally , the @xmath497 are simply expressed in terms of the following transport integrals @xmath498 , \end{aligned}\ ] ] via @xmath499 and @xmath500 . substituting these values for @xmath497 into ( [ transport - quantities - lij - g]-[transport - quantities - lij - k ] ) results in the expressions ( [ transport - expressions - g]-[transport - expressions - k ] ) given in the text . c. grenzenbach , f. b. anders and g. czycholl , in _ properties and applications of thermoelectric materials _ nato advanced reasearch workshop , series b : physics and biophysics , edited by v. zlati and a. c. hewson , ( springer , dordrecht , 2009 ) . b. kubala , j. knig , and j. pekola , phys . lett . * 100 * , 066801 ( 2008 ) ; a. garg , d. rasch , e. shimshoni , and a. rosch , phys . 103 * , 096402 ( 2009 ) ; r. wirkowicz , m. wierzbicki , and j. barna , phys . rev . b * 80 * , 195409 ( 2009 ) . the quantity @xmath273 is the linear coefficient of specific heat of the quantum dot . the specific heat , @xmath501 , of a quantum dot is difficult to measure , but theoretically it is easily calculated from the free energy @xcite . the linear coefficient of specific heat , @xmath502 , is then obtained via @xmath503 .	the thermoelectric properties of strongly correlated quantum dots , described by a single level anderson model coupled to conduction electron leads , is investigated using wilson s numerical renormalization group method . we calculate the electronic contribution , @xmath0 , to the thermal conductance , the thermopower , @xmath1 , and the electrical conductance , @xmath2 , of a quantum dot as a function of both temperature , @xmath3 , and gate voltage , @xmath4 , for strong , intermediate and weak coulomb correlations , @xmath5 , on the dot . for strong correlations and in the kondo regime , we find that the thermopower exhibits two sign changes , at temperatures @xmath6 and @xmath7 with @xmath8 . we find that @xmath9 , where @xmath10 is the position of the kondo induced peak in the thermopower , @xmath11 is the kondo scale , and @xmath12 , where @xmath13 is the level width . the loci of @xmath6 and @xmath7 merge at a critical gate voltage @xmath14 , beyond which no sign change occurs at finite gate voltage ( measured relative to mid - valley ) . we determine @xmath15 for different @xmath16 finding that @xmath15 coincides , in each case , with entry into the mixed valence regime . no sign change is found outside the kondo regime , or , for weak correlations @xmath17 , making such a sign change in @xmath18 a particularly sensitive signature of strong correlations and kondo physics . the relevance of this to recent thermopower measurements of kondo correlated quantum dots is discussed . the results for quantum dots are compared also to the relevant transport coefficients of dilute magnetic impurities in non - magnetic metals : the electronic contribution , @xmath19 , to the thermal conductivity , the thermopower , @xmath1 , and the impurity contribution to the electrical resistivity , @xmath20 . in the mixed valence and empty orbital regimes , we find , as a function of temperature , two peaks in @xmath0 as compared to a single peak in @xmath19 , and similarly , @xmath21 exhibits a finite temperature peak on entering the mixed valence regime , whereas such a pronounced peak is absent in @xmath22 even far into the empty orbital regime . we compare and contrast the figure of merit , power factor and the extent of violation of the wiedemann - franz law in quantum dots and dilute magnetic impurities . the extent of temperature scaling in the thermopower and thermal conductance of quantum dots in the kondo regime is discussed .
oriented knots in @xmath0 are usually represented by knot diagrams . projecting a knot to a plane or 2-sphere in a generic direction gives an immersed oriented planar or spherical curve with finitely many double points , or crossings . a knot diagram is obtained by marking a neighborhood of each crossing to indicate which strand lies above the other . the higher strand is called the overcrossing and the lower one the undercrossing . starting with a knot diagram , one can recover the original knot up to isotopy by constructing a curve with the overcrossing arcs pushed slightly above the plane of the diagram . a central issue is to determine whether two knot diagrams represent the same knot , i.e. whether the curves in @xmath0 corresponding to each diagram are isotopic . if they represent the same knot we say that the two diagrams are _ alexander and briggs @xcite and independently reidemeister @xcite showed that equivalent diagrams can be connected through isotopy and a series of three types of moves , usually referred to as reidemeister moves . the number of such moves required to connect two equivalent diagrams , is difficult to estimate . an exponential upper bound for the number of reidemeister moves required to connect two equivalent diagrams is obtained in @xcite . we can get some lower bounds by looking at crossing numbers , writhes and winding numbers of diagrams since each reidemeister move changes these numbers by 0 , 1 or 2 . less obvious bounds are obtained in @xcite and @xcite . in this paper we define a new family of knot diagram invariants , and focus on one of them in particular . we found these invariants by following the program of arnold and vassiliev for finite order invariants . our invariants are of order one . as an application , for each @xmath1 we present two diagrams @xmath2 for the unknot , each with @xmath3 crossings . for these two diagrams , which are almost identical , the writhe , cowrithe , crossing number and winding number give a lower bound of 2 for the number of reidemeister moves required to pass from one to the other . using our new invariant , we show that the minimal number of reidemeister moves required to pass from @xmath4 to @xmath5 is @xmath6 . we also obtain restrictions on which reidemeister moves may appear in any sequence of reidemeister moves which realizes this minimum . our invariant takes values in a very large abelian group . it is natural to investigate @xmath7 valued invariants obtained by composing it with homomorphisms into @xmath7 . the `` cowrithe '' introduced in @xcite is obtained in this way . we obtain a relation of the cowrithe to arnold s spherical curve invariants and the alexander - conway polynomial , which clarifies the limitations of the cowrithe for studying reidemeister moves . in what follows we consider two different types of geometric objects . the first objects , which are the subject of study in this paper , are knot diagrams in @xmath8 . two such diagrams are considered the same if they differ by an ambient isotopy of @xmath8 . we denote the set of all such diagrams by @xmath9 . our goal is to construct invariants of knot diagrams . towards that end we construct from a diagram a second geometric object , namely a two component link in @xmath0 . this is a smooth embedding of @xmath10 in @xmath0 . two such embeddings are considered the same if they differ by an ambient isotopy of @xmath0 . we denote the set of all two component links by @xmath11 , and the term _ links _ in this paper always refers to two component links . our basic construction relating knot diagrams to links is the following . given a knot diagram @xmath12 and a crossing @xmath13 in @xmath14 , define the smoothing @xmath15 , to be the link obtained by smoothing the crossing @xmath13 , i.e. performing a cut and paste on the four strands at the crossing that preserves the orientation of the arcs . the smoothing operation is independent of the orientation of the curve , since reversing orientation results in a change of orientation of both strands at the crossing . the diagram resulting from the smoothing is the diagram of an oriented 2-component link in @xmath0 , the link @xmath16 . we order the components so that the first @xmath17 in @xmath10 is the component that enters a neighborhood of the crossing along the overcrossing arc and leaves it along the undercrossing arc , see figure [ smoothing ] . we define the sign of a crossing in a diagram in the usual way using the right - hand rule , so that each crossing in figure [ smoothing ] is negative . we now show how each invariant of 2-component links gives rise to an invariant of knot diagrams . given a knot diagram @xmath14 , denote by @xmath18 the set of all positive crossings in @xmath14 and by @xmath19 the set of all negative crossings . given an invariant @xmath20 where @xmath21 is any set , let @xmath22 denote the free abelian group with basis @xmath23 . we then define the invariant @xmath24 as follows : @xmath25 a particularly interesting example occurs when @xmath26 is taken to be @xmath27 , the linking number of the two components of a link in @xmath11 . the resulting invariant is applied in the next section . we now compute how @xmath28 changes under reidemeister moves on @xmath14 . reidemeister moves ri , rii , riii , are illustrated in figure [ reidmoves ] . ri : the contribution of all previously existing crossings is unchanged , and one new term is added . the added term is @xmath29 if the new crossing created by the reidemeister move is positive and is @xmath30 if the new crossing is negative , where @xmath31 is the value of @xmath26 on the link . see figure [ linkab](a ) . in the case where @xmath32 , the added term is @xmath33 or @xmath34 , since the linking number of the smoothed link is @xmath35 . rii : again the contribution of all previously existing crossings is unchanged , but this time two new crossings are added . there are two cases , depending on whether the orientations of the two strands participating in the reidemeister move coincide or are opposite . we call these a _ matched _ or _ unmatched _ rii move , respectively . for an unmatched rii move each of the two smoothings gives the same link ( see figure [ linkab](b ) ) , so the addition to the value of @xmath36 due to the reidemeister move is of the form @xmath37 . for @xmath32 this gives @xmath38 , where @xmath1 is the linking number of this two - component link . for a matched rii move , two different links appear from the two smoothings , differing from each other by one crossing change between the two components ( see figure [ linkab](c ) ) . the addition to the value of @xmath36 is of the form @xmath39 where @xmath31 and @xmath40 are the values of @xmath26 on these two links . when @xmath32 , the negative smoothing produces a link with linking number greater by 1 than that produced by the positive smoothing , and so the added term is of the form @xmath41 . riii : to each crossing before the reidemeister move there corresponds a crossing after the move . for all crossings other than the three crossings participating in the move , the contribution to @xmath36 is clearly unchanged . as to these three crossings , the crossing between the top and middle strand also gives the same contribution before and after the move , since one can slide the bottom strand below the smoothed crossing , to show that the same link type is produced . an instance of this is demonstrated in figure [ link3 ] where diagrams @xmath42 are the diagrams before and after an riii move . the crossing between the top and middle strands is marked @xmath13 and @xmath43 are seen to be the same link , via an isotopy corresponding to the original riii move between @xmath14 and @xmath44 . by the same argument , the same is true for the crossing between the middle and bottom strands . . the only contribution that changes is that of the crossing between the top and bottom strands . if the crossing of the top and bottom strands is positive , then a term @xmath29 is replaced by some term @xmath45 , and so the change in the value of the invariant is of the form @xmath46 . in the same way , if the crossing of the top and bottom strands is negative , the change is of the form @xmath47 . for the case @xmath48 , one checks directly that the linking numbers of these two links differ by precisely 1 . characteristic examples of the two basic cases appear in figure [ lktopbot ] . for the links @xmath43 in figure [ lktopbot ] , three strands appear . the two that participated in the smoothing belong to different components , while the third strand ( corresponding to the middle strand in @xmath42 ) belongs to one or the other component . from this it is clear that @xmath49 and @xmath50 differ by @xmath51 . a similar analysis applies to the links @xmath52 in figure [ lktopbot ] . it follows that the change in the value of @xmath53 due to an riii move is of the form @xmath54 or @xmath55 . we may use the above analysis to show that for any @xmath26 , the invariant @xmath36 is an order one invariant of knot diagrams . rather than giving a general definition of order @xmath1 invariants and then taking @xmath56 , we give the following equivalent definition . + * definition . * an invariant on knot diagrams has _ order one _ if whenever we may simultaneously perform two reidemeister moves on a diagram , in two disjoint discs @xmath57 ( that is , configurations as in figure [ reidmoves ] appear in @xmath58 and @xmath59 ) , then the change in the invariant due to the move in @xmath58 is not affected by whether we first perform the move in @xmath59 . [ too ] for any link invariant @xmath20 , the invariant @xmath24 is an order one invariant . as seen in the analysis above , the change in the value of @xmath60 due to the reidemeister move in @xmath58 , is determined by the link types obtained by smoothing either one or two crossings in @xmath58 , before and/or after the move in @xmath58 . the reidemeister move performed in @xmath59 , if done first , does not affect the link type of these smoothed links , since a reidemeister move in @xmath59 corresponds to an isotopy on the smoothed links . in the cases where @xmath26 is unchanged when permuting the two components of a link and when reversing the orientation of a link , as is true for the invariant @xmath61 , then @xmath28 is independent of the orientation of @xmath14 . this is true since reversing the orientation of @xmath14 does not effect the smoothing and the sign at each crossing . it only reverses the orientation of the smoothed links , and interchanges their two components . so , though the orientation of @xmath14 is used in the computation of @xmath53 , the invariant @xmath53 is an invariant of unoriented diagrams . the mirror image of a diagram @xmath14 is the diagram obtained from @xmath14 by reversing all of its crossings . taking the mirror image of a knot diagram @xmath14 has the following effect on @xmath53 . the smoothing at each crossing is the same , but the sign of all crossings is reversed . this interchanges the @xmath62 s and @xmath63 s , and reverses all linking numbers . so , the effect of taking mirror image is given by mapping @xmath64 and @xmath65 . we also note that @xmath53 is additive with respect to the operation of connected sum of diagrams . let @xmath66 be the set of elements in @xmath67 of the form @xmath33 , @xmath34 , @xmath38 , @xmath41 , @xmath68 , @xmath69 , and their negatives . by the analysis of section b , the elements of @xmath66 are precisely the elements of @xmath67 that may appear as the change in the value of @xmath70 when performing a reidemeister move on @xmath14 . note that @xmath66 generates @xmath67 . the length of an element of @xmath67 with respect to the generating set @xmath66 is called its _ @xmath66-length_. given two diagrams @xmath71 of the same knot , the @xmath66-length of @xmath72 is a lower bound on the number of reidemeister moves needed to get from @xmath14 to @xmath44 . in particular , if @xmath14 is a diagram of the unknot , then the @xmath66-length of @xmath70 gives a lower bound on the number of reidemeister moves needed to get from @xmath14 to the trivial diagram . we point out that though we are working in the setting of spherical diagrams , the same lower bounds apply , a fortiori , to planar diagrams . we now use the above procedure to determine the minimal number of reidemeister moves needed to pass between certain pairs of diagrams of the unknot . and its mirror image @xmath5,width=480 ] let @xmath4 be the knot diagram appearing in figure [ newdnen ] ( for @xmath73 ) . it has @xmath3 crossings , where the first @xmath74 crossings , from left to right , are negative , and the following @xmath1 crossings are positive . let @xmath5 be the knot diagram obtained from @xmath4 by reversing the middle crossing , so that @xmath5 has from left to right , @xmath1 negative crossings followed by @xmath74 positive crossings . note that @xmath5 is a mirror image of @xmath4 . the writhes of @xmath4 and @xmath5 differ by two . their winding numbers , crossing numbers and cowrithes are the same ( for definition of cowrithe see section d ) . so these invariants can only tell us that at least two reidemeister moves are needed to pass between these diagrams . it is easy to see how to arrive from @xmath4 to @xmath5 with @xmath6 reidemeister moves , namely , perform @xmath1 rii moves and one ri move to arrive from @xmath4 to the trivial diagram , and then one ri move and @xmath1 rii moves to create @xmath5 . we now prove that there is no shorter way : [ dnen ] let @xmath75 be the two knot diagrams with @xmath3 crossings appearing in figure [ newdnen ] . then the minimal number of reidemeister moves required to arrive from @xmath4 to @xmath5 is @xmath6 . furthermore , any sequence of reidemeister moves realizing this minimum involves precisely two ri moves . the other @xmath76 moves are of type corresponding to the following four elements of @xmath66 : @xmath77 , @xmath78 , @xmath79 , @xmath80 . in particular , no unmatched rii move may appear . by direct computation we see that @xmath81 and @xmath82 so @xmath83 we denote this element by @xmath84 . we first show that the @xmath66-length of @xmath84 is @xmath6 . let @xmath85 be the homomorphism defined by setting @xmath86 , @xmath87 , and @xmath88 for all @xmath89 . then @xmath90 , and @xmath91 for any @xmath92 . it follows that the @xmath66-length of @xmath84 is at least @xmath6 . the sequence of reidemeister moves described above realizes this minimum , namely @xmath93 . we now show that any presentation of @xmath84 as a sum of @xmath6 elements of @xmath66 involves precisely two terms of the form @xmath33 and/or @xmath94 , and all other @xmath76 terms are of the four types appearing in the statement of the theorem . let @xmath95 be the homomorphism defined by @xmath96 for all @xmath97 . then @xmath98 and the value of @xmath99 on all other elements of @xmath66 is 0 . since @xmath100 we see we need at least two terms of the form @xmath33 or @xmath94 to present @xmath84 . ( this is simply a writhe argument , since @xmath101 is the writhe . ) now , let @xmath102 be the homomorphism defined by setting @xmath103 , @xmath104 , and @xmath105 for all other @xmath106 . then @xmath107 . we have @xmath108 , and for all elements of @xmath66 , @xmath109 , where the only elements in @xmath66 for which the value of @xmath110 is precisely 2 are @xmath77 , @xmath78 , @xmath79 , @xmath80 . we already know that at least two terms of the form @xmath111 appear in any presentation of @xmath84 , each contributing only 1 to @xmath110 , and so the other @xmath76 terms of a minimal presentation must each contribute @xmath112 in order to get to @xmath113 . so , the other @xmath76 terms must be of the form @xmath77 , @xmath78 , @xmath79 , @xmath80 . as to the concluding remark , since none of these four elements are of the form @xmath114 , no unmatched rii move may appear . in the proof of theorem dnen we have used three different homomorphisms @xmath115 in order to analyze the element @xmath84 . @xmath116 and @xmath110 were ad hoc choices constructed especially for this specific @xmath84 . but as noted , @xmath99 is of general interest , @xmath101 being the writhe of the diagram . another such substitution is @xmath117 for all @xmath1 . the resulting diagram invariant @xmath118 is then the crossing number of the diagram . in this section we are interested in the substitution @xmath119 given by @xmath120 for all @xmath1 . for a knot diagram @xmath14 we denote @xmath121 . this invariant of knot diagrams appears in @xcite and in @xcite ( with opposite sign ) where it is named the _ cowrithe_. we now study the properties of @xmath122 , and understand its limitations in establishing lower bounds for the number of reidemeister moves between knot diagrams . we will see that @xmath122 may be presented as @xmath123 where @xmath124 depends only on the knot type of @xmath14 , and @xmath125 depends only on the underlying spherical curve of @xmath14 . now , if we are interested in the number of reidemeister moves between two diagrams @xmath42 , then all diagrams considered are of the same knot type , so for the sake of this analysis @xmath124 is a constant and we are left with @xmath125 which is only an invariant of the underlying spherical curves . so any lower bound obtained from @xmath122 to the number of reidemeister moves required to pass between @xmath14 and @xmath44 is in fact a lower bound to the number of moves on curves required to pass between the underlying curves of @xmath14 and @xmath44 . for example , for the pair @xmath75 of section c , @xmath126 since they are each diagrams of the same knot type and have the same underlying curve , so @xmath122 gives 0 as lower bound . the invariants @xmath127 referred to , are well known invariants . @xmath124 is @xmath128 times the coefficient of @xmath129 in the conway polynomial , and @xmath125 is arnold s invariant of spherical curves @xmath130 , normalized to have value 0 on the two curves appearing in figure [ normalize ] , the circle and the figure eight curve . ( these are representatives of the two regular homotopy classes of immersions of @xmath17 into @xmath8 . ) we denote by @xmath58 the invariant @xmath131 with this normalization . we mention that arnold originally considered curves in the plane in @xcite , but then observed that his invariants may be defined for curves in @xmath8 . indeed they are well defined on @xmath8 , since they are locally well defined , as in the plane , and since the components of the space of `` normed '' immersions of @xmath17 into @xmath8 defined in @xcite are simply connected . let @xmath132 denote the underlying spherical curve of a knot diagram @xmath14 , and let @xmath133 denote the coefficient of @xmath129 in the conway polynomial of the knot represented by @xmath14 . then the relation mentioned is @xmath134 this relation follows from @xcite , as we will show . ( various similar observations appear in @xcite . ) we then give a self contained proof of this relation , which demonstrates the special properties of @xmath122 and @xmath58 and their relation to the properties of @xmath53 . we first show that @xmath122 indeed coincides ( up to sign ) with the cowrithe defined in @xcite . to a knot diagram @xmath14 there is attached a chord diagram , where we connect two points of the domain @xmath135 by a straight chord in @xmath136 , if they are mapped into the same point in the diagram @xmath14 . for two crossings @xmath137 in @xmath14 , @xmath138 is a crossing between the two distinct components of @xmath16 precisely when the chords corresponding to @xmath13 and @xmath138 cross in the chord diagram . we denote this ( symmetric ) relation between @xmath13 and @xmath138 by @xmath139 . so @xmath138 contributes @xmath140 to @xmath49 according to the sign of @xmath138 which we denote @xmath141 , precisely when @xmath139 . it follows that @xmath142 from this it follows that @xmath143 ( note that @xmath144 appearing under the summation sign is a set , not an ordered pair . ) we see that indeed @xmath122 is minus the cowrithe . from these formulas for @xmath49 and @xmath145 we can see how @xmath122 changes under a crossing change . let @xmath146 be two diagrams differing only by a crossing change at @xmath13 , the crossing at @xmath13 for @xmath147 being positive and negative , respectively , and let @xmath16 be the common smoothing of @xmath148 at @xmath13 . then the following is clear from the above two formulas : if an invariant @xmath150 of diagrams satisfies the skein relation appearing in proposition skein , and in addition @xmath151 for any diagram of the unknot , then by k chapter iii it must coincide with @xmath152 . in fact it is enough that @xmath151 for any descending knot diagram . define @xmath153 our goal is to show that @xmath154 . since @xmath155 then by proposition skein we have @xmath156 . it remains to show that @xmath151 for descending diagrams . this follows directly from @xcite corollary 2 , if one chooses the basepoint required there at the initial point of descent . we now present a self contained proof that @xmath150 is a knot invariant , by showing that it is invariant under all reidemeister moves . the proof is independent of @xcite and of the fact that @xmath157 is a knot invariant . knowing that @xmath158 is a knot invariant will also reprove that @xmath159 using the above argument , since @xmath151 for the trivial diagram , and therefore for any diagram of the unknot . we first note the values of @xmath160 on @xmath66 : @xmath161 , @xmath162 , @xmath163 , @xmath164 , @xmath165 and @xmath166 . we thus see that @xmath122 remains unchanged when a diagram changes by an ri move and an unmatched rii move , and increases by 1 when a diagram changes by a matched rii move . furthermore , @xmath122 changes by @xmath51 when the diagram changes by an riii move , and inspection shows that the change of @xmath167 or @xmath168 precisely coincides with the change of @xmath167 or @xmath168 of arnold s `` strangeness '' invariant @xmath169 for the underlying curve . two cases are indicated in figure [ lktopbot ] . for each case one needs to check both possible cyclic orderings of the three strands along @xmath17 . all other cases are obtained from these two by reversing the orientation of the middle strand , and by reversing the crossing between the top and bottom strands while adjusting the middle strand accordingly . so , using the same names for the moves on curves as for the corresponding moves on knot diagrams , we have shown that the change in @xmath145 and @xmath170 is the same under moves rii , riii . as to ri moves , the change in @xmath122 is 0 , and we now check the change for @xmath58 . we point out that an ri move on spherical curves changes the regular homotopy class of the curve , and so asking about the change under an ri move is only meaningful when a specific normalization is chosen for the two regular homotopy classes . now given a smooth general position curve @xmath171 , there is a regular homotopy @xmath172 in @xmath8 taking @xmath173 either to the circle or to the figure eight . such regular homotopy involves only rii and riii moves , and the sum of contributions of all these moves along @xmath172 is precisely @xmath174 . let @xmath175 be a curve obtained from @xmath173 by an ri move , and apply the same regular homotopy @xmath172 to @xmath175 , carrying along the added kink . we experience the same moves along @xmath172 except that occasionally a strand needs to pass the additional kink as in figure [ 2moves ] . by proposition [ p1 ] this occurrence does not change the value of @xmath58 , and so the total contribution of all moves is again @xmath174 . we arrive at a curve which is obtained from the embedded circle or figure eight by an ri move , and we have verified that the value of @xmath58 on such a curve is 0 . we conclude that @xmath176 . proposition p2 establishes that the change due to an ri move is the same for @xmath145 and @xmath170 . together with the above similar observation regarding rii and riii moves , we see that @xmath177 is invariant under all reidemeister moves , proving that @xmath158 is a knot invariant .	we construct a new order 1 invariant for knot diagrams . we use it to determine the minimal number of reidemeister moves needed to pass between certain pairs of knot diagrams .
storing and querying two - dimensional points sets is fundamental in computational geometry , geographic information systems , graphics , and many other fields . most researchers have aimed at designing data structures whose size , measured in machine words , is linear in the number of points . that is , data structures are considered small if they store a set of @xmath1 points on a @xmath2 grid in @xmath3 words of @xmath4 bits each . using @xmath5 bits is within a constant factor of optimality when the points are distributed sparsely and randomly over the grid , but we can often do better on real - word point sets because they tend to be clustered and , therefore , compressible . quadtrees @xcite tend to have @xmath6 nodes when the points are clustered , but pointer - based quadtree data structures can still take @xmath7 bits . one way to avoid storing pointers is to store the points coordinates instead @xcite , but that also takes @xmath7 bits . hudson @xcite gave a structure that uses @xmath3 bits when the points are spaced appropriately and we are willing to tolerate some distortion of the points positions . recently , de bernardo et al . @xcite and venkat and mount @xcite independently proposed similar structures based on static and dynamic succinct tree representations , respectively ( see , e.g. , @xcite ) . both structures use @xmath0 bits per node in the quadtree and have the same asymptotic query times as traditional structures , which support only edge - by - edge navigation . venkat and mount noted , however , that _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` a method for compressing paths or moving over multiple edges at once using a succinct structure may speed up the many algorithms that rely on traversal of the quadtree . '' _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in section [ sec : space ] we review the ideas behind quadtrees and prove a simple upper bound on the number of nodes in terms of the points clustering . in section [ sec : structure ] we describe a quadtree data structure that uses @xmath0 bits per node in the quadtree and allows us to move over multiple edges at once . in section [ sec : membership ] we show how this lets us perform faster membership queries . finally , in section [ sec : experiments ] we present experimental evidence that our structure is practical . we leave as future work making our structure dynamic . let @xmath8 be a set of @xmath1 points on a @xmath2 grid . if @xmath1 is 0 or @xmath9 points then the quadtree for the grid is only a root storing either 0 or 1 , respectively ; otherwise , the root stores 1 and has four children , which are the quadtrees of the grid s four quadrants . figure [ fig : tree ] shows an example , taken from @xcite . notice the order of the quadrants is top - left , top - right , bottom - left , bottom - right , instead of the counterclockwise order customary in mathematics . this is called the morton or z - ordering and it is useful because , assuming @xmath10 is a power of 2 and the origin is at the top right without loss of generality , since we can manipulate the coordinate system to make it so the obvious binary encoding of a root - to - leaf path is the interleaving of the binary representations of the corresponding point s @xmath11- and @xmath12-coordinates . grid ( left ) ; the quadtree for those points ( right ) . the heavy lines in the quadtree indicate the path to the leaf corresponding to the shaded point on the grid.,scaledwidth=90.0% ] for example , if we imagine the edges descending from each internal node in figure [ fig : tree ] are labelled @xmath13 from left to right , then the thick edges are labelled @xmath14 ; the obvious binary encoding for this path is @xmath15 . the coordinates for the shaded point , which corresponds to the leaf at the end of this path , are @xmath16 , so interleaving the binary representations 1001 and _ 0110 _ of its @xmath11- and @xmath12-coordinates also gives @xmath17 . we can interleave a point s coordinates in @xmath0 time on a ram @xcite ; this operation is also fast in practice on real machines , e.g. , using pre - computed tables . the quadtree has height at most @xmath18 and @xmath5 nodes , and a subtree rooted at depth @xmath19 encodes the points on a @xmath20 square . given a query rectangle @xmath21 , we can find all the points in @xmath22 by starting at the root and visiting all the nodes whose subtrees squares overlap @xmath21 , recording the leaves storing 1s . this is called range reporting or , in the special case @xmath23 , a membership test for @xmath24 . if we report @xmath25 points then we visit @xmath26 nodes . if the points in @xmath8 are clustered , however , then intuitively the root - to - leaf paths in the quadtree will share many nodes and we will use less space and time . [ thm : space ] suppose we can partition @xmath8 into @xmath27 clusters , not necessarily disjoint , with @xmath28 points and diameters @xmath29 . then the quadtree has @xmath30 nodes . let @xmath31 be an @xmath32 square on the grid , let @xmath33 , and let @xmath34 be the set of ancestors in the quadtree of the points in @xmath35 . ( for simplicity , we identify points in @xmath8 with their corresponding leaves in the quadtree . ) let @xmath36 be the ancestors of only the corners of @xmath31 ( which may or may not be in @xmath8 ) . notice @xmath37 for any ancestor @xmath38 of a point in @xmath35 that has depth at most @xmath39 , @xmath38 s subtree contains all the points in a square of size at least @xmath40 . therefore , the square must contain at least one corner of @xmath31 , so @xmath41 . it follows that @xmath42 so @xmath43 . the proof above is something like a two - dimensional analogue of gupta , hon , shah and vitter s @xcite analysis of tries . in the full version of this paper we will consider higher dimensions and give bounds with respect to hierarchical clustering . the structure by de bernardo et al . @xcite mentioned in section [ sec : introduction ] is a variation of brisaboa , ladra and navarro s @xcite @xmath44-tree structure . brisaboa , ladra and navarro designed @xmath44-trees to compress the web graph , and de bernardo et al . adapted it to other domains , such as geographic data . the main difference is that if @xmath8 contains all the @xmath20 points encoded by the subtree of a node at depth @xmath19 , then in de bernardo et al.s structure that subtree is only the node itself , conforming to the definition of a quadtree ; in brisaboa , ladra and navarro s structure , that subtree has height @xmath45 and @xmath46 leaves . thus , the original @xmath44-tree can have more nodes than the quadtree . for many applications it is rare that point sets contain large squares that are completely filled . therefore , in this version of this paper we make the simplifying assumption that each internal node of the quadtree has at least one descendant storing a 0 , so both versions of the @xmath44-tree have the same number of nodes as the quadtree . we will remove this assumption in the full version . to store a quadtree , we first replace each internal node by a binary tree of height 2 and remove any node that has no descendant storing a 1 ; this increases the size of the whole tree by a factor of at most @xmath47 . let @xmath48 be the resulting binary tree . in addition to simplifying our construction , this modification makes quadtrees more practical in higher dimensions ( see @xcite ) , which we will also consider in the full version of this paper . we then perform a heavy - path decomposition @xcite of @xmath48 . that is , we partition @xmath48 into root - to - leaf paths , called heavy paths , such that the path containing a node @xmath38 also contains the child of @xmath38 with the most leaf descendants ( breaking ties arbitrarily ) . one well - known property of this decomposition is that each root - to - leaf path in @xmath48 consists of @xmath49 initial segments of heavy paths . figure [ fig : decomposition ] shows the heavy - path decomposition of the binary tree for our example from figure [ fig : tree ] . . nodes storing 1s are black ; nodes storing 0s are shown hollow , and discarded ; thick edges belong to heavy paths . the numbers below the black leaves indicate our ordering of the paths.,scaledwidth=90.0% ] we encode each heavy path @xmath50 as a binary string whose 0s and 1s indicate which of @xmath50 s nodes are left children and which are right children ( considering the root as a left child , say , for simplicity ) in increasing order by their depths . we sort the encodings into decreasing order by length , breaking ties such that if two paths @xmath50 and @xmath51 have the same length and their topmost nodes are @xmath38 and @xmath52 , then the encodings of @xmath50 and @xmath51 appear in the same order as the encodings of the paths containing the parents of @xmath38 and @xmath52 . ( notice @xmath38 and @xmath52 can not have the same parent , since they have the same height and the tree is binary . ) the numbers below the leaves in figure [ fig : decomposition ] indicate how we order the paths in our example . we store the concatenation @xmath53 of the paths encodings , which consists of @xmath54 bits . we say the bit @xmath55 $ ] corresponds to the node @xmath38 if @xmath55 $ ] indicates whether @xmath38 is a left child or a right child . for each depth @xmath56 ( considering the root to have depth 0 and leaves to have depth @xmath18 ) , we store a bitvector @xmath57 with 1s indicating which nodes at that depth in @xmath48 have two children ; see , e.g. , @xcite for a discussion of bitvectors . these bitvectors have as many bits as there are internal nodes in @xmath48 . for our example , @xmath58 dashes and spaces are only for legibility . storing @xmath53 and all the @xmath57s takes @xmath0 bits per node in @xmath48 and , therefore , also @xmath0 bits per node in the original quadtree . for each length , we also store the starting position in @xmath53 of the first encoding with that length ; this also takes a total of @xmath59 bits . in the full version of this paper we will give more details and discuss ways in which we can slightly reduce our space usage . suppose @xmath55 $ ] corresponds to node @xmath38 in @xmath48 . given @xmath60 , in @xmath0 time we can determine the length of the heavy path containing @xmath38 , that path s rank in our ordering , @xmath38 s depth in @xmath48 , and whether @xmath38 is the top node in its path . if @xmath38 is the top node in the @xmath61th path starting at depth @xmath19 then , by our choice of ordering , @xmath38 s parent @xmath10 is the @xmath61th node at depth @xmath62 that @xmath63 indicates has two children . it follows that we can compute in @xmath0 time which bit of @xmath53 corresponds to @xmath10 , via a select query on @xmath63 and some arithmetic . ( since select queries are quite slow in practice , however , even in applications involving ascents it may be better for us simply to backtrack . ) for example , if @xmath38 is the top node in the ninth path in our ordering , which is the third path starting at depth 5 , then @xmath10 is the third node at depth 4 which @xmath64 indicates has two children . since the third 1 in @xmath64 is its fourth bit , @xmath10 is the node at depth 4 in @xmath48 in the fourth path in our ordering . once we know @xmath38 s depth @xmath19 in @xmath48 and its path s ranking in our ordering , we know immediately whether @xmath38 is a leaf and we can check @xmath57 to see whether @xmath38 has two children . if @xmath38 is the @xmath61th node at depth @xmath19 which has two children then the child of @xmath38 that is not in the same path as @xmath38 , is the top node in the @xmath61th path starting at depth @xmath65 . it follows that we can compute in @xmath0 time which bit of @xmath53 corresponds to @xmath66 , via a rank query on @xmath57 and some arithmetic ; the other child of @xmath38 corresponds to @xmath67 $ ] , and that bit tells us which child is which . reversing the example in the previous paragraph , if @xmath38 is the node at depth 4 in @xmath48 in the fourth path in our ordering , then it is the third node at depth 4 which @xmath64 indicates has two children . therefore , one of @xmath38 s children is the node at depth 5 in the same path , while @xmath38 s other child is the top node in the third path starting at depth 5 , which is the ninth path in our ordering . suppose we want to perform a membership query for @xmath24 . we compute the label on the path to @xmath24 in @xmath0 time , as described in section [ sec : introduction ] . we set @xmath38 to be the root of @xmath48 , then repeat the following steps until we reach @xmath24 or can descend no further : we find the longest common prefix of the remainder of the path label for @xmath24 and the encoding of the heavy path starting at @xmath38 ( except that we ignore the first bit of @xmath53 , which is 0 and corresponds to the root ) , which takes @xmath0 time because the path label and the encoding are @xmath4 bits ; we descend the initial segment of the heavy path encoded by that common prefix ; if we reach @xmath24 , then we report @xmath68 ; if the node we are currently visiting has only one child , then we report @xmath69 ; otherwise , we set @xmath38 to be the child of the node we are currently visiting that is not in the same heavy path , and continue . in total , the query takes time proportional to the number of initial segments we traverse . since we only descend , we never need select queries . to perform a membership query for @xmath16 in our example , we compute the path label 10010110 and set @xmath38 to be the root ; we find that this label does not share any non - empty prefix with the encoding of the heavy path starting at @xmath38 ( ignoring the leading 0 because @xmath38 is the root , although in this case it makes no difference ) ; and we set @xmath38 to be the root s right child . we then find that the remainder of the path label ( which is all of it ) shares a prefix of length 6 , 100101 , with the encoding of the heavy path starting at @xmath38 ; we descend to the 6th node on that path ; we set @xmath38 to be that node s right child . finally , we find the remainder of the path label , 10 , shares a prefix of length 2 , all of 10 , with the encoding of the heavy path starting at @xmath38 ; we descend to the 2nd node on that path ; and we report @xmath70 . since each root - to - leaf path in @xmath48 consists of @xmath49 initial segments of heavy paths , our data structure obviously supports membership queries in @xmath49 time . if the query point is isolated , we use even less time . [ thm : membership ] we can store @xmath8 using @xmath0 bits per node in the quadtree such that a membership query for @xmath24 takes @xmath71 time , where @xmath72 is the number of points in @xmath8 within distance @xmath73 of @xmath24 . any node @xmath38 at depth at least @xmath74 in @xmath48 whose subtree s square contains @xmath24 , has at most @xmath72 leaf descendants . it follows that the path from @xmath38 to the deepest node @xmath66 of @xmath48 whose subtree s square contains @xmath24 , consists of @xmath75 initial segments of heavy paths . to see why , consider that if we ascend from @xmath66 to @xmath38 , every time we move from the top - most node in one heavy path to its parent in another heavy path , the number of leaf descendants in the subtree below us at least doubles . since the path from the root to @xmath38 has length @xmath76 , the path from the root to @xmath66 consists of @xmath77 initial segments of heavy paths . we note that combining theorems [ thm : space ] and [ thm : membership ] suggests our structure should be particularly suited to applications in which , e.g. , points are highly clustered but queries are chosen according to a very different distribution . [ cor : combination ] suppose we can partition a set @xmath8 of @xmath1 points into @xmath27 clusters , not necessarily disjoint , with @xmath28 points and diameters @xmath29 . then we can store @xmath8 in @xmath30 bits such that a membership query for @xmath24 takes @xmath71 time , where @xmath72 is the number of points in @xmath8 within distance @xmath73 of @xmath24 . we have implemented the data structure from section [ sec : structure ] and compared it experimentally with the @xmath44-tree . due to space constraints , we present the results only for membership queries ; in the full paper we will present results for range reporting . all the experiments presented here were performed in an intel core i7 - [email protected] , 32 gb ram , running ubuntu server ( kernel 3.13.0 - 35 ) . we compiled with gnu / g++ version 4.6.3 using -o3 directive . in our implementation we make use of the bitvector implementations available in libcds ( https://github.com/fclaude/libcds ) . specifically , we use three types of bitvectors . our first variant , a heavy - path implementation with plain bitvectors , we call hp@xmath78 ; for our second variant , named hp@xmath79 , we use compressed bitmaps . libcds provides several implementations of compressed bitmaps , and we use either the raman , raman and rao ( rrr ) implementation or sadakane s sdarray ( for each dataset , we select the one achieving better compression ) . we compare these two variants with two configurations of @xmath44-trees . the first variant , named @xmath44-tree@xmath80 , consists of a basic version of the @xmath44-tree where the degree @xmath81 for all the levels of the tree . the second variant , named @xmath44-tree@xmath82 , is the configuration considered as optimal for the @xmath44-tree @xcite , which includes a hybrid approach with different @xmath25 values for the levels of the tree and a vocabulary of leaf submatrices to obtain better compression . we did not compare to venkat and mount s structure because they have not yet made an implementation available . we also did not compare to the classical quadtree representations since the @xmath44-tree is an order of magnitude smaller and has better access time @xcite . for our experimental evaluation we use grid datasets from different domains : geographic information systems ( gis ) , social networks ( sn ) , web graphs ( web ) and rdf datasets ( rdf ) . for gis data we use the geonames dataset , which contains more than 9 million populated places , and convert it into three grids with different resolutions : geo - sparse , geo - med , and geo - dense . ( the higher the resolution , the sparser the matrix . ) for sn and web we consider the grid associated with the adjacency matrix of two web graphs ( indochina-2004 , uk-2002 ) and two social networks ( dblp-2011 , enwiki-2013 ) obtained from the laboratory for web algorithmics @xcite . finally , we use rdf data obtained from the dbpedia dataset . this rdf dataset contains triples ( s , p , o ) indicating subjects that are related to objects with a specific predicate . thus , each predicate defines a binary relation among subjects and objects that can be represented in a grid with points . we create three different grids for our experiments , selecting predicates with different numbers of related objects ( triples - sparse , triples - med , and triples - dense ) . table [ tab : space ] gives the main characteristics of the datasets used : name of the dataset , size of the grid ( @xmath10 ) , number of points it contains ( @xmath1 ) and the space achieved by the four representations compared : @xmath44-tree@xmath80 , @xmath44-tree@xmath82 , hp@xmath78 , and hp@xmath79 . the space is measured in bits per points ( bpp ) , dividing the total space of the structure by the number of points ( @xmath1 ) in the grid . we can observe that hp@xmath83 obtains the worst compression among all the alternatives , but hp@xmath79 obtains better results than @xmath44-tree@xmath82 for some of the datasets , which is remarkable as this configuration of the @xmath44-tree exploits several compression techniques that may be considered for future extensions of this proposal and may allow the hp@xmath79 variant to reduce its space . hp@xmath79 clearly outperforms @xmath44-tree@xmath82 for very sparse grids . .description of the datasets and space comparison [ cols= " < , < , > , > , > , > , > , > " , ] [ tab : space ] we now analyze the time performance of our proposed structure . we distinguish three different types of membership queries : empty cells , filled cells and isolated filled cells ( top 100,000 most isolated filled cells ) , and measure average times per query in nanoseconds . we show in figure [ fig : times ] the results obtained by the four representations over two kinds of grid datasets , gis and sn . ( due to space constraints we omit results for the web and rdf datasets . ) we can observe that for both scenarios we obtain similar performance . the @xmath44-tree representation obtains better results when querying empty cells , as the computation for reaching a zero node in the @xmath44-tree is lighter than using heavy paths . however , hp becomes the best alternative when querying filled cells : our non - compressed data structure is always the fastest one for cells with values , and much faster for isolated points . in this latter case , even the compressed variant of our structure outperforms the most optimized @xmath44-tree both in time and space . -tree variants , for the gis ( left ) and sn ( right ) datasets and each type of membership query : empty cells ( top ) , filled cells ( center ) and isolated filled cells ( bottom ) . curves for the gis datasets show results for geo - sparse , geo - med , and geo - dense ; curves for the sn datasets show results for dblp-2011 and enwiki-2013 . for example , consider the graph for the gis dataset and queries on isolated filled cells ( bottom left ) : the middle point on the curve for hp@xmath79 is to the left of and below the bend in the curve for @xmath44-tree@xmath82 ; this means that , for the geo - med dataset , the hp@xmath79 structure uses less space than the @xmath44-tree@xmath82 structure and also answers faster . ( notice the query type does not affect space usages.),title="fig:",scaledwidth=48.0%]-tree variants , for the gis ( left ) and sn ( right ) datasets and each type of membership query : empty cells ( top ) , filled cells ( center ) and isolated filled cells ( bottom ) . curves for the gis datasets show results for geo - sparse , geo - med , and geo - dense ; curves for the sn datasets show results for dblp-2011 and enwiki-2013 . for example , consider the graph for the gis dataset and queries on isolated filled cells ( bottom left ) : the middle point on the curve for hp@xmath79 is to the left of and below the bend in the curve for @xmath44-tree@xmath82 ; this means that , for the geo - med dataset , the hp@xmath79 structure uses less space than the @xmath44-tree@xmath82 structure and also answers faster . ( notice the query type does not affect space usages.),title="fig:",scaledwidth=48.0% ] -tree variants , for the gis ( left ) and sn ( right ) datasets and each type of membership query : empty cells ( top ) , filled cells ( center ) and isolated filled cells ( bottom ) . curves for the gis datasets show results for geo - sparse , geo - med , and geo - dense ; curves for the sn datasets show results for dblp-2011 and enwiki-2013 . for example , consider the graph for the gis dataset and queries on isolated filled cells ( bottom left ) : the middle point on the curve for hp@xmath79 is to the left of and below the bend in the curve for @xmath44-tree@xmath82 ; this means that , for the geo - med dataset , the hp@xmath79 structure uses less space than the @xmath44-tree@xmath82 structure and also answers faster . ( notice the query type does not affect space usages.),title="fig:",scaledwidth=48.0%]-tree variants , for the gis ( left ) and sn ( right ) datasets and each type of membership query : empty cells ( top ) , filled cells ( center ) and isolated filled cells ( bottom ) . curves for the gis datasets show results for geo - sparse , geo - med , and geo - dense ; curves for the sn datasets show results for dblp-2011 and enwiki-2013 . for example , consider the graph for the gis dataset and queries on isolated filled cells ( bottom left ) : the middle point on the curve for hp@xmath79 is to the left of and below the bend in the curve for @xmath44-tree@xmath82 ; this means that , for the geo - med dataset , the hp@xmath79 structure uses less space than the @xmath44-tree@xmath82 structure and also answers faster . ( notice the query type does not affect space usages.),title="fig:",scaledwidth=48.0% ] -tree variants , for the gis ( left ) and sn ( right ) datasets and each type of membership query : empty cells ( top ) , filled cells ( center ) and isolated filled cells ( bottom ) . curves for the gis datasets show results for geo - sparse , geo - med , and geo - dense ; curves for the sn datasets show results for dblp-2011 and enwiki-2013 . for example , consider the graph for the gis dataset and queries on isolated filled cells ( bottom left ) : the middle point on the curve for hp@xmath79 is to the left of and below the bend in the curve for @xmath44-tree@xmath82 ; this means that , for the geo - med dataset , the hp@xmath79 structure uses less space than the @xmath44-tree@xmath82 structure and also answers faster . ( notice the query type does not affect space usages.),title="fig:",scaledwidth=48.0%]-tree variants , for the gis ( left ) and sn ( right ) datasets and each type of membership query : empty cells ( top ) , filled cells ( center ) and isolated filled cells ( bottom ) . curves for the gis datasets show results for geo - sparse , geo - med , and geo - dense ; curves for the sn datasets show results for dblp-2011 and enwiki-2013 . for example , consider the graph for the gis dataset and queries on isolated filled cells ( bottom left ) : the middle point on the curve for hp@xmath79 is to the left of and below the bend in the curve for @xmath44-tree@xmath82 ; this means that , for the geo - med dataset , the hp@xmath79 structure uses less space than the @xmath44-tree@xmath82 structure and also answers faster . ( notice the query type does not affect space usages.),title="fig:",scaledwidth=48.0% ] we have presented a fast space - efficient representation of quadtrees , answering in the affirmative to the conjecture of venkat and mount @xcite . our structure has nice theoretical bounds and it is practical . space requirements are similar to other space - efficient representations of quadtrees , e.g. , the @xmath44-trees , but our structure is faster handling isolated filled cells . in the full version of this paper we will generalize our structure to higher dimensions , give bounds in terms of hierarchical clustering and present experimental results for range reporting . many thanks to timothy chan and yakov nekrich for directing us toward venkat and mount s paper . the first author is also grateful to the late ken sevcik for introducing him to some concepts used in this paper . d. arroyuelo , r. cnovas , g. navarro , and k. sadakane . succinct trees in practice . in _ proc . alenex _ , pages 8497 , 2010 . + n. bereczky , a. duch , k. nmeth , and s. roura . quad - k - d trees . in _ proc . latin _ , pages 743754 , 2014 . + p. boldi , m. rosa , m. santini , and s. vigna . layered label propagation : a multiresolution coordinate - free ordering for compressing social networks . in _ proc . www _ , pages 587596 , 2011 . + p. boldi and s. vigna . the webgraph framework i : compression techniques . in _ proc . www _ , pages 595601 , 2004 . + n. r. brisaboa , s. ladra , and g. navarro . compact representation of web graphs with extended functionality . , 39:152174 , 2014 . + p. davoodi and s. s. rao . succinct dynamic cardinal trees with constant time operations for small alphabet . in _ proc . tamc _ , pages 195205 , 2011 . + g. de bernardo , s. lvarez - garca , n. r. brisaboa , g. navarro , and o. pedreira . compact querieable representations of raster data . in _ proc . spire _ , pages 96108 , 2013 . + f. e. fich . constant time operations for words of length w. 1999 . + i. gargantini . an effective way to represent quadtrees . 25:905910 , 1982 . + a. gupta , w .- k . hon , r. shah , and j. s. vitter . compressed data structures : dictionaries and data - aware measures . , 387:313331 , 2007 . + b. hudson . succinct representation of well - spaced point clouds . technical report abs/0909.3137 , corr , 2009 . + g. m. morton . a computer oriented geodetic data base ; and a new technique in file sequencing . technical report , ibm ltd . + g. navarro and k. sadakane . fully functional static and dynamic succinct trees . , 10:16 , 2014 . + m. ptracu . succincter . in _ proc . focs _ , pages 305313 , 2008 . + d. d. sleator and r. e. tarjan . a data structure for dynamic trees . , 26:362391 , 1983 . + p. venkat and d. m. mount . a succinct , dynamic data structure for proximity queries on point sets . in _ proc . cccg _ , 2014 . to appear .	real - world point sets tend to be clustered , so using a machine word for each point is wasteful . in this paper we first bound the number of nodes in the quadtree for a point set in terms of the points clustering . we then describe a quadtree data structure that uses @xmath0 bits per node and supports faster queries than previous structures with this property . finally , we present experimental evidence that our structure is practical .
magnetohydrodynamic ( mhd ) equations can be used in modeling phenomena in a wide range of applications including laboratory @xcite , astrophysical @xcite , and space plasmas @xcite . for example , 3d mhd simulations have been widely adopted in space weather simulations . the historical review and current status of the existing popular 3d mhd models can be found in @xcite and @xcite , respectively . however , mhd equations form a nonlinear system of hyperbolic conservation laws , which is so complex that high - resolution methods are necessary to solve them in order to capture shock waves and other discontinuities . these high - resolution methods are in general computationally expensive and parallel computational resources such as beowulf clusters or even supercomputers are often utilized to run the codes that implemented these methods @xcite @xcite @xcite @xcite @xcite . in the last few years , the rapid development of graphics processing units ( gpus ) makes them more powerful in performance and more programmable in functionality . by comparing the computational power of gpus and cpus , gpus exceed cpus by orders of magnitude . the theoretical peak performance of the current consumer graphics card nvidia geforce gtx 295 ( with two gpus ) is 1788.48 g floating - point operations per second ( flops ) per gpu in single precision while a cpu ( core 2 quad q9650 3.0 ghz ) gives a peak performance of around 96gflops in single precision . the release of the _ compute unified device architecture ( cuda ) _ @xcite hardware and software architecture is the culmination of such development . with cuda , one can directly exploit a gpu as a data - parallel computing device by programming with the standard c language and avoid working with a high - level shading language such as cg @xcite , which requires a significant amount of graphics specific knowledge and was previously used for performing computation on gpus . detailed performance studies on gpus with cuda can be found in @xcite and @xcite . cuda is a general purpose parallel computing architecture developed by nvidia . it includes the cuda instruction set architecture ( isa ) and the parallel compute engine . an extension to c programming language and its compiler are provided , making the parallelism and high computational power of gpus can be used not only for rendering and shading , but also for solving many computationally intensive problems in a fraction of the time required on a cpu . cuda also provides basic linear algebra subroutines ( cublas ) and fast fourier transform ( cufft ) libraries to leverage gpus capabilities . these libraries release developers from rebuilding the frequently used basic operations such as matrix multiplication . graphics cards from g8x series support the cuda programming mode ; and the latest generation of nvidia gpus ( gt2x0 series or later ) unifies vertex and fragment processors and provides shared memory for interprocessor communication . a increasing number of new gpu implementations with cuda in different astrophysical simulations have been proposed . et al . _ @xcite re - implemented the direct gravitational @xmath0-body simulations on gpus using cuda . for @xmath1 , they reported a speedup of about 100 compared to the host cpu and about the same speed as the grape-6af . a library ` sapporo ` for performing high precision gravitational @xmath0-body simulations was developed on gpus by gaburov _ et al . _ this library achieved twice as fast as commonly used grape6a / grape6-blx cards . et al . _ @xcite @xcite implemented a particle - in cell ( pic ) code on gpus for plasmas simulations and visualizations and demonstrated a speedup of 11 - 22 for different grid sizes . sainio @xcite presented an accelerated gpu cosmological lattice program for solving the evolution of interacting scalar fields in an expanding universe , achieving speedups between one and two orders of magnitude in single precision . in the above works , no discussion on using double precision on gpus was reported . in mhd simulations , the support of double precision is important , especially for nonlinear problems . we will evaluate the performance and accuracy of double precision on gpus in this work . in this paper , we present an efficient implementation to accelerate computation of mhd simulations on gpus , called _ gpu - mhd_. to our knowledge , this is the first work describing mhd simulations on gpus in detail . the goal of our work is to perform a pilot study on numerically solving the ideal mhd equations on gpus . in addition , the trend of today s chip design is moving to streaming and massively parallel processor models , developing new mhd codes to exploit such architecture is essential . _ gpu - mhd _ can be easily ported to other many - core platforms such as intel s upcoming larrabee @xcite , making it more flexible for the user s choice of hardware . this paper is organized as follows : a brief description of the cuda programming model is given in section 2 . the numerical scheme in which _ gpu - mhd _ adopted is presented in section 3 . in section 4 , we present the gpu implementation in detail . numerical tests are given in section 5 . accuracy evaluation by comparing single and double precision computation results is given in section 6 . performance measurements are reported in section 7 and visualization of the simulation results is described in section 8 . we conclude our work and point out the future work in section 9 . the compute unified device architecture ( cuda ) was introduced by nvidia as a general purpose parallel computing architecture , which includes gpu hardware architecture as well as software components ( cuda compiler and the system drivers and libraries ) . the cuda programming model @xcite @xcite @xcite consists of functions , called _ kernels _ , which can be executed simultaneously by a large number of lightweight _ threads _ on the gpu . these threads are grouped into one- , two- , or three - dimensional _ thread blocks _ , which are further organized into one- or two - dimensional _ grids_. only threads in the same block can share data and synchronize with each other during execution . thread blocks are independent of each other and can be executed in any other . a graphics card that supports cuda , for example , the gt200 gpu @xcite , consisting of 30 streaming multiprocessors ( sms ) . each multiprocessor consists of 8 streaming processors ( sps ) , providing a total of 240 sps . threads are grouped into batches of 32 called _ warps _ which are executed in single instruction multiple data ( simd ) fashion independently . threads within a warp execute a common instruction at a time . for memory access and usage , there are four types of memory , namely , _ global memory _ , _ constant memory _ , _ texture memory _ as well as _ shared memory_. global memory has a separate address space for obtaining data from the host cpu s main memory through the pcie bus , which is about 8 gb / sec in the gt200 gpu . any valued stored in global memory can be accessed by all sms via load and store instructions . constant memory and texture memory are cached , read - only and shared between sps . constants that are kept unchanged during kernel execution may be stored in constant memory . built - in linear interpolation is available in texture memory . shared memory is limited ( 16 kb for gt200 gpu ) and shared between all sps in a mp . for detailed information concerning memory optimizations , we refer the reader to cuda best practice guide " @xcite . double precision is one important concern in many computational physics applications , however , support of double precision is limited to the nivdia cards having compute capability 1.3 ( see appendix a in @xcite ) such as the gtx 260 , gtx 280 , quadro fx 5800 ( contains one gt200 gpu ) , and tesla c1060 ( contains one gt200 gpu ) and s1070 ( contains four gt200 gpus ) . in gt200 gpu , there are eight single precision floating point ( fp32 ) arithmetic logic units ( alus ) ( one per sp ) in sm , but only one double precision floating point ( fp64 ) alu ( shared by eight sps ) . the theoretical peak performance of gt200 gpu is 936 gflops in single precision and 78 gflops in double precision . in cuda , double precision is disabled by default , ensuring that all double numbers are silently converted into float numbers inside kernels and any double precision calculations computed are incorrect . in order to use double precision floating point numbers , we need to call ` nvcc ` : ` -arch ` `` = ` sm_13 ` " . the flag ` -arch ` `` = ` sm_13 ` " in the command tells ` nvcc ` " to use the compute capability 1.3 which means enabling the double precision support . the recent fermi architecture @xcite ( gtx 480 , for example ) significantly improves the performance of double precision calculations by introducing better memory access mechanisms . in sections 6 and 7 we compare the accuracy and actual performance of _ gpu - mhd _ in single and double precision on both gt200 and fermi architectures . the ideal mhd equations with the assumption of the magnetic permeability @xmath2 can be represented as hyperbolic system of conservation laws as follows @xcite @xmath3 here , @xmath4 is the mass density , @xmath5 the momentum density , @xmath6 the magnetic field , and @xmath7 the total energy density . the total pressure @xmath8 where @xmath9 is the gas pressure that satisfies the equation of state , @xmath10 . in addition , the mhd equations should obey the divergence - free constraint @xmath11 . over the last few decades , there has been a dramatic increase in the number of publications on the numerical solution of ideal mhd equations . in particular the development of shock - capturing numerical methods for ideal mhd equations . we do not provide an exhaustive review of the literature here . a comprehensive treatment of numerical solution of mhd equations can be found in @xcite , for example . et al . _ @xcite proposed a free , fast , simple , and efficient total variation diminishing ( tvd ) mhd code featuring modern high - resolution shock capturing on a regular cartesian grid . this code is second - order accuracy in space and time and enforces the @xmath12 constraint to machine precision and it was successfully used for studying nonradiative accretion onto the supermassive black hole @xcite and fast magnetic reconnection @xcite . due to these advantages and convenience for gpu verse cpu comparison , the underlying numerical scheme in _ gpu - mhd _ is based on this work . a detailed comparison of shock capturing mhd codes can be found in @xcite , for example . we plan to explore other recent high - order godunov schemes such as @xcite and @xcite for _ gpu - mhd _ as our future work . we briefly review the numerical scheme @xcite we adopted in _ gpu - mhd _ here . in this numerical scheme , the magnetic field is held fixed first and then the fluid variables are updated . a reverse procedure is then performed to complete a one time step . three dimensional problem is split into one - dimensional sub - problems by using a strang - type directional splitting @xcite . firstly , we describe the fluid update step in which the fluid variables are updated while holding the magnetic field fixed . the magnetic field is interpolated to cell centers for second - order accuracy . by considering the advection along the @xmath13 direction , the ideal mhd equations can be written in flux - conservative vector form as @xmath14 where the flux vector is given by @xmath15 equation ( [ eq : euler ] ) is then solved by jin & xin s relaxing tvd method @xcite . with this method , a new variable @xmath16 is defined , where @xmath17 is a free positive function called the _ flux freezing speed_. for ideal mhd equations , we have @xmath18 and equations @xmath19 \frac{\partial { { { \mbox{\boldmath $ w$}}}}}{\partial t } + \frac{\partial}{\partial x}(c{{{\mbox{\boldmath $ u$ } } } } ) & = 0\end{aligned}\ ] ] these equations can be decoupled through a change of left- and right - moving variables @xmath20 and @xmath21 @xmath22 \frac{\partial { { { \mbox{\boldmath $ u$}}}}^l}{\partial t } - \frac{\partial}{\partial x}(c{{{\mbox{\boldmath $ u$}}}}^l ) & = 0\end{aligned}\ ] ] the above pair of equations is then solved by an upwind scheme , separately for right- and left - moving waves , using cell - centered fluxes . second - order spatial accuracy is achieved by interpolating of fluxes onto cell boundaries using a monotone upwind schemes for conservation laws ( muscl ) @xcite with the help of a flux limiter . runge - kutta scheme is used to achieve second - order accuracy of time integration . we denote @xmath23 as the cell - centered values of the cell @xmath24 at time @xmath25 , @xmath26 as the cell - centered flux in cell @xmath24 , as an example , we consider the positive advection velocity , negative direction can be obtained in a similar way . we obtain the first - order upwind flux @xmath27 from the averaged flux @xmath28 in cell @xmath24 . two second - order flux corrections can be defined using three local cell - centered fluxes as follows @xmath29 \delta{{\mbox{\boldmath $ f$}}}_{n + 1/2}^{r , t } & = \frac{{{\mbox{\boldmath $ f$}}}_{n + 1}^t - { { \mbox{\boldmath $ f$}}}_n^t}{2}\end{aligned}\ ] ] when the corrections have opposite signs , there is no second - order correction in the case of near extrema . with the aid of a flux limiter @xmath30 we then get the second - order correction @xmath31 the van leer limiter @xcite @xmath32 is used in _ gpu - mhd_. by adding the second - order correction to the first - order fluxes we obtain second - order fluxes . for example , the second - order accurate right - moving flux @xmath33 can be calculated @xmath34 the time integration is performed by calculating the fluxes @xmath35 and the freezing speed @xmath36 in the first half time step is given as follows @xmath37 where@xmath38 is computed by the first - order upwind scheme . by using the second - order tvd scheme on @xmath39 , we obtain the full time step @xmath40 @xmath41 to keep the tvd condition , the flux freezing speed @xmath42 is the maximum speed information can travel and should be set to @xmath43 as the maximum speed of the fast mhd wave over all directions is chosen . as the time integration is implemented using a second - order runge - kutta scheme , the time step is determined by satisfying the cfl condition @xmath44\\ \delta t & = cfl / c_{max } \end{aligned}\ ] ] where @xmath45 is the courant - number and @xmath46 is generally set to @xmath47 for stability , and @xmath48 is the magnitude of the magnetic field . constrained transport ( ct ) @xcite is used to keep the @xmath49 to machine precision . therefore , the magnetic field is defined on cell faces and it is represented in arrays @xcite @xmath50 where the cell centers are denoted by @xmath51 , and faces by @xmath52 , @xmath53 , and @xmath54 , etc . the cells have unit width for convenience . secondly , we describe the update of the magnetic field in separate two - dimensional advection - constraint steps along @xmath13-direction while holding the fluid variables fixed . the magnetic field updates along @xmath55 and @xmath56-directions can be handled in a similar matter . we follow the expressions used in @xcite . for example , we can calculate the averaging of @xmath57 along @xmath13 direction as follows @xmath58\ ] ] a first - order accurate flux is then obtained by @xmath59 where the velocity average is @xmath60\ ] ] @xmath61 is updated by constructing a second - order - accurate upwind electromotive force ( emf ) @xmath62 using jin & xin s relaxing tvd method @xcite in the advection step . then this same emf is immediately used to update @xmath63 in the constraint step . extension to three dimensions can be done through a strang - type directional splitting @xcite . equation ( [ eq : euler ] ) is dimensionally split into three separate one - dimensional equations . for a time step @xmath64 , let @xmath65 be the fluid update along @xmath13 , @xmath66 be the update of @xmath61 along @xmath55 , and @xmath67 be the update operator of @xmath68 to @xmath69 by including the flux along @xmath70 direction . each @xmath67 includes three update operations in sequence , for example , @xmath71 includes @xmath65 , @xmath72 , and @xmath73 . a forward sweep and a reverse sweep are defined as @xmath74 and @xmath75 , respectively . a complete update combines a forward sweep and reverse sweep . the dimensional splitting of the relaxing tvd can be expressed as follows @xcite @xmath76 { { \mbox{\boldmath $ u$}}}^{t_3 } & = { { \mbox{\boldmath $ u$}}}^{t_2 + 2\delta t_2 } = l_zl_xl_yl_yl_xl_z{{\mbox{\boldmath $ u$}}}^{t_2}\\[8pt ] { { \mbox{\boldmath $ u$}}}^{t_4 } & = { { \mbox{\boldmath $ u$}}}^{t_3 + 2\delta t_3 } = l_yl_zl_xl_xl_zl_y{{\mbox{\boldmath $ u$}}}^{t_3}\end{aligned}\ ] ] where @xmath77 , @xmath78 , and @xmath79 are sequential time steps after each double sweep . for cartesian coordinate system , it is easy to apply strang - type directional splitting @xcite on a high - dimensional problem and split it into one - dimensional sub - problems in cartesian coordinate system @xcite . in principle , we can also apply directional splitting for cylindrical or spherical coordinate systems . we may need to split the edges of grid in any direction into equal - distance pieces and determine the positions of the cell centers and face centers . similar techniques from li and li @xcite can be utilized to extend the usage of directional splitting for cylindrical or spherical coordinate systems . this extension will be left as our future work . in this section , we provide the implementation details of _ gpu - mhd_. with _ gpu - mhd _ , all computations are performed entirely on gpus and all data is stored in the gram of the graphics card . currently , _ gpu - mhd _ works on a regular cartesian grid and supports both single and double precision modes . considering the rapid development of graphics hardware , our gpu implementation was design general enough for gt200 architecture ( gtx 295 in our study ) and fermi architecture ( gtx 480 in our study ) . therefore , _ gpu - mhd _ can be used on newer architectures without significant modification . before we explain our gpu implementation in detail , the consideration and strategy of our design is presented first . during the computational process , the tvd numerical scheme for solving the mhd equations will generate many intermediate results such as the flux " and some interpolated values of each grid point . these intermediate results will then be used in the next calculation step . one important thing is not only these intermediate results of the current grid point but also those of the neighboring grid points are needed to be stored . this means the intermediate results of the neighboring grid points have to be calculated before going to the next calculation step . as a result , each calculation step in the algorithm was designed with one or several kernels and huge amount of data should be stored . in order to avoid the data transmission between cpu and gpu during the computation , _ gpu - mhd _ was designed to be run entirely on gpus . to reduce the memory usage , the storage for the intermediate results will be reused to store the intermediate results generated by the next step . the eight components ( @xmath80 ) for solving the mhd equations are stored in the corresponding eight arrays . each component of a grid point is stored close to the same component of the neighboring gird points . in any calculation step , only the necessary component of a calculation ( kernel ) will be accessed , thus providing more effective i / o . the strategy of our design is summarized as follows : * each step of the numerical scheme is handled with one or several kernels to exploit the parallelism of gpus ; + * storage of the intermediate results are reused to reduce memory usage ; + * components of the mhd equations are stored in separate arrays to provide effective memory access . although shared memory provides much faster access rate than global memory , its size is very limited ( 16 kb in gtx 295 and 48 kb in gtx 480 ) . as we have to process many intermediate results in each calculation step , shared memory is too small to fit in our gpu implementation . of course there are some techniques of using shared memory , the basic idea is to copy the data from global memory to the shared memory first , and then use the data in shared memory to do calculations . after the calculations have been completed , write these results back to global memory . this will benefit those computations that need many data accesses during the calculation period . however , as we mentioned in the beginning of this section , due to the nature of the algorithm , _ gpu - mhd _ was designed with many separated cuda kernels . calculation of each kernel actually is simple and variables of grid points in each kernel are mostly accessed only once ( read ) or twice ( read and then write the result ) . in order to provide fast access speed , parameters and temporary results ( generated and used only within kernel ) in each kernel are stored with registers . the parameters for the whole simulation such as the data size and size of dimensions are stored using constant memory . thus in our case , shared memory does not show its advantages . on the other hand , the size of shared memory is too small for our problem , especially when double precision is used in the calculations . we did try to use shared memory in _ gpu - mhd _ by coping the capable amount of data to shared memory for the calculations , but there is no speedup compared to our current approach . therefore , our code mainly uses global memory . there are three phases in our code : transform of data from the host memory into the global memory , execution of the kernels , and transfer of data from the gpu into the host memory . for global memory , if the data is well organized in global memory with the form that a load statement in all threads in a warp accesses data in the same aligned 128-byte block , then the threads can efficiently access data from the global memory . the process of organizing the data in such a form is so called coalescing @xcite @xcite . actually , gt200 architecture ( with compute capability 1.2 or 1.3 ) has more flexible in handling data in global memory than those cards with compute capability 1.1 or lower . coalescing of loading and storing data that are not aligned perfectly to 128-byte boundaries is handled automatically on this architecture ( see appendix g.3.2.2 in @xcite ) . we illustrate this new feature in figure [ fig : gtx200_coalescing ] . gt200 architecture supports 32 bytes memory block and has less limitation to memory address , which is accessed by the header ( first ) thread . even without shifting " the address to aligned 64 bytes or 128 bytes , the gpu kernels can still keep good performance , especially when we only process with @xmath81 data . the memory arrangement of _ gpu - mhd _ is presented here . the most intuitive way to write a parallel program to solve a multidimensional problem is to use multidimensional arrays for data storage and multidimensional threads for computation . however , the ability of the current cuda is limited in supporting multidimensional threads , therefore , we could not implement our code in such a straightforward way . especially in three dimensions or higher dimensions , there are still some limitations in handling multidimensional arrays and multidimensional threads . as a result , the most primitive way is to store data in one - dimension and perform parallel computation with one - dimension threads . by using an indexing technique , our storage and threading method can be extended to to solve multidimensional problems . our data storage arrangement is expressed in fig . [ fig : matrix ] and in equations ( [ index1 ] ) to ( [ index2 ] ) . @xmath82 } / size_{z } \\ = index & \mathbf{mod } & size_{z } \end{array}\right .\ ] ] @xmath83\label{index2 } index_{y } \pm 1 & = index \pm size_{z}\\[8pt ] index_{z } \pm 1 & = index \pm 1\label{index3}\end{aligned}\ ] ] here @xmath84 , @xmath85 , and @xmath86 are the indexes of a 3d matrix . @xmath87 is the 1d index used in _ gpu - mhd _ , @xmath88 , and @xmath89 are the matrix size ( number of grid points in our study ) of a 3d matrix . equation ( [ index1 ] ) expresses the mapping of three - dimensional ( 3d ) indexes to one - dimensional ( 1d ) indexes . equations ( [ index2 ] ) to ( [ index3 ] ) express the shift operations . shift operations are very important in numerical solution of conservation laws because some calculations are based on the neighboring grid points . the above indexing technique is used to prepare suitable values ( vectors ) as input values for the calculation kernels we implemented in cuda . as an example , we give a conceptual calculation kernel for a calculation in @xmath13-dimension to show how the indexing technique works for this task in the following . this kernel calculates the result with the grid point itself and neighboring gird points in @xmath13-dimension . the calculations in @xmath55- or @xmath56-dimension have the similar form . calculate_x(data , result ) index = getid ( ) ; //self - increment index for multi - threading grid_point_index = index ; //(x , y , z ) neighbor_1 = grid_point_index + ( sizey * sizez ) ; //(x + 1 , y , z ) neighbor_2 = grid_point_index - ( sizey * sizez ) ; //(x - 1 , y , z ) calculate_kernel(data , result , grid_point_index , neighbor_1 , neighbor_2 , ... ) ; ...... the indexing technique is a common way to represent multidimensional arrays using 1d arrays by mapping a 3d index @xmath90 to an 1d index @xmath91 . the gpu kernels of tvd were designed such that each kernel calculates using the actual index of a particular grid point and its neighbors . for example , if the calculation needs the information in a particular gird point and its neighboring grid points in @xmath56-dimension , then the indexing operation will retrieve @xmath92 $ ] , @xmath93 $ ] and @xmath94 $ ] and proceed the calculation . if the calculation needs the information in a particular grid point and its neighboring grid points in @xmath55-dimension , then the indexing operation will retrieve @xmath95 $ ] , @xmath93 $ ] and @xmath96 $ ] . then these resulting indexes from indexing operation will pass to the gpu kernels of tvd for performing the calculation . as a result , for @xmath0-dimension problem , what we need are @xmath0-dimension indexing operation kernels while only one tvd kernel is needed at all the time . for efficiency , _ gpu - mhd _ only processes the problem with the number of grid points satisfying @xmath81 condition . one reason is that the size of a warp of gpu contains 32 threads , problems with grid point number of @xmath81 are easier to determine the number of threads and blocks to fit in multiple of a warp before the gpu kernel is called . that means we do not need to check if the i d of the grid point being processed ( calculated by the block i d and thread i d ) is out of the range . it is very helpful in making the gpu code run more efficient . on the other hand , it is also effective to reduce logical operations in a gpu kernel , which is known to be a little bit slow in the current gpu architecture . as a result , warp divergence caused by the number of the data is avoided ( there is still a little bit warp divergence caused by if " operation in the calculation of our algorithm ) . similar method is used in the cuda sdk code sample reduction " . the actual memory pattern used in _ gpu - mhd _ will be presented at the end of next subsection after introducing our algorithm . a cuda kernel " is a function running on gpu @xcite @xcite @xcite . noted that the cuda kernel will process all grid points in parallel , therefore , a ` for ` instruction is not needed for going through all grid points . _ gpu - mhd _ includes the following steps : 1 . cuda initialization 2 . setup the initialize condition for the specified mhd problem : + @xmath97 of all grid points , @xmath98 of cell faces , and set parameters such as time @xmath25 , etc . 3 . copy the the initialize condition @xmath99 , @xmath100 to device memory ( cuda global memory ) 4 . for all grid points , calculate the @xmath101 by equation ( [ equat_cfl ] ) ( implemented with a cuda kernel ) 5 . use ` cublasisamax ` ( in single precision mode ) function or ` cublasidamax ` ( in double precision mode ) function of the cublas library to find out the maximum value of all @xmath101 , and then determine the @xmath64 6 . since the value of @xmath64 is stored in device memory , read it back to host memory ( ram ) 7 . sweeping operations of the relaxing tvd ( calculation of the @xmath102 , implemented with several cuda kernels , will be explained in the next subsection ) 8 . @xmath103 9 . ` if ` @xmath25 reaches the target time , go to next step + ` else ` repeats the procedure from step ( 4 ) 10 . read back data @xmath99 , @xmath100 to host memory 11 . output the result the program flow of _ gpu - mhd _ is shown in fig . [ fig : flowchart ] . after the calculation of the cfl condition , the sweeping operations will be performed . the sweeping operation @xmath104 will update both the fluid variables and orthogonal magnetic fields along @xmath70 dimension . this is a core computation operation in the relaxing tvd scheme described in section [ mhd ] . the cfl condition for the three - dimensional relaxing tvd scheme is obtained by equation ( [ equat_cfl ] ) . the procedure is to calculate all the @xmath101 of each grid point and find out the maximum value . in _ gpu - mhd _ , parallel computation power of cuda is exploited to calculate the @xmath101 of each grid point in parallel and all the @xmath101 values are stored in a matrix . then the ` cublasisamax ` function is used ( in double precision mode , the ` cublasidamax ` function is used ) to find out the maximum @xmath101 of the matrix in parallel ( called the reduction operation ) . the ` cublasisamax ` function is provided in the cublas library a set of basic operations for vector and matrix provided by nvidia with the cuda toolkit @xcite . the reason we read the @xmath64 back and store both @xmath64 and @xmath25 in host memory is due to the data in device memory can not be printed out directly in the current cuda version . this information is useful for checking if there is any problem during the simulation processing . the implementation of sweeping operations will be explained in the next subsection . before we start to describe the sweeping operations , consideration of memory arrangement is presented first in the following . implementing parallel computation using cuda kernels is somewhat similar to parallel implementation on a cpu - cluster , but it is not the same . the major concern is the memory constrain in gpus . cuda makes parallel computation process on gpus which can only access their graphics memory ( gram ) . therefore , data must be stored in gram in order to be accessed by gpus . there are several kinds of memory on graphics hardware including registers , local memory , shared memory , and global memory , etc . , and they have different characteristics and usages @xcite , making memory management of cuda quite different compared to parallel computation on a cpu - cluster . in addition , even the size of gram in a graphics card increases rapidly in newer models ( for example , the latest nvidia graphics card geforece gtx 295 has 1.75 g gram ) , but not all the capacity of gram can be used to store data arbitrarily . shared memory and local memory are flexible to use , however , their sizes are very limited in a block and thus they can not be used for storing data with large size . in general , numerical solution of conservation laws will generate many intermediate results ( for example , @xmath105 , @xmath106 , @xmath42 , @xmath107 , etc . ) during the computation process , these results should be stored for subsequent steps in the process . therefore , global memory was mainly used in _ gpu - mhd_. after the maximum value of @xmath101 in equation ( [ equat_cfl ] ) is found , we can get the @xmath108 by determining the courant - number ( @xmath45 ) . the sequential step is the calculation of @xmath104 ( @xmath109 ) . the implementation of @xmath104 includes two parts : update the fluid variables and update the orthogonal magnetic fields . as an example , the process for calculating @xmath71 is shown in fig . [ fig : lx ] where each block was implemented with one or several cuda kernels . the process for calculating @xmath110 or @xmath111 is almost the same as @xmath71 except that the dimensional indexes are different . .,width=384 ] the first part of the @xmath71 calculation process is @xmath65 . the fluid variables will be updated along @xmath13 . * algorithm 1 * shows the steps and gpu kernels of this process ( the data of @xmath99 and @xmath100 are already copied to device memory ) , all the steps are processed on all grid points with cuda kernels in parallel . load @xmath99 , @xmath100 and @xmath64 memory allocation for the storage of the intermediate results : @xmath112 , @xmath113 , @xmath114 , @xmath115 , ( @xmath115 includes the storage of @xmath106 , @xmath42 , @xmath107 , etc ) @xmath116 results obtained by equation ( [ equat_b ] ) with @xmath100 , ( @xmath100 stored the magnetic field of the cell faces ) @xmath117 results obtained by equations ( [ eq : flux ] ) and ( [ eq : lr ] ) with @xmath99 @xmath118 the flux of a half time step : difference calculation ( first - order upwind scheme of fluid " cuda kernels in fig . [ fig : lx ] ) obtained by equation ( [ eq : delta_u_half ] ) using @xmath115 @xmath119 calculate the intermediate result ( @xmath120 ) using equation ( [ eq : delta_u_half ] ) with @xmath99 and @xmath114 @xmath117 results obtained by equations ( [ eq : flux ] ) and ( [ eq : lr ] ) with @xmath113 ( the same algorithm and same cuda kernels in step 4 ) @xmath118 the flux of another half time step : difference calculation ( second - order tvd scheme of fluid " cuda kernels in fig . [ fig : lx ] ) obtained by equation ( [ eq : delta_u_full ] ) and the limiter ( equation ( [ eq : limiter ] ) ) using @xmath115 calculate the result of @xmath121 with @xmath114 using equation ( [ eq : delta_u_full ] ) and save it back to @xmath99 free the storage of the intermediate results ( continue to the second part of @xmath122 , update the orthogonal magnetic fields ) in this process , we have to calculate the magnetic fields of the grid point ( equation [ equat_b ] ) first because all the magnetic fields are defined on the faces of the grid cell @xcite . to update the fluid variables of @xmath71 , the main process , which includes one or even several cuda kernels , is to calculate the affect of the orthogonal magnetic fields to the fluid variables of equations ( [ eq : flux ] ) , ( [ eq : lr ] ) and ( 10 ) . one such main process gives the flux of the @xmath123 step . after two main processes of flux calculation and the other difference calculations , the value of fluid @xmath99 is updated from @xmath124 to @xmath125 in one @xmath122 process . the second part of the @xmath71 calculation process is to update the orthogonal magnetic fields in @xmath55-dimension ( @xmath126 ) , and @xmath56-dimension ( @xmath73 ) with the fluid along @xmath13-dimension . the strategy and implementation are similar to those in the first part but with a different algorithm for the orthogonal magnetic fields . ( after the processes of fluid , we obtain an updated @xmath99 ) load @xmath127 ( density @xmath4 ) , @xmath128 ( @xmath129 ) , @xmath100 and @xmath64 memory allocation for the intermediate results : @xmath112 , @xmath114 , @xmath130 and @xmath131 @xmath132 determine the fluid speed with the updated @xmath127 and @xmath128 in @xmath65 , with the difference calculated in @xmath55-dimension @xmath133 results obtained by equation ( [ eq : averaging_x ] ) @xmath118 the flux of a half time step : difference calculation of flux of magnetic field in @xmath55-dimension " ( first - order upwind scheme of magnetic field " cuda kernels in fig . [ fig : lx ] ) obtained by equations ( [ eq : delta_u_half ] ) and ( [ eq : first_flux ] ) ) @xmath116 calculate the intermediate result ( @xmath120 ) by applying equation ( [ eq : delta_u_half ] ) to @xmath134 ( not by applying equation ( [ eq : delta_u_half ] ) to @xmath99 ) with @xmath63 and @xmath114 @xmath118 the flux of another half time step : difference calculation ( second - order tvd scheme of magnetic field " cuda kernels in fig . [ fig : lx ] ) obtained by equation ( [ eq : delta_u_half ] ) , the limiter of equation ( [ eq : limiter ] ) and equation ( [ eq : first_flux ] ) calculate the result of @xmath135 and @xmath136 with @xmath137 by applying equation ( [ eq : delta_u_full ] ) ) to @xmath134 , and save it back to @xmath100 ( the following steps is similar to above steps but the affected orthogonal magnetic field is changed from @xmath55 to @xmath56 ) @xmath132 determine the fluid speed with the updated @xmath127 and @xmath128 in @xmath65 , with the difference calculated in @xmath56-dimension @xmath133 results obtained with equation ( [ eq : averaging_x ] ) using index of @xmath70 , @xmath138 , @xmath139 @xmath118 the flux of a half time step : difference calculation of flux of magnetic field in @xmath56-dimension " ( first - order upwind scheme of magnetic flied " cuda kernels in fig . [ fig : lx ] ) obtained by equations ( [ eq : delta_u_half ] ) and ( [ eq : first_flux ] ) @xmath140 calculate the intermediate result ( @xmath141 ) by applying equation ( [ eq : delta_u_half ] ) to @xmath142 ( not by applying equation ( [ eq : delta_u_half ] ) to @xmath99 ) with @xmath143 and @xmath114 @xmath118 the flux of another half time step : difference calculation ( second - order tvd scheme of magnetic flied " cuda kernels in fig . [ fig : lx ] ) obtained by equation ( [ eq : delta_u_half ] ) , the limiter of equation ( [ eq : limiter ] ) and equation ( [ eq : first_flux ] ) calculate the results of @xmath144 and @xmath145 with @xmath137 by applying equation ( [ eq : delta_u_full ] ) ) to @xmath142 , and save it back to @xmath100 free the storage of the intermediate results in * algorithm 1 * , the calculations in steps ( 4 ) to ( 9 ) are the steps for @xmath126 , and steps ( 11 ) to ( 16 ) are the steps for @xmath73 . the steps for @xmath126 and @xmath73 are almost the same , and the only different parts are the dimensional indexes of the difference calculations , and the affected magnetic fields : @xmath134 and @xmath142 . after the first part of @xmath122 the fluid @xmath124 is updated to @xmath125 . this change of the fluid affects to the orthogonal magnetic fields . therefore , the corresponding change ( flux ) of orthogonal magnetic fields can be calculated with the density and velocity of the updated fluid @xmath146 . then the orthogonal magnetic fields are also updated to @xmath147 and @xmath148 , and also , these changes give effects to @xmath149 . after one process of @xmath122 , both fluid and magnetic fields are updated to @xmath150 with the affect of the flow in @xmath13-dimension . and a sweeping operation sequence includes two @xmath122 , @xmath151 , and @xmath152 ( see equations ( [ eq : l_sequence ] ) ) . so we actually get the updated fluid and magnetic fields of @xmath153 after one sweeping operation sequence . note that the second @xmath122 in the sequence is a reverse sweeping operation , the order of @xmath65 , @xmath154 and @xmath73 has to be reversed : @xmath126 and @xmath155 first , and @xmath65 second . as we mentioned before , numerical solution of conservation laws needs lots of memory because there are many intermediate results generated during the computation process . these intermediate results should be stored for the next calculation steps which need the information of the neighboring grid points obtained in the previous calculation steps . otherwise , in order to avoid the asynchronous problem in parallel computation , we have to do many redundant processes . this is due to the processors on gpus will not automatically start or stop working synchronously . without storing the intermediate results , it will be hard to guarantee the values of the neighboring grid points updated synchronously . with the purpose to minimizing the memory usage , not only the calculation process of @xmath71 is divided into several steps ( cuda kernels ) , but also the intermediate results are stored as little as possible . the processes dealing with the difference calculations are also divided into several steps to minimize the storage of the intermediate results and to guarantee there is no wrong result caused by asynchronous problem . it should be realized that most of the processes in the three - dimensional relaxing tvd scheme with the dimensional splitting technique is similar . et al . _ @xcite swapped the data of @xmath13 , @xmath55 , and @xmath56-dimensions while _ gpu - mhd _ used one - dimensional arrays . but the similar swapping technique can be applied in our case with some indexing operations . instead of transposing or swapping the data , we implemented each calculation part of the flux computation with two sets of cuda kernels : one set is the cuda kernels for calculating the relaxing tvd scheme ( we call it tvd kernel here ) and the other set is the cuda kernels actually called by @xmath104 operations ( we call them @xmath104 kernels here ) . indexing operations are contained in all @xmath104 kernels . after the index is calculated , tvd kernels are called and the indexes are passed to the tvd kernels , letting the tvd kernels calculate the flux of corresponding dimension . therefore , the difference among @xmath122 , @xmath151 , and @xmath152 is the dimensional index . the flux computation of _ gpu - mhd _ is shown in fig . [ fig : indexingkernels ] . the indexing operation swaps the target that will be updated and the neighboring relationship will also be changed accordingly . for example , the calculation that uses @xmath156 as the neighboring element in @xmath71 will be changed to @xmath157 in @xmath110 . as transposing the data in a matrix needs more processing time , it is efficient and flexible to extend the code to multidimensional by dividing the indexing operation and flux calculation . as we mentioned in section 4.1 , the data is stored in 1d array , the data accesses of @xmath71 , @xmath110 , and @xmath111 are depicted in fig . [ fig : dataaccess ] . in @xmath71 , the data of @xmath158 are used to calculate and update the data of @xmath159 . the data of @xmath160 are used to calculate and update the data of @xmath161 , and so on . similarly , in @xmath110 , the data of @xmath162 are used to calculate and update the data of @xmath90 . the data of @xmath163 are used to calculate and update the data of @xmath164 , and so on . in @xmath111 , the data of @xmath165 are used to calculate and update the data of @xmath90 . the data of @xmath166 are used to calculate and update the data of @xmath167 , and so on . it seems that the data accesses of @xmath71 and @xmath110 will slow down the performance since these accesses are not in so called coalescing " pattern . however , experimental results show that the computational times spending on calculating each dimensional component such as @xmath65 and @xmath126 in @xmath71 , @xmath110 , and @xmath111 are very close in our current arrangement ( see tables [ table_1d_part ] [ table_2d_part ] , and [ table_3d_part ] in section 7 ) . this is due to the fact that gt200 and fermi gpu are more flexible to handle the data access that is not perfectly coalesced ( see section 4.1 ) . thus we did not further perform the coalescing to make these data accesses in optimal coalescing pattern . , @xmath110 and @xmath111,width=480 ] after the whole pipeline of fig . [ fig : flowchart ] is completed , the mhd simulation results will be stored in gram and these results are readily to be further processed by the gpu for visualization or read back to the cpu for other usage . due to the data - parallel nature of the algorithm and its high arithmetic intensity , we can expect our gpu implementation will exhibit a relatively good performance on gpus . in this section , several numerical tests in one - dimensional ( 1d ) , two - dimensional ( 2d ) , and three - dimensional ( 3d ) for validation of _ gpu - mhd _ are given . two graphics cards nvidia geforce gtx 295 and gtx 480 were used . gtx 295 has two gpus inside but only one was used in these numerical tests . the results shown in this section are computed with single precision mode in _ gpu - mhd _ on gtx 295 . the difference between single precision and double precision computation results will be discussed in section [ sec : accuracy ] . 1d brio - wu shock tube problem @xcite which is a mhd version of the sod problem @xcite , consists of a shock tube with two initial equilibrium states as follows left side @xmath168 @xmath169 @xmath170 @xmath171 right side @xmath172 @xmath173 @xmath174 @xmath175 constant value of @xmath176 was used and the problem was solved for @xmath177 $ ] with 512 grid points . numerical results are presented at @xmath178 in fig . [ fig : briowu ] and fig . [ fig : briowu2 ] , which include the density , the pressure , the energy , the @xmath55- and @xmath56-magnetic field components , and the @xmath13- , @xmath55- and @xmath56-velocity components . the results are in agreement with those obtained by brio and wu @xcite and zachary _ et al . _ @xcite . . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] the second 1d test is the mhd shock tube problem considered in @xcite . left side @xmath168 @xmath179 @xmath180 @xmath181 right side @xmath172 @xmath182 @xmath183 @xmath184 constant value of @xmath185 was used and the problem was solved for @xmath177 $ ] with 512 grid points . numerical results are presented at @xmath186 in fig . [ fig : rj952a ] and fig . [ fig : rj952a2 ] , which include the density , the pressure , the energy , the @xmath55- and @xmath56-magnetic field components , and the @xmath13- , @xmath55- and @xmath56-velocity components . the results are in agreement with those obtained by @xcite and @xcite . . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] hfill . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] . the result computed with 512 grid points is shown with circles and solid line shows reference high resolution result of 4096 grid points.,title="fig:",width=240 ] the first 2d test is orszag - tang problem @xcite , which is used to study incompressible mhd turbulence . in our test , the boundary conditions are periodic everywhere . the density @xmath4 , pressure @xmath187 , initial velocities @xmath188 , and magnetic field @xmath189 are given by @xmath190 @xmath191 @xmath192 the orszag - tang vertex test was performed in a two - dimensional periodic box with 512 @xmath193 512 grid points . the results of the density and gas pressure evolution of the orszag - tang problem at @xmath194 and @xmath195 are shown in fig . [ fig : ot ] , where the complex pattern of interacting waves is perfectly recovered . the results agree well with those in lee _ et al . _ @xcite . ( left ) and @xmath195 ( right ) computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] ( left ) and @xmath195 ( right ) computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] ( left ) and @xmath195 ( right ) computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] ( left ) and @xmath195 ( right ) computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] the second 2d test is the mhd blast wave problem . the mhd spherical blast wave problem of zachary _ et al . _ @xcite is initiated by an over pressured region in the center of the domain . the result is a strong outward moving spherical shock with rarified fluid inside the sphere . we followed the test suite @xcite of athena @xcite . the condition for 2d mhd blast wave problem is listed as follows @xcite @xmath196 @xmath197 @xmath198 @xmath199 in fig . [ fig:2dblast ] , we present images of the density and gas pressure at @xmath200 computed with 512 @xmath193 512 grid points . the results are in excellent agreement with those presented in @xcite . , computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] , computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] the third 2d test is the mhd rotor problem . the problem was taken from @xcite . it initiates a high density rotating disk with radius @xmath201 of fluid measured from the center point @xmath202 . the ambient fluid outside of the spherical region of @xmath203 has low density and @xmath204 , and the fluid between the high density disk fluid and ambient fluid ( @xmath205 where @xmath206 ) has linear density and angular speed profile with @xmath207 and @xmath208 where @xmath209 . two initial value sets of @xmath210 and @xmath211 provided in @xcite and @xcite were tested . the initial condition for 2d mhd rotor problem is listed as follows @xmath212 @xmath213 @xmath214 @xmath215 @xmath216 @xmath217 @xmath218 @xmath219 first rotor problem : @xmath220 second rotor problem : @xmath221 in fig . [ fig:2drotor ] , we present images of the density , gas pressure of the two rotor problems computed with 512 @xmath193 512 grid points . the results are in excellent agreement with those presented in @xcite and @xcite . , results of the density ( bottom - left ) , gas pressure ( bottom - right ) of the second mhd rotor test at @xmath222 , both computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] , results of the density ( bottom - left ) , gas pressure ( bottom - right ) of the second mhd rotor test at @xmath222 , both computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] , results of the density ( bottom - left ) , gas pressure ( bottom - right ) of the second mhd rotor test at @xmath222 , both computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] , results of the density ( bottom - left ) , gas pressure ( bottom - right ) of the second mhd rotor test at @xmath222 , both computed with 512 @xmath193 512 grid points.,title="fig:",width=240 ] the 3d version of mhd spherical blast wave problem was also tested . the condition is listed as follows @xcite @xmath223 @xmath224 @xmath225 @xmath226 fig . [ fig:3dblast01 ] and fig . [ fig:3dblast02 ] show the results of 3d blast wave problem , which include the density , gas pressure , and magnetic pressure at @xmath227 and @xmath186 sliced along @xmath13-@xmath55 plane at @xmath56 = 0.5 . the test was computed with 128 @xmath193 128 @xmath193 128 grid points . due to the scarcity of published 3d test results , we do not make direct contact with results presented in the literature here . considering only the @xmath99 and @xmath100 , the memory requirement of @xmath228 mhd problem is about 512 mb gram for single precision and 1024 mb gram for double precision , respectively . if the storage of intermediate results such as @xmath112 , @xmath113 , @xmath114 and @xmath229 etc . ( see section [ sec : sweeping ] ) are considered , the amount of memory requirement will be about 2.25 gb ( single precision ) . as we mentioned in section [ sec : sweeping ] , not all the capacity of gram can be used to store data arbitrarily . as we said in the beginning of this section , there are actually two gpus inside the gtx 295 and the 1.75 gb gram is the total amount of the gram shared by two gpus , so that only less than @xmath230 gb gram can be used . as a result , the test of 3d problem with @xmath228 resolution are not able to be provided on a graphics card . sliced along @xmath13-@xmath55 plane at @xmath56 = 0.5 and computed with 128 @xmath193 128 @xmath193 128 grid points.,title="fig:",width=240 ] sliced along @xmath13-@xmath55 plane at @xmath56 = 0.5 and computed with 128 @xmath193 128 @xmath193 128 grid points.,title="fig:",width=240 ] sliced along @xmath13-@xmath55 plane at @xmath56 = 0.5 and computed with 128 @xmath193 128 @xmath193 128 grid points.,title="fig:",width=240 ] sliced along @xmath13-@xmath55 plane at @xmath56 = 0.5 and computed with 128 @xmath193 128 @xmath193 128 grid points.,title="fig:",width=240 ] sliced along @xmath13-@xmath55 plane at @xmath56 = 0.5 and computed with 128 @xmath193 128 @xmath193 128 grid points.,title="fig:",width=240 ] sliced along @xmath13-@xmath55 plane at @xmath56 = 0.5 and computed with 128 @xmath193 128 @xmath193 128 grid points.,title="fig:",width=240 ] in mhd simulations , accuracy is always to be considered since the error may increase fast and crash the simulation if low precision is used for computation . scientific computations such as mhd simulation mostly use double precision to reduce errors . in this section , the results generated by _ gpu - mhd _ using single precision and double precision modes are shown and compared . the difference between the results of double precision and single precision computation of the @xmath231 one - dimensional brio - wu shock tube problem is shown in fig . [ fig : briowu_diff ] . two curves are almost the same but there are actually some differences with the amount of @xmath232 : ( top ) , @xmath233 ( middle ) and @xmath234 ( bottom ) of 1d brio - wu shock tube problem at @xmath178 with 512 grid points , title="fig:",width=240 ] ( top ) , @xmath233 ( middle ) and @xmath234 ( bottom ) of 1d brio - wu shock tube problem at @xmath178 with 512 grid points , title="fig:",width=240 ] ( top ) , @xmath233 ( middle ) and @xmath234 ( bottom ) of 1d brio - wu shock tube problem at @xmath178 with 512 grid points , title="fig:",width=240 ] ( top ) , @xmath233 ( middle ) and @xmath234 ( bottom ) of 1d brio - wu shock tube problem at @xmath178 with 512 grid points , title="fig:",width=240 ] ( top ) , @xmath233 ( middle ) and @xmath234 ( bottom ) of 1d brio - wu shock tube problem at @xmath178 with 512 grid points , title="fig:",width=240 ] ( top ) , @xmath233 ( middle ) and @xmath234 ( bottom ) of 1d brio - wu shock tube problem at @xmath178 with 512 grid points , title="fig:",width=240 ] in 2d cases , the absolute difference between the results of double precision and single precision computation of mhd rotor test ( @xmath235 ) and orszag - tang vortex test ( @xmath236 and @xmath195 ) are shown in fig . [ fig : rotor_diff ] and fig [ fig : ot_diff ] , respectively . the double precision computation results of both tests are also shown in the left - hand side of these figures . for the mhd rotor test , even the resulting image ( left in fig . [ fig : rotor_diff ] ) looks similar to the single precision resulting image ( top - left of fig . [ fig:2drotor ] ) , the high differences at the dense region can be found . experimental result shows that the maximum @xmath237 is larger than @xmath238 . ( left ) and @xmath239 ( right ) of mhd rotor problem at @xmath235 with @xmath240 grid points , title="fig:",width=240 ] ( left ) and @xmath239 ( right ) of mhd rotor problem at @xmath235 with @xmath240 grid points , title="fig:",width=240 ] fig . [ fig : ot_diff ] shows the absolute difference between the results of double precision and single precision computation of orszag - tang test at @xmath236 and @xmath195 . as the simulation time increases , the maximum @xmath237 increases from about @xmath241 to @xmath242 . ( left ) and @xmath239 ( right ) of orszag - tang problem at @xmath236 ( top ) and @xmath195 ( bottom ) with @xmath240 grid points , title="fig:",width=240 ] ( left ) and @xmath239 ( right ) of orszag - tang problem at @xmath236 ( top ) and @xmath195 ( bottom ) with @xmath240 grid points , title="fig:",width=240 ] ( left ) and @xmath239 ( right ) of orszag - tang problem at @xmath236 ( top ) and @xmath195 ( bottom ) with @xmath240 grid points , title="fig:",width=240 ] ( left ) and @xmath239 ( right ) of orszag - tang problem at @xmath236 ( top ) and @xmath195 ( bottom ) with @xmath240 grid points , title="fig:",width=240 ] fig . [ fig:3dblast1_diff ] and fig . [ fig:3dblast2_diff ] show the resulting images of the simulation using double precision and the contours of the absolute differences between the results of double precision and single precision computation of 3d blast wave test with @xmath243 grid points at @xmath227 and @xmath186 . as it is a high dimension computation in low resolution , the differences between them are clear . the number of grid points having higher difference value increases , and the @xmath237 is still less than @xmath244 . small difference value makes the double precision resulting images ( fig . [ fig:3dblast1_diff ] and fig . [ fig:3dblast2_diff ] ) looked similar to the single precision resulting images ( fig . [ fig:3dblast01 ] and fig . [ fig:3dblast02 ] ) . ( top - left ) , @xmath245 ( bottom - left ) and @xmath246 ( top - right ) , @xmath239 ( bottom - right ) of 3d blast wave problem at @xmath227 with @xmath243 grid points , title="fig:",width=240 ] ( top - left ) , @xmath245 ( bottom - left ) and @xmath246 ( top - right ) , @xmath239 ( bottom - right ) of 3d blast wave problem at @xmath227 with @xmath243 grid points , title="fig:",width=240 ] ( top - left ) , @xmath245 ( bottom - left ) and @xmath246 ( top - right ) , @xmath239 ( bottom - right ) of 3d blast wave problem at @xmath227 with @xmath243 grid points , title="fig:",width=240 ] ( top - left ) , @xmath245 ( bottom - left ) and @xmath246 ( top - right ) , @xmath239 ( bottom - right ) of 3d blast wave problem at @xmath227 with @xmath243 grid points , title="fig:",width=240 ] ( top - left ) , @xmath245 ( bottom - left ) and @xmath246 ( top - right ) , @xmath239 ( bottom - right ) of 3d blast wave problem at @xmath186 with @xmath243 grid points , title="fig:",width=240 ] ( top - left ) , @xmath245 ( bottom - left ) and @xmath246 ( top - right ) , @xmath239 ( bottom - right ) of 3d blast wave problem at @xmath186 with @xmath243 grid points , title="fig:",width=259 ] ( top - left ) , @xmath245 ( bottom - left ) and @xmath246 ( top - right ) , @xmath239 ( bottom - right ) of 3d blast wave problem at @xmath186 with @xmath243 grid points , title="fig:",width=240 ] ( top - left ) , @xmath245 ( bottom - left ) and @xmath246 ( top - right ) , @xmath239 ( bottom - right ) of 3d blast wave problem at @xmath186 with @xmath243 grid points , title="fig:",width=259 ] an important point can be realized that not only the grid points at the high density region has high difference value , but also the number of grid points having high difference value and the amount of the difference values are increasing along with the increase of the simulation time . higher dimension is another factor to introduce noticeable differences between the computation results with different precisions because higher dimension means a grid point has more neighbors and more neighbors need more computation steps in one time step . as a result the differences become more obvious . therefore , for a long - term simulation , double precision computation is a must . the original fortran code @xcite is a second - order accurate high - resolution tvd mhd code . theoretically , we consider that _ gpu - mhd _ is sufficient to capture forward and reverse shocks as well as any other discontinuities such as contact discontinuities which are important in space physics . as _ gpu - mhd _ is a dimensional - splitting based code , there are two drawbacks : ( i ) the code is unable to evolve the normal ( to the sweep direction ) magnetic field during each sweep direction @xcite , and ( ii ) splitting errors will generally be introduced due to the linearized jacobian flux matrices do not commute in most of the nonlinear multidimensional problems @xcite . the performance measurements of the gpu and cpu implementations as well as the computation using double precision and single precision are carried out in this section . different numbers of grid points and different dimensions were used in the performance tests . we run both _ gpu - mhd _ and pen _ et al . _ s fortran / cpu mhd code @xcite to perform the simulations on a pc with intel core i7 965 3.20 ghz cpu , 6 g main memory , running microsoft windows xp 64-bit professional . two graphics cards were tested : nvidia geforce gtx 295 with 1.75 g video memory and gtx 480 ( fermi ) with 1.5 g video memory . the fortran compiler and gpu development toolkit we used are g95 stable version 0.92 and nvidia cuda 3.2 , respectively . _ gpu - mhd _ was designed for three - dimensional problems , thus the dimensions are expressed in three - dimensional form in all figures . for 1d test , 1d brio - wu shock tube problem ( see section [ sub : brio - wu ] ) was used . for 2d test , 2d orszag - tang problem ( see section [ sub : o - t ] ) was used was used . for 3d test , 3d blast wave problem ( see section [ sub:3dblast ] ) was used . [ fig : speedup1d ] reports the comparison of _ gpu - mhd _ and the fortran / cpu code of 1d test with different numbers of grid points in single and double precisions . in single precision mode , basically there is only about 10 times speedup ( @xmath247 case ) since the number of grid points is small . and it should be realized that the amount of speedup is increased as long as the resolution is increased but dropped when the resolution reaches @xmath248 . it is because the `` max threads per block '' of gtx 295 is 512 , all the computations are handled within one block and a very high processing speed can be archived . on gtx 480 , there is about 80 times speedup ( @xmath247 case ) and the amount of speedup is increased linearly due to the higher floating point capability ( 512 mad ops / clock ) . in double precision mode , around 10 times and 60 times speedup ( @xmath247 case ) is achieved on gtx 295 and gtx 480 , respectively . table [ table_1d_precision ] gives the comparison of _ gpu - mhd _ using single precision and double precision of 1d test with different numbers of grid points . on gtx 295 , a similar speed drop happened both in single and double precision modes , but it occurred in different resolutions : @xmath248 in single precision and @xmath249 in double precision . this is not strange and not difficult to understand since the double precision has double size of data to be handled by the processors . except the special case of @xmath248 resolution , the processing speed in both modes are very closed . on gtx 480 , the performance between single and double precision is quite close . .the performance results of 1d test between single precision and double precision of _ gpu - mhd _ at different resolutions [ cols="^,^,^,^,^ " , ] there is a need to visualize the mhd simulation data , for examples , daum @xcite developed a toolbox called _ visan mhd _ in matlab for mhd simulation data visualization and analysis . with the help of gpus , stantchev _ et al . _ @xcite used gpus for computation and visualization of plasma turbulence . in _ gpu - mhd _ , using the parallel computation power of gpus and cuda , the simulation results of one time step can be computed in dozens or hundreds milliseconds . according to the efficiency of _ gpu - mhd _ , near real - time visualization is able to be provided for 1d and 2d problems . the motion or attributes of the magnetic fluid can be computed and rendered on the fly . so the changes of the magnetic fluid during the simulation can be observed in real - time . by adding the real - time visualization , the flow of _ gpu - mhd _ , fig . [ fig : flowchart ] is extended as fig . [ fig : flowchart_v ] : _ gpu - mhd _ provides different visualization methods for one - dimensional , two - dimensional and three - dimensional problems . to visualize one - dimensional problems for each time step , the simulation results are copied to the cuda global memory that mapped to the vertex buffer object ( vbo ) @xcite . for all grid points , one grid point is mapped to one vertex . the position of each grid point is mapped as the @xmath13-position of the vertex and the selected physical value ( @xmath4 , @xmath187 , etc . ) is mapped as the @xmath55-position of the vertex . then a curve of these vertices is drawn . since the vbo is mapped to cuda global memory and simulation results are stored in gram , the copying and mapping operations are fast . experimental result shows that _ gpu - mhd _ with real - time visualization can achieve 60 frame per second ( fps ) in single precision mode and 30 fps in double precision mode on gtx 295 . on gtx 480 , around 60 fps in both single and double precisions is achieved . [ fig:1dvis ] shows two example images of 1d visualizations using _ gpu - mhd_. ) of brio - wu shock tube problem with @xmath248 grid points using _ gpu - mhd_.,title="fig:",width=240 ] ) of brio - wu shock tube problem with @xmath248 grid points using _ gpu - mhd_.,title="fig:",width=240 ] the operational flow of visualization of 2d problems is similar to that in 1d visualization . however , instead of vertex buffer object ( vbo ) , pixel buffer object(pbo ) @xcite is used . for each time step , the simulation results are copied to the cuda global memory that are then mapped to the pbo . for all grid points , one grid point is mapped to one pixel . the @xmath13 and @xmath55 position of each grid point are mapped as the corresponding @xmath13-position and the @xmath55-position of the vertex and the selected physical value ( @xmath4 , @xmath187 , etc . ) is mapped as the color of the pixel to form a color image . to render this color image , a transfer function is set to map the physical value to the color of the pixel and then the resulting image is drawn . similar to vbo , pbo is also mapped to cuda global memory and the simulation results are stored in gram , so the copying and mapping operations are also fast and do not affect too much to the performance . although the number of grid points in 2d problem is much larger than those in the one - dimension problem , the fps still reaches 10 in single precision mode and 6 in double precision mode on gtx 295 when the number of grid points is @xmath240 , still giving acceptable performance to the user . on gtx 480 , 22 fps in single precision and 17 fps in double precision are achieved and thus interactive rates are available . [ fig:2dvis ] shows two example images of 2d visualizations using _ gpu - mhd_. ) of orszag - tang vortex problem with @xmath240 grid points using _ gpu - mhd_.,title="fig:",width=240 ] ) of orszag - tang vortex problem with @xmath240 grid points using _ gpu - mhd_.,title="fig:",width=240 ] however , visualization of 3d problem is different to 1d and 2d problems . gpu - based volume visualization method @xcite and texture memory ( or video memory ) are used . unfortunately , the current version ( version 2.3 ) of cuda does not provide the feature to copy the data from the cuda global memory to texture memory directly , even both of them are in gram . on the other hand , texture memory is readable but is not rewritable in cuda . so the simulation results have to be copied to the main memory first , and then be copied to texture memory . in addition , the number of grid points is usually large compared to 2d problems and volume visualization techniques are somewhat time - consuming . as a result , on gtx 295 , _ gpu - mhd _ only gets 2 fps in single precision mode and 1 fps in double precision mode when the number of grid points is @xmath243 , and it is far from real - time . nevertheless , we still get 10 fps ( single precision mode ) and 6 fps ( double precision mode ) for performing the simulation of problems with resolution of @xmath250 and about 20 fps ( single and double precision modes ) for problems with resolution of @xmath251 . on gtx 480 , we can get 60 fps for both single and double precision for @xmath251 grid points , 20 fps ( single ) and 16 fps ( double ) for @xmath250 grid points , and 6.1 fps ( single ) and 3.6 fps ( double ) for @xmath243 grid points . [ fig:3dvis ] shows two example images of 2d visualizations using _ gpu - mhd_. ) of 3d blast wave problem with @xmath243 grid points using _ gpu - mhd_.,title="fig:",width=240 ] ) of 3d blast wave problem with @xmath243 grid points using _ gpu - mhd_.,title="fig:",width=240 ] in this paper we present , to the author s knowledge , the first implementation of mhd simulations entirely on gpus with cuda , named _ gpu - mhd _ , to accelerate the simulation process . the aim of this paper is to present a gpu implementation in detail , demonstrating how a tvd based mhd simulations can be implemented efficiently for nvidia gpus with cuda . a series of numerical tests have been performed to validate the correctness of our code . accuracy evaluation by comparing single and double precision computation results is also given , indicating that double precision support on gpus is a must for long - term mhd simulation . performance measurements of both single and double precision modes of _ gpu - mhd _ are conducted on gt200 architecture ( gtx 295 ) and fermi architecture ( gtx 480 ) . these measurements show that our gpu - based implementation achieves between one and two orders of magnitude depending on the used graphics card , problem size , and precision when comparing to the original serial cpu mhd implementation . in order to provide the user better understanding of the problems being investigated during the simulation process , we have extended _ gpu - mhd _ to support visualization of the simulation results . with _ gpu - mhd _ , the whole mhd simulation and visualization process can be performed entirely on gpus . there are two directions in our future work , firstly , we are going to extend _ gpu - mhd _ for multiple gpus and gpu cluster @xcite to fully exploit the power of gpus . secondly , we will investigate implementing other recent high - order godunov mhd algorithms such as @xcite and @xcite on gpus . these gpu - based algorithms will be served as the base of our gpu framework for simulating large - scale mhd problems in space weather modeling . this work has been supported by the science and technology development fund of macao sar ( 03/2008/a1 ) and the national high - technology research and development program of china ( 2009aa122205 ) . xueshang feng is supported by the national natural science foundation of china ( 40874091 and 40890162 ) . the authors would like to thank dr . ue - li pen and bijia pang at the canadian institute for theoretical astrophysics , university of toronto for providing the fortran mhd code . thanks to dr . yuet ming lam for his suggestions on the revision of the paper . special thanks to anonymous reviewers for their constructive comments on the paper . d. s. balsara , d. s. spicer , a staggered mesh algorithm using high order godunov fluxes to ensure solenoidal magnetic fields in magnetohydrodynamic simulations , _ j. comput . phys . _ * 149 * ( 1999 ) 270 - 292 . r. g. belleman , j. bdorf , s. f. portegies zwart , high performance direct gravitational n - body simulations on graphics processing units ii : an implementation in cuda , _ new astronomy _ * 13 * ( 2008 ) 103 - 112 . s. che , m. boyer , j. meng , d. tarjan , j. w. sheaffer , k. skadron , a performance study of general - purpose applications on graphics processors using cuda , _ j. parallel distrib . * 68 * ( 2008 ) 1370 - 1380 . a. ciardi , s. lebedev , a. frank , e. blackman , d. ampleford , c. jennings , j. chittenden , t. lery , s. bland , s. bott , g. hall , j. rapley , f. vidal , a. marocchino , 3d mhd simulations of laboratory plasma jets , _ astrophysics and space science _ * 307 * ( 2007 ) 17 - 22 . m. dryer , space weather simulations in 3d mhd from the sun to earth and beyond to 100 au : a modeler s perspective of the present state of the art ( invited review ) , _ asia j. of physics _ * 16 * ( 2007 ) 97 - 121 . j. c. hayes , m. l. norman , r. a. fiedler , j. o. bordner , p. s. li , s. e. clark , a. ud - doula , m .- m. low , simulating radiating and magnetized flows in multiple dimensions with zeus - mp , _ astrophys . j. supp . _ * 165 * ( 2006 ) 188 - 228 . d. lee , a. e. deane , a parallel unsplit staggered mesh algorithm for magnetohydrodynamics , in _ parallel computational fluid dynamics theory and applications _ , edited by a. deane _ et al . _ , elsevier ( 2006 ) 243 - 250 . p. mininni , e. lee , a. norton , j. clyne , flow visualization and field line advection in computational fluid dynamics : application to magnetic fields and turbulent flows , _ new j. phys . _ * 10 * ( 2008 ) 125007 . l. seiler , d. carmean , e. sprangle , t. forsyth , m. abrash , p. dubey , s. junkins , a. lake , j. sugerman , r. cavin , r. espasa , e. grochowski , t. juan , p. hanrahan , larrabee : a many - core x86 architecture for visual computing , _ acm trans . _ * 27 * ( 2008 ) article 18 . g. stantchev , d. juba , w. dorland , a. varshney , using graphics processors for high - performance computation and visualization of plasma turbulence , _ computing in science and engineering _ * 11 * ( 2009 ) 52 - 59 . g. tth , d. odstr@xmath252il , comparison of some flux corrected transport and total variation diminishing numerical schemes for hydrodynamic and magnetohydrodynamic problems , _ j. comput . * 128 * ( 1996 ) 82 - 100 .	magnetohydrodynamic ( mhd ) simulations based on the ideal mhd equations have become a powerful tool for modeling phenomena in a wide range of applications including laboratory , astrophysical , and space plasmas . in general , high - resolution methods for solving the ideal mhd equations are computationally expensive and beowulf clusters or even supercomputers are often used to run the codes that implemented these methods . with the advent of the compute unified device architecture ( cuda ) , modern graphics processing units ( gpus ) provide an alternative approach to parallel computing for scientific simulations . in this paper we present , to the author s knowledge , the first implementation of mhd simulations entirely on gpus with cuda , named _ gpu - mhd _ , to accelerate the simulation process . _ gpu - mhd _ supports both single and double precision computation . a series of numerical tests have been performed to validate the correctness of our code . accuracy evaluation by comparing single and double precision computation results is also given . performance measurements of both single and double precision are conducted on both the nvidia geforce gtx 295 ( gt200 architecture ) and gtx 480 ( fermi architecture ) graphics cards . these measurements show that our gpu - based implementation achieves between one and two orders of magnitude depending on the used graphics card , problem size , and precision when comparing to the original serial cpu mhd implementation . in addition , we extend _ gpu - mhd _ to support the visualization of the simulation results and thus the whole mhd simulation and visualization process can be performed entirely on gpus . mhd simulations , gpus , cuda , parallel computing
in @xcite it was argued that the transport barrier near the core of the austral polar night jet can be explained by a mechanism different from the potential vorticity ( pv ) barrier mechanism @xcite . the new barrier mechanism , which was subsequently referred to as `` strong kam stability '' @xcite , follows from an argument that does not make use of dynamical constraints on the streamfunction . this necessitates that dynamical constraints be considered separately . interestingly , decoupling of the dynamical constraints from the barrier mechanism leads to the possibility that transport barriers in pv - conserving flows may occur at locations that do not coincide with pv - barriers . @xcite predicted that barriers of this type should be present in close proximity to the cores of westward zonal jets in planetary atmospheres . in this paper we demonstrate that transport barriers of this type are present in a numerically simulated pv - conserving flow . we also argue that barriers of the type described are present in jupiter s midlatitude weather layer and in the earth s summer hemisphere subtropical stratosphere . in the following section passive tracer transport in a numerically simulated perturbed pv - staircase flow is investigated . it is shown that robust meridional transport barriers in close proximity to the cores of both eastward and westward zonal jets are present . the surprise is that westward jets , at which the background pv - gradient vanishes , act as transport barriers . essential elements of the strong kam stability argument are reviewed to explain this behavior . in section 3 we discuss the relevance of transport barriers of the strong kam stability type to : 1 ) the interpretation of jupiter s midlatitude weather layer belt - zone structure ; and 2 ) the earth s summer hemisphere subtropical stratosphere . in the final section we briefly discuss our results . in this section we consider passive tracer transport in a perturbed pv - staircase flow . we assume quasigeostrophic dynamics in a one - layer reduced - gravity setting and make use of a local cartesian coordinate system @xmath0 where @xmath1 and @xmath2 increase to the east and north , respectively , and the constant @xmath3 is the local @xmath2-derivative of the coriolis parameter . the zonal and meridional components of the velocity field are @xmath4 and @xmath5 , respectively , where @xmath6 is the streamfunction . the flow is constrained to satisfy @xmath7 where @xmath8 is the quasigeostrophic potential vorticity and @xmath9 is the deformation radius . recent theoretical , numerical and experimental work , including extensions involving spherical geometry , shallow - water dynamics and inclusion of weak forcing and dissipation , has shown that flows satisfying ( [ qcons])([qdef ] ) with periodic boundary conditions in @xmath1 tend to evolve toward a state of the form @xmath10 where @xmath11 is a small perturbation to @xmath12 @xcite . the background zonal flow is characterized by an approximately piecewise constant pv distribution that has been appropriately described as a pv - staircase @xcite . the corresponding zonal velocity profile @xmath13 is periodic in @xmath2 . taking the period to be @xmath14 , the jump in @xmath15 at each step is @xmath16 , and the zonal flow is the periodic extension of @xmath17 consisting of a periodic sequence of alternating narrow eastward and broad westward zonal jets with @xmath15 piecewise constant between adjacent eastward jets , as illustrated in fig . note that at the center of each constant @xmath18 band lies a westward jet . in the limit @xmath19 ( [ u_0 - 1 ] ) reduces to @xmath20 whose qualitative features are identical to those of the finite @xmath21 case . the flows ( [ u_0 - 1 ] ) and ( [ u_0 - 2 ] ) are normalized so that the integral of @xmath22 over @xmath2 vanishes . the rhines scale @xmath23 , where @xmath24 is a characteristic velocity , is an approximate measure of the separation between adjacent eastward jets . corresponding to ( [ u_0 - 2 ] ) this separation is exactly @xmath25 where @xmath26 is the wind speed at the core of one of the eastward jets . before presenting numerical simulations of passive tracer transport in a perturbed pv - staircase flow we describe predictions based on two different arguments of the expected locations of transport barriers in this flow . first , the pv - barrier argument @xcite leads to the expectation that transport barriers should be present only near the cores of the eastward jets . the basic elements of the pv - barrier argument were used originally by @xcite to explain the mechanism by which the eastward jet at the perimeter of the austral stratospheric polar vortex , sometimes referred to as the austral polar night jet , during the late winter and early spring serves to trap ozone - depleted air inside the polar vortex . the essential elements of the argument are that at eastward jets the large gradient of @xmath15 is associated with a large rossby wave restoring force ( `` rossby wave elasticity '' ) which inhibits meridional exchange of fluid at larger scales and that shear @xmath27 acts to inhibit meridional exchange at smaller scales . ( but note that in @xcite it is argued that increasing meridional shear acts , on average , to increase meridional exchange . ) an alternative argument , based on the strong kam stability mechanism @xcite , leads to the expectation that transport barriers should be present near the cores of both eastward and westward jets in a pv - staircase flow . the argument leading to this expectation will now be reviewed . the lagrangian ( particle trajectory ) equations of motion , @xmath28 constitute a nonautonomous one - degree - of - freedom hamiltonian system , with @xmath0 the canonically conjugate coordinate momentum pair and @xmath6 the hamiltonian . this allows results from studies of integrable and nonintegrable hamiltonian systems to be applied . in the background steady flow , with @xmath29 , equations ( [ lagrange ] ) are integrable and the motion is describable using a transformed hamiltonian @xmath30 where @xmath31 are action angle variables . each trajectory lies on a torus which is labeled by its @xmath32-value . motion is @xmath33-periodic in @xmath34 with angular frequency @xmath35 . according to each of many variants of the kolmogorov arnold moser ( kam ) theorem @xcite , many of the unperturbed tori survive in the perturbed system ( [ psi ] ) , albeit in a slightly distorted form , provided certain conditions are met . surviving tori can not be traversed and serve as transport barriers . ( for reasons described in @xcite , the process known as arnold diffusion does not occur in the systems under study . ) torus destruction is caused by the excitation and overlapping of resonances . each resonance has a characteristic width @xmath36 . nondegenerate @xmath37 resonance widths are proportional to @xmath38 . degenerate @xmath39 resonance widths do not vanish but are generally narrower than nondegenerate resonance widths . ( quantitative estimates of both degenerate and nondegenerate resonance widths are given in @xcite ; for our purposes it suffices to note the general trend . ) for most moderate strength perturbations to the background , small resonance widths near degenerate tori lead to nonoverlapping resonances and thus unbroken tori that serve as transport barriers . this constitutes the strong kam stability barrier argument . in our model ( [ psi ] ) the connection between @xmath0 and @xmath31 is particularly simple : @xmath40 , @xmath41 where @xmath42 is the distance around the planet along a constant latitude circle at the latitude at which @xmath3 is defined . the period of motion @xmath43 is @xmath44 , so @xmath45 . at the cores of both eastward and westward jets @xmath46 , so @xmath47 at these locations and the strong kam stability argument predicts that robust meridional transport barriers should be present . barriers of this type may be broken if the transient perturbation @xmath48 strongly excites a low - order resonance with the frequency @xmath49 of the background flow near the core of the jet . three final issues are noteworthy . first , because the strong kam stability argument is a kinematic argument ( based on ( [ lagrange ] ) alone ) , dynamical consistency consistency between ( [ qcons])([psi ] ) and ( [ u_0 - 1 ] ) or ( [ u_0 - 2])and flow stability must be considered separately . second , our emphasis on jets is unnecessarily restrictive inasmuch as the strong kam stability argument holds at all locations where @xmath50 . third , the stated results on kam theory assume that @xmath48 can be expressed as a multiperiodic ( generically quasiperiodic ) function of @xmath51 . we now describe a set of numerical experiments that were performed to investigate passive tracer transport in a perturbed pv - staircase flow . the streamfunctions on which our tracer transport simulations are based were constructed by numerically solving ( [ qcons])([qdef ] ) using as an initial state a perturbation to the background pv - staircase ( [ u_0 - 1 ] ) . before presenting the results , it is appropriate to make two comments about what we expect to learn from these simulations . first , these simulations provide a test of the assertion that the decomposition ( [ qcons])([u_0 - 1 ] ) is dynamically consistent . second , given a positive outcome of the first test , these simulations test whether westward zonal jets , across which there is no pv - barrier , serve as robust meridional transport barriers . the quasigeostrophic equation ( [ qcons])([qdef ] ) was solved numerically using a standard pseudospectral technique on a @xmath52 grid in the @xmath53 computational domain . periodic boundary conditions in @xmath54 were applied in both the @xmath1 and @xmath2 directions . the arakawa representation of the advective terms , which are often written as a jacobian , the solution was marched forward in time using a second - order adams bashforth scheme with a dimensionless timestep @xmath55 . to control the spurious amplification of high wavenumber modes , we applied a weak exponential cutoff filter and included a small amount of hyperviscosity . two types of initial perturbation to the background pv - staircase were used in our simulations . the first type of perturbation consisted of a superposition of periodic displacements of pv contours with random phases uniformly distributed on @xmath56 . the second type of perturbation was a doubly periodic perturbation to the streamfunction consisting of a sum of a product of fourier modes with random phases . in all of our simulations the separation between adjacent eastward jets @xmath14 was taken to be 8000 km and @xmath57 km@xmath58 s@xmath58 was used . the simulations shown correspond to @xmath59 . in anticipation of our discussion of jupiter in the following section , these parameters were chosen to approximately reproduce conditions on jupiter . note , however , that in our simulations the period in @xmath1 is 32000 km , which is approximately one - tenth the midlatitude distance around jupiter on a line of constant latitude . many one - year model simulations were run . for some parameter values ten - year model simulations were performed . for the parameter values given , the change in both energy and enstrophy throughout the duration of the simulations performed was less than @xmath60 , giving us confidence in the accuracy of the simulations . both types of initial perturbation gave similar results . figure [ uq ] shows plots of instantaneous , at @xmath61 year , zonally - averaged zonal velocity @xmath62 , zonally averaged potential vorticity @xmath63 and potential vorticity @xmath64 . comparison of figs . [ u0q0 ] and [ uq ] provides strong support for the dynamical consistency of the decomposition ( [ psi])([u_0 - 1 ] ) , but these plots provide no insight into whether transport barriers are present . to address the latter question we have used the year - long records of computed velocity fields to : 1 ) follow the evolution of distributions of passive tracers , which evolve according to ( [ lagrange ] ) ; and 2 ) compute finite - time lyapunov exponents ( ftles ; see , for example , * ? ? ? * ) . typical results are shown in fig . [ ftle ] . as shown in the figure , the initial positions of the passive tracers fell on the lines @xmath65 , which lie midway between the unperturbed eastward and westward jets . it is seen that after a one - year integration the regions between jets are well - mixed , but that there is no meridional transport across the cores of the zonal jets . it should be emphasized that in this and other simulations tracer particles spread meridionally to fill the domains shown in about two weeks . throughout the reminder of the one - year integration no additional meridional tracer spreading occurs . with this in mind , [ ftle ] clearly shows that both eastward and westward zonal jets act as meridional transport barriers , consistent with the strong kam stability argument . calculation of ftles provides an additional test of the correctness of the strong kam stability argument . lyapunov exponents are a measure of the rate of divergence of neighboring trajectories . according to the strong kam stability argument , the transport barriers at the cores of zonal jets coincide with generally thin bands of kam invariant tori on which lyapunov exponents are zero . finite time estimates of lyapunov exponents will not be identically zero on kam invariant tori , but these structures should be readily identifiable as thin bands of low ftle estimates . this is precisely what is seen in fig . [ ftle ] ; both westward and eastward jets are identifiable as thin bands of low ftle estimates , consistent with the strong kam stability argument . typical computed values of ftle , in units of @xmath66 , shown in fig . [ ftle ] are 2 at the cores of the westward jets , 7 at the cores of the eastward jets , and 17 in the well - mixed regions . we have performed many numerical experiments based on pv - staircase flows of the type described here . these simulations support the conclusion that both eastward zonal jets ( where @xmath67 is very large ) and westward zonal jets ( where @xmath67 is very small ) act as robust meridional transport barriers . this conclusion is not sensitive to the choice of parameter values or details of the initial perturbation . two general trends are noteworthy . first , for a small perturbation the width of the barrier region near westward jets is greater than the width of the barrier region near eastward jets . this behavior is consistent with the strong kam stability argument : @xmath68 is small over a larger @xmath2-domain near westward jets than near eastward jets . second , as the perturbation strength increases , transport barriers near westward jets generally break before eastward jet barriers break . in our simulations the westward jet barriers broke when the initial rms meridional pv - contour displacement exceeded approximately @xmath69 , while the eastward jet barriers broke when the initial rms displacement was approximately twice this value . possible explanations for the somewhat more robust nature of the eastward jets are : 1 ) the pv - barrier mechanism and the strong kam stability barrier mechanism act in tandem to strengthen the barriers near eastward jets ; and 2 ) we have performed a linear theory rossby wave analysis of pv - staircase flows that reveals that rossby wave critical layers are precluded at the eastward jets ( see also * ? ? ? * ) , which suggest that the eastward jet barriers may be more robust . in the previous section it was demonstrated that transport barriers may exist in a pv - conserving flow at locations that do not coincide with pv - barriers . in this section we discuss observational evidence that suggests the existence of transport barriers in the absence of pv - barriers . we consider two examples : 1 ) jupiter s weather layer ; and 2 ) the earth s stratosphere . in both cases , conclusions drawn should be regarded as tentative inasmuch as we do not treat either system in enough depth to make a definitive statement . in spite of our incomplete treatment of these topics , we feel that it is important to point out that , consistent with the theoretical and numerical results presented in the previous section , there is observational evidence in planetary atmospheres that suggests the existence of transport barriers in the absence of pv - barriers that can be explained by the strong kam stability mechanism . both the examples considered are , in our view , sufficiently important that the connection between the observations discussed and the strong kam stability barrier mechanism is worthy of a much more thorough investigation . it is natural to focus on flows consisting of a sequence of alternating eastward and westward zonal jets because many of the arguments used in the previous section are then applicable . in particular , for this class of flows , meridional transport barriers of the strong kam stability type are predicted to occur at the latitudes where @xmath50 . furthermore , in alternating zonal jet flows one may anticipate that eastward and westward jets are associated with large and small background pv - gradients , respectively . note , however , that for the purpose of identifying transport barriers in the absence of pv - barriers it is not necessary that the background pv distribution in the flows considered be that of an idealized pv - staircase . this point is discussed in more detail below . the following simple scaling argument shows why an alternating zonal jet flow is well - developed in jupiter s weather layer but is only marginally identifiable in the earth s stratosphere . these systems are then discussed in turn . recall that in a pv - staircase flow the separation between adjacent eastward jets is @xmath25 . we shall assume that this estimate approximately holds for general midlatitude multiple zonal jet mean flow patterns . let @xmath70 denote planetary radius and @xmath71 planetary rotation rate . then @xmath72 and the number of eastward ( or westward ) jets one expects to observe in each hemisphere at midlatitudes ( whose extent in latitude is taken here to be half the equator to pole distance ) is approximately @xmath73 . in jupiter s weather layer ( @xmath74 m s@xmath58 , @xmath75 km , @xmath76 h ) @xmath77 , in good agreement with fig . [ beltzone ] , as described below . in the earth s stratosphere ( @xmath74 m s@xmath58 , @xmath78 km , @xmath79 h ) @xmath80 . thus , conditions for the formation of a multiple zonal jet mean flow pattern are only marginally satisfied in the earth s stratosphere . in the earth s troposphere @xmath26 is smaller , suggesting more favorable conditions , but mountain ranges and thermal exchange processes between the atmosphere and irregularly shaped oceans and continents constitute significant hindrances to the formation of zonal flows . conditions are favorable in the earth s oceans ( @xmath81 m s@xmath58 , corresponding to @xmath82 ) but there the presence of lateral boundaries dictates that zonal jets be embedded in recirculation gyres @xcite . the most striking feature of jupiter s weather layer circulation @xcite is that it is organized in a sequence of alternating eastward and westward zonal jets whose meridional excursion is very small . figure [ beltzone ] shows zonally - averaged zonal wind speed on jupiter at the cloud top level as a function of latitude . regions with dark and light shading in this figure are referred to as belts and zones , respectively . belts and zones correspond to regions in which the background motion is cyclonic @xmath83 and anticyclonic @xmath84 , respectively . the boundaries between adjacent belts and zones coincide with the cores of the zonal jets . at these boundaries @xmath50 . belts and zones have different radiative transfer properties ( analyses are not limited to the visible band of the electromagnetic spectrum ) which is attributed to differences in chemical composition @xcite . assuming that the weather layer flow is approximately two dimensional and that chemical species are long - lived , one may infer from the observation that adjacent belts and zones have different chemical constituents that there is very little fluid exchange between adjacent belts and zones , implying that both eastward and westward zonal jets act as robust meridional transport barriers . figure [ uqq_y ] shows @xmath22 , @xmath15 , and @xmath85 in jupiter s weather layer . the same data are shown ( but are plotted differently ) in @xcite . the question of whether a pv - staircase is a useful approximate model of jupiter s weather layer has been considered by many authors @xcite . for our purposes the answer to this question is not critical . our focus is on identifying transport barriers that can not be explained by the pv - barrier mechanism . figure [ uqq_y ] shows that while all eastward jets are associated with large meridional pv - gradients , most of the westward jets are associated with small meridional pv - gradients . ( here and above we are using the term westward jet somewhat loosely to include minima of @xmath22 , even when @xmath86 at the minimum . ) in other words , most of the transport barriers near the cores of the westward jets can not be explained by the pv - barrier mechanism . but all of the belt - zone boundaries the apparent meridional transport barriers coincide with latitudes at which @xmath50 , consistent with the strong kam stability barrier mechanism . thus all of the apparent transport barriers can be explained by the strong kam stability barrier mechanism . note , however , that there is no apparent barrier on the equator , where @xmath50 . this is probably due to a combination of anomalous equatorial dynamics @xcite and the manner by which chemical constituents are pumped into the near - equatorial weather layer . some caveats relating to our interpretation of observations from jupiter should be emphasized , however . first , we have assumed that the weather layer flow is nearly two dimensional and horizontally nondivergent , being only weakly forced by convection . although these assumptions are generally accepted ( see , for example , * ? ? ? * ) , it should be noted that our explanation of the apparent transport barrier rests on their validity . a second assumption that we have made is that chemical species in jupiter s weather layer are long - lived . another possible explanation of the apparent transport barriers between adjacent belts and zones is that chemical species are short - lived , being continuously pumped into the weather layer by convective overturning @xcite . we can not rule out this possibility . our argument shows , however , that the apparent lack of fluid exchange between adjacent belts and zones _ can _ be explained using dynamical arguments . the simple scaling argument given above predicts that conditions for the formation of a stable alternating multiple jet zonal mean flow pattern are only marginally satisfied in the earth s stratosphere . in qualitative agreement with this prediction , in each hemisphere there is one readily identifiable eastward zonal jet and one westward jet , and the appearance of these jets is seasonal ( see , for example , * ? ? ? the stronger jets are the high latitude eastward polar night jets which appear in the winter hemisphere . ( the austral polar night jet is particularly strong , persisting throughout most of the stratosphere during the austral fall , winter and spring . ) the westward jets are present in the subtropics throughout most of the stratosphere during the summer months in each hemisphere . the eastward polar night jets , especially in the southern hemisphere , act as transport barriers and are associated with strong pv - gradients . because the focus in the present study is on identifying transport barriers in the absence of pv - barriers , these jets are not of interest here . in contrast , the westward subtropical jets in the summer hemisphere are very much of interest because these are not associated with a strong pv - gradient . the properties just described are illustrated in fig . [ era40 ] . that figure shows a 7-year ( 19921998 ) monthly average of zonally - averaged potential vorticity ( in color ) and zonal wind ( as contours ) on the 460 k isentropic surface ( which lies in the lower stratosphere ) based on the european center for medium - range weather forecasts ( ecmwf ) 40-year re - analysis ( era-40 ) product . both the eastward polar night jets and the westward subtropical jets are readily identifiable . as we have noted , the eastward winter hemisphere polar night jets are associated with strong meridional pv - gradients while the westward subtropical jets are associated with nearly homogenous pv - distributions . the westward subtropical jets have the dynamical properties that we seek zonal jets in the absence of strong pv - gradients . we note also that , consistent with arguments made originally by @xcite , rossby wave perturbations to the background in these regions are weak . from the standpoint of applicability of the strong kam stability argument , weak perturbations are advantageous . we turn our attention now to the question of whether the cores of these jets serve as robust meridional transport barriers . studies of stratospheric transport based on effective diffusivity have been carried out by @xcite and @xcite . the effective diffusivity is large ( small ) in regions where fluid is well ( poorly ) mixed . fluid in the vicinity of a transport barrier is poorly mixed ; these regions are thus characterized by a small effective diffusivity . both stratospheric effective diffusivity analyses reveal that the westward subtropical jet in the summer hemisphere coincides with a region of anomalously low effective diffusivity ; see plates 1 and 4 in @xcite and 1 through 4 in @xcite . this suggests that these westward jets act as meridional transport barriers . previous work by @xcite and @xcite had focused on this `` subtropical barrier . '' indeed , this barrier comprises a critical element of the `` tropical pipe '' model @xcite of stratospheric transport . observational evidence that suggests the presence of subtropical transport barriers is presented in @xcite , @xcite , @xcite , @xcite and @xcite . @xcite provides a recent review of stratospheric transport , including a discussion of subtropical transport barriers . the evidence that we have pointed out strongly suggests that : 1 ) the stratospheric subtropical barrier is a robust meridional transport barrier that coincides with the core of a westward jet ; 2 ) the associated meridional pv - gradient is very small so this barrier can not be explained by the pv - barrier mechanism ; and 3 ) because @xmath50 at this barrier , the barrier is predicted by the strong kam stability barrier mechanism . to test these tentative conclusions more rigorously , a study based on realistic synoptic winds that tracks both potential vorticity and tracer distributions should be conducted . in the first part of this paper we presented numerical simulations of passive tracer transport in a perturbed pv - staircase flow and showed that both eastward and westward jets in this flow act as meridional transport barriers . the surprise is that westward jets , where the background pv gradient vanishes , act as transport barriers . this behavior was explained as being a consequence of the strong kam stability barrier mechanism . we then briefly discussed the applicability of the strong kam stability mechanism to explaining observations of jupiter s weather layer and the earth s subtropical stratosphere . in both of these systems westward jets are present that appear to act as robust meridional transport barriers in the absence of a background meridional pv - barrier . these barriers are predicted by the strong kam stability mechanism . in both cases the evidence presented should be regarded as suggestive . more thorough investigations of both problems is recommended . the principal weakness of our explanation of the apparent lack of fluid exchange between adjacent belts and zones in jupiter s midlatitude weather layer is that we can not exclude the possibility that the observed chemical composition differences in adjacent belts and zones are caused by a combination of strong convective overturning and short - lived chemical species . in spite of this caveat , it is important to emphasize that we have shown that maintenance of the apparent chemical composition differences between adjacent belts and zones _ can _ be explained using a dynamical argument ( as opposed to a chemistry - based argument ) in which the weather layer flow is only weakly convectively forced . the principal weaknesses in our discussion of the earth s subtropical stratospheric transport barrier were that all of the properties noted were based on averaged winds rather than synoptic winds and that tracer transport and potential vorticity distributions were not estimated in a way that was guaranteed to be self consistent . it should not be difficult to overcome these shortcomings using model - based synoptic winds . we thank t. dowling and r. morales - juberias for providing the data used to construct figs . [ beltzone ] and [ uqq_y ] . comments on the manuscript by anonymous reviewers , t. shepherd , t. zgkmen , and j. willemsen are sincerely appreciated . the ecmwf era-40 data used in this study were obtained from the ecmwf data server . support for this work was provided by the u.s . national science foundation under grants cmg0417425 and oce0648284 . , p. l. , y. h. yamazaki , s. r. lewis , p. d. williams , r. wordsworth , k. miki - yamazaki , j. sommeria , and h. didelle , 2007 : dynamics of convectively driven banded jets in the laboratory . , * 64 * , 40314052 . rypina , i. i. , m. g. brown , f. j. beron - vera , h. kocak , m. j. olascoaga , and i. a. udovydchenkov , 2007 : on the lagrangian dynamics of atmospheric zonal jets and the permeability of the stratospheric polar vortex . , * 64 * ,	the connection between transport barriers and potential vorticity ( pv ) barriers in pv - conserving flows is investigated with a focus on zonal jets in planetary atmospheres . a perturbed pv - staircase model is used to illustrate important concepts . this flow consists of a sequence of narrow eastward and broad westward zonal jets with a staircase pv structure ; the pv - steps are at the latitudes of the cores of the eastward jets . numerically simulated solutions to the quasigeostrophic pv conservation equation in a perturbed pv - staircase flow are presented . these simulations reveal that both eastward and westward zonal jets serve as robust meridional transport barriers . the surprise is that westward jets , across which the background pv gradient vanishes , serve as robust transport barriers . a theoretical explanation of the underlying barrier mechanism is provided . it is argued that transport barriers near the cores of westward zonal jets , across which the background pv gradient is small , are found in jupiter s midlatitude weather layer and in the earth s summer hemisphere subtropical stratosphere .
the big bang cosmological model rests upon the cosmological principle , which determines the form of the metric of the world , and allows the explanation of the redshift phenomenon as a pure geometrical effect directly related to the nature of that metric . the microwave background radiation or the abundance of the light elements strongly support the standard model . thus , given the overall consistency of the model , some authors consider as superfluous any attempt to test the basic hypothesis , i.e. , the form of the metric . furthermore , its explanatory capacity is often used to rule out any other theoretical alternative . recently , van dokkum and franx ( 1996 ) have argued that the microwave background radiation would not be planckian if the surface brightness ( sb ) dimming were different of @xmath1 . we remark that this is true from within the standard model , otherwise it should have been abandoned long ago . but the fact that data can be accommodated within the standard model , important as it is , can not be considered as a formal proof of the underlying hypothesis . it is worth to remember that , formally , a false statement can have consequences that are true . direct tests have been proposed since long ( see hubble & tolman 1935 ) , and recently reformulated from different angles ( moles 1991 ; sandage & perelmuter 1991 ; kjrgaard , jrgensen & moles 1993 ) . among the tests proposed by hubble & tolman , the sb test has been considered the most powerful and _ clean_. indeed , the sb of a standard candle only depends on the red - shift : it would decrease as @xmath1 in an expanding universe . otherwise , i.e. , in a static , non - expanding universe , with some ( unknown ) operating mechanism to produce the observed redshift , the sb dimming would go like @xmath2 . the test can not be considered as able to select one among different models since there are no known alternatives to the standard cosmology based on a non - expanding metric . here , as stated by hubble & tolman , the non - expanding case is taken to gauge the standard model . in practice , when the fundamental plane ( fp ) or kormendy relations are used , it is necessary to use the metric to transform the angles we measure into linear sizes , to compare data at different z values . thus , what we actually have is a mixture of the sb and angular size tests . this is discussed in the next section , where the test is formulated in terms of some relation ( assumed universal ) between structural parameters of early type galaxies . the evolutionary aspects and corrections are discussed in 3 . the results , based on good quality data recently published ( pahre et al . 1996 ; van dokkum and franx 1996 ; fasano et al . 1997 ; kelson et al 1997 ; bender et al 1997 ) are presented in 4 . a discussion of the results together with the conclusions can be found in 5 . previous claims ( pahre et al 1996 ) that the static case is discarded at the 5@xmath3 confidence level are shown to be incorrect . particular emphasis is made on the almost - lack of decidability of the test up to @xmath41 , except perhaps in the k - band . kjrgaard et al ( 1993 ) have discussed the possibility of using the structural relations found for early type galaxies to perform the sb test . they considered the sb intercept of the fundamental plane ( fp ) or kormendy relation as a mark for the tolman test . it s assumed that an intrinsic relation of the form @xmath5 does exist for early type galaxies , where @xmath6 is the average sb inside the metric radius r@xmath7 . the coefficients _ a _ , _ b _ and _ c _ are given by the fit to a particular set of data , at z@xmath0 0 . at redshift z , the surface brightness will be @xmath8 where @xmath9 is the cosmic dimming ( the same for all galaxies at a given redshift ) , and @xmath10 and @xmath11 are the evolution and k - effect ( both different , in principle , from galaxy to galaxy ) corrections . to compare data at different redshift , the sb has to be referred to the same metric size , which implies the transformation of the observed angular sizes , @xmath12 , into the corresponding linear sizes . for the expanding case , the dimming is given by @xmath9 = 10 log(1+z ) . assuming that the evolution and k - corrections at a given z are the same ( in magnitudes ) for all the galaxies defining the relation , and that r@xmath7 and @xmath3 do not change with z , the structural relation will have the same form at any z , @xmath13 = a log@xmath12 + b log@xmath3 + @xmath14 , where @xmath15 with ( for @xmath16 ) , @xmath17 } { \omega_0 ^ 2(1+z)^2}\ ] ] for the einstein static case , for which there is neither evolution correction ( see later ) nor size correction with z , @xmath18 with ( see moles 1991 ) @xmath19 the @xmath20 term has the same form for any scaling relation involving luminosities and sizes . besides the dimming , cosmic evolution ( if any ) , and k - correction , @xmath20 includes another term that depends on the metric ( i.e. , on the assumed cosmological model ) that is being tested . that term can be neglected when only expanding models are compared among themselves , but has a sizeable influence when the static case is also considered . at @xmath21 , the difference between the static and expanding predictions is reduced by about 0.4 magnitudes ( for @xmath22 0.2 ) with respect to the case of a pure sb test . as we will see in section 4 , the conclusions by pahre et al ( 1996 ) are affected by this problem . any test based on the comparison of ( assumed ) similar objects at different distances has to deal with all evolutionary effects . regarding the static case , the stationarity hypothesis implies that galaxies would be formed at any moment . the same range of evolutionary stages should be observed in any ( large enough ) volume , and at any redshift . this is a statement of statistical nature since it is the population itself that remains the same at any redshift , not each individual system . no evolutionary corrections have to be applied , and any structural relation that could be established at z = 0 should be observed at any redshift : in the static framework , the existence of the fp or the kormendy relation at every z is a test in itself . for the expanding model , the observation of the same structural relation at different z values implies that the evolution would follow well defined patterns . unfortunately the processes of galaxy formation and evolution are still too poorly constrained to explore the consequences of the fact that the fp or kormendy relations are well satisfied at different redshifts . given that situation , we will consider here that the evolution only affects the luminosity and colors of the galaxies . past claims that elliptical galaxies may have formed through a continuous merging process lasting till very low red - shift ( e.g. broadhurst et al . 1992 ) are now seriously challenged by detailed observations of elliptical galaxies in clusters at low - to - moderate redshift . the tight correlations between metallicity and velocity dispersion or color and magnitude ( e.g .. bower et al . 1992 ; ellis et al . 1997 ) strongly favor the simpler models of _ passive luminosity evolution_. elliptical galaxies would seem to have formed at red - shift @xmath23 , with the bulk of star formation taking place in relatively short ( @xmath24 gyr ) periods of time . since then , their luminosities would have passively evolved . though the existence of merging , expected in hierarchical models for structure formation , is not completely ruled out by the observations , this would have to happen at high ( @xmath25 ) redshift . therefore we choose to consider only passive evolution , and adopted the models by bruzual & charlot ( 1993 ; new version of 1995 ) , with @xmath26 = 4 , to compute the k- and evolutionary corrections . we notice that those models are compatible with other results and interpretations , as they are able to provide reasonable fits to the counts , as well as to the color and redshift distribution of galaxies , to the faintest levels ( metcalfe et al . 1996 ; pozzetti at al 1996 ; campos 1997 ) . fortunately , the details of the star formation processes at that remote epoch are expected to be of little importance for the z range at which the test can be presently performed . we have explored a decreasing star formation rate ( sfr ) with e - folding times of 0.3 and 1 gyr , in combination with two initial mass functions ( imf ) . the results do confirm that they have no sizeable influence on the predictions , and the k- and evolutionary corrections are similar up to z@xmath01 , except in the b - band , where the corrections become strongly model dependent from z = 0.6 on . the kind of evolution we consider here imposes that the coefficients of the scaling relation ( other than the zero point ) should be the same at any redshift . regarding the fp relation , we have adopted the values used by van dokkum & franx ( 1996 ) for coma . the @xmath20 values predicted by the different models , for the redshifted @xmath27 band , are shown in figure 1 . ( we use h@xmath28 = 50 and @xmath29 = 0.2 ) . the expansion prediction incorporates the corrections for evolution after bruzual and charlot ( 1993 ) models . it is clear that the predictions of the static and expanding models are very close till z@xmath01 , the difference in the cosmic dimming being almost completely compensated by the metric term and the evolutionary corrections in the expanding case . the same is true for the other visible photometric bands ( see below ) . for the b - band , the differences between the different evolutionary models become even larger than those between the static and expanding cases for @xmath30 0.6 . figure 1 shows that data by van dokkum and franx ( 1996 ) , and kelson et al ( 1997 ) does fit very well the predictions , as does the data by bender et al ( 1997 ) for the redshifted b band ( not shown here ) . we have also considered the kormendy relation as a tool for the tolman test . the value adopted for the coefficient a is 2.77 for the visible bands , and of 2.5 for the k - band , to be consistent with pahre et al ( 1996 ) . as before , the value of a can have some influence on the predictions , the difference between the static and expansion cases being further reduced for higher a values . not surprisingly , the results are similar to those found for the fp , in the sense that the predictions of the expanding and static models for the visible photometric bands ( shown in figure 2a , and b ) are very similar in the considered redshift range . the b- and r - band data , taken from pahre et al ( 1995 , and 1996 ) and from fasano et al ( 1997 ) , are remarkably well fitted by the predictions . this is at variance with pahre et al , who found that their b - band intercept for a851 was out of the prediction in the expanding case by 0.3 magnitudes . the difference is probably due to the use of different models to evaluate the evolutionary corrections . we therefore conclude that the tolman test can not be used to distinguish between the expanding and non - expanding world models in the optical passbands . the situation looks however more promising in the k - band , where the expected ( model ) luminosity evolutionary corrections are smaller . now , the difference between the static and expanding predictions amounts to @xmath0 0.4 magnitudes at z = 0.4 , and it reaches 0.8 magnitudes at z = 1 . unfortunately , the k - band data is not fitted by any model . the discrepancy amounts to @xmath00.2 magnitudes for the expanding case , and to @xmath00.6 magnitudes for the static case . trying to understand the situation , we notice that the sb values at the 1 kpc intercept given for coma by pahre et al imply @xmath31 4.56 , which is significantly redder than expected . indeed , persson et al ( 1979 ) gave an average value @xmath32 = 4.22 for a sample of 52 galaxies in coma , a result that has been confirmed by more recent studies ( recillas - cruz et al 1991 ) , and agrees with the value predicted by bruzual & charlot models at the redshift of coma . if charged on the k - band data , this discrepancy could account for the misfit found for the k - band data and would shift the data points somewhere between the static and expansion predictions . more data are needed to resolve this issue . we have shown that the dependence on cosmology of the kormendy relation or the fp includes , besides the sb cosmic dimming and the evolutionary ( if any ) and k - corrections , a factor depending on the metric . this metric term , that has been generally overlooked , has a sizeable effect , when the static predictions are also considered . it explains the discrepancy between the results reported by pahre et al ( 1996 ) and ours , as they did not take it into account . in principle , the sb test could be formulated in such a way to avoid the presence of the metric term , by using the @xmath33-@xmath3 relation . however , looking at this relation for the clusters considered here , it is found that it is too loosely defined , and can not be used with any confidence for the tolman test . our main result is that as far as scaling relations are used , the tolman test can not be conclusive in any practicable range of redshift for visible photometric bands data , due to the conspiracy of the metric term and the evolution corrections to reduce the difference between the static and expanding world models . in the v- and r- bands those differences are nearly zero till z@xmath01 this is also the case for b - band data till [email protected] . at larger redshifts the static case is enclosed in the spread of possible evolutionary corrections . the b- , v- , and r - band data are shown to fit very well the ( similar ) predictions of the different models . only in the k - band the predictions are separated , even if only by modest amounts till z@xmath01 , but on the empirical side , the situation is more ambiguous . the data are not fitted by any prediction , even if they are closer to the expanding case . we have noticed that there is a clear mismatch between the average @xmath32 color index obtained from the 1kpc intercepts of the kormendy relation , and the measured values for similar galaxies , or even the predictions at z@xmath00 by the evolutionary models . the situation needs to be clarified before a conclusion based on the tolman test could be attempted . the data confirm that the fp and kormendy relations are well defined at least till z@xmath34 . in the static case they are automatically satisfied since the stationarity hypothesis implies that there should be always and in every cluster a population of galaxies satisfying a given relation . in the expanding case it represents only a constraint on the allowed evolutionary paths of early type galaxies , which should be checked independently of the tolman test . it is clear that , if the change of the coefficients of the structural relation with z hinted by van dokkum and franx ( 1996 ) were confirmed , the static , non - evolving case would be ruled out . in that sense , the scaling relations constitute a test in themselves to prove the evolving character of the world . _ acknowledgments_. this research was partly supported by dgicyt ( spain ) grant pb93 - 0139 , and the danish natural science research council grant 9401635 . carlos barcel is acknowledged for helpful discussions on the nature of the tests . we acknowledge the referee for enlightening comments that did help to appreciably improve the manuscript .	the use of relations between structural parameters of early type galaxies to perform the tolman test for the expansion of the universe is reconsidered . scaling relations such as the fundamental plane or the kormendy relation , require the transformation from angular to metric sizes , to compare the relation at different z values . this transformation depends on the assumed world model : galaxies of a given angular size , at a given z , are larger ( in kpc ) in a non - expanding universe than in an expanding one . furthermore , the luminosities of galaxies are expected to evolve with z in an expanding model . these effects are shown to conspire to reduce the difference between the predicted surface brightness change with redshift in the expanding and non expanding cases . we have considered expanding models with passive luminosity evolution . we find that their predictions for the visible photometric bands are very similar to those of the static model till z@xmath01 , and therefore , the test can not distinguish between the two world models . recent good quality data on the kormendy relation and the fundamental plane at intermediate redshifts are consistent with the predictions from both models . in the k - band , where the expected ( model ) luminosity evolutionary corrections are smaller , the differences between the expanding and static models amount to @xmath00.4 ( 0.8 ) magnitudes at z = 0.4 ( 1 ) . it is shown that , due to that small difference between the predictions in the covered z - range , and to the paucity and uncertainties of the relevant sb photometry , the existing k - band data is not adequate to distinguish between the different world metrics , and can not be yet used to discard the static case . it is pointed out that the scaling relations could still be used to rule out the non - evolving case if it could be shown that the coefficients change with the redshift . _ subject headings : _ galaxies : fundamental parameters cosmology : theory cosmology : observations
more than 50 years ago hans bethe has remarked about the nature of nucleon - nucleon in his famous review `` what holds the nucleus together ? '' : _ in the past quarter century physicists have devoted a huge amount of experimentation and mental labor to this problem probably more man - hours than have been given to any other scientific question in the history of mankind _ " . and this claim is still valid today . despite of the enormous efforts of many labs , groups and personal researchers over the world during the last 50 years and despite of the great progress in experiments and fundamental theory provided by developments in qcd , we are still not having in our disposal a consistent and reliable dynamical theory for the @xmath0-interaction at intermediate and especially at short distances , which can form a solid basis for the explanation of numerous precise experimental data in meson - production in @xmath10 , @xmath11 , collisions , in few - body physics and in many two - proton photo - emission nuclear experiments , like @xmath12 , @xmath13 etc . , at rather high momentum transfer . we enumerate here some of basic points where the conventional description of the short - range @xmath0-interaction in terms of meson exchanges ( e.g. @xmath3 , @xmath4 , @xmath5 , @xmath14 , @xmath15 etc . ) fails or looks to be quite inconsistent . first of all , the value of the @xmath16 coupling constant ( which is responsible for the short - range @xmath0 repulsion and the spin - orbit splitting ) accepted in all obe - models turns out to be in the range @xmath17 which is much larger than the respective su(3 ) value ( @xmath18 ) while all other coupling constants used in obe - models are in a good agreement with the respective su(3 ) predictions . another evident inconsistency is that almost all meson - exchange nucleon - nucleon models use a too high value for the tensor - to - vector @xmath4-meson coupling to the nucleon , @xmath19 . at the same time , an analysis of the data on @xmath20 scattering yields a much more modest value , @xmath21 for this parameter ( see , e.g. @xcite ) . thus the short - ranged central and tensor nucleon - nucleon forces in boson - exchange models are enhanced by hand to reproduce the experimental @xmath0 phase shifts . another fundamental difficulty in the traditional obe - approach , also tightly related to the short - range nucleon - nucleon interaction , are the very large values of cut - off parameters @xmath22 , @xmath23 , in meson - baryon vertices , which are responsible for short - range limit of the @xmath0-interaction . in fact , the values of @xmath24 , @xmath25 etc . accepted in traditional obe - models ( @xmath26 gev ) are at least three times larger than those derived from any dedicated experiments ( e.g. from @xmath20 scattering ) , or from all fundamental theories ( @xmath27 gev ) . moreover , the values of @xmath22 or @xmath25 necessary to fit the charged pion production in @xmath10-collision is strongly different from those values needed to fit the _ elastic _ @xmath10-scattering . and those both values are different from respective values which are adopted in @xmath28-force operator to fit the @xmath28 binding energy within @xmath29- and @xmath28-force models driven by one - boson exchange @xcite . cross section as a function of the energy transfer @xmath5 at the value of momentum transfer @xmath30 mev/_c _ and @xmath31 mev/_c _ ( left ) and @xmath32 mev/_c _ ( right ) . the crosses give the experimental data while the curves represent calculations with a one - body current only ( solid ) , including mecs ( dashed ) and including mecs and static isobar currents ( dot - dashed).[fig1a],height=264 ] cross section as a function of the energy transfer @xmath5 at the value of momentum transfer @xmath30 mev/_c _ and @xmath31 mev/_c _ ( left ) and @xmath32 mev/_c _ ( right ) . the crosses give the experimental data while the curves represent calculations with a one - body current only ( solid ) , including mecs ( dashed ) and including mecs and static isobar currents ( dot - dashed).[fig1a],title="fig:",width=253 ] cross section as a function of the energy transfer @xmath5 at the value of momentum transfer @xmath30 mev/_c _ and @xmath31 mev/_c _ ( left ) and @xmath32 mev/_c _ ( right ) . the crosses give the experimental data while the curves represent calculations with a one - body current only ( solid ) , including mecs ( dashed ) and including mecs and static isobar currents ( dot - dashed).[fig1a],title="fig:",width=264 ] a direct consequence of the improper treatment for the short - range nuclear force and respective short - range two - nucleon correlations in nuclei is a quite serious disagreements between the predictions of most realistic modern few - body theories and data observed in a few key e .- . experiments like @xmath33 , @xmath34 etc . at rather high momentum and energy transfers @xcite . to illustrate the current situation in this field we present in figs . [ fig1 ] and [ fig1a ] ( taken from ref . @xcite ) the comparison between the most elaborated modern calculations for @xmath28-system ( both in bound state and continuum ) which include also the respective mec - contributions @xcite and the results of recent experiments at nikhef on @xmath35 reaction . it is clearly seen from the comparison between theoretical and experimental results that the underlying mechanism of the reaction should be quite different as compared to that assumed in the existing theoretical approach ( because the energy and momentum dependence derived from the theory have opposite trends than the experiments showing see fig . [ fig1a ] ) . a similar disagreement has also been observed in electromagnetic knock - out of a @xmath36-pair from @xmath37 @xcite in the mainz experiments . so one can expect a quite different mechanism for such reactions at rather high momentum transfers as compared to that underlying the obe - picture of @xmath29 and @xmath28 interactions . the key problem in a consistent theoretical description for such experiments is that the large incident momentum of a high - energy projectile has to be shared among several nucleons in nuclei , which requires a very strong short - range correlation of nucleons @xcite . but the conventional picture of the correlations induced by the @xmath6-channel meson exchanges can transfer the high momentum only with high values of cut - off parameters @xmath38 , or , alternatively , by incorporation of the numerous diagrams with multiple rescatterings in intermediate state @xcite , which are out of the control . in the last decade , numerous attempts have been made to describe the intermediate and short - range @xmath0 interaction within the framework of a cluster quark model employing the resonating group method ( rgm ) approach @xcite . however all these attempts were only partially successful and unable to extend the energy range much larger than 300400 mev . another serious problem with these qcd - inspired approaches is their inability to describe consistently the inelastic processes , e.g. the one- or two - pion productions . they focused only on the elastic @xmath0 scattering at rather low energies . thus we have now `` on the market '' two extreme types of models for the short and intermediate range @xmath0 interaction : either the traditional obe - type models which do not include any quarks , or the quark - motivated models which include the mesons only as the carriers of @xmath39 interaction @xcite , i.e. with no explicit mesonic degrees of freedom . thus , with such a limited basis the quark models immediately meet serious problems in the description of meson production in @xmath0 collisions ( which do require the incorporation of mesonic degrees of freedom in _ an explicit form _ ) . after the well - known work of weinberg @xcite the effective field theory ( eft ) approach became very popular and has been applied to the @xmath0 interaction and related phenomena in two forms : ( i ) in the `` pionless '' form which generalises somehow ( in field - theory language ) the effective range expansion @xcite and ( ii ) in the form of chiral perturbation theory ( chpt ) . in both directions great progress has been made and full calculations up to high orders are now feasible . within the framework of chpt the calculations have been performed @xcite ( see also @xcite ) for @xmath0 elastic scattering with energies up to 290 mev ( in lab . system ) and for deuteron properties . the authors @xcite used the chpt expansion up to fourth order ( i.e. @xmath40 ) and have derived the respective @xmath0-potential which leads to the same quality of fit for the @xmath0 phase shifts ( with @xmath41 ) as for the best modern phenomenological @xmath0-potentials . for this fit some 29 free parameters has been employed , 26 of them being related to parametrization of contact terms ( in @xmath42 ) . these contact terms , however , are out of the control by chiral theory and thus the description of short - range interaction is still on a purely phenomenological level . ) taken from ref . @xcite , ( @xmath43 ) ref . @xcite and ( @xmath44 ) ref . [ fig2],height=377 ] on the other hand , recent very powerful efforts of the juelich group to describe quantitatively the very abundant and precise new data on the neutral pion production cross sections and analyzing powers in @xmath45-collisions obtained in iucf @xcite have revealed a rather strong disagreement between well developed theory based on the meson - exchange approach and the polarization observables ( see fig . [ fig2 ] ) @xcite . this means that the traditional models still fail to describe the @xmath9-production in @xmath10-collisions ( especially for the spin observables ) . there are also quite serious problems in the description of _ two - pion _ production in @xmath10-collisions found in recent experiments at uppsala @xcite . here the main focus is on the _ simultaneous description _ of @xmath46 , @xmath47 and @xmath48 pair production . while the existing theory is able to predict the data in one channel , e.g. in the @xmath46 channel , it fails to give a reasonable description in the other two channels ( see fig . [ fig3 ] ) . so , the existing theory seems not to include some important contribution(s ) in the two - pion production sector . and @xmath49 ( quoted from ref . @xcite ) shown together with the previously existing data in the intermediate energy region @xcite . the curves represent theoretical predictions @xcite . [ fig3],title="fig:",width=245,height=226 ] and @xmath49 ( quoted from ref . @xcite ) shown together with the previously existing data in the intermediate energy region @xcite . the curves represent theoretical predictions @xcite . [ fig3],title="fig:",width=245,height=226 ] all these problems are tightly interrelated to the short- and intermediate - range @xmath0 interaction . so , the understanding of this part of @xmath0-interaction should be essentially improved . it is worth to add here that some interesting new proposal dedicated to the experimental and theoretical study for short - range @xmath10 correlations in @xmath37 and other light nuclei have been elaborated very recently by a team in the us @xcite . but in their project the authors avoided any reference to microscopic or field - theory approaches and employed only some phenomenological estimations based on @xmath10 scattering in the high - energy region . on the other hand , the short - range @xmath0 correlations are very crucial for our understanding the nuclear properties at moderate and high momentum transfer and more deep insight into phase transitions in nuclear matter . it should be stressed here while the existing theory can treat the averaged nuclear properties like single - particle momentum distribution on the quantitative level it meets serious problems when explaining the two - body short - range correlations even for the lightest nuclei like @xmath37 , where we can treat now both bound states and also the three - nucleon continuum fully consistently ( i.e. with complete account of fsi and rescattering terms ) . thus all the above facts ( and also other arguments omitted here ) and experimental findings have motivated us to develop some alternative approach for the intermediate- and short - range nucleon - nucleon force and alternative mechanisms for single- and two - meson production and also for short - range correlations in nuclei . the present study is devoted to developing an alternative and new way for the treatment of the intermediate- and short - range correlations @xcite . this novel approach starts with an important observation that the @xmath6-channel nucleon - nucleon interaction induced by two - pion exchange with full account of @xmath8 correlation in @xmath50 state ( i.e. in the @xmath3 meson channel ) , being treated consistently , leads inevitably to strong short - range _ repulsion _ and very weak intermediate - range attraction @xcite instead of the strong attraction induced by the conventional @xmath3 meson exchange in obe - models where the @xmath3 is treated as a stable particle @xcite . this amazing result has been obtained recently by two independent methods @xcite and thus looks to be established rather reliably . moreover , in the latter work @xcite it was found that even the @xmath0 spin - orbit force induced by scalar @xmath8-exchange is almost totally _ repulsive _ , i.e. it is in sharp contrast to that which is required by the fit to @xmath0 phase shifts . so , we need some other general source for intermediate - range strong attraction and the short - range repulsion for the central @xmath0 force and also an another source for the strong attractive spin - orbital @xmath0 interaction . the new force model proposed recently by two groups jointly : the moscow state university group and the tbingen university group @xcite , treats the intermediate- and short - range @xmath0-interaction as that proceeding through an intermediate dressed six - quark bag ( i.e. via @xmath1-channel mechanism ) : @xmath51 where the dressing by the meson fields ( mainly through the @xmath3-field ) surrounding the dense six - quark core strongly stabilizes the @xmath52-bag and shifts its mass downward to the @xmath0 threshold . the physical pattern underlying this new @xmath1-channel mechanism can be shortly outlined as follows @xcite . due to a strong attraction of the @xmath3-meson to the quarks in the @xmath7\rangle$ ] configuration and spherical symmetry of the @xmath53 six - quark bag , the @xmath3-field will compress somehow the bag . this bag compression may induce very short - range scalar diquark correlations in the bag and enhances further the quark - meson interaction . all these attractive effects will compensate completely the energy losses originated from @xmath3-meson generation and rather high kinetic energy of quarks in the bag . so the net result of this interplay has been estimated to be strongly attractive . then we employed the quark - meson microscopic model @xcite to estimate these effects and to construct the respective new model for the @xmath0 interaction at short ranges . our extensive tests have demonstrated @xcite that this model works very well and makes it possible to explain many long - standing puzzles in the field , e.g. it uses the ope terms with proper cut - off parameters @xmath54 gev . the new model predicts also the appearance of a strong three - nucleon scalar force induced by @xmath3-exchange between the dressed bag and the third nucleon . to study the effects of the new @xmath28 scalar force when calculating the properties of @xmath55 and @xmath37 nuclei , we have developed @xcite some multi - component formalism which made it possible to include both nucleonic and dibaryon components into @xmath28 wave functions . it has been found @xcite the latter components in @xmath28 wave function has a quite noticeable weight ( @xmath56 or even higher ) and these contribute very essentially to the total @xmath28 binding energy . thus the interplay between two - body and three - body force contributions peculiar to the conventional force models is changed drastically in the new approach . however , despite of this , we were still able to describe fully quantitatively , with no adjustable parameters , the basic static properties of @xmath55 and @xmath37 nuclei , especially the coulomb displacement energy @xmath57 which was a long - standing puzzle in the field . thus we can summarize here that albeit the new force model predicts some novel features of nuclear force it leads , at least , for static observables , largely to the same @xmath28 observables as for other modern approaches like cd - bonn @xmath58 uix or tm forces . however many other effects in strong or electroweak interaction phenomena will be described in fully different manner with the new force model . in previous versions of the model @xcite , we extensively employed the semi - relativistic quark - meson microscopic model and @xmath3-meson dressing and have fitted the lower partial @xmath0 phase shifts with the model up to 1 gev . however , in order to work reliably in the range 13 gev we need a consistent fully relativistic approach . another important application of the new force model was a treatment of short - range @xmath59 interaction within the same dressed bag model @xcite . this part of @xmath59 interaction is at the moment still poorly known , however it can be crucially important in the description of numerous processes in two- and few - nucleon systems going through @xmath60-isobar current and accompanied by a large momentum transfer , like one- or two - pion photoproduction @xmath61 , @xmath62 etc . in our new field theoretical approach the @xmath59 and @xmath63 channels are considered on the same footing as @xmath0 channel by calculating the respective @xmath59 and @xmath63 loops of the dressed dibaryon @xcite . on the other hand , the formalism developed makes it possible , in general , to treat the elastic and inelastic channels in a unified way ( see the subsequent section of the article ) . the only possible way that meets the above general requirements which the present authors can presently imagine is a fully covariant field - theoretical approach . effective field - theory approaches became very popular in this field in last decade . such an effective field approach is ideally suited for our purposes because it includes in an absolutely natural manner the relativistic formalism and the processes ( in intermediate and final states ) with generation of other particles ( mesons , isobars etc . ) . it can deal also quite naturally with meson and baryon loops in the intermediate states . thus we propose in the current work to develop a unified effective field theoretical approach to treat intermediate- and short - range @xmath0-interaction at low and intermediate energies of 13 gev jointly with description of meson production processes at @xmath10 and @xmath36 collisions . the clear distinction of the present force model from the existing ones in this aspect is the appearance of new ( dressed ) dibaryon components in all nuclei with quite significant weight @xmath64 . this component can absorb , contrary to the conventional single - nucleon picture , a large momentum and energy . so , many processes associated , e.g. with high momentum transfer , can proceed easily through this new dibaryon component , the latter being excited by external probe can produce e.g. a few mesons , or an isobar etc . and these production processes through the intermediate dressed dibaryon should have much a higher probability than the corresponding processes by absorption on one nucleon or on a meson - exchange current . our first calculations made within the framework of this approach to treat some electromagnetic processes in @xmath29 and @xmath28 systems have lead to quite promising results @xcite . the structure of the paper is as follows . in section 2 , we explain the new physical model for intermediate and short - range @xmath0 interaction . in section 3 , an effective field - theoretical approach is developed , which describes in the field - theoretical language the above physical model with @xmath1-channel dibaryon generation . section 4 is devoted to derivation of a relativistic @xmath0-potential using the formalism developed above . the semi - phenomenological nonrelativistic model proposed by our joint group previously is compared with the present treatment . in section 5 , we dwell briefly on a generalization of our approach to other hadronic processes like single- and two - meson production in @xmath10-collisions . in conclusion , we summarize the content of the work . here we will outline a multiquark model for the intermediate dressed dibaryon in @xmath1-channel , whose formation is assumed to be responsible in our approach for intermediate- and short - range @xmath0 interaction . we start with an observation done a decade ago @xcite that from two possible space symmetries , viz . @xmath65_x\otimes[3]_x=[6]_x\oplus[42]_x$ ] , the @xmath7_x\rangle$]-component has comparable projections onto three possible two - cluster components , viz . @xmath66 and @xmath67 , while the second possible component @xmath68_x\rangle$ ] , which corresponds to the @xmath69 excited six - quark configuration ( more exactly a coherent sum of many such components ) , has a large projection onto the @xmath70 channel only . thus , the fully symmetric part of the total six - quark wave function @xmath7_x\rangle$ ] can be identified with the bag - like structure while the mixed symmetry part @xmath68_x\rangle$ ] should be associated with the cluster - like @xmath70 component . hence , from the six - quark symmetry point of view the @xmath69-excited six - quark part corresponds to the proper initial @xmath0 channel while the ground @xmath7_x\rangle$ ] part includes mainly the intermediate bag - like components . therefore the whole @xmath1-channel process going through the intermediate dressed dibaryon can be represented by the following chain of steps : 1 . fusion of two three - quark nucleon clusters into @xmath69 excited six - quark system with the space symmetry @xmath68_x\rangle$ ] . emission of two @xmath71-wave pions from the excited mixed symmetry bag with subsequent formation of the scalar isoscalar @xmath3-meson due to very strong correlation of these two @xmath71-wave pions : @xmath68_x\rangle\rightarrow\|\mathrm{s}^6\rangle+(\pi\pi)_{j = i=0}$ ] . generation of the spherical dense six - quark bag @xmath72 in which the spherical core is surrounded by strong scalar @xmath3-field that compresses the quark bag due to strong attraction between @xmath3-field and quarks . in this new phase , one has two very important features characterizing the six - quark dynamics of the bag : * partial restoration of the chiral symmetry , that leads to strong reduction in both @xmath3-meson and constituent quark masses ; * essential enhancement of diquark correlations in the bag due to the bag compression fm @xcite . ] . this enhancement should lead to additional energy gain in the process . these two strong effects lead to a strong effective attraction in @xmath0-channel at intermediate range . and this two - stage complicated mechanism replaces in our approach the simple @xmath6-channel @xmath3-exchange in conventional obe - models ( which is invalid as has been proved by recent studies @xcite see the introduction ) and is the origin for intermediate range @xmath0-attraction . however , this dressed bag state , being singlet in color space , has still a large overlap with the @xmath0-channel , so that this state can decay fast back into the initial @xmath0-channel . hence , this intermediate dibaryon has a large width ( @xmath73 mev ) and thus can not be visualized as a resonance - like bump in @xmath0 phase shifts or cross sections . our previous papers @xcite have demonstrated that this two - stage mechanism is quite able to explain with reasonable @xmath74 and @xmath75 coupling constants @xmath0 phase shifts in low partial waves up to 1 gev . so that , quite similar to the nucleon dressing by a pion field , the dibaryon is dressed mainly by the scalar - isoscalar @xmath3-field and the physical dibaryon should exist partially in the bare state and in the dressed state . resorting to a diquark language for description of the @xmath76 ( the dressed bag state ) transition , one can assume that two unpaired quarks from both nucleons form a diquark pair ( of scalar or axial - vector type ) and thus , the intermediate stage in the transition is a state composed from three diquarks . then such a three - diquark state gets dressed further with the @xmath3-field that squeezes the bag . as a consequence , it can become favourable for diquarks to condense @xcite at large energies . at the same time , the chiral symmetry gets partially restored , i.e. we are faced to a phase transition from the usual chiral - broken phase to the so - called color superconducting phase . in the present study , we are , of course , far from the regime of color superconductivity but the quark - quark correlations can still play a significant role already in low energy physics . in some way diquarks can serve as useful constituents and building blocks in investigating the properties of multi - quark systems @xcite . anyway , the generation of the @xmath3-field in the symmetric @xmath52-bag leads to a large energy gain and the physical mass of the dressed bag shifts strongly downwards . effectively it corresponds to an appearance of a strong attractive force in the @xmath0-channel . another important issue in the suggested mechanism for @xmath76 transition is a domination of the excited mixed symmetry @xmath68 , lsi\rangle$ ] in the initial stage of the transition , so that the resulting @xmath0-wave function has a nodal structure . that the @xmath69 excited mixed symmetry state dominates in the @xmath0-wave function over a fully symmetric non - excited @xmath7\rangle$ ] configuration can be argued from various points of view . one argument is related to the specific spin - flavor nature of @xmath39 interaction in the goldstone - boson - exchange model @xcite . however , in this model for the @xmath39 force , both configurations lie rather high in energy , so that the effective @xmath0-interaction in both symmetries corresponds to repulsion . the second argument is that the @xmath3-dressing of the mixed symmetry bag : @xmath68\rangle\rightarrow\|\mathrm{s}^6[6]+\sigma\rangle$ ] leads to much stronger energy shift of the bare initial @xmath52-configuration than the respective shift of the symmetric state : @xmath7\rangle\rightarrow\|\mathrm{s}^6[6]+\sigma\rangle$ ] , because the latter process is in close analogy with the lamb shift ( due to interaction of quarks in the bag with vacuum fluctuations of the scalar field around the bag ) while in the first process the @xmath3-field is generated via a real deexcitation of two @xmath77-shell quarks and their transition to the @xmath78-orbit . the next section is devoted to the more formal and strict formulation of this qualitative pattern in terms of a covariant effective field theory . we are aiming to develop a consistent covariant field - theoretical description for the dibaryon model of nuclear forces , which allows us to go up to the intermediate energy region 13 gev inaccessible to chpt . this approach leads to the important modification of the force model : there arises a consistent scheme for the dressing of @xmath52-bag in terms of effective field theory . here the additional account of mesonic degrees of freedom in the dbs leads to an addition of corresponding loops in the complete polarization operator of the dibaryon , instead of a total replacement of the vertices in the transition @xmath79 ( @xmath80 ) , ( as was in our previous version of the model @xcite and was described briefly in the preceding section ) , see fig . [ fig4 ] . -channel @xmath0 interaction through the intermediate dressed dibaryon @xmath81 ( the peripheral one - pion exchanges are also shown ) . the graphs in the second row illustrate some possible diagrams for the dibaryon dressing.[fig4],width=491 ] it is important to stress that such a field theoretical picture allows to carry out a natural matching between the field theory approach at low - energies ( where chpt is fully applicable ) and the high energy region . in both regions one can now formulate a field theoretical language . in such an approach , the peripheral , i.e. the `` external '' part of @xmath0-interaction is described by the conventional @xmath2- and @xmath8-exchanges in the @xmath6-channel while the intermediate and short range interaction is described via generation of an intermediate dressed dibaryon which can be represented as a color quantum string able to vibrate and rotate . in the present model the intermediate dibaryon is produced by the color exchange between two quarks belonging to different nucleons with subsequent formation of a double string ( possibly with diquark clusterization due to a strong @xmath3-field generated on later stages of the process ) . the schematic evolution for this process is depicted in fig . [ fig5 ] . to be more specific , we assume when the two quarks belonging to different @xmath82-clusters interact and exchange a color , this can lead to the formation of either a nonconfined six - quark state ( i.e. @xmath0 , @xmath59 , etc . ) or a confined @xmath52-state locking by a hexagon - type string @xcite . the first non - resonance process is , in fact , analogous in action to conventional meson exchange , while the second leads to formation of a hidden color @xmath83-state . in diquark language , the first mechanism can be represented by the diagram with two interacting quarks forming a color representation @xmath84 of su(3 ) group , because the one - gluon exchange or instanton induced interaction between quarks in this channel is repulsive . but if two interacting quarks form the system in a @xmath85 color representation , the attractive correlations may lead to patching the quarks and as a consequence to the diquark formation . in this latter case , the dibaryon can be modeled as a system of three diquarks connected by y - type string . means that there are possible diquark correlations in dibaryon . [ fig5],width=491 ] because the value of quark radius of a nucleon is around 0.6 fm and the interaction between quarks can proceed also at the peripheral regions of both nucleons , the size of the intermediate dibaryon can be around 1 fm , so that the dibaryon can not be considered as a point - like object . thus , the transition amplitude @xmath86 must be described through the non - local lagrangian density : @xmath87 \|2,1\rangle-\mathrm{i}\langle 4,3\|\mathrm{i}\|2,1\rangle.\ ] ] here t means time - ordering while the bra and ket @xmath88 and @xmath89 relates to the initial and final nucleons with 4-momenta @xmath90 and @xmath91 respectively . the second term in ( [ eq0 ] ) , where @xmath92 is the unit matrix , corresponds to the propagation of noninteracting nucleons and thus must be subtracted . the nonlocal lagrangian density in eq . ( [ eq0 ] ) corresponding to transition from the two - nucleon to a bare dibaryon state , and the subsequent dibaryon dressing can be written in the form : @xmath93 where the lagrangian @xmath94is odd in dibaryon field operators @xmath95 and @xmath96 , and the lagrangian @xmath97 describing the interaction of @xmath52-bag with its meson cloud , is even in @xmath95 s . the function @xmath98 describes the dibaryon with spin @xmath99 , while @xmath100 corresponds to the @xmath101 dibaryon ( in both cases the isospin is assumed to have the value @xmath102 , for the case of isospin @xmath103 we should make a replacement @xmath104 in ( [ eqdd ] ) ) . the bispinor @xmath105 has the same transformation properties as the dirac - conjugated nucleon - field operator @xmath106 in the lorentz and isospin groups , @xmath107 is the charge - conjugation operator , @xmath108 is a pauli matrix , and @xmath109 stands for a transposed matrix . we use the notation and definitions borrowed from ref . @xcite . -state ( or dibaryon ) bound by double string.[fig6],width=302 ] the interaction @xmath110 related to the transition vertex @xmath111 can be illustrated by the graph shown in fig . this interaction can be found , in general , from the quark - microscopic model , e.g. assuming a diquark substructure of the dibaryon . but , since in this work , we are interesting in developing a general effective field - theory approach to the dynamics of the intermediate dressed dibaryon we will ignore any details of quark substructure of the dibaryon . so that , the potential @xmath112 is taken from the overlap of the inner nucleonic wave functions ( we postpone more elaborated calculation of this transition to the near future ) : @xmath113 where we choose for the @xmath114 the covariant harmonic oscillator wavefunction : @xmath115},\ ] ] @xmath116 is an effective coupling parameter and @xmath117 denotes the nucleon mass . in non - relativistic limit @xmath118},\ ] ] so we get a covariant form for @xmath119 : @xmath120},\ ] ] where @xmath121 is the c.m . 4-momentum of two nucleons . from eqs . ( [ eq0 ] ) and ( [ eql3 ] ) , it is evident that the transition amplitude must be even in dibaryon field operator and at least of the second order in interaction with a nucleon current . as a result , one has the following expression for @xmath122 ( in @xmath102 channel ) : @xmath123 where @xmath124 is an exact dibaryon propagator that takes into account the dressing of the dibaryon with meson fields . the vertices @xmath125 are equal : @xmath126 the dressed dibaryon propagator is found by solving the respective dyson equation @xmath127 where @xmath128\|0\rangle\ ] ] is a bare dibaryon propagator and @xmath129 stands for a polarization operator of the dibaryon and @xmath63 loop diagrams . thus , all other interactions in the multi - quark system , like one - gluon exchange , instanton - induced , confinement etc . should be included into the bare dibaryon propagator . ] . here , the time - ordering is performed with respect to the coordinates @xmath130 and @xmath131 of the dibaryon center of mass . it is worth to note that the t - ordering on the coordinates ( of the two color clusters ) @xmath132 and @xmath133 in amplitude @xmath122 corresponds only to the positive and negative signs for the time - like components of the relative coordinates @xmath134 and @xmath135 in the integration over @xmath136 ( @xmath137 ) , while the mixed ordering , for example , on the coordinates @xmath138 and @xmath139 results in two types of exchange interaction displayed in fig . the first graph describes the ( resonance - like ) @xmath1-channel exchange by the dibaryon whereas the second graph describes non - resonance @xmath6- and @xmath140-channel exchanges by an exotic dibaryon ( which corresponds to a heavy @xmath141-meson ) . this exotic contribution is omitted in eq . ( [ eq1 ] ) and will be further neglected . ) . the first diagram corresponds to @xmath1-channel dibaryon exchange , while the second diagram describes the @xmath140- or @xmath6-channel exchange of a heavy @xmath142 meson having an exotic @xmath143 structure.[fig7],width=491 ] we consider the dibaryon wavefunctions as a convolution of orbital and total spin parts : @xmath144 , where the orbital part @xmath145 is taken as a superposition of covariant harmonic oscillator ( cho ) eigenfunctions @xmath146 ( see below ) . although there is presently no direct connection between the cho - formalism and qcd , its application appears quite successful in describing the baryon spectra and the systematics of @xmath147 meson states @xcite ; in addition , it leads to hadron form factors whose behavior agrees well with experimental data . as for the microscopic basis of the cho - functions one should note , that these eigenfunctions are closely related to the solution of the dirac equation with a linear vector potential @xcite which corresponds to a string - like interaction ( between two colored clusters ) . the total spin wavefunctions @xmath148 satisfy to the standard klein gordon equation : @xmath149 while the @xmath146 is a tensor of rank @xmath150 ( @xmath150 is the number of excitation quanta of the string ) in the space @xmath151 and satisfies to the equation : @xmath152 with the mass operator squared is defined as @xmath153 here @xmath154 , @xmath155 and @xmath156 are kinematical masses of @xmath157 clusters on the ends of the string . @xmath158 is defined by the string potential energy and also by non - perturbative interactions of instanton type : @xmath159 while the perturbative term @xmath160 ( the effect of which can be reduced mainly to some deviation of the respective regge - trajectory from the straight line ) is omitted here . this leads to the following cho eigenfunctions : @xmath161,\label{eq9a}\\ & f_{n_e}(x;p)=\langle x\|a^\dag_{\nu_{n_e}}\ldots a^\dag_{\nu_1}\|g(p)\rangle , \\ & m_{n_e}^2=dc+({n_e}+1)\omega = m^2_0+{n_e}\omega,\end{aligned}\ ] ] where @xmath162 is the creation operator for oscillator quanta , @xmath163 squared oscillator radius and @xmath164 is the oscillator frequency . for the mass spectrum , we thus obtain a linear regge trajectory . the operators @xmath165 are defined in such a way that a physical state satisfies the condition , @xmath166 which rules out the appearance of an unphysical spectrum with respect to the time variable @xcite . for example , the function corresponding to a single - quantum excitation of the oscillator can be represented in the form @xmath167 this function is normalized by the condition @xmath168 in the case of a two - quantum excitation , two excited states are possible , @xmath169f_0(x;p),\\ f_{\mu\mu}(x;p)=\left(3 + 2\beta x^2 - 2\beta\frac{(px)^2}{p^2}\right)f_0(x;p)\nonumber,\end{gathered}\ ] ] which are normalized as follows : @xmath170 hereinafter , we shall restrict our consideration only to even - parity dibaryons , including the states with zero and two quanta only . moreover , in the current work we shall consider only those partial waves whose coupling to @xmath59-channel is absent or suppressed below 1 gev in the lab . system ( i.e. we shall limit ourselves to consider in this paper @xmath171 , @xmath172 and @xmath173 partial channels only ) . then for corresponding dibaryon wavefunctions @xmath174 ( where @xmath175 and @xmath176 are the total angular and orbital momenta of the dibaryon ) , one has : + for the @xmath171-partial wave : @xmath177 where @xmath178 , @xmath179 and the normalization is chosen for a single particle state in a unit volume ; + for the @xmath180-partial wave and @xmath181 partial waves will be considered below when the dressing procedure will be described . ] : @xmath182 + for the @xmath181-partial wave : @xmath183 where @xmath184 and @xmath185 ; + and for the @xmath173-partial wave : @xmath186 now we can expand the bare dibaryon green functions into the string eigenstates : @xmath187\|0\rangle.\ ] ] in this subsection we will show that , instead of a complicated integral dyson equation , we can get its matrix analog ( with a very low matrix dimension ) for the projection of all propagators and polarization operators onto the above cho basis . simple estimates give the energy value of the string excitation quantum @xmath188 mev so that it is sufficient to take into account only two quanta excitations of the dibaryon string when the incident energy in @xmath0-channel does not exceed 1 gev ( in lab . system ) . by restricting ourselves to zero and two - quanta excitations of the string , one can write the bare partial propagator for the @xmath171-channel in the form : @xmath189 then , by substituting this bare propagator into the dyson equation ( [ eq13 ] ) one can get the following representation for the dressed propagator : @xmath190 where @xmath191 in which @xmath192 are polarization operators projected onto the cho basis : @xmath193 then , in order to pursue the renormalization procedure , the dressed propagator @xmath194 must be diagonalized in the @xmath195 space . to do this , one can represent the propagator ( [ eq16 ] ) in matrix form : @xmath196 where we introduced the rotation matrix @xmath197 which diagonalizes the matrix propagator . the mixing angle @xmath198 , which mixes the different six - quark configurations @xmath199 and @xmath200 , can be found from the condition @xmath201 finally , we get for the @xmath202 : @xmath203 after the diagonalization of @xmath194 , the `` diagonal '' propagators take the `` canonical '' form : @xmath204 where @xmath205 are renormalized matrix elements of polarization operator @xmath206 and @xmath207 the physical dibaryon masses are found from the transcendental equations : @xmath208 we omit here rather lengthy expressions for projected polarization operators and eventual formulas for the dressed propagators . it is useful , however , to illustrate here the calculations of polarization operators in different channels by some basic graphs ( see fig . [ fig8 ] ) . these graphs include both @xmath2- and @xmath3-meson dressing of the intermediate dibaryon . , width=491 ] in this case , three six - quark configurations , viz . @xmath199 , @xmath200 and @xmath210 can mix with each other . while the mixing between the first and the second configurations is the same as for the previous case of @xmath171 partial wave , their mixing with the latter @xmath52-configuration is due to the tensor force . because our focus here is to describe the tensor mixing of @xmath180 and @xmath181 channels , induced by the intermediate dibaryon , we neglect now the mixing between the states with different string excitation energy , which is assumed to be small due to the quite large energy gap ( @xmath211 mev ) between them . thus , the dressed dibaryon propagator can be represented as a sum of two terms : @xmath212 where the first and the second terms correspond to @xmath213 and @xmath214 dibaryons respectively . owing to a possible mixing of @xmath200 and @xmath210 configurations , the propagator can be represented by a @xmath215 matrix . since we neglect here the mixing between @xmath213 and @xmath214 configurations , for the first term in eq . ( [ 3sd1 ] ) we have : @xmath216 taking into account the expression for the bare dibaryon propagator of @xmath213-bag @xmath217 one can write the exact propagator , namely the @xmath218 matrix element , in the form : @xmath219 making further a decomposition @xmath220 into two independent tensors of rank 2 : @xmath221 and @xmath185 that are projectors on the states with spin 0 and 1 , and solving the respective matrix dyson equation , one gets eventually : @xmath222 where in analogy with the @xmath171-channel , the renormalized polarization operator is @xmath223 and the physical mass of the dibaryon can be found from the condition @xmath224 the expressions for polarization operators projected on cho basis and spin state are defined in eqs . ( [ eqgr ] ) and ( [ eqgl ] ) as follows : @xmath225 since the functions @xmath226 are analytic , the pole singularity can appear only in that part of the dibaryon propagator which correspond to spin 1 propagation ( that is , the first term in eq . ( [ eqgr ] ) ) . because of the dressing of the dibaryon with a meson cloud , the pole position appears to be shifted from the real axis into the complex @xmath227 plane and it corresponds to a resonance state of the dressed six - quark bag . the contribution to the nucleon - nucleon potential from `` unphysical '' states of spin zero , which are developed for an off - shell vector particle and which are inherent in a field - theoretical description of particles having higher spins , vanishes in the amplitude ( [ eq1 ] ) due to nucleon current conservation . in any case , it is of order @xmath228 and therefore is very small . one can write the propagator of the @xmath229-bag state in the following matrix form : @xmath230 then again , by decomposing the functions @xmath231 and using condition ( [ cond ] ) , we arrive at the following form for the propagator , which has to be diagonalized : @xmath232 and @xmath233 the first term in eq . ( [ eq30 ] ) , corresponding to off - shell spin - zero propagation , vanishes exactly after substituting this expression into the amplitude ( [ eq1 ] ) . by solving the respective matrix dyson equation with the bare propagators @xmath234 we eventually arrive at the following expressions for the functions @xmath235 : @xmath236 where , as well as just above , after renormalization one has : @xmath237 ( the same equation is valid for @xmath238 with a replacement @xmath239 ) . here the diagonalized projected polarization operators are @xmath240 the mixing angle @xmath198 is defined by the same expression as in the case of the @xmath171-channel ( [ chi ] ) , where we should make a replacement @xmath241 , but with the different cho - projected polarization operators : @xmath242 the physical masses of dibaryons in the oscillator states @xmath243 and @xmath244 can be found as usually through the equality @xmath245 now , we shall briefly outline the final formulas for the dressed dibaryon propagator in @xmath173 partial channel , as the mostly all steps of the derivation are the same as for the previous cases . after substituting the bare @xmath246 dibaryon propagator into the matrix dyson equation , its solution can be found in the following form : @xmath247 here @xmath248 is a projection operator onto the momentum @xmath246 state . we omit in eq . ( [ 3d2 ] ) the terms corresponding to off - shell lower spin propagation of the dibaryon , since they vanish completely in the amplitude @xmath122 due to nucleon current conservation and condition ( [ cond ] ) as well . we would like to note here , that this is a general result , independent of the value of total angular momentum @xmath175 . so , our amplitude is free from the `` unphysical '' lower spin state contributions inherent in field - theoretical formulations for higher ( @xmath249 ) spin particles . for the renormalized projected polarization operator , we have @xmath250 where @xmath251 and the physical @xmath173-dibaryon mass is extracted from the condition : @xmath252 in principle , the dibaryon polarization operator @xmath253 ( or @xmath254 ) is determined in even partial waves mainly by the @xmath255 and @xmath256 loops , while its imaginary part is largely determined by the @xmath0 loops and these loops are responsible for dibaryon decay back to two real nucleons . hence , in our case , if we want to construct the @xmath0-potential , we must exclude coupling to an intermediate nucleon - nucleon state . this is because of the fact that in the scattering equation ( e.g. bethe salpeter or any quasipotential equation ) , such channels are automatically taken into account by virtue of the requirement that the scattering amplitude satisfies two - particle unitarity . therefore , the main contribution to the polarization operator comes from the dressing of the dibaryon with its meson cloud . this interaction induces a transition of the dibaryon ( featuring @xmath257 string - excitation quanta ) to the ground state @xmath213 at low energies . at the same time , the interaction with @xmath4 , @xmath5 and the other mesons adds a much smaller contribution because of their rather large mass and due to the fact that these mesons can propagate only in odd partial waves with respect to the dibaryon @xmath213 or in even partial waves with respect to the @xmath258-excited configuration @xmath259 , which has one excitation quantum @xmath260 . demanding the parity and total - isospin conservation , one can write the transitions between different @xmath261 states that are to be accounted for in polarization operators under consideration : @xmath262 here , we emphasize an important fact that follows directly from the above quantum numbers for the allowed transitions : the @xmath1-channel transitions forming the intermediate states of the dressed dibaryon are in a full correspondence with ordinary @xmath6-channel transitions . _ it is precisely for this reason that the @xmath1-channel mechanisms considered here make those contributions to various high - momentum - transfer processes that were missing in the traditional obe - models . _ we would also like to indicate that , as it is evident from ( [ tens ] ) , a @xmath263 dibaryon state may be a result of the dressing of the bare dibaryon @xmath264 with a pion cloud where the pion is in @xmath265- ( or other odd ) partial wave in respect to the six - quark core . since the color - dependent interaction between quarks ( that produces a hidden @xmath83 state ) of tensor type is rather weak , the decay of this state directly to the nucleon - nucleon channel should be strongly suppressed , but on the other side , the @xmath263 state has quite a large coupling to the @xmath59-channel . it is well known that just the behavior of the @xmath266 phase shift in @xmath0 scattering ( which corresponds to the @xmath263 dibaryon state ) is indicative of the presence of a strong @xmath59 correlation , which is sizable even at energies of about 400 mev in the laboratory frame . in contrast to the conventional meson - exchange models , where the formation of this state in the @xmath59 channel is due to @xmath6-channel pion exchange and requires hard pion form factors , our model proposes a totally new mechanism that is responsible for the formation of a @xmath59 state at short and intermediate distances and which is caused by the dressing of the dibaryon with the pion field ; that is , @xmath267 thus , the assumption of an intermediate dressed dibaryon gives a very natural explanation for strong coupling between @xmath0 and @xmath59 channels in some partial waves . we also note that , in the present model , the polarization operators ( and consequently the @xmath0-potential as well ) are complex - valued functions . their imaginary parts are related to open inelastic channels and are determined by the discontinuities of these quantities at the unitary cut in the complex energy plane . in the example considered here , we have two inelastic channels corresponding to real intermediate states , @xmath268 and @xmath269 , where the latter is associated with the mechanism of @xmath3-meson decay to two real pions . finally we illustrate here how to derive the relativistic @xmath0 potential from the formalism developed . as an example , we consider here the potentials in the lowest partial waves , namely the isotriplet @xmath171 and isosinglet @xmath180-@xmath181 channels . to this aim we use , in the amplitude @xmath122 ( [ eq1 ] ) , the relative 4-momenta of nucleons @xmath270 and @xmath271 , and employ of the relation between bispinors with positive and negative energies @xmath272 moreover , the transition amplitude taken in the form ( [ eq1 ] ) is more appropriate for the reaction @xmath273 in the crossed channel . thus to pass to the @xmath0 channel one should employ the fierz transformation : @xmath274[\overline{\phi}\gamma_j\psi]=\frac{1}{16}\sum_{k , l=1}^{16 } \mathrm{tr}(\gamma_i\gamma_k\gamma_j\gamma_l)[\overline{\psi}\gamma^k\psi][\overline{\phi}\gamma^l\phi],\ ] ] where @xmath275 and @xmath276 denotes the dirac bispinors and @xmath277 corresponds to one of the 16 dirac matrices @xmath278 forming a basis in the space of @xmath279 matrices . the @xmath0 potential in this channel can now be written in the following form ( separating out the @xmath280-function factor , arising from the integration in eq . ( [ eq1 ] ) on the coordinate of the dibaryon center of mass ) : @xmath281 [ \overline{v}({\bf q})\gamma_5u(-{\bf q})]f^{(00)}(q',q;p),\ ] ] where the factor 2 in the potential arises due to antisymmetrization over nucleon variables scattering and 2 for @xmath10 scattering , since in contrast to the @xmath6-channel interaction , where antisymmetrization occurs at the level of feynman diagrams , it is achieved here even at the level of vertices , with the result that the amplitude @xmath122 appears to be automatically antisymmetrized . in calculating the @xmath10 phase shifts , it must therefore be divided by a factor of 2 ( whereupon we obtain isotopic invariance ) with the net result being the same as for the @xmath36 potential . ] . @xmath282 plays the role of the invariant amplitude . we omit here rather lengthy expression for the @xmath282 in the case of mixing @xmath195 states , which is straightforward to write down employing the results of preceding section for the dibaryon propagator . for brevity , we present here the simplified formula for the case of the vanishing mixing , i.e. we put here @xmath283 : @xmath284 where @xmath285 substituting further the expressions for @xmath111 transition potential ( [ potent ] ) and dibaryon wave function into the formula for the dibaryon formfactor @xmath286 one gets : @xmath287 @xmath288\widetilde{\phi}_{(0\mathrm{s})}(q , p).\ ] ] then , using some properties of the dirac matrices after the fierz transformation one can write : @xmath289 [ \overline{v}({\bf q})\gamma_5u(-{\bf q})]=\\ = -\frac{(1-\vec{\sigma}_1\vec{\sigma}_2)}{2}w \left(1+\frac{{\bf q}'{}^2}{(e_{q'}+m_n)^2}\right)\left(1+\frac{{\bf q}^2}{(e_{q}+m_n)^2}\right)\!,\end{gathered}\ ] ] where the factor is @xmath290 , @xmath291 and the spin operator @xmath292 corresponds to the total spin of the @xmath0 system . further , if one employs the partial - wave expansion of the @xmath0-potential : @xmath293 with @xmath294 where @xmath295 represents a standard spin - angular part , one finally gets in the channel @xmath296 @xmath103 the potential in the form : @xmath297 according to the results for the dibaryon propagator in the mixed channel , the @xmath0 potential should be a @xmath215 matrix : @xmath298 [ \overline{v}({\bf q})\gamma^\nu u(-{\bf q})]f_{\mu\nu}(q',q;p),\]]where the elements of the matrix tensor @xmath299 are @xmath300\mathcal{p}^1_{\mu\nu},\tag{\theequation a}\\ & f^{(02)}_{\mu\nu}=\frac{3}{\sqrt{2}}\widetilde{\phi}_{(2\mathrm{s})}(q',p)\widetilde{\phi}_{(2\mathrm{d})\mu\nu}(q , p ) \left[\widetilde{\mathcal{g}}_2(p^2)-\widetilde{\mathcal{g}}_0(p^2)\right]\sin\chi\cos\chi,\tag{\theequation b}\\ & f^{(20)}_{\mu\nu}=\frac{3}{\sqrt{2}}\widetilde{\phi}_{(2\mathrm{d})\mu\nu}(q',p)\widetilde{\phi}_{(2\mathrm{s})}(q , p ) \left[\widetilde{\mathcal{g}}_2(p^2)-\widetilde{\mathcal{g}}_0(p^2)\right]\sin\chi\cos\chi,\tag{\theequation c}\\ & f^{(22)}_{\mu\nu}=\frac{9}{2}\widetilde{\phi}_{(2\mathrm{d})\mu\alpha}(q',p)\widetilde{\phi}_{(2\mathrm{d})\alpha\nu}(q , p ) \left(\widetilde{\mathcal{g}}_0(p^2)\sin^2\chi+\widetilde{\mathcal{g}}_2(p^2)\cos^2\chi\right),\tag{\theequation d}\end{aligned}\ ] ] where formfactor @xmath301 is defined by eq . ( [ formf ] ) and one has in explicit form : @xmath302 after making the respective fierz transformations , one arrives eventually at the following expressions for the matrix elements @xmath303 : @xmath304 @xmath305 where @xmath306 stands for the tensor operator and functions @xmath307 are defined as follows @xmath308\sin\chi\cos\chi,\tag{\theequation b}\\ & f^{(20)}=\frac{3}{\sqrt{2}}\widetilde{\phi}_{(2\mathrm{d})}(q',p)\widetilde{\phi}_{(2\mathrm{s})}(q , p ) \left[\widetilde{\mathcal{g}}_0(p^2)-\widetilde{\mathcal{g}}_2(p^2)\right]\sin\chi\cos\chi,\tag{\theequation c}\\ & f^{(22)}=\frac{9}{2}\widetilde{\phi}_{(2\mathrm{d})}(q',p)\widetilde{\phi}_{(2\mathrm{d})}(q , p ) \left(\widetilde{\mathcal{g}}_0(p^2)\sin^2\chi+\widetilde{\mathcal{g}}_2(p^2)\cos^2\chi\right),\tag{\theequation d}\end{aligned}\ ] ] and @xmath309 it should be noted here that superscripts in the potentials @xmath310 _ refer to the dibaryonic channels _ , i.e. they denote the angular quantum numbers of intermediate dibaryon . however , as is clear from eqs . ( [ eq53])([eq56 ] ) , due to the relativistic tensor mixing , the angular momentum of the two - nucleon system is not conserved . this means that the initial ( or final ) two - nucleon angular momentum ( which is marked in the phase shift ) may not coincide with the angular momentum of the dibaryon . therefore partial nucleon - nucleon potential should be a sum of partial potentials corresponding to four different dibaryonic channels ( [ eq53])([eq56 ] ) : @xmath311 similar , albeit quite lengthy purely algebraic formulas have been derived for relativistic @xmath0-potentials in all other channels . it has been shown @xcite that when doing the nonrelativistic reduction of the above relativistic potentials we get the formulas very similar to those obtained in our previous fully microscopic semi - relativistic approach @xcite . after the complete derivation of the covariant @xmath0-potential in various partial waves we should add this short - range interaction to the peripheral one - pion and two - pion exchange potentials derived previously in chpt ( but with the short - range contact terms being parameterized in a purely phenomenological manner ) or alternatively , we should replace the contact terms in the chpt - approach with the covariant short - range potential derived in our field - theoretical approach . the resulting full @xmath0-potential can be fitted first in the low - energy region , @xmath312 mev and by this way the main input parameters of the short - range part of interaction ( e.g. the coupling constants of the meson fields with the string ) can be established . this procedure is similar in some degree to fitting the parameters of contact terms in higher - order chpt . thus , by combining the low - energy and high - energy effective field theories one can reach a very consistent and fully dynamical description of elastic and inelastic @xmath0-interaction from zero energy up to the gev region . this way is quite analogous to the treatment of nucleon(or any other projectile)-nucleus scattering : at low energies and in lowest partial waves the @xmath1-channel ( or compound - state ) scattering is dominating while at higher partial waves the @xmath6-channel mechanisms should be prevailing . so , for a complete description of this interaction we should combine somehow these two aspects . we can now illustrate the applicability of the above eft - approach by treating the lowest partial @xmath0 phase shifts with a simplified model derived from the general formulas ( [ eq21])-([eq85 ] ) after the nonrelativistic reduction . for this purpose we will consider the description for the @xmath171 and @xmath180-@xmath181 partial channels in the energy range from zero up to 1 gev . the total potential takes the form : @xmath313 where @xmath314 and @xmath315 are the @xmath6-channel one- and two - pion exchange potentials which are responsible for the peripheral part of @xmath0-interaction . the ope - potential is taken here with soft cutoff constant @xmath316 mev @xcite which is dictated by all dynamical models of @xmath20-interaction . the @xmath315 is approximated here , for the sake of simplicity , by a simple local gaussian function @xmath317 the term @xmath318 in eq.([eq86 ] ) describes the intermediate- and short - range interaction ( @xmath319 fm ) induced by the dressed dibaryon formation in intermediate state . after the nonrelativistic reduction of the general expression ( [ eq21 ] ) , the @xmath318 part of interaction in the uncoupled @xmath171 channel can be rewritten in the form : @xmath320 where @xmath321 , @xmath322 are nonrelativistic h.o . functions of 0s and 2s form , @xmath323 is an arbitrary , but a large positive constant ( @xmath324 mev ) ) , according to our general approach , serves to represent the strong @xmath0 repulsion when the intermediate six - quark bag has @xmath325_x\rangle$ ] structure @xcite . ] . the energy - dependent strength parameter @xmath326 is calculated straightforwardly from the @xmath3-meson loop shown in fig . it was shown earlier , the above loop integral can be well approximated in the energy range 01 gev by a simple pade - approximant : @xmath327 with parameter values , @xmath328 and @xmath329 , presented in refs . @xcite , while the strength constant , @xmath330 , is proportional to the @xmath331 coupling constant which determines the coupling of the @xmath3-field and six - quark core . this constant is taken in this illustrative model as a free parameter . so we have in the model only four free parameters , two of which , @xmath332 and @xmath333 , are related to the two - pion exchange potential in the peripheral region , and the other two , @xmath330 and @xmath334 are related to the dressed dibaryon properties . two parameters of @xmath315 are fixed easily by fitting two effective range parameters which describe the @xmath171 phase shifts until 15 mev . so , we are left with only two basic parameters related to the properties of the @xmath52-bag to describe the @xmath171 phase shifts in a very large energy range 151000 mev . by varying these two parameters we arrive at the high quality fit displayed in the fig . [ fig9 ] ( upper left panel ) . for the @xmath180-@xmath181 coupled channels we can derive the model from the general formulas by a manner similar to the uncoupled @xmath171 case . now , the dibaryon part of interaction , @xmath318 , can be presented in the matrix form : potential model similar to that given in eq . ( [ eq21 ] ) ) for some lower @xmath0 phase shifts and respective phase shift analysis ( taken from said).[fig9 ] ] @xmath335 where the matrix operator @xmath336 takes the form : @xmath337 one can attain a quite good description of the triplet phase shift in the whole energy range 01 gev even when choosing the one common value of the oscillator radius for the @xmath1-wave , @xmath338 , and @xmath339-wave , @xmath340 , potential form factors in eq . ( [ eq90 ] ) . the parameter @xmath329 , that determines the energy variation of all the strength parameters @xmath341 is kept the same as in @xmath171-channel . so that in the @xmath180-@xmath181 coupled channel case , one has ( except of two parameters in @xmath315 ) only three adjustable strength parameters , viz . @xmath342 and @xmath343 which make it possible to fit with a good quality the phase shifts together with the respective tensor mixing parameter in a very large energy range 151000 mev ( see fig . [ fig9 ] , upper right and two lower panels)-wave oscillator parameter can be allowed to vary , the quality of the fit may become almost perfect . ] . thus , the illustrative model derived from eft - approach developed in the work makes it possible , using physically transparent interaction model with a minimal number of free parameters , to fit very reasonably the phase shifts in @xmath171 and @xmath180-@xmath181 channels in the large energy range and the deuteron structure as well . this description can be compared to the fit quality for the phase shifts given by a purely phenomenological separable interaction ( of conventional type ) of graz group @xcite . the latter model includes more than twenty five free parameters to fit the phase shifts in the same channels until 500 mev only . thus the description of the elastic @xmath0 scattering given by the present illustrative model derived from eft - formalism looks highly superior the description attained with the phenomenological separable models . it is worth to emphasize here that , at the value of the string excitation quantum @xmath188 mev , taking up to two - quanta excitations of the string will be sufficient for description of @xmath0-interaction from zero energy until 1 gev or even higher , while account of four - quanta excitations is sufficient for description of interaction in @xmath0-channel or the @xmath0 meson - production processes up to energies of @xmath344 gev . having developed the field - theoretical formalism for the description of the @xmath0 elastic scattering in the gev region one can apply this also to describe inelastic processes within the same framework , i.e. with the same input constants etc . and this possibility is one of main advantage of this unified approach . for the sake of brevity we will omit many details here and illustrate the description of inelastic processes by some graphs only . in the above field - theoretical formalism the inelastic processes are tightly interrelated to the elastic ones above the respective thresholds through the unitarity relation : @xmath345 which corresponds in our language to the respective cuts of the diagrams across the meson loops . so , the respective meson production amplitudes can be illustrated by the graphs shown in fig . [ fig10 ] . the left graph relates to @xmath8-production in @xmath10-collisions through the intermediate @xmath3-meson in the dressed bag , while the graph on the right relates to the near - threshold @xmath5-meson production to the intermediate @xmath346-isobar generation . it should be stressed here that the @xmath0 `` distorted waves '' in such meson production amplitudes are generated _ from the same diagrams _ ( with the same parameter values ) as these meson productions . so the approach can be made quite consistent and universal both for elastic and inelastic processes . -meson emission in a nucleon - nucleon collision . [ fig10],title="fig:",width=226 ] -meson emission in a nucleon - nucleon collision . [ fig10],title="fig:",width=226 ] another very important issue which can be answered on the basis of the developed approach is multi - meson production and generation of the so called cumulative pions in high - energy @xmath10 collisions . according to the relativistic picture presented here , the multi - meson production should correspond to @xmath347 excitation of the intermediate string ( @xmath348 ) with its subsequent deexcitation by multi - meson emission or by hadronic jets . by a similar mechanism one can get the emission of cumulative pions ( i.e. forbidden in a single - nucleon emission ) , e.g. in the process @xmath349 , where the energy of outgoing pion is larger then half energy of the incident deuterons . the appearance of the dibaryon component in a nucleus can modify noticeably the nucleon momentum distribution probed at large momentum transfer . e.g. one gets some new contributions to electromagnetically induced two - nucleon knock - out processes in which the emerging two nucleons ( from the excited dibaryon decay ) escape from the dibaryon state in `` back - to - back '' kinematics . in fact , just this kinematics was observed for the two - nucleon ejection in many experiments of this type . it is evident that such contributions for the @xmath350 and @xmath12 processes at high ( virtual ) momenta or energies of @xmath351-quantum must contribute to the reaction cross sections . in such two - nucleon processes , when treated within our approach , the virtual high - energy ( or momentum ) can be absorbed by the string ( i.e. by the dibaryon as a whole ) resulting in its excitation followed by either meson emission or break up of the excited dibaryon into the two high - energy nucleons ( see e.g. the fig . also , the inverse process of bremsstrahlung in @xmath10-scattering : @xmath352 should be analysed with the same new currents . ] , width=491 ] another good test of the force model developed here will be in prediction of the main characteristic features of e .- processes like @xmath353 , @xmath354 , or @xmath355 with linearly or circularly polarized @xmath351-quanta which are planned to be measured in mainz @xcite . for calculations of such processes our very accurate @xmath37 wavefunctions @xcite found with the above force model can be employed jointly with new e .- currents tightly interrelated to the dressed dibaryon degree of freedom . some diagrams for these new @xmath1-channel currents in two - nucleon systems are displayed in fig . [ fig12 ] . the @xmath1-channel meson - exchange currents for @xmath356 or @xmath357 processes should be confronted with the traditional @xmath6-channel currents for @xmath358-subsystem ( see the fig . [ fig13 ] ) and we can conclude from this comparison that both types of currents , i.e. @xmath6- and @xmath1-channel ones can be considered as a complementary to each other . in other words , those @xmath6-channel currents which disappear in the @xmath10-system ( i.e. pion - in - flight etc . ) will appear in form of @xmath1-channel currents . the same occurs also for @xmath36-system . thus the processes like @xmath359 or @xmath360 or @xmath361 should be a very good test for the new electromagnetic currents . hence , the dressed dibaryon component and new currents associated with it should modify the results of the theoretical interpretations of many dedicated experiments done at moderate to high momentum and energy transfers , especially those which are now going on or are planned at mainz and bonn . bremsstrahlung within traditional obe - models of the nucleon - nucleon interaction . [ fig13],width=491 ] the first attempts to apply the dbs - model for description of properties of three - nucleon systems @xcite was very successful . in particular , it has been clearly demonstrated that the new short - range three - body attractive force induced by @xmath3-meson exchange between the dbs and third nucleon ( see the fig . [ fig11 ] ) with the @xmath362 coupling constant adjusted to give the @xmath55 binding energy gives an important contribution to the @xmath28 binding energy and other properties of @xmath28-system , and can explain quantitatively all static properties of @xmath37 and @xmath55 ground states , including a precise parameter - free description of the coulomb displacement energy of @xmath37@xmath55 @xcite , and all the charge distributions in these nuclei . in these @xmath28 calculations done within the above dibaryon @xmath0-force model with additional incorporation of three - nucleon scalar forces shown by the graphs in the fig . [ fig11 ] , the weight of the dibaryon component in the @xmath28 wave function reaches as much as @xmath363% and its contribution to the total @xmath28 binding energy is about 50% ( ! ) . in other words , half of the total nuclear binding comes from the strong interaction of dibaryon with surrounding nucleons via the @xmath3-meson exchange @xcite . another important conclusion of these @xmath28 calculations is a strong density dependence of the effective two- and three - body forces induced by the intermediate dibaryon generation . the density dependence has a repulsive character and is qualitatively similar to that assumed in the phenomenological skyrme model . it is very likely that just the interplay between the above density dependent repulsion and @xmath29 and @xmath28 attraction related to the scalar meson exchange provides the saturation property of nuclear matter . we developed in the work some covariant effective field - theoretical approach toward the description of intermediate- and short - range components of @xmath0 interaction and nuclear forces . the approach assumes that the nucleon - nucleon interaction at intermediate and short ranges has a very complicated and multistep character . in the first step , the two - nucleon system goes , by color exchange , to the confined two - cluster @xmath83 configuration where the quark clusters are moving at the end of a quantum string ( inside of which the gluon fields are localized ) . excitations ( i.e. vibrations and rotations ) of this string can be described by a relativistic covariant harmonic oscillator with a linear confinement potential which leads quite naturally to the linear regge - trajectory of the respective excited levels of the string . in the next step , the string interacts with vacuum meson fields ( mainly @xmath364 and @xmath5 ) and thus it goes to the dibaryon stage by the respective dressing . it is shown that @xmath69-excitation of the string generates a strong scalar @xmath3-field which , due to the strong attraction to the six - quark core , will compress and stabilize the resulting dressed six - quark bag . in some aspect the emerged picture should be rather similar to the concept of `` small chiral bag '' suggested long ago by brown and rho @xcite . in their model the @xmath82-mit bag is compressed by the strong pressure of its pion ( chiral ) field to the small size where the additional inner kinetic pressure of free quarks will compensate the external pressure of the pion field to the bag surface . in our case , in contrast to the above `` small bag model '' , the main meson field is a strong scalar @xmath3-field which interacts with the six quark system in the whole volume of the bag rather than on the its surface only . another great difference from the brown - rho model is the origin of the stabilizing field ( @xmath3 in our case ) . in our approach the main component of this scalar field is emerged in deexcitation of the string from @xmath69 to @xmath365 states . thus , in this process one observes the transformation of the `` color '' energy of gluon fields ( inside the string ) to the `` white '' energy of scalar meson field . in the next stage of the process , the @xmath3 field will compress and stabilize the six - quark bag and shift strongly its mass down to the pion - production threshold . thus this shift should be included to the polarization operator @xmath366 of the total string propagator . further , keeping in mind a strong coupling of the @xmath3-meson to the two - pion channel one can predict from this specific dressing mechanism an enhanced two - pion production in @xmath0 collision as compared to the traditional obe - models , the feature just observed experimentally . moreover , almost the same mechanism can enhance quite naturally the single - pion production in @xmath10-collision and especially the near threshold @xmath367 production . the consistent theoretical explanation of which met so serious problems in the recent decade @xcite . moreover , a quite similar two - stage mechanism ( in first stage , the excitation of the color string with the subsequent transformation of the `` color '' string energy to the meson fields ) should be responsible for multi - meson production in @xmath0 and heavy ion collisions . so , by this way , the multi - meson production in high - energy heavy - ion collision may give a good test for the proposed color string model . the authors plan to return to this point in their next papers . the authors appreciate very much the fruitful discussions in the course of this work with p. grabmayr , m. kaskulov , i.t . obukhovsky and v.n . the work of the v.i.k . and m.a.s . is partially supported financially by the dfg - grant no . fa-67/20 - 1 and rfbr - grants nos . 02 - 02 - 16612 and 04 - 02 - 04001 . 99 a.d . lahiff , i.r . afnan , phys . c 60 ( 1999 ) 024608 . a. nogga , h. kamada , w. glckle , phys . 85 ( 2000 ) 944 . kukulin , i.t . obukhovsky , v.n . pomerantsev , a. faessler , phys . at . 64 ( 2001 ) 1667 . d. plaemper , j. flender , m.f . gari , phys . c 49 ( 1994 ) 2370 . t. sasakawa , s. ishikawa , few body syst . 1 ( 1986 ) 3 . hicks et al . , c 67 ( 2003 ) 064004 ; f. moschini , proc . of 6th workshop on `` e - m induced two - hadron emission '' , pavia , september 24 - 27 , 2003 , p. 156 . groep et al . , phys . rev . c 63 ( 2001 ) 014005 . s. ishikawa , j. golak , h. witala , h. kamada , w. glckle , d. hber , phys . c 57 ( 1998 ) 39 ; r. skibiski et al . , c 67 ( 2003 ) 054002 ; w. glckle et al . , proc . of 6th workshop on `` e - m induced two - hadron emission '' , pavia , september 24 - 27 , 2003 , p. 166 . e. jans , p. barneo , proc . of 6th workshop on `` e - m induced two - hadron emission '' , pavia , september 24 - 27 , 2003 , p. 183 . weinstein , proc . of 5th workshop on `` e - m induced two - hadron emission '' , lund , june 13 - 16 , 2001 , p. 93 ; e. piasetzky , r. gilman , m. sargsian , proc . of 6th workshop on `` e - m induced two - hadron emission '' , pavia , september 24 - 27 , 2003 , p. 211 . laget , nucl . phys . a 579 ( 1994 ) f. stancu , few body syst . 14 ( 2003 ) 83 ; d. bartz , f. stancu , phys . c 63 ( 2001 ) 034001 ; nucl . phys . a 699 ( 2002 ) 316 . mau , prog . part . 50 ( 2003 ) 561 . m. oka , k. shimizu , k. yazaki , prog . 137 ( 2000 ) 1 . entem , f. fernndez , a. valcarce , phys . rev . c 62 ( 2000 ) 034002 ; c 67 ( 2003 ) 014001 . s. weinberg , phys . b 251 ( 1990 ) 288 ; nucl . phys . b 363 ( 1991 ) 3 . d.b . kaplan , nucl - th/9506035 ; nucl . b 494 ( 1997 ) 471 ; d.b . kaplan , m. savage , m.b . wise , nucl . b 478 ( 1996 ) 629 ; phys . lett . b 424 ( 1998 ) 390 . d.r . entem , r. machleidt , phys . c 68 ( 2003 ) 041001 . e. epelbaum , w. glckle , u.g . meissner , eur . j. a 19 ( 2004 ) 125 . r. higa , m.r . robilotta , phys . c 68 ( 2003 ) 024004 ; r. higa , m.r . robilotta , c.a . da rocha , phys . c 69 ( 2004 ) 034009 . meyer et al . , c 63 ( 2001 ) 064002 . c. hanhart , phys . 397 ( 2004 ) 155 . b. von przewoski et al . c 58 ( 1998 ) 1897 . daehnick et al . , phys . c 65 ( 2002 ) 024003 . bo histad , nucl . a 721 , ( 2003 ) 570c ; j. johanson et al . , phys . a 712 ( 2002 ) 75 . f. shimizu et . al . , nucl . phys . a 386 ( 1982 ) 571 . l. alvarez - ruso , e. oset , e. hernandez , nucl . phys . a 633 ( 1998 ) 519 . brodsky , l. frankfurt , r. gilman , j.r . hiller , g.a . miller , e. piasetzky , m. sargsian , m. strikman , phys . b 578 ( 2004 ) 69 . v.i . kukulin , v.n . pomerantsev , a. faessler , j. phys . g 27 ( 2001 ) 1851 . kukulin , i.t . obukhovsky , v.n . pomerantsev , a. faessler , int . e 11 ( 2002 ) 1 . nyman , d.o . riska , phys . scripta 34 ( 1986 ) 533 ; phys . b 203 ( 1988 ) 13 . e. oset , h. toki , m. mizobe , t.t . takahashi , progr . 103 ( 2000 ) 351 . kaskulov , h. clement , phys . c 70 ( 2004 ) 014002 . n. kaiser , u.g . meissner , phys . b 233 ( 1989 ) 457 . v.i . kukulin , few body syst . suppl . 14 ( 2003 ) 71 . kukulin , v.n . pomerantsev , m.m . kaskulov , a. faessler , j. phys . g 30 ( 2004 ) 287 . kukulin , v.n . pomerantsev , a. faessler , j. phys . g 30 ( 2004 ) 309 . obukhovsky , v.i . kukulin , m.m . kaskulov , p. grabmayr , a. faessler , j. phys . g 29 ( 2003 ) 2207 . kukulin , m.a . shikhalev , phys . at . 67 ( 2004 ) 1536 . kaskulov , p. grabmayr , v.i . kukulin , int . j. mod e 12 ( 2003 ) 449 ; m.m . kaskulov , v.i . kukulin , p. grabmayr , few body syst . 14 ( 2003 ) 101 . a.m. kusainov , v.g . neudatchin , i.t . obukhovsky , phys . c 44 ( 1991 ) 2343 . p. faccioli , e.v . shuryak , phys . d 65 ( 2002 ) 076002 . r. rapp , t. schafer , e. shuryak , m. velkovsky , phys . 81 ( 1998 ) 53 . m. alford , k. rajagopal , f. wilczek , phys . b 422 ( 1998 ) 247 . m. huang , p.f . zhuang , w.q . chao , phys . d 65 ( 2002 ) 076012 . f. wilczek , hep - ph/0409168 . d. bartz , fl . stancu , phys . c 59 ( 1999 ) 1756 . bjrken , s.d . drell , relativistic quantum mechanics , mcgraw hill , new york , 1964 . feynman , m. kislinger , f. ravndal , phys . d 3 ( 1971 ) 2706 ; s. ishida , m. ishida , t. maeda , prog . 104 ( 2000 ) 785 . m. moshinsky , a. szczepaniak , j. phys . a 22 ( 1989 ) l817 ; a. szczepaniak , a.g . williams , phys . d 47 ( 1993 ) 1175 . t. takabayasi , phys . rev . b 139 ( 1965 ) b1381 ; suppl . 67 ( 1979 ) 1 . l. mathelitsch , w. plessas , m. schweiger , phys . c 26 ( 1982 ) 65 . powrie et al . , phys . c 64 ( 2001 ) 034602 . brown , m. rho , comments nucl . 18 ( 1988 ) 1 ; see also s.w . hong and b.k . jennings , phys . c 64 ( 2001 ) 038203 .	the present paper is aimed to developing an effective field theory in the gev region to describe consistently both elastic and inelastic @xmath0-scattering in a fully covariant way . in this development we employ our novel @xmath1-channel mechanism of @xmath0-interaction at intermediate and short ranges assuming a formation of the six - quark bag dressed with basic meson fields such as @xmath2 , @xmath3 , @xmath4 and @xmath5 in the intermediate state . the peripheral part of the interaction is governed by @xmath6-channel one- and two - pion exchanges described fairly well by chiral perturbation theory . the @xmath3-meson dressing should be especially important for the stabilization of the bag due to a high coupling constant of the @xmath3-meson with the bag and also a strong interaction between the @xmath3-field and space symmetric quark core @xmath7\rangle$ ] . this stabilization of the intermediate dressed bag ( which shifts the mass of the bag downward to @xmath0 threshold ) will be crucially important for the description of @xmath2 , @xmath8 , @xmath4 and @xmath5-meson production in @xmath0-collisions , and , in particular , for the correct description of @xmath9 production near threshold . so that , we expect that the results of the present work will lead to a deeper understanding the short- and intermediate - range nuclear forces and internucleon correlations in nuclei . , @xmath0 interaction ; effective - field theory ; dressed bag ; dibaryon 13.75.cs ; 11.10.-z ; 14.20.pt
we consider a delay differential equation that occurs in the study of periodic chronic myelogenous leukemia @xcite , @xcite @xmath0x(t)+k\frac{\beta_0x(t - r)}{1+x(t - r)^n}.\ ] ] here @xmath1 are positive parameters . parameter @xmath2 is of the form @xmath3 , with @xmath4 also positive . we chose here to take @xmath2 as an independent parameter , instead of @xmath5 keeping in mind the fact that , due to its definition , @xmath6 . we do not insist here on the physical significance of the model , this is largely presented in @xcite , @xcite . as usual in the study of differential delay equations , we consider the banach space @xmath7\mapsto \mathbb{r},\ , \psi\ , \mathrm{is\ , continuous\ , on\,}[-r,0 ] \right\},$ ] with the sup norm , and for a function @xmath8 and a @xmath9 we define the function @xmath10 by @xmath11 in @xcite we proved that with the initial condition @xmath12 has an unique , defined on @xmath13 bounded solution . the equilibrium points of the problem are @xmath14 the second one is acceptable from the biological point of view if and only if @xmath15 condition that implies @xmath16 in @xcite we studied in detail the stability of equilibrium solutions of ( [ eq ] ) . the results therein are briefly recalled in section 2 . the second equilibrium point can present hopf bifurcation for some points in the parameter space @xcite , @xcite , @xcite . in @xcite we considered a typical hopf bifurcation point , constructed an approximation of the center manifold in that point and found the normal form of the hopf bifurcation ( by computing the first lyapunov coefficient , @xmath17 ) . from this normal form we obtained the type of stability of the periodic orbit emerged by hopf bifurcation . in section 3 we remind the ideas concerning the approximation of the center manifold , necessary for the computation of @xmath18 that were developed in @xcite . in order to give a visual representation of the hopf bifurcation points we fix two parameters ( @xmath19 and @xmath20 ) and in the three dimensional parameter space @xmath21 we represent the surface of hopf bifurcation points . let us denote by @xmath22 the vector of parameters @xmath23 . if @xmath24 is on the hopf bifurcation surface and @xmath25 the hopf bifurcation is a non - degenerate one , and @xmath26 is named _ codimension one hopf point _ @xcite . if at a certain @xmath27 , @xmath28 we have a degenerate hopf bifurcation and if we fix three of the five parameters and vary the other two parameters in a neighborhood of the bifurcation point in the parameter space , we obtain a bautin type bifurcation @xcite ( if the second lyapunov coefficient , @xmath29 , is not zero ) . if @xmath30 then @xmath31 is also named a _ codimension two hopf point _ @xcite ( since bautin bifurcation is a codimension two bifurcation @xcite ) . in section 4 of the present work we present the method used by us to explore the existence of points with @xmath32 for equation then , for a typical such point with @xmath32 , we develop the procedure to compute the second lyapunov coefficient , @xmath33 . for this we compute a higher order approximation of the center manifold . this involves solving a set of differential equations and algebraic computation ( done in maple ) . in section 5 we present the results obtained for our problem by using the methods exposed in section 4 . we found that the considered problem presents points with @xmath34 and we identify such points ( to a certain approximation ) by the interval bisection technique applied with respect to one of the parameters . we found that all hopf codimension two points previously determined have @xmath35 . we give tables with the values of all the parameters at the points with @xmath32 that we previously determined and we plot these points on the hopf bifurcation points surface . section 6 presents the conclusions of our work , while section 7 is the appendix containing the differential equations for the determination of the approximation of the center manifold ( and their solutions ) . the linearized equation around one of the equilibrium points is @xmath36z(t)+kb_1z(t - r),\ ] ] where @xmath37 or @xmath38 @xmath39 , and@xmath40 the nonlinear part of equation , written in the new variable @xmath41 , will be denoted by @xmath42 the characteristic equation corresponding to is @xmath43 for the equilibrium point @xmath44 we have @xmath45 . in @xcite we pointed out that for @xmath46 @xmath47 for all eigenvalues @xmath48 iff the condition @xmath49 is satisfied . it follows that @xmath50 is locally asymptotically stable in this zone of the parameter space . when @xmath51 @xmath52 is an eigenvalue of the linearized ( around @xmath50 ) problem . for this case , by constructing a lyapunov function , we proved that @xmath50 is stable @xcite . hence , @xmath50 is stable iff it is the only equilibrium point ( see condition ) . when the second equilibrium point occurs , @xmath50 loses its stability . for the equilibrium point @xmath53 @xmath54.\ ] ] the study of stability performed in @xcite , that relies on the theoretical results of @xcite , reveals the following distinct situations for the stability of @xmath55 . * i. * @xmath56 that is equivalent to @xmath57 with two subcases : * i.a . * @xmath56 and @xmath58 in this case @xmath59 for all eigenvalues @xmath60 if and only if @xmath61 and @xmath62 where @xmath63 is the solution in @xmath64 of the equation @xmath65 * remarks . * * 1 . * _ the condition _ @xmath66 _ is equivalent to _ @xmath67 _ that implies _ @xmath68 _ and becomes _ * 2 . * _ the condition _ @xmath70 _ is , in this case , equivalent to _ @xmath71 _ that is the condition of existence of _ @xmath53 _ hence only condition _ _ brings relevant information . _ * i.b . * @xmath56 and @xmath72 in this case , @xmath73 for all eigenvalues @xmath60 if and only if @xmath74 where @xmath63 is defined as above . hence , when these conditions are satisfied , @xmath55 is locally asymptotically stable . * remarks . * _ the condition _ @xmath66 _ is equivalent to _ _ condition _ @xmath76 _ is equivalent to _ @xmath77 _ if _ @xmath78 _ the above inequality is satisfied since the expression in the left hand side is positive . _ _ if _ @xmath79 _ then two cases are possible : either is satisfied and then _ @xmath80 _ for every eigenvalue , or _ @xmath81 _ and the stability condition is satisfied if _ @xmath82 * ii . * @xmath83 that is equivalent to @xmath84 in this case , we can only have @xmath85 and , by @xcite , @xmath86 for all eigenvalues @xmath60 if and only if @xmath87 but this inequality is equivalent to @xmath88 that is already satisfied , since @xmath55 exists and is positive . it follows that if @xmath83 then @xmath55 is stable . in the stability discussion of @xmath55 shortly presented above , in * i. * , the case @xmath89 occurs on the frontier of the stability domain . relations and @xmath90 imply @xmath91 and that the pair @xmath92 represents a solution of ( [ chareq ] ) . all others eigenvalues have strictly negative real parts . thus the points in the parameter space where relations and hold are candidates for hopf bifurcation . in order to have an image of the set of such points , we fixed @xmath19 and @xmath20 and in the space @xmath93 we represented the surface of equation @xmath94 both the numerator and the denominator of the ratio above show that the condition @xmath61 must be fulfilled . in fig . 1 we present the image of the obtained surface for @xmath95 and a projection of the surface on the plane @xmath96 indicating the domain of definition of the function in the rhs of . assuming that only one parameter varies ( denote it by @xmath22 ) while the others are kept fixed , a point @xmath97 for which is satisfied , is a nondegenerate hopf bifurcation point if @xmath98 and @xmath25 where @xmath17 is the first lyapunov coefficient , to be defined below . if @xmath99 the hopf bifurcation is degenerate . in the sequel we study the existence of points @xmath27 in the parameter space , satisfying and and @xmath100 for the equilibrium point @xmath55 . in order to do this , we have to construct an approximation of the local invariant center manifold and the approximation of the restriction of problem to the local center manifold . in @xcite , by using the method of @xcite , the problem was formulated as an equation in the banach space @xmath101\mapsto \mathbb{r},\ , \psi\ , \mathrm{is\ , continuous\ , on\,}[-r,0)\wedge\,\exists \lim_{s\rightarrow 0}\psi(s)\in \mathbb{r } \right\},\ ] ] that is @xmath102 where @xmath103 @xmath104 and , for @xmath105 @xmath106,\;\;\ ; l(\varphi)=-(\delta+b_1)\varphi(0)+kb_1\varphi(-r).\ ] ] the developments of this subsection are done for a point @xmath97 in the parameter space , for which relations and are satisfied . hence the problem has two eigenvalues @xmath107 and all other eigenvalues have strictly negative eigenvalues . the eigenfunctions corresponding to eigenvalues @xmath107 are given by @xmath108.$ ] since the eigenfunctions are complex functions , we need to use the complexificates of the spaces @xmath109 that we denote by @xmath110 respectively . we denote by @xmath111 the subspace of @xmath112 generated by @xmath113 in @xcite , by using the ideas in @xcite , we constructed a projector @xmath114 that induces a projector defined on @xmath115 with values also in @xmath111 . the space @xmath116 is decomposed as a direct sum @xmath117 where @xmath118 we do not present here the details of this construction . the projection of problem on @xmath111 leads us to the equation @xmath119 where @xmath120 ^ 2 + \omega^{2}r^2}.\ ] ] the local center manifold is , for our problem , a @xmath121 invariant manifold , tangent to the space @xmath111 at the point @xmath122 ( that is @xmath123 ) , and it is the graph of a function @xmath124 defined on a neighborhood of zero in @xmath111 and taking values in @xmath125 . a point on the local manifold has the form @xmath126 the restriction of problem with @xmath127 to the invariant manifold is @xmath128 with the initial condition @xmath129 where @xmath130 the real and the imaginary parts of this complex equation , represent the two - dimensional restricted to the center manifold problem . we can study this problem with the tools of planar dynamical systems theory ( see , e.g. @xcite ) . we set @xmath131 and write @xmath132 where @xmath133 @xmath134 by using , we write @xmath135 equation ( [ proj - eq1 ] ) becomes @xmath136 where @xmath137 all the computation below are made at @xmath138 but , for simplicity , the parameter will not be written . the first lyapunov coefficient at @xmath138 is given by @xcite @xmath139 in order to compute it , we have to compute the corresponding coefficients @xmath140 of , and , for this , we must find the coefficients of the series of @xmath141 in . we consider the function @xmath142 and its derivatives of order @xmath143 in @xmath53 i.e. @xmath144 let us also consider the function @xmath145 . the nonlinear part of eq . is @xmath146 or , in terms of @xmath147 the function @xmath141 of ( [ nonlineq3 ] ) is @xmath148 we denote by @xmath149 the semigroup of operators generated by eq . . we have ( by putting @xmath150 , and by assuming that the initial function , @xmath151 is on the invariant manifold ) @xmath152(0))+kf([\widetilde{t}(t)\phi](-r))=\ ] ] @xmath153 ^ 2-\ ] ] @xmath154 ^ 3- .... +\ ] ] @xmath155 ^ 2+\ ] ] @xmath156 ^ 3+ ... =\ ] ] @xmath157 the coefficients @xmath158 involved in the expression of @xmath18 ( computed also in @xcite ) , are obtained by equating the same order terms in the two series from @xmath159 @xmath160 @xmath161 @xmath162@xmath163 from where @xmath164 will be easily obtained from ( with @xmath165 given in ( [ psi ] ) ) . since @xmath166 depends on some values of @xmath167 and @xmath168 , we have to determine these functions first . the functions @xmath169 are determined from the relation @xcite , @xcite , @xcite @xmath170 @xmath171 that , by matching the same type terms yields differential equations for each @xmath172 while the integration constants are determined from @xmath173 @xmath174 e. g. in order to determine @xmath175 we equate the terms containing @xmath176 and obtain @xmath177 we integrate between @xmath178 and @xmath179 and obtain @xmath180 by equating the coefficients of @xmath176 in , the relation @xmath181 results . from the two relations above , we find @xmath182 and @xmath183 the equations for all other functions @xmath184 needed in the computation of @xmath17 and @xmath33 are all listed in the appendix . they were obtained by symbolic computations in maple . remark . * _ the functions @xmath185 were determined ( for other problems ) by many authors @xcite , @xcite , @xcite , in connection to the study of hopf bifurcations . _ once @xmath186 computed , we determine the value of @xmath187 and that of @xmath17 . the sign of the first lyapunov coefficient @xmath17 determines the type of the hopf bifurcation @xcite . that is , if @xmath188 we have a supercritical hopf bifurcation , while if @xmath189 - a subcritical hopf bifurcation . the hopf points of codimension two are among the points of the surface of equation , surface that , for @xmath19 and @xmath20 fixed can be represented in @xmath190 ( as in fig.1 ) . since they obey a new constraint , @xmath34 they must lie on a curve on that surface . if we could write this constraint in a simple algebraical form , then we could attempt to represent the curve defined by and @xmath191 unfortunately , this is not possible , since the expression of @xmath18 written in terms of the parameters of the problem is quite complex . in this situation , our strategy for finding hopf points of codimension two ( bautin type bifurcation points ) is the following . * as in @xcite , we chose @xmath192 in a zone of parameters acceptable from biological point of view ; * with the chosen values of the parameters we compute @xmath193 , ( the value of @xmath194 at these parameters ) , next + - if @xmath195 then @xmath55 is stable for the chosen parameters and we stop ; + - if @xmath196 we compute @xmath197 ; * - if @xmath198 we stop ; + - if @xmath199 we determine @xmath200 and @xmath201 such that condition ( [ r - h ] ) is fulfilled i.e. we set @xmath202 @xmath203 * for the found values of parameters , the first lyapunov coefficient , @xmath17 , is computed ; * then we vary a parameter different of @xmath204 , such that the absolute value of @xmath17 decreases - we have chosen to vary @xmath205 ; * the above computations are repeated for several values of @xmath206 until we find two values @xmath207 of this parameter such that the values of @xmath17 have opposite signs - if such values exist ( obviously , at each computation , the value of @xmath201 changes , but the condition @xmath208 is maintained ) ; * then by using the bisection of the interval technique ( with respect to @xmath205 , starting from the interval @xmath209 $ ] ) , we find a set of parameters such that @xmath32 ( to a certain approximation ) . the obtained numerical results are presented in section 5 . in order to see if the hopf points found by the above algorithm are points of codimension two , or have a higher degeneracy , we must compute the second lyapunov coefficient in such a point . the formula for the second lyapunov coefficient , at a hopf codimension two point is @xcite @xmath210+\ ] ] @xmath211 @xmath212 + 3im(g_{20}g_{11})img_{21}\right\}+\ ] ] @xmath213+im(g_{20}g_{11})[3re(g_{20}g_{11})-2\|g_{02}\|^2]\right\}.\ ] ] we list below the expressions of the @xmath158s corresponding to the involved @xmath214s ( not already given in subsection 3.2 ) : @xmath215 + @xmath216 @xmath217 @xmath218 and , finally , @xmath219-b_3\left[6w_{11}(0)^2 + 3w_{12}(0)+3w_{02}(0)w_{20}(0)+w_{30}(0)+\right.\\ \,&\left.+6w_{11}(0)w_{20}(0)+6w_{21}(0)\right]+b_4\left[3w_{20}(0)+6w_{11}(0)+w_{02}(0)\right]-\\ \,&-b_5+kb_2\left[6w_{11}(-r)w_{21}(-r)+3w_{12}(-r)w_{20}(-r)+3e^{-i\omega r}w_{22}(-r)+ \right . \\ \,&\left.+w_{02}(-r)w_{30}(-r)+2e^{i\omega r}w_{31}(-r)\right]+kb_3\left[3w_{02}(-r)e^{-i\omega r)}w_{20}(-r)+\right.\\ \,&+ 6e^{-i\omega r}w_{11}(-r)^2 + 3e^{-2i\omega r}w_{12}(-r)+e^{2i\omega r}w_{30}(-r)+\\ \ , & \left.+6e^{i\omega r}w_{11}(-r)w_{20}(-r)+6w_{21}(-r)\right]+kb_4\left[3e^{i\omega r}w_{20}(-r)+\right.\\ \,&\left.+6e^{-i \omega r}w_{11}(-r)+w_{02}(-r ) e^{-3i\omega r}\right]+kb_5e^{-i\omega r}. \end{array}\ ] ] the values in 0 and @xmath178 of the involved @xmath184s are computed by solving the differential equations and by using the additional conditions listed in the appendix . there is a function @xmath220 that requires a special treatment , i.e. @xmath221 . as shown in @xcite , the two equations that determine @xmath222 are dependent , and the value of @xmath223 may be obtained by considering a perturbed problem ( depending on a small @xmath224 ) , computing the corresponding @xmath225 and by taking the limit as @xmath226 the formula obtained in @xcite is @xmath227 where @xmath228 is the bilinear form @xcite , @xcite , @xcite defined on @xmath229,\mathbb{r})\times c([-r,0],\mathbb{r}),$ ] by @xmath230 @xmath231 $ ] and @xmath232.$ ] after determining @xmath233 we compute @xmath234 by using any of the two relations that connect the two values . all other necessary values of @xmath220 , @xmath235 are determined by using the relations in the appendix . as we asserted above , we explored the equilibrium point @xmath55 with regard to the occurrence of hopf degenerate bifurcations . as first step we search for points in the parameter space , where @xmath191 we explored the zone of parameters @xmath236 given by : @xmath237 ( we remind that @xmath238 in order that @xmath55 exist , and @xmath6 ) . for every @xmath239 fixed , we looked for a @xmath240 and a @xmath201 such that @xmath241 and @xmath191 we easily see that the case @xmath242 is not interesting since , for @xmath243 by , @xmath244 and the points with @xmath83 are points where @xmath55 is stable and no hopf bifurcations can occur . interesting results were found for @xmath245 and @xmath246 for each @xmath247 and each @xmath248 we encountered the following behavior : at small values of @xmath205 ( i.e. @xmath249 ) , we found @xmath250 hence we could detect hopf points . for these points @xmath17 was proved to be positive , it decreased with increasing @xmath205 and became negative for some value of @xmath205 . we then applied the bisection of interval technique as described in section 4.1 and actually found hopf points with @xmath32 . for each such points we computed the second lyapunov coefficient and we found that all points with @xmath34 determined before , have @xmath251 hence they are codimension two hopf points . for @xmath252 the hopf points occur for very large values of @xmath204 ( @xmath253 ) and , since such values for @xmath204 are not realistic ( as asserted in @xcite , @xcite ) , we present only the results obtained for @xmath254 in figures 2 - 6 we present the sets of parameters @xmath27 with @xmath255 found by us , in tables and plotted on the surfaces of hopf bifurcation points . these points were determined as the points where @xmath256 is approximately equal to @xmath257 a better precision may be attained by continuing the bisection interval procedure further . * * [ cols="^,^,^,^",options="header " , ] ] by connecting the points with @xmath32 we determine approximately a curve of points with @xmath32 on each of the surfaces in figs . let us denote by @xmath258 the projection of this curve on the plane @xmath259 . our computations of @xmath17 showed that on each of the surfaces in figs . 2 - 6 , @xmath189 for the points of the surface that are projected on the zone @xmath260 of the plane @xmath259 , and hence in this zone , the hopf bifurcation is subcritical . for the points that are projected on the zone @xmath261 we found @xmath262 , hence in this zone , the hopf bifurcation is supercritical . finally , for @xmath263 the hopf points found had only negative @xmath18 and thus , we could not find any degenerate hopf points . the existence of hopf codimension two points was investigated for the non - zero equilibrium point of the equation that models the periodic chronic myelogenous leukemia , presented in @xcite , @xcite . we searched such points in a zone of the 5-dimensional parameters space , of biological significance . we found that points @xmath22 with @xmath264 actually exist and we present tables with values of the parameters where they occur . for each of these points we computed the corresponding second lyapunov coefficient @xmath265 and listed the obtained values in the above mentioned tables . we remark that for all the considered sets of parameters , @xmath266 we also remark that the values of @xmath33 are close to zero , and , at least for @xmath267 and the chosen values of @xmath20 the behavior of @xmath33 as function of @xmath2 is the same , i.e. @xmath33 decreases when @xmath2 increases . since @xmath268 it can not be greater than 2 , and for @xmath6 we found only negative values of @xmath33 ( we also considered in our computation the limit case @xmath269 - that would imply @xmath270 - and for @xmath271 and for every @xmath20 considered we found still negative @xmath33 in the points where @xmath32 ) . hence among the considered points no higher order degeneracies can be found ( in the zones with biological significance ) . the points where @xmath32 and @xmath272 are , if we vary two parameters , points of bautin bifurcation ( provided a certain non - degeneracy condition is satisfied @xcite ) . a numerical investigation of the solutions of the equation for the parameters in and around the points with @xmath34 is in course and will be soon submitted to publication . we list below the differential equations , their solutions , and the supplementary conditions for the functions @xmath220 with @xmath273 excepting @xmath167 that is treated in section 3 . the right hand side of the differential equations as well as the supplementary conditions are obtained by symbolic computation in maple . s. a. campbell , _ calculating centre manifolds for delay differential equations using maple _ , in balachandran b. , kalmar nagy t. and gilsinn d. ( eds ) , delay differential equations : recent advances and new directions , springer , new york , 2009 . t. faria , _ normal forms for rfde in finite dimensional spaces _ -section 8.3 of j. hale , l.t . magalhaez , w. oliva , _ dynamics in infinite dimensions _ , applied mathematical sciences , * 47 * , springer , new york , 2002 . a. v. ion , _ new results concerning the stability of equilibria of a delay differential equation modeling leukemia _ , proceedings of the 12th symposium of mathematics and its applications , timioara , november 5 - 7 , 2009 , 375 - 380 ; arxiv:1001.4658 . a. v. ion , _ on the computation of the third order terms of the series defining the center manifold for a scalar delay differential equation _ , accepted in journal of dynamics and differential equations , doi 10.1007/s10884 - 012 - 9243 - 8 .	this paper continues the work contained in two previous papers , devoted to the study of the dynamical system generated by a delay differential equation that models leukemia . here our aim is to identify degenerate hopf bifurcation points . by using an approximation of the center manifold , we compute the first lyapunov coefficient for hopf bifurcation points . we find by direct computation , in some zones of the parameter space ( of biological significance ) , points where the first lyapunov coefficient equals zero . for these we compute the second lyapunov coefficient , that determines the type of the degenerate hopf bifurcation . + * acknowledgement . * work partially supported by grant 11/05.06.2009 within the framework of the russian foundation for basic research - romanian academy collaboration . + * keywords : * delay differential equations , stability , hopf bifurcation , normal forms . + * ams msc 2000 : * 65l03 , 37c75 , 37g05 , 37g15 .
the amplitude of the @xmath0-mode component of the cosmic microwave background ( cmb ) polarization on large scales provides the current best constraint on the thomson scattering optical depth to reionization , @xmath2 . using the first three years of data from the _ wilkinson microwave anisotropy probe _ ( _ wmap _ ) and making the simple assumption that the universe was reionized instantaneously , @xcite find @xmath3 . theoretical studies suggest that the process of reionization was too complex to be well described as a sudden transition ( e.g. , * ? ? ? previous studies have examined how the constraint on @xmath2 depends on the evolution of the globally - averaged ionized fraction during reionization , @xmath4 , for a variety of specific theoretical scenarios . if the assumed form of @xmath4 is incorrect , the estimated value of @xmath2 can be biased ; this bias can be lessened by considering a wider variety of reionization histories at the expense of increasing the uncertainty in @xmath2 @xcite . the angular scales on which cmb polarization from reionization is correlated depend on the horizon size at the redshift of the free electrons : the higher the redshift , the higher the multipole , @xmath5 ( e.g. , * ? ? ? * ; * ? ? ? varying @xmath4 changes the relative contributions to the polarization coming from different redshifts , and therefore changes the shape of the large - scale @xmath0-mode angular power spectrum , @xmath6 . because of this dependence , measurements of the low-@xmath5 @xmath0-mode spectrum should place at least weak constraints on the global reionization history in addition to the constraint on the total optical depth . recent studies suggest that _ wmap _ data provide little information about @xmath4 beyond @xmath2 ( e.g. , * ? ? ? * ) , but it is worth asking what we can ultimately expect to learn about reionization from cmb polarization . @xcite proposed using a principal component decomposition of the ionization history to quantify the information contained in the large - scale @xmath0-mode polarization . the effect of any ionization history on the @xmath0-modes can be completely described by a small number of eigenmode parameters , unlike a direct discretization of @xmath7 in redshift bins . here we extend the methods of @xcite using markov chain monte carlo techniques to find constraints on the principal components of @xmath4 using both the 3-year _ wmap _ observations and simulated future data . analytic studies and simulations indicate that reionization is an inhomogeneous process , and this inhomogeneity is expected to contribute to the small - scale cmb temperature and polarization anisotropies ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? here we focus on the large - scale @xmath0-modes only ( @xmath8 ) where such effects can be neglected , so we only consider the evolution of the globally - averaged ionized fraction as a function of redshift . in the following section , we describe the principal component method for parameterizing the ionization history and show that the effects of @xmath4 on large - scale @xmath0-mode polarization can be encapsulated in a small set of parameters . the method allows @xmath4 to be a free function of redshift that is not bounded by physical considerations , so in [ sec : phys ] we derive limits that can be placed on the model parameters to eliminate most of the unphysical models where @xmath9 or @xmath10 . we outline several ways to apply the principal component approach to constrain the reionization history with large - scale @xmath0-mode data in [ sec : appl ] . using markov chain monte carlo methods , we examine some of these applications in more detail in [ sec : mcmc ] using both 3-year _ wmap _ data and simulated future cmb polarization data . we summarize our findings and conclude in [ sec : discuss ] . models with similar total optical depth but different ionization histories can produce markedly different predictions for the @xmath0-mode power spectrum . in an instantaneous reionization scenario , the contribution to the optical depth from @xmath4 is concentrated at the lowest redshifts possible for a given @xmath2 , and the @xmath0-mode power spectrum for such a model is sharply peaked on large scales , at @xmath11 . the main effect of shifting some portion of the reionization history to higher redshifts while keeping @xmath2 fixed is to reduce the @xmath0-mode power on the largest scales and increase it on smaller scales . this redistribution of power is illustrated in figure [ fig : cl ] by the ionization histories and @xmath6 for two models , one with nearly instantaneous reionization and the other with comparable optical depth but with @xmath4 concentrated at higher @xmath12 . in general , a flatter large - scale @xmath0-mode spectrum with power extending out to @xmath13 - 20 is a sign of a large ionized fraction at high redshift . however , there is not a one - to - one correspondence between @xmath5 and @xmath12 ; the ionized fraction within any particular narrow redshift bin affects @xmath6 over a wide range of angular scales . ionized fractions in adjacent redshift bins have highly correlated effects on @xmath6 , which makes it difficult to extract from @xmath0-mode polarization data a constraint on @xmath14 at a specific redshift . as suggested by @xcite , one can use principal components of the reionization history as the model parameters instead of @xmath14 in redshift bins . these components are defined to have uncorrelated contributions to the @xmath0-mode power ; since each has a unique effect on @xmath6 , the amplitudes of the components can be inferred from measurements of the large - scale power spectrum . principal component analysis of @xmath4 also indicates which components can be determined best from the data . in this section , we describe the principal component method and introduce the notation that we will use throughout the paper . consider a binned ionization history @xmath15 , @xmath16 , with redshift bins of width @xmath17 spanning @xmath18 , where @xmath19 and @xmath20 so that @xmath21 . throughout this paper we assume that the ionized fraction is @xmath22 for redshifts @xmath23 and @xmath24 at @xmath25 . we take @xmath26 , consistent with observations of quasar spectra @xcite . the principal components of @xmath15 are eigenfunctions of the fisher matrix , computed by taking derivatives of @xmath6 with respect to @xmath15 : @xmath27 assuming full sky coverage and neglecting noise . since @xmath15 only significantly contributes to the @xmath0-mode spectrum at small @xmath5 , we typically truncate the sum in equation ( [ eq : fisher1 ] ) at @xmath28 where @xmath6 is dominated by the first acoustic peak . the derivatives are evaluated at a fiducial reionization history , @xmath29 . following @xcite we typically choose @xmath29 to be constant during reionization , although other functional forms may be used . since the effects on @xmath6 of @xmath15 in adjacent redshift bins are highly correlated , the fisher matrix contains large off - diagonal elements . the principal components @xmath30 are the eigenfunctions of @xmath31 , @xmath32 where the factor @xmath33 is included so that the eigenfunctions and their amplitudes have certain convenient properties . the inverse eigenvalues , @xmath34 , give the estimated variance of each principal component eigenmode from the measurement of low-@xmath5 @xmath0-modes . we order the modes so that the best - constrained principal components ( smallest @xmath34 ) have the lowest values of @xmath35 , starting at @xmath36 . the noise level and other characteristics of an experiment can be included in the construction of the eigenfunctions , but since the effect on @xmath37 is small we always use the noise - free eigenfunctions here . the eigenfunctions satisfy the orthogonality and completeness relations @xmath38 the normalization of @xmath37 is chosen so that the eigenfunctions are independent of bin width as @xmath39 . in equation ( [ eq : orthog1 ] ) and elsewhere in this paper where there are sums over redshift , we assume the continuous limit , replacing @xmath40 by @xmath41 . as long as the bin width is chosen to be sufficiently small , the final results we obtain are independent of the redshift binning . we adopt @xmath42 as the default bin width . the three lowest - variance eigenfunctions for two different fiducial models are shown in figure [ fig : eigfn ] . the lowest eigenmode ( @xmath36 ) is an average of the ionized fraction over the entire redshift range , weighted at high @xmath12 . the @xmath43 mode can be thought of as a difference between the amount of ionization at high @xmath12 and at low @xmath12 , and higher modes follow this pattern with weighted averages of @xmath4 that oscillate with higher and higher frequency in redshift . eigenfunctions of fiducial models with different values of @xmath44 have similar shapes with the redshift axis rescaled according to the width of @xmath45 . the eigenfunctions are mostly insensitive to the choice of ionized fraction in the constant-@xmath14 fiducial histories . an arbitrary reionization history can be represented in terms of the eigenfunctions as @xmath46 where the amplitude of eigenmode @xmath35 for a perturbation @xmath47 is @xmath48 [ note that our conventions for the normalization of @xmath49 and @xmath37 differ from those of @xcite by factors of @xmath50 . ] any global ionization history @xmath4 over the range @xmath51 is completely specified by a set of mode amplitudes @xmath49 . if perturbations to the fiducial history are small , @xmath52 , then the mode amplitudes are uncorrelated , with covariance matrix @xmath53 . for a fixed fiducial model , however , arbitrary reionization histories generally have @xmath54 , in which case the amplitudes of different modes can become correlated as we discuss in [ sec : fisher ] . the main advantage of using the principal component eigenmodes of @xmath4 instead of some other parameterization is that most of the information relevant for large - scale @xmath0-modes is contained in the first few modes . this means that if one constructs @xmath4 from equation ( [ eq : mmutoxe ] ) keeping only the first few terms in the sum over @xmath35 , then the @xmath0-mode spectrum of the resulting ionization history will closely match that of @xmath4 with all modes included in the sum . the effect of each eigenmode on @xmath6 becomes smaller as @xmath35 increases , as shown in figure [ fig : zmaxa ] for the first two modes . @xcite demonstrated that for a specific fiducial ionization history @xmath55 and assumed true history , only the first three modes of @xmath4 are needed to produce @xmath6 indistinguishable from the true @xmath0-mode spectrum . the ionization histories and corresponding @xmath0-mode spectra in figure [ fig : clcomp ] demonstrate this completeness for a fairly extreme model in which the first ten eigenmodes all have significant amplitudes . even in the simplest case where a single eigenmode is used in place of the original @xmath4 , the error in @xmath6 is only @xmath56 . with 3 - 5 modes , the error is a few percent or less at all multipoles and safely smaller than the cosmic variance of @xmath57 the top panel of figure [ fig : clcomp ] shows that this completeness does not extend to the ionization history itself : @xmath4 constructed from as many as five eigenmodes is a poor approximation to the full reionization history . from this and other similar tests on the completeness in @xmath6 of the lowest - variance eigenmodes we conclude that the first 3 - 5 modes contain essentially all of the information about the reionization history that is relevant for large - scale @xmath0-mode polarization . this fact is particularly useful for constraining the global reionization history with cmb polarization data using markov chain monte carlo techniques . since the number of parameters that must be added to a monte carlo chain to describe an arbitrary @xmath4 is relatively small , we can obtain constraints from the data that are independent of assumptions about the reionization history with minimal added computational expense @xcite . the exact number of modes required varies depending on the true ionization history and the fiducial history , so when analyzing data it is a good idea to check that the results do not change significantly when the next modes are included in the sum in equation ( [ eq : mmutoxe ] ) . the main caveat to this model independence is that in practice we must set some maximum redshift for histories in any particular chain of monte carlo samples , ignoring any contribution to the observed low-@xmath5 @xmath6 from ionization at @xmath58 . since the eigenfunctions of the ionization history are stretched in redshift as @xmath44 increases ( fig . [ fig : eigfn ] ) , at higher @xmath44 more modes are needed to accurately represent any particular feature in @xmath4 . for example , take the true ionization history to be instantaneous reionization at @xmath59 . figure [ fig : zmaxb ] shows the error in @xmath6 if we truncate the eigenmode sum of equation ( [ eq : mmutoxe ] ) at @xmath60 modes using fiducial histories with @xmath61 and @xmath62 . for each fiducial model , the error decreases as the number of modes in the sum increases , but the error at fixed @xmath60 is larger for @xmath61 than @xmath62 . the requirement of retaining a larger set of parameters as @xmath44 increases makes it less practical to study models with significant reionization at extremely high redshifts [ @xmath63 ; e.g. , @xcite ] , but even for @xmath44 as high as @xmath64 the number of eigenmodes needed is reasonably small ( @xmath65 ) . while theories of reionization provide useful priors on @xmath44 in the context of specific models , it would be better to be able to constrain @xmath44 empirically by measuring the @xmath0-mode power accurately up to @xmath66 . the current 3-year _ wmap _ polarization data have high enough signal - to - noise to be useful for parameter constraints only at @xmath67 @xcite , so their sensitivity to high-@xmath12 reionization is limited . given that ionization at some redshift generates polarization out to a maximum @xmath5 , _ wmap _ data can still place weak bounds on the total optical depth contribution above a certain redshift , even if it can not distinguish whether these contributions arise from redshifts above a chosen @xmath44 as we discuss in [ sec : wmap ] . the values we choose here for @xmath44 are partly influenced by the fact that few theoretical reionization scenarios predict an ionized fraction at @xmath68 that would significantly affect large - scale @xmath6 . future data should better constrain @xmath6 at higher multipoles , allowing useful limits to be placed on @xmath44 from the data alone . the need to consider a limited range in redshift is shared by other methods , for example those that constrain binned @xmath15 instead of the eigenmode amplitudes @xcite . as already mentioned , the principal component approach has the unique advantages that results can be made independent of bin width and that relatively few extra parameters are required . however , this approach also has a unique difficulty in that the physicality of the reionization history [ @xmath69 is not built in to the method . hence constraints derived from measurements of the eigenmodes can be weaker than those from a method that enforces physicality . we can , however , place some prior constraints on the set of eigenmode amplitudes that must be satisfied by any physical model , as we discuss in the next section . although the actual ionized fraction must be between 0 and 1 ( neglecting helium reionization and the small residual ionized fraction after recombination ) , there is nothing in the construction of @xmath4 from the eigenmodes in equation ( [ eq : mmutoxe ] ) that ensures that the ionized fraction will obey these limits . whether or not @xmath14 has a physical value at a particular redshift depends on the amplitudes of all of the principal components . even for a physical ionization history , the truncated sum up to mode @xmath60 , @xmath70 is not necessarily bounded by 0 and 1 at all redshifts . while formally it is possible to evaluate @xmath6 for reionization histories with unphysical values of @xmath14 , we would like to eliminate as much as possible those models for which the full sum of the eigenmodes , @xmath71 , is unphysical . we find the largest and smallest values of @xmath49 that are consistent with @xmath72 $ ] for all @xmath12 using the definition of @xmath49 in equation ( [ eq : xetommu ] ) . we are free to choose @xmath73 $ ] so that @xmath74 , where the lower limit is strictly negative or zero and the upper limit is positive or zero . for a particular mode @xmath35 , the choice of @xmath75 that maximizes @xmath49 is @xmath76 using this in equation ( [ eq : xetommu ] ) gives an upper limit on @xmath49 . similarly , a lower limit can be obtained by reversing the signs of the inequalities in equation ( [ eq : dxemax ] ) . the resulting physicality bounds are @xmath77 , where @xmath78\pm \|s_{\mu}(z)\|}{2({z_{\rm max}}-{z_{\rm min}})}. \label{eq : physbounds}\ ] ] if @xmath49 violates these bounds for any @xmath35 , the reionization history is guaranteed to be unphysical for some range in redshift . the opposite is not true , however : even if all @xmath49 satisfy equation ( [ eq : physbounds ] ) , @xmath4 may still be unphysical for some @xmath12 . the parameter space that physical models may occupy is restricted further by an inequality that must be satisfied by all eigenmodes simultaneously . assume for simplicity that the fiducial model has a constant ionized fraction , @xmath79 $ ] , for @xmath80 . any physical reionization history @xmath4 must satisfy @xmath81 ^ 2 \leq ( { z_{\rm max}}-{z_{\rm min } } ) f , \label{eq : phys2}\ ] ] where @xmath82 $ ] . using equation ( [ eq : mmutoxe ] ) , the left side of the inequality can also be written @xmath83 ^ 2&=&\int _ { { z_{\rm min}}}^{{z_{\rm max } } } dz \left[\sum_{\mu } m_{\mu } s_{\mu}(z ) \right]^2 \nonumber\\ & = & ( { z_{\rm max}}-{z_{\rm min } } ) \sum_{\mu } m_{\mu}^2 \ , , \label{eq : phys3}\end{aligned}\ ] ] where the second line follows from the orthogonality of the eigenfunctions ( eq . [ [ eq : orthog1 ] ] ) . comparing equations ( [ eq : phys2 ] ) and ( [ eq : phys3 ] ) we obtain a constraint on the sum of the squares of the mode amplitudes , @xmath84 where @xmath85 , depending on the value of @xmath86 . as with the physicality bounds of equation ( [ eq : physbounds ] ) , this upper limit is a necessary but not sufficient condition for physicality . since in practice we can only constrain a limited set of eigenmodes , the uncertainty in modes higher than the first few prevents us from simply excluding all models where @xmath10 or @xmath9 at any redshift because the higher modes can have a significant effect on @xmath4 . however , as shown in [ sec : eigenmodes ] , the higher modes do not affect the polarization power spectrum since the high - frequency oscillations in redshift of higher modes are averaged out . similarly , such eigenmodes have a small effect on the optical depth from a sufficiently large range in redshift . for example , the optical depth from @xmath87 due to modes @xmath88 subject to the physicality constraints of this section can be no larger than @xmath89 , and is likely to be smaller for realistic reionization scenarios . because of this , we assume that we can place priors on the _ optical depth _ that correspond to @xmath90 over the relevant range in redshift . reionization histories with an unphysical optical depth over a large range in redshift are considered to be unphysical models since the addition of higher modes can not perturb the optical depth enough to give it a physical value . once constraints on the principal components of the reionization history have been obtained from cmb polarization data , there are several ways to use those constraints to place limits on observables such as @xmath2 or to test theories of reionization @xcite . we describe some possible applications in this section , and in [ sec : mcmc ] we put these ideas into practice using the 3-year _ wmap _ data and simulated future data . as mentioned in [ sec : intro ] , the constraint on the total optical depth to reionization depends on the assumed model for @xmath4 . the principal component method allows us to explore all globally - averaged ionization histories within a chosen redshift range , @xmath80 . for a given set of eigenmode amplitudes , @xmath91 , equation ( [ eq : mmutoxe ] ) yields the corresponding ionization history which can then be integrated to find the optical depth between any two redshifts @xmath92 and @xmath93 , @xmath94 in particular , the total optical depth to reionization is @xmath95 where @xmath96 for @xmath97 . the principal component approach provides a model - independent way to constrain @xmath2 , so we expect the results to be unbiased and the uncertainty in @xmath2 to accurately reflect the present uncertainty about @xmath4 . moreover , the information about @xmath98 is encapsulated in the first few eigenmodes , i.e. the truncated ionization history of equation ( [ eq : trunc ] ) , @xmath99 , with @xmath100 - 5 for typical fiducial models . the total optical depth and the first principal component amplitude , @xmath101 , are similar quantities in that they are both averages of @xmath4 weighted at high @xmath12 . as we show in the next section , cmb @xmath0-mode polarization data can constrain higher eigenmodes as well ( @xmath102 ) . in particular , the second mode should be the next best constrained quantity since it is constructed to have the smallest variance after @xmath101 . since @xmath103 is related to the difference in @xmath14 between high redshift and low redshift ( see fig . [ fig : eigfn ] ) , we might guess that besides total @xmath2 the data are mainly constraining the fraction of optical depth coming from high redshift versus low redshift . given any set of mode amplitudes defining an ionization history , we can compute a low-@xmath12 optical depth , @xmath104 , and a high-@xmath12 optical depth , @xmath105 , for some choice of intermediate redshift , @xmath106 . note that if either of the redshift ranges @xmath107 $ ] or @xmath108 $ ] is too narrow , constraints on the partial optical depths will be influenced by the physicality priors ( since @xmath109 sets limits on @xmath2 in redshift intervals as discussed in [ sec : phys ] ) and by the uncertainty in modes higher than those included in the chain , which have greater effect on the optical depth in narrower redshift intervals . for these reasons , @xmath106 should be chosen to be not too close to either @xmath110 or @xmath44 . given an appropriately chosen value of @xmath106 , constraints on the principal components can be converted into constraints on the optical depths at high and low redshift . these partial optical depths are observables in the sense that high-@xmath12 and low-@xmath12 optical depth affect the large - scale @xmath0-modes over different ranges of multipoles . for example , compare the two reionization models in figure [ fig : cl ] . both have similar total optical depth , but in one case @xmath2 only comes from @xmath111 , resulting in more power at @xmath112 and less power at @xmath113 than the other model in which the optical depth primarily comes from @xmath114 . besides learning about the relative amount of ionization at @xmath115 and @xmath116 , the partial optical depth constraints also provide a way to empirically set the maximum redshift , @xmath44 . if some set of data are found to place a tight upper bound on @xmath105 for fairly conservative ( i.e. large ) values of @xmath106 and @xmath44 , then for analyses of future data this value of @xmath106 can be used as a new , lower value for @xmath44 since optical depth from higher redshifts is small . this approach assumes that there is essentially no reionization earlier than the original @xmath44 , but as long as this initial maximum redshift is taken to be large the presence or absence of high-@xmath12 ionization can be tested in this way over a wide range of redshifts . we explore this idea further in [ sec : mcmc ] . while we do not examine constraints on specific reionization models in this paper , the principal component approach is well suited to model testing . consider a model of the global reionization history , @xmath117 , parameterized by @xmath118 . this could be a simple toy model ( for example , instantaneous reionization where @xmath118 is a single parameter , the redshift of reionization ) or a more physical model where @xmath118 might include parameters that govern properties of the ionizing sources . for a particular choice of @xmath55 , the principal component amplitudes of the reionization model follow from equation ( [ eq : xetommu ] ) , giving a set of mode amplitudes that depend on the model parameters , @xmath119 . then constraints on the mode amplitudes from cmb polarization data ( defined using the same fiducial history ) can be mapped to constraints on the parameters of the reionization model . applying this method to a model that has as one of its parameters a maximum redshift allows one to obtain constraints on @xmath44 within a class of theoretical models . although obtaining good constraints on model parameters does not necessarily imply validation of the model class , this procedure provides a straightforward method by which different model classes can tested and falsified within a single analysis . generating monte carlo chains to find constraints on @xmath91 as described in the next section can be a somewhat time - consuming process , but it only needs to be done once per redshift range , after which any model of the global reionization history within this range @xmath80 can be tested using the same parameter chains . in cases where constraints from the data turn out to be close to gaussian , the covariance matrix of the principal components can be used in place of the full monte carlo chains , reducing the amount of information needed for model testing from @xmath120 numbers in chains of monte carlo samples to only @xmath121 numbers when @xmath60 eigenmodes are included in the chains [ @xmath60 mean values plus @xmath122 entries in the covariance matrix , @xmath123 . however , near - gaussian constraints on @xmath91 are likely to be possible only for certain realizations of future data , as discussed in the next section . we find reionization constraints from cmb polarization data using markov chain monte carlo ( mcmc ) techniques to explore the principal component parameter space ( see e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? chains of monte carlo samples are generated using the publicly available code cosmomc @xcite , which includes the code camb @xcite for computing the theoretical angular power spectrum at each point in the @xmath124 parameter space . we have modified both codes to allow specification of an arbitrary reionization history calculated from a set of mode amplitudes using equation ( [ eq : mmutoxe ] ) . the principal component amplitudes are the only parameters allowed to vary in the chains . since nearly all the information in @xmath6 from reionization is contained within the first few eigenmodes of @xmath4 , we include the modes @xmath124 in each chain with @xmath125 . we use only the @xmath0-mode polarization data for parameter constraints , and assume that the values of the standard @xmath126cdm parameters ( besides @xmath2 ) are fixed by measurements of the cmb temperature anisotropies . this leads to a slight underestimate of the error on @xmath2 as we discuss later , but to a good approximation the effect of @xmath4 on the large - scale @xmath0-modes is independent of the other parameters . specifically , we take @xmath127 , @xmath128 , @xmath129 ( corresponding to @xmath130 ) , @xmath131 , and @xmath132 , consistent with results from the most recent version of the _ wmap _ 3-year likelihood code . when computing the optical depth to reionization we take the primordial helium fraction to be @xmath133 . we also assume that @xmath97 , so that the optical depth contributed at lower redshifts is fixed at @xmath134 . the remaining total optical depth from reionization , @xmath98 , is determined by the values of @xmath91 for each sample in the chains . the default bin width for our fiducial models @xmath55 is @xmath135 , which is small enough that numerical effects related to binning should be negligible . to get accurate results from mcmc analysis , it is important to make sure that the parameter chains contain enough independent samples covering a sufficient volume of parameter space so that the density of the samples converges to the actual posterior probability distribution . for each scenario that we study , we run 4 separate chains until the gelman and rubin convergence statistic @xmath136 , corresponding to the ratio of the variance of parameters between chains to the variance within each chain , satisfies @xmath137 @xcite . the convergence diagnostic of @xcite is used to determine how much each chain must be thinned to obtain independent samples . both of these statistics and other diagnostic measures are computed automatically by cosmomc . since @xmath4 during reionization must match onto @xmath22 at @xmath138 and @xmath24 at @xmath139 , there are often sharp transitions at these redshifts since nothing in equation ( [ eq : mmutoxe ] ) forces @xmath4 to satisfy these boundary conditions . to avoid problems with the time integration in camb , we smooth the reionization history by convolving @xmath4 with a gaussian of width @xmath140 . as described in previous sections , we assume that @xmath109 between @xmath110 and @xmath44 . this assumption neglects helium reionization , which can make @xmath14 slightly larger than unity , and the small residual ionized fraction remaining after recombination that prevents @xmath14 from ever being exactly zero . these are relatively small effects , especially since we are not placing constraints on @xmath4 directly but rather on weighted averages of @xmath14 over redshift . the residual ionized fraction at @xmath58 is accounted for in the monte carlo exploration of reionization histories . in [ sec : wmap ] , we examine the current constraints from the 3-year _ wmap _ data . we then provide forecasts for principal component constraints with idealized , cosmic variance - limited , simulated data in [ sec : cvdata ] . in each case the likelihood computation includes only the @xmath0-mode polarization data , up to @xmath141 for simulated data and @xmath142 for _ wmap _ ; the likelihood code for _ wmap _ does not use @xmath6 at smaller scales due to low signal - to - noise . at multipoles @xmath143 the global reionization history only affects the amplitude of the angular power spectra , which we fix by setting @xmath144 constant in the monte carlo chains . a comparison of the mcmc results with the fisher matrix approximation follows in [ sec : fisher ] . in our analysis of _ wmap _ data we use the 3-year _ wmap _ likelihood code , with settings chosen so that likelihoods include only contributions from the low-@xmath5 @xmath0-mode polarization . in this regime , the code computes model likelihoods with a pixel - based method instead of using the angular power spectrum @xcite . since the maximum redshift at which there is still a significant ionized fraction is uncertain , we generate monte carlo chains using different fiducial models with @xmath145 . to avoid possible bias due to neglecting the possibility of high - redshift reionization , we focus here on the results obtained using the more conservative values of @xmath61 and 40 . the mcmc constraints on the first three principal components from _ wmap _ are shown in figure [ fig : mcmcwmap ] using fiducial models with @xmath61 and @xmath146 . the marginalized constraints are plotted within the physicality bounds of equation ( [ eq : physbounds ] ) , @xmath77 , so the size of the contours inside each box gives an idea of the constraining power of the data within the space of potentially physical models . for both fiducial histories , the data place a strong upper limit on @xmath101 and weakly constrain @xmath103 . it is important to note that although the parameters @xmath147 have the same names in both the @xmath61 and @xmath146 plots , they are defined with respect to different fiducial models and so we do not expect the contours in the left and right plots in figure [ fig : mcmcwmap ] to agree exactly . the qualitative similarity between the contours simply reflects the fact that the eigenfunctions for different @xmath44 have similar shapes ( figure [ fig : eigfn ] ) . since @xmath101 and @xmath2 are both averages of @xmath14 from @xmath110 to @xmath44 with more weight at high @xmath12 than low @xmath12 , the strong constraint on @xmath101 mostly reflects the ability of the data to constrain the total optical depth to reionization . indeed , @xmath101 and @xmath2 are strongly correlated in the monte carlo chains , while the correlations between @xmath2 and higher principal components of @xmath4 are weaker . relative to the physicality bounds , the constraints in figure [ fig : mcmcwmap ] are stronger for the @xmath146 chains than for @xmath61 ; this is to be expected since the physicality bounds on @xmath91 permit a greater variety of ionization histories , and in particular a wider range in @xmath2 , when @xmath44 is larger . the range of eigenmode amplitudes allowed by the limits of equation ( [ eq : physbounds ] ) is nearly independent of @xmath44 , but the effect of a unit - amplitude principal component on @xmath6 increases with @xmath44 as illustrated in figure [ fig : zmaxa ] . as described in [ sec : appl ] , principal component constraints from mcmc yield model - independent constraints on the total optical depth . for each monte carlo sample of the principal components we compute @xmath2 using equation ( [ eq : tau ] ) . the constraints on @xmath2 are listed in table [ tab : tau ] for various values of @xmath44 and @xmath60 , the number of modes in the monte carlo chains . the uncertainty in @xmath2 from the monte carlo chains is slightly underestimated because we fix all cosmological parameters that are not directly related to reionization . to estimate the effect of this assumption , we compare constraints on @xmath2 in two cases where instantaneous reionization is assumed instead of using the model - independent principal component approach : one in which the non - reionization parameters are fixed as in the rest of our monte carlo chains , and the other , from the 3-year _ wmap _ analysis of @xcite , in which these parameters are allowed to vary . these constraints , listed in rows 4 and 5 of table [ tab : tau ] , suggest that fixing @xmath126cdm parameters besides @xmath2 reduces @xmath148 by @xmath56 . llcclc & use pcs & & & fix other & + data & of @xmath4 ? & @xmath44 & @xmath60 & parameters ? & @xmath2 + wmap3 & yes & 20 & 3 & yes & @xmath149 + wmap3 & yes & 30 & 5 & yes & @xmath150 + wmap3 & yes & 40 & 5 & yes & @xmath151 + wmap3 & no & & & yes & @xmath152 + wmap3 & no & & & no & @xmath153 + cv & yes & 20 & 3 & yes & @xmath154 + cv & yes & 30 & 5 & yes & @xmath155 + cv & no & & & yes & @xmath156 + [ tab : tau ] in all cases where we fix the other cosmological parameters , the uncertainty in @xmath2 is similar regardless of whether we use principal components to explore a variety of reionization histories or restrict the analysis to instantaneous @xmath4 . this is somewhat surprising , since in general one would expect that expanding the model space for @xmath4 would increase the estimated error on @xmath2 . the physicality priors may be responsible for reducing @xmath148 slightly top - hat priors on @xmath91 induce a prior on @xmath2 that is flat over a certain range but falls off approximately linearly at the edges but even after accounting for priors , the optical depth constraint is robust to replacing the instantaneous reionization assumption with a model - independent analysis . the insensitivity of the constraint on @xmath2 to the set of models considered is partly due to the fact that the degeneracy in the eigenmode constraints is aligned in the direction of constant @xmath2 , as shown in figure [ fig : tauinst ] in the @xmath157 plane . the set of instantaneous reionization models , plotted as a curve in figure [ fig : tauinst ] , cuts across the _ wmap _ constraints on more general models in a region of high posterior probability where the distribution of samples varies slowly along lines of constant @xmath2 . this large overlap between the general reionization histories favored by the data and the line of instantaneous models is the main reason why the probability distributions @xmath158 , plotted in the right panel of figure [ fig : tauinst ] , are similar for the two classes of models . [ the sharp cutoff at low @xmath2 in the instantaneous reionization @xmath158 comes from our @xmath97 prior . ] note that the fact that the instantaneous reionization curve passes through the middle of the 68% confidence region also indicates that models with rapid reionization are not at all disfavored by the 3-year _ wmap _ data . we use two methods to find the posterior distribution for @xmath2 in the instantaneous reionization case : one is the usual approach of varying the optical depth ( or reionization redshift ) in a monte carlo chain , and the other involves computing @xmath158 for a subset of samples from chains in which principal components of @xmath4 are varied , selecting only those samples with @xmath91 values close to the 1d instantaneous reionization curve . both approaches produce consistent probability distributions ; the former method is used for @xmath158 plotted in figure [ fig : tauinst ] . the weak @xmath103 constraint visible in figure [ fig : mcmcwmap ] suggests that in addition to determining the total optical depth , _ wmap _ data may provide useful limits on high-@xmath12 and low-@xmath12 optical depth as defined in the previous section . the 68 and 95% posterior probability contours for the partial optical depths from _ wmap _ are shown in figure [ fig : tauwmap ] for @xmath159 and @xmath160 . the @xmath161 physicality prior cuts off the contours at @xmath162 ; in both panels , the upper limits set by @xmath163 are outside the plotted area , so the contours are not strongly influenced by those priors . the error `` ellipses '' are narrowest in the direction along which the total optical depth is constrained , as shown by dashed lines of constant total optical depth at @xmath164 and @xmath165 [ approximately the upper and lower 1 @xmath166 limits from the 3-year _ wmap _ analysis of @xcite ] . comparison of the two panels in figure [ fig : tauwmap ] reveals that the choice of @xmath44 does not have a large effect on constraints on the optical depth above and below @xmath167 . this suggests that @xmath168 should be interpreted as @xmath169 . we provide further justification for this interpretation in [ sec : cvdata ] . the 95% upper limits on the high - redshift optical depth , @xmath169 , are @xmath170 and @xmath171 for @xmath61 and 40 , respectively , after marginalizing over all other parameters . this is not a particularly strong constraint , since this result is only marginally inconsistent with all of the optical depth from reionization coming from @xmath1 , but with future data it should be possible to either reduce the upper limit on high-@xmath12 optical depth or to detect the presence of a substantial ionized fraction at high redshift ( see [ sec : cvdata ] ) . models of reionization can be tested by computing the principal components of proposed ionization histories and comparing with constraints on @xmath91 from monte carlo chains . the _ wmap _ constraints in figure [ fig : mcmcwmap ] are non - gaussian , in part because the physicality priors on @xmath91 intersect the posterior probability contours where the likelihood is large . because the constraints do not have a simple gaussian form , it appears that the full parameter chains are necessary for accurate model testing , although the viability of various models can be estimated by comparing their eigenmode amplitudes with marginalized constraints such as those in figure [ fig : mcmcwmap ] , keeping in mind that models favored by marginalized constraints could be disfavored in the full @xmath60-d parameter space . to forecast how well principal components of the reionization history could be measured by future cmb polarization experiments , we repeat the analysis of [ sec : wmap ] using simulated realizations of the full - sky , noiseless @xmath0-mode angular power spectrum instead of _ wmap _ data . any real experiment will of course involve sky cuts , noise , foregrounds , and other complications , so the results presented here represent an optimistic limit on what we can learn about the global reionization history from low-@xmath5 @xmath0-mode polarization . at the end of this section , we estimate how much the constraints might be degraded from this idealized case for an experiment with characteristics similar to those proposed for the _ planck _ satellite . we generate simulated realizations of @xmath6 drawn from @xmath172 distributions with cosmic variance determined by the theoretical angular power spectra that we compute using camb . for the @xmath173th sample in a chain , the likelihood including only @xmath0-mode polarization is @xmath174 where @xmath175 is the spectrum of the simulated data and @xmath176 is the theoretical spectrum calculated with the parameter values at step @xmath173 in the chain . we have run monte carlo chains for multiple realizations of @xmath6 drawn from spectra computed assuming a variety of `` true '' reionization histories , @xmath177 . we start by taking the true history to be a model with nearly instantaneous reionization and @xmath178 . as a contrasting model for comparison we also use an extended , double reionization model with @xmath179 . figure [ fig : cl ] shows @xmath6 and @xmath4 for each of these models . since parameter constraints derived using a single draw of @xmath6 from the underlying power spectrum may contain features unique to that realization , we generated monte carlo chains for 10 random realizations of the instantaneous reionization power spectrum . two of these realizations are plotted as points in figure [ fig : clcv ] , along with the theoretical spectrum with cosmic variance bands . the thick curves in figure [ fig : clcv ] are spectra of the best - fit models from the monte carlo chains for each realization . the overall best - fit models are similar for the two realizations and both agree closely with the theoretical @xmath6 . the dashed curve in the right panel of figure [ fig : clcv ] shows an `` alternative '' best - fit model that we discuss later . the 2d marginalized mcmc constraints on principal components for these two realizations are shown in figure [ fig : mcmccv ] . the plotted regions are bounded by the physical top - hat priors on @xmath91 as in figure [ fig : mcmcwmap ] , but the eigenmodes are different here with @xmath62 instead of 30 or 40 for the fiducial history . since we have constructed @xmath177 to have no ionization at @xmath180 , we know that @xmath62 should be a large enough maximum redshift ; this choice of @xmath44 also reflects the fact that improved empirical determination of @xmath44 will allow it to be set at lower redshift for future data analysis , assuming that high-@xmath12 reionization is not detected . the contours in figure [ fig : mcmccv ] show that cosmic variance - limited data can constrain models within the physically allowed parameter space much better than current data can . the chains used for the results shown here vary the first 3 eigenmodes ; by running chains with @xmath181 we find that cosmic variance - limited data can provide @xmath182 constraints that are tighter than the physicality bounds for the first 4 eigenmodes of a @xmath62 fiducial model and for at least 5 eigenmodes when @xmath61 . although the only difference in the mcmc setup for the left and right panels of figure [ fig : mcmccv ] is the realization of @xmath6 ( fig . [ fig : clcv ] ) , there are some significant differences in the resulting eigenmode constraints , particularly in the @xmath183 plane . this kind of variation is typical among the realizations of simulated data , and is related to whether features in the theoretical @xmath6 remain intact with the inclusion of cosmic variance . the relevant features in the spectrum are the small oscillations in the low - power `` valley '' in @xmath6 on scales @xmath184 . in the example shown here , the feature that matters most is the bump in the theoretical spectrum at @xmath185 . in the left panel of figure [ fig : clcv ] , the @xmath6 realization retains this bump , while in the right panel the bump is washed out by cosmic variance . the result is that the left realization is better able to pick out those reionization models in monte carlo chains that reproduce the true @xmath6 , leading to the tight constraints on the left side of figure [ fig : mcmccv ] . the realization in the right panel of figure [ fig : clcv ] lacks this important `` fingerprint '' for identifying models that match the theoretical @xmath6 , so the data allow a wider variety of models as reflected in the constraints on the right side of figure [ fig : mcmccv ] . in this case , there are two best - fit models corresponding to the two @xmath186 contours separated in the @xmath103 direction . the spectrum of the overall best - fit model ( with smaller @xmath103 ) , which is close to the theoretical spectrum , is plotted as a thick solid curve in the right panel of figure [ fig : clcv ] , while the other best - fit model ( at larger @xmath103 ) is plotted as a dashed curve . these two models have significantly different spectra , but because of the scatter in the random draw of @xmath6 they are each able to fit the data better at some multipoles and worse at others in such a way that the high-@xmath103 best - fit model is only a slightly worse fit than the low-@xmath103 best - fit model . among the realizations of simulated data that we explored with mcmc methods , constraints like those in the right panel of figure [ fig : mcmccv ] are a fairly extreme case , occurring about 10 - 20% of the time . in general the realizations form a sort of continuum between the two presented here , with about half showing some displacement of the @xmath187 contour towards larger @xmath103 but not as much as in the second realization that we show as an example . the extent to which differences between realizations show up in principal component constraints depends on @xmath177 . for example , the @xmath6 for the double reionization model ( fig . [ fig : cl ] ) have more pronounced oscillations at @xmath188-@xmath189 than the instantaneous reionization spectrum , so they are not as easily erased by cosmic variance or noise and the resulting principal component constraints tend to be more consistent from one realization to the next and more like the left panel of figure [ fig : mcmccv ] . the differences between constraints from various realizations of simulated data suggest the interesting possibility that our ability to learn about the reionization history from large scale @xmath0-mode polarization may ultimately depend on the luck of the draw of @xmath6 at our particular vantage point . in an unlucky draw , cosmic variance can distort subtle features in the power spectrum in a way that would limit constraints on reionization eigenmodes to be worse than we might expect from the fisher approximation or a more typical draw . the good news is that even a realization like the one in the right panels of figures [ fig : clcv ] and [ fig : mcmccv ] would permit constraints that are far better than what is currently possible . it is also important to note that although such constraints are weaker than in the best - case scenario , they are still consistent with the true parameter values and in general would not lead us to rule out the true reionization history based on a principal component analysis of the data . as with the principal components of @xmath4 , mcmc analysis indicates that constraints on the total optical depth to reionization would be greatly improved with cosmic variance - limited data : @xmath190 , down a factor of six from the 3-year _ wmap _ value , @xmath191 . the constraints on @xmath2 using the realization of instantaneous reionization @xmath6 in the left panel of figure [ fig : clcv ] are listed in table [ tab : tau ] for two fiducial models with @xmath62 and 30 . optical depth constraints from the other draw of @xmath6 are similar , so estimates of @xmath2 and @xmath148 appear not to be biased for atypical realizations . it makes sense that differences in realizations do not significantly affect @xmath2 since the optical depth constraint comes mainly from the peak in @xmath6 on the largest scales whereas differences in principal component constraints between realizations arise from the low - power part of the spectrum at higher @xmath5 . the model - independent constraints on @xmath2 are neither biased nor significantly weaker relative to the constraint obtained assuming instantaneous reionization ( bottom row in table [ tab : tau ] ) , as is true for the _ wmap _ data . this should not be surprising since we have assumed that @xmath177 is an instantaneous reionization model . however , if @xmath177 is in fact very different from the instantaneous model , the estimate of @xmath2 obtained under the assumption of instantaneous reionization will be incorrect . for example , if we take @xmath177 to be the double reionization model of figure [ fig : cl ] , the model - independent approach yields an optical depth constraint consistent with the true value of @xmath179 . on the other hand , mcmc analysis restricted to instantaneous reionization histories gives a significantly biased estimate , @xmath192 . in this case , the 1d curve of instantaneous reionization models in the space of eigenmodes never lies near the principal component values favored by the data , so any analysis that only considers such models would never find a good fit to the data . this illustrates the importance of a model - independent approach : although results from current data may not be significantly affected by the assumption of a specific form of @xmath4 , future data will be sensitive enough to the reionization history that the choice of a specific model can greatly impact the results @xcite . though constraints on the contributions to @xmath2 from high and low @xmath12 are currently limited to weak upper bounds , future polarization data will likely enable more definitive detections of high-@xmath12 reionization if it is present , or tighter upper limits if absent . the left and right panels of figure [ fig : taucv1 ] show these constraints from monte carlo chains using the @xmath6 plotted in the left and right panels of figure [ fig : clcv ] , respectively . the true ionization history has zero ionized fraction at @xmath114 , and the mcmc analysis places a 95% cl upper limit on @xmath193 of @xmath194 . figure [ fig : taucv2 ] shows mcmc constraints on the same partial optical depths using simulated data where the true model is instead taken to be the extended , double reionization history plotted as a thin curve in the inset of figure [ fig : cl ] . in this case , most of @xmath98 comes from @xmath114 , and using the principal components of @xmath4 the high-@xmath12 optical depth can be detected at a high level of confidence and measured to an accuracy of @xmath89 . note that the constraints in both figures [ fig : taucv1 ] and [ fig : taucv2 ] are consistent with the _ wmap _ constraints in figure [ fig : tauwmap ] . as with _ wmap _ data , most of the constraints from simulated data in figures [ fig : taucv1 ] and [ fig : taucv2 ] do not appear to depend strongly on the choice of @xmath44 . for the instantaneous reionization model used for figure [ fig : taucv1 ] , this is to be expected since we know that the simulated @xmath6 have no contribution from @xmath195 . each set of contours in figure [ fig : taucv1 ] is consistent with the values of @xmath196 and @xmath197 for @xmath177 . the constraints in the second panel in figure [ fig : taucv1 ] extend to larger high-@xmath12 optical depth than those in the first panel because the effect of the physicality priors is not as strong at larger @xmath44 , as noted in [ sec : wmap ] ; the leftmost @xmath62 contours would stretch to larger @xmath198 and smaller @xmath196 if we did not apply the physicality bounds . the two panels on the right side of figure [ fig : taucv1 ] show weaker constraints on the partial optical depths because the principal component constraints for that realization of @xmath6 are weaker . in the extended , double reionization model used for figure [ fig : taucv2 ] , reionization actually starts at @xmath199 , so the choice of @xmath62 is not appropriate for this model . the result of such an error is that the constraint on @xmath198 is inconsistent with the true value , marked by a cross in the left panel of figure [ fig : taucv2 ] . the @xmath61 constraints in the right panel are consistent with the expected value , although not in perfect agreement due to cosmic variance . it is interesting that although the value chosen for @xmath44 in the left panel is too low for this model , the constraints on @xmath193 for @xmath62 and @xmath61 are still consistent with each other . this supports the interpretation of constraints on @xmath105 as @xmath200 in [ sec : wmap ] while the cosmic variance - limited simulated data are useful for determining how well the reionization history could possibly be constrained by large - scale @xmath0-mode polarization measurements , it is also interesting to ask how well we can do with future experiments that fall somewhat short of the idealized case that we have considered so far . in particular , the upcoming _ planck _ satellite is expected to improve our knowledge of the large - scale @xmath0-mode spectrum substantially @xcite ; what does this imply for constraints on @xmath4 ? to estimate what might be possible with _ planck _ data , we assume that after subtracting foregrounds a single foreground - free frequency channel remains for constraining the low-@xmath5 @xmath0-mode polarization . we take this to be the 143 ghz channel with a white noise power level of @xmath201k@xmath202 and beam size @xmath203 , and we assume that the sky coverage is @xmath204 after cutting out the galactic plane @xcite . we compute the likelihood of monte carlo samples using the routines provided in cosmomc and analyze parameter chains with the principal component method as described for _ wmap _ and cosmic variance - limited data . for various choices of @xmath55 and @xmath177 we find that going from full - sky cosmic variance - limited data to _ planck_-like data increases the uncertainty in @xmath4 principal components , @xmath205 , and in the optical depth , @xmath148 , by roughly a factor of two . based on these results , it should be possible to constrain about three eigenmodes at @xmath206 within the space of physical models , and we expect @xmath2 to be determined to an accuracy of about @xmath207 . this is consistent with previous estimates of the _ planck _ optical depth uncertainty when considering more general reionization models than instantaneous reionization ( e.g. , * ? ? ? * ; * ? ? ? the fisher matrix analysis of [ sec : eigenmodes ] predicts that the principal components of @xmath4 should be uncorrelated , with errors given by @xmath205 from equation ( [ eq : fisher2 ] ) ( in the idealized limit of full sky , noiseless observations ) . these characteristics rely on the assumption that the difference between the true and fiducial reionization histories is small : @xmath208 . clearly this assumption will not hold for any single fiducial history if one wants to consider a variety of possible true histories , and the consequence is that @xmath209 differs from @xmath210 . as an example of the difference between the fisher approximation and the full monte carlo analysis , in figure [ fig : fisher2d ] the marginalized constraints in the @xmath157 plane from the left panel of figure [ fig : mcmccv ] are compared with the fisher matrix error ellipses , centered on the true @xmath211 for the same choices of @xmath177 ( instantaneous reionization with @xmath178 ) and @xmath55 ( constant @xmath212 out to @xmath62 ) . the mcmc constraints have correlated values of @xmath101 and @xmath103 with somewhat larger uncertainties than the fisher matrix @xmath213 . the eigenfunctions , @xmath37 , are constructed to have orthogonal effects on @xmath6 in the vicinity of the fiducial model , but for large @xmath75 the orthogonality breaks down . fortunately the errors on @xmath2 remain largely unaffected because the @xmath214 correlation is oriented such that the increased errors are along the direction of constant total @xmath2 ( see figure [ fig : tauinst ] ) . one can compute new eigenfunctions by evaluating @xmath215 at @xmath177 and see that their dependence on redshift differs from the original @xmath37 . it is possible to decorrelate the principal components by diagonalizing the covariance matrix from the monte carlo chains and rotating the eigenmodes into the new basis . however , this procedure is unnecessary for most applications ; the most important property of the principal components for constraining @xmath4 is that the first few modes form a complete basis for @xmath6 on large scales , accurate within cosmic variance limits . completeness of the eigenmodes holds true even if they are not exactly orthogonal . finally , note that if the true ionization history and the draw of @xmath6 permit constraints on principal components similar to those in the left panel of figure [ fig : mcmccv ] , then these constraints may be close enough to gaussian so that the covariance matrix would be sufficient to describe the information in the @xmath6 for finding constraints on reionization models . however , if what we see looks more like the right panel of figure [ fig : mcmccv ] then the full chains of monte carlo samples may be necessary for further applications even if we can measure the @xmath0-mode polarization perfectly . observations of the large - scale @xmath0-mode polarization of the cmb in the near future are expected to yield new information about the spatially - averaged reionization history of the universe . the principal components of the reionization history are a promising tool for extracting as much of that information as possible from the data . we have shown that the principal component method can be usefully applied to real , currently available data , and forecasts from simulated data suggest that there is room to substantially improve constraints on the reionization history using this method as measurements of the large - scale @xmath0-modes improve . we find that the key features of the principal component analysis put forward by @xcite continue to apply when we go from the fisher matrix approximation to an exploration of the full likelihood surface using markov chain monte carlo methods . for fairly conservative choices of the maximum redshift of reionization ( @xmath216 - 40 ) , only the first five principal components at most are needed for a complete representation of the @xmath0-mode angular power spectrum to within cosmic variance . to account for arbitrary reionization histories in the analysis of cmb data , only a few additional parameters must be included in chains of monte carlo samples if those parameters are taken to be the lowest - variance principal components of @xmath4 . specific models of reionization can be tested easily by computing their eigenmode amplitudes and comparing with constraints on the eigenmodes from the data . constraints on derived parameters , such as total optical depth or the optical depth from a certain range in redshift , represent other applications of the mcmc constraints on principal components of @xmath4 . often , estimates of the optical depth to reionization are computed assuming instantaneous reionization or some other simple form for @xmath4 . here we extend the analysis of the 3-year _ wmap _ data to allow a more general set of models of the global reionization history . we find that expanding the model space does not significantly widen the uncertainty in @xmath2 beyond the instantaneous reionization value of @xmath217 . robust @xmath2 constraints are important for tests of the dark energy based on the growth of structure since they control the uncertainty on the amplitude of the initial spectrum . moreover , even with current data the principal component constraints are beginning to show the possibility of determining properties of reionization in addition to @xmath2 . by comparing the optical depth from high @xmath12 with that from low @xmath12 , we obtain an upper limit on the contribution to the optical depth from high redshift : @xmath218 at 95% confidence , assuming that there is no significant episode of reionization at @xmath219 . due to the limitations of noise and foreground contamination , only the first two eigenmodes of the reionization history , @xmath101 and @xmath103 , can be determined with present polarization data to any reasonable degree of accuracy . constraints on these two modes come primarily from the main , broad peak in @xmath0-mode power at low @xmath5 from reionization . as measurements of the @xmath0-mode polarization improve , for example from additional _ wmap _ data or through planned future experiments such as _ planck _ , better knowledge of the low - power `` trough '' in @xmath6 between the main reionization peak and the first acoustic peak should enable constraints on the third and higher principal components , up to about @xmath220 for near cosmic variance - limited data . since constraints on these higher modes rely on the ability to identify subtle features in the trough of @xmath6 , the ultimate accuracy to which the eigenmodes can be determined may depend on whether or not the necessary features are well reproduced in the particular random draw of @xmath6 that is available to us . however , even if we are unlucky enough to have a realization in which some of the important features of the spectrum are washed out by randomness , it should still be possible to measure several of the principal components to better accuracy than is currently possible . knowledge of the @xmath221 eigenmodes of @xmath4 , along with improved constraints on @xmath101 and @xmath103 , will allow more stringent tests of reionization models and a better understanding of the global reionization history . : we thank cora dvorkin , gilbert holder , dragan huterer , hiranya peiris , and jochen weller for useful discussions . mjm was supported by a national science foundation graduate research fellowship . wh was supported by the kicp through the grant nsf phy-0114422 , the doe through contract de - fg02 - 90er-40560 and the david and lucile packard foundation . , a. , bernstein , g. , cahn , r. , freedman , w. l. , hewitt , j. , hu , w. , huth , j. , kamionkowski , m. , kolb , e. w. , knox , l. , mather , j. c. , staggs , s. , & suntzeff , n. b. 2006 , arxiv astrophysics e - prints astro - ph/0609591 , l. , hinshaw , g. , komatsu , e. , nolta , m. r. , spergel , d. n. , bennett , c. l. , barnes , c. , bean , r. , dor , o. , halpern , m. , hill , r. s. , jarosik , n. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , peiris , h. v. , tucker , g. s. , verde , l. , weiland , j. l. , wollack , e. , & wright , e. l. 2006 , arxiv astrophysics e - prints astro - ph/0603450 , d. n. , bean , r. , dor , 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 , arxiv astrophysics e - prints astro - ph/0603449	on large angular scales , the polarization of the cmb contains information about the evolution of the average ionization during the epoch of reionization . interpretation of the polarization spectrum usually requires the assumption of a fixed functional form for the evolution , e.g. instantaneous reionization . we develop a model - independent method where a small set of principal components completely encapsulate the effects of reionization on the large - angle @xmath0-mode polarization for any reionization history within an adjustable range in redshift . using markov chain monte carlo methods , we apply this approach to both the 3-year _ wmap _ data and simulated future data . _ wmap _ data constrain two principal components of the reionization history , approximately corresponding to the total optical depth and the difference between the contributions to the optical depth at high and low redshifts . the optical depth is consistent with the constraint found in previous analyses of _ wmap _ data that assume instantaneous reionization , with only slightly larger uncertainty due to the expanded set of models . using the principal component approach , _ wmap _ data also place a 95% cl upper limit of 0.08 on the contribution to the optical depth from redshifts @xmath1 . with improvements in polarization sensitivity and foreground modeling , approximately five of the principal components can ultimately be measured . constraints on the principal components , which probe the entire reionization history , can test models of reionization , provide model - independent constraints on the optical depth , and detect signatures of high - redshift reionization .
twenty years after the discovery @xcite of the fractional quantum hall effect ( fqhe ) dyakonov @xcite wrote a rather sharp critique of the present state of its theory . among his criticisms were the absence of any simple or beautiful first - landau - level ( fll ) formulation of the hierarchy wave functions ( @xmath7 ) , the lack of an explicit representation of conjugate states ( e.g. , fillings where @xmath8 , so @xmath0 = 2/3 , 3/5 , etc . ) , the absence of a justification for wave functions in terms of underlying principle such as minimizing the interaction energy , and the lack of a sound theoretical foundation for concepts such as composite fermions ( cfs ) , which he argued had been neither derived nor even adequately defined . while recognizing the phenomenological success of jain s wave function engineering " from which the notion of cfs @xcite derives , he argued that far too many simple questions about the fqhe remain without reasonable answers . here we address some of dyakonov s concerns by providing an explicit mapping from the strongly - correlated - electron form of fqhe wave functions to a quasi - electron ( or cf ) form , yielding wave functions algebraically identical to those of the noninteracting integer qhe . this is done for both the sphere and plane . the quasi - electrons have a simple analytic form and are eigenstates of the angular momentum operators @xmath9 and @xmath10 , and also carry a label @xmath5 related to haldane s @xmath11-wave pseudopotential @xcite . the quasi - electrons occupy fine - structure fll sub - shells distinguished by the same label . the wave functions for @xmath12 and @xmath13 correspond to configurations where @xmath11 sub - shells are fully filled by their respective quasi - electrons . the sub - shell structure is induced by the coulomb interaction , with the gap between neighboring sub - shells reflecting the energy cost of removing one @xmath11-wave coupling between quasi - electron 1 and one of its @xmath14 neighbors . we discuss the relevance of the construction to a number of current questions about the fqhe , including the relationship between cf and hierarchical descriptions of wave functions @xcite ; the structure of the @xmath6 state @xcite , where we argue there exist alternative majorana and pseudo - dirac descriptions associated with the symmetries at this filling , and where we identify an operator quite similar to a pfaffian ; and possible systematic improvements of wave functions , in the spirit of an effective theory , that may help address certain open questions about the fqhe . jain constructed his 2/5 , 3/7 , 4/9 , ... hierarchy wave functions by first operating with a multiply - filled integer quantum hall state on the half - filled symmetric state , producing a wave function spread over multiple higher landau levels , which then was projected numerically onto the fll , eliminating the unwanted components . when tested against numerical solutions obtained by diagonalizing the coulomb interaction , excellent overlaps were found , comparable to those for laughlin s @xcite @xmath0= 1/3 and 1/5 wave functions . however , as dyakonov discusses , the jain construction is troubling on several grounds . laughlin s @xmath0= 1/3 and 1/5 states are supported by certain variational arguments . for example , at short - distances the wave functions vary as @xmath15 and @xmath16 , eliminating the most repulsive multipoles of the electron - electron coulomb interaction @xmath17 . jain s construction makes no reference to the electron - electron interaction , but instead employs an operator taken from the noninteracting integer quantum hall effect ( iqhe ) , with electrons occupying higher lls characterized by large magnetic gaps , which at the end are eliminated numerically a procedure dyakonov termed bizarre . " laughlin s wave function can be expressed analytically as simple products of closed - shell operators , while jain s final result has no such representation . laughlin s construction is based on the rescaling of all inter - electron correlations , for @xmath18 , by factors of @xmath19 , @xmath20 , a procedure that reflects the scale invariance of the underlying coulomb potential . in 1996 ginocchio and haxton @xcite ( gh ) generalized this approach to successively larger groups of electrons : recognizing that higher density defects " would necessarily arise beyond @xmath21 beyond this filling @xmath11-wave electron pairs must start to appear they introduced operators to create such defects , then looked for solutions that would distribute these over - dense regions uniformly . on the sphere the closed - shell operators laughlin employed can be labeled by the quantum number @xmath22 , where @xmath23 is the electron number . the gh construction produces a larger class of such scalar operators , defined by two quantum numbers @xmath22 and @xmath24 . the gh and laughlin operators coincide for @xmath24=0 . the gh operators generate the full set of hierarchy states ( the jain states ) when @xmath25 is allowed to run , constrained by @xmath26 ; and they generate hierarchy states and their conjugates when @xmath27 are varied , without constraints . these quantum numbers are related to the electron number by @xmath28 . the gh operators were later used by jain and kamilla @xcite , but otherwise have not been broadly applied . one reason may be the limitation to the sphere : gh used this geometry because spherical @xmath29-electron wave functions generated from scalar operators have total angular momentum @xmath30 , guaranteeing both translational invariance and homogeneity ( uniform one - body density ) . but most investigators work in the plane , where simple polynomials replace angular momentum couplings . second , while the gh treatment of higher - order electron correlations led to operators with a manifest sub - shell structure , implying a fine - structure splitting of the fll , this structure was not obvious in the final wave functions : the gh operators contain spherical tensor derivatives ( operators that destroy magnetic flux ) that can be evaluated analytically , but a compact form of the scalar wave function with all such derivates eliminated by such evaluation was not provided . this shortcoming complicates the construction of analogous wave functions in the plane , as derivatives on the sphere are associated with raising and lowering operators that operate only within finite hilbert spaces , unlike the case of the plane . although the gh construction addresses two issues dyakonov raised producing an analytic generalization of laughlin s construction with the same variation justification , and generating the full sets of hierarchy and conjugate states the final wave functions lack the elegant simplicity of laughlin s results . gh , in fact , did not seek the most simple application of their operators , as their goal was to account for jain s surprising construction . thus they applied their @xmath31 operators as jain did his iqhe operators , acting directly on the half - filled shell . this procedure yields results numerically identical to jain s , and demonstrates that correlations within the fll generate an su(2 ) sub - shell algebra distinct from but algebraic identical to that of multiply filled iqhe states . this algebraic similarity accounts for jain s success , though his construction misses the connection between broken scalar pairs and the generation of angular momentum that is responsible for the gh fll sub - shell structure . in this paper we discuss an alternative use of gh operators , as creators of quasi - electrons . the resulting gh@xmath32 hierarchy and conjugate states are closed - shell configurations of families of quasi - electrons , all of which have a common form as tensor products of two spinors , one creating an electron and one creating a single unit of magnetic flux . as the construction eliminates all gh derivatives , wave functions can be readily expressed in either spherical or planar geometry . the gh@xmath32 quasi - electron wave function of filling @xmath33 and magnetic field strength @xmath34 is identical in form to the iqhe electron wave function of the same electron number , but in a reduced magnetic field @xmath35 . the plan of this paper is as follows . in sec . ii we review properties of laughlin s wave functions , providing benchmarks for subsequent discussions of other hierarchy states . we use the laughlin wave function to illustrate how planar operators can be written as analogs of spherical ones , and use this mapping to define states of uniform density in the plane otherwise an ill - defined concept , in our view . this involves reorganizing the usual planar degrees of freedom into single - particle and schur - polynomial spinors , with the latter representing the addition of a unit of magnetic flux . we describe how vector products can be formed in the plane to produce quasi - electrons with good angular momentum , and consequently how to generate planar scalars " that are both translationally invariant and uniform in density . in sec . iii we discuss the gh operator construction in more detail than was possible in the original letter @xcite . we describe the @xmath31 form , connected with electron correlations , and the @xmath36 form , connected with fll shell fine structure . the two representations allow one to understand the connections between fll electron correlations , energy minimization , and sub - shell structure . in section iv we show how the gh operators can be used to produce hierarchy and conjugate states in their gh@xmath32 form , noninteracting quasi - electrons occupying filled sub - shells . we describe properties of the quasi - electrons , then properties of the low - momentum representation of the many - electron states that can be built as closed - shell configurations of the quasi - electrons . we illustrate the novel properties of the sub - shell structure , which is not static but evolves as electrons are added , considering various trajectories " in the two - parameter hilbert space , such as fixed @xmath0 with increasing @xmath29 , or fixed magnetic field strength with increasing @xmath29 . we describe symmetries associated with in the exchange of the particle and flux spinors for the @xmath6 case , where the quasi - electrons exhibit special symmetries : we find majorana - like and dirac - like solutions at @xmath6 , the latter associated with spin flip . we also show the connection to the pfaffian . we evaluate the overlaps of gh@xmath32 wave functions with results from exact diagonalizations of the coulomb interaction . the concluding section v includes a discussion of issues for further study . the formalism is suggestive of an effective theory , and we remark on work that could be undertaken to explore this possibility , including potential connections to states at fillings like @xmath37 . in the appendix , we present a more technical discussion of correlations , to contrast the current construction ( which is guided by the scale invariance of the coulomb potential ) with alternative variational schemes focused of the short wave behavior of wave functions , e.g. , some generalization of the haldane pseudopotential for interactions among multiple electrons . * single electron states in the plane : * the hamiltonian for an electron moving in a plane under the influence of a perpendicular magnetic field @xmath38 is @xmath39 where @xmath40 is the electron mass and @xmath41 is the electron charge . we take the direction of electron rotation to be clockwise in the @xmath42 plane ( as viewed in a right - handed coordinate system from positive @xmath43 ) , a choice that requires b to be negative the field points in the @xmath44 direction . thus @xmath45 is positive . in the symmetric gauge we employ @xmath46 . * single electron states on the sphere : * the original gh operator construction was done on the sphere , a fqhe geometry introduced by haldane @xcite : the electrons move on the sphere s surface under the influence of a radial magnetic field generated by a dirac monopole at the origin . this geometry provides two advantages . first , in contrast to the plane , there is both a defined surface area and a fixed number of fll single - particle states , determined by the number of monopole quanta . thus densities and fractional fillings can be defined unambiguously . second , on the sphere many - electron states with total @xmath9=0 are both homogeneous uniform density over the sphere and translationally invariant ( displacements generated by @xmath59 and @xmath60 ) @xcite . states with @xmath30 can be constructed from single - particle spinors of good @xmath9 , @xmath10 ( rotation matrices ) , using standard angular momentum coupling methods . dirac s monopole quantization condition requires the total magnetic flux through the sphere of radius @xmath61 to be an integral multipole of the elementary flux @xmath62 , @xmath63 . the hamiltonian for an electron confined to the sphere is @xmath64 where @xmath65 is the dynamical angular momentum , @xmath66 is the cyclotron frequency , and @xmath67 where @xmath68 . the angular momentum operators @xmath69 satisfy the commutation relations @xmath70=i \epsilon_{ijk } l_k$ ] . as @xmath71 is normal to the surface while @xmath72 is radial , @xmath73 and @xmath74 . these relations give @xmath75 , where @xmath76 . consequently the eigenvalues corresponding to the landau levels are @xmath77 the single - particle wave functions are the wigner d - functions @xmath78 thus there are @xmath79 degenerate single - particle states in the fll , @xmath80 in the second ll , etc . the fll wave functions can be written as a monomial of power @xmath81 in the elementary spinors @xmath82 , @xmath83^{1/2 } u_{1/2}^{s+m } u_{-1/2}^{s - m}\equiv [ u]^{s}_m \label{eq : us}\ ] ] where @xmath84 + * jain s operator in spherical notation : * as we will discuss in sec . iii , jain used the antisymmetric iqhe state consisting of the lowest @xmath85 shells , fully occupied , as an operator , acting on the half - filled shell . thus @xmath9 in eq . ( [ eq : d ] ) runs over @xmath86 , where @xmath24 is an integer or half integer , @xmath87 . defining @xmath88 , so that @xmath89 can also be an integer or half integer , the single - electron spinors ( eq . ( [ eq : d ] ) ) for these shells are @xmath90 note that @xmath91 . there are @xmath92 allowed values of @xmath93 . + * the interacting problem : * the planar hamiltonian responsible for the fqhe effect is obtained by adding the electron - electron coulomb interaction to the n - electron version of eq . ( [ eq : ham0 ] ) , as well as a uniform neutralizing electrostatic background field @xmath94 @xmath95 on the sphere , @xmath96 can be identified with the chord separation of electrons @xmath97 and @xmath98 . the many - electron hilbert space consists of slater determinants form from these single - particle wave functions defined above . the degeneracy among these states is the broken by the interaction . in the case of the sphere , as the fll single - particle basis is of dimension @xmath79 , the fraction of the single particles states filled is @xmath99 . the fractional filling @xmath0 of a series of related states , such as the laughlin @xmath100 series discussed below , is defined as the large-@xmath29 limit of this ratio . + * plane - sphere relationships and scalar contractions : * on the sphere many - body states of definite total @xmath30 are both rotationally invariant ( that is , invariant under small displacements along the sphere s surface ) and homogeneous ( uniform one - electron density ) . in order to generalize the gh spherical construction to the plane , it is important to find a procedure for generating analogous scalar states on the plane . such states will then be automatically translationally invariant and can also be considered homogeneous , as we discuss below . this requires an analog of the spherical tensor product , in which objects of definite rotational symmetry are combined to produce new objects with such symmetry , @xmath101_{l_0 m_0}.\ ] ] the spherical and planar geometries are related by the mapping of the three - dimensional rotation group into the two - dimensional euclidean group @xmath102 the correspondence between @xmath103 and @xmath104 means that the scalar product we seek will automatically produce translationally invariant states in the plane . ( that is , the slater determinants one forms will have polynomials that depend only on coordinate differences , while the center - of - mass of the @xmath29-electron system will be in the lowest harmonic oscillator state . consequently , while the wave functions technically involve @xmath105 spatial coordinates , in fact they can be considered functions of just @xmath106 intrinsic coordinates . ) while we have noted that the notion of a homogeneous finite-@xmath29 state in the plane is ambiguous the electrons are not strictly confined to any definite area - the mapping between sphere and plane can be used to define homogeneity : a planar state is homogeneous if it corresponds to a homogeneous spherical state under this mapping . one can define a scalar product between two such planar vectors of the same rank @xmath57 by @xmath113^{k } \odot [ z_2]^{k } = \sum_{i =- k}^{k}~(-1)^{k - i}~ [ z_1]^{k}_i ~ [ z_2]^{k}_{-i } ~.\ ] ] it follows simply that this scalar quantity is translationally invariant it can not be lowered , @xmath114^{k } \odot [ z_2]^{k } = 0.\ ] ] simple examples are found in laughlin s two - electron building blocks of the next subsection @xmath115^{1/2 } \odot [ z_2]^{1/2 } = z_1-z_2 \nonumber \\ & & [ z_1]^{m/2 } \odot [ z_2]^{m/2 } = ( [ z_1]^{1/2 } \odot [ z_2]^{1/2 } ) ^m=(z_1-z_2)^m \nonumber \label{eq : sc}\end{aligned}\ ] ] the anti - aligned coupling in eq . ( [ eq : quasiel ] ) is favored energetically , producing a factor of @xmath143 , for all @xmath97 . the flux creation operator @xmath144 $ ] is symmetric among particle exchange the components are the elementary schur polynomials for @xmath14 coordinates . thus the correlations that laughlin builds in to his wave function treat all particles equivalently , even though in a given configuration some particles @xmath97 will be closer to particle 1 , and some farther away . his construction respects the scale invariance of the coulomb potential classically , given a solution at one density , others could be obtain by a simple rescaling of the magnetic length . for a system of quantum mechanical fermions , this rescaling is restricted to odd @xmath40 . the gh and gh@xmath32 constructions are guided by very similar considerations . + , depicted as a quasi - electron closed sub - shell . the laughlin quasi - electron is an anti - aligned product.,scaledwidth=46.0% ] * some properties of laughlin wave function : * figure [ fig : fig13 ] depicts laughlin s @xmath100 wave function geometrically , as a closed sub - shell of quasi - electrons , as @xmath29 is incremented . + _ recursion relation : _ for fixed @xmath0 , with each increment in @xmath29 , one electron and @xmath40 units of magnetic flux are added overall to the wave function . the scalar recursion operator describing this process acts on the @xmath29-particle wave to produce the @xmath145-particle wave function , @xmath146= r_{n+1}^m ( n+1 ) e^{-\|z_{n+1}\|^2/2 } l[n , m ] \label{eq : lrecur}\end{aligned}\ ] ] where for m=3 @xmath147^{3 n \over 2 } \odot \left [ z_1 ^ 3 z_2 ^ 3 ... z_n^3 \right]^{3 n \over 2}\ ] ] and where @xmath148^{n m \over 2}_{n m \over 2 } \equiv { 1 \over \sqrt{(mn ) ! } } z_1^m \cdots z_n^m\ ] ] with the remaining components obtained by lowering , as described previously . the operator is ( necessarily ) a scalar and symmetric among exchange of coordinates of the pre - existing electrons . in is easy to convert this recursion operator into its second quantized form , then employ it to recursively generate the @xmath29-electron , starting from the trivial @xmath149 state . + _ fractional charges : _ experiments have identified fractionally charged fqhe excitations @xcite , while theory can generate corresponding wave functions from recursion relations @xcite . labeling the location of a hole by a coordinate @xmath150 , one can define the wave functions @xmath151^{n \over 2 } \odot \left [ z_1 z_2 \cdots z_n \right]^{n \over 2 } l[n , m ] & q=-{e \over m } \\ ~&~\\ \left [ z_0 \right]^{n } \odot \left [ z_1 ^ 2 z_2 ^ 2 \cdots z_n^2 \right]^n l[n , m ] & q=-{2e \over m } \\ \vdots & \\ \left [ z_0 \right]^{mn \over 2 } \odot \left [ z_1^m z_2^m \cdots z_n^m \right]^{m n \over 2 } l[n , m ] & q =- e \end{array } \right . \label{eq : fc}\end{aligned}\ ] ] that insert a hole in all possible locations in the shell . as the last step is identical to the recursion relation that creates the @xmath152-electron state , apart for the substitution @xmath153 , one identifies the charge of the corresponding hole state as @xmath154 , relative to the @xmath152-electron state : that charge would be cancelled by adding an electron . the remaining states , as cycles that accumulate to the last state , then carry the charges indicated . note that each component of the scalar products above corresponds to the insertion of a hole at a specific location , breaking translational invariance , which is then restored when the dot - product sum is taken over all such components . a similar cycle of substitutions can be carried out , given a quasi - electron single - determinant form for the wave function . for the case of @xmath100 , the determinant in eq . ( [ eq : laughlin2 ] ) evolves as following where the notation is @xmath156^ { n \over 2 } \odot \left [ z_2 z_3 ... z_n z_0 \right]^ { n \over 2 } \nonumber \\ r_{n+1}(z_0 ) & \equiv & \left [ z_{0}\right]^ { n \over 2 } \odot \left [ z_1 z_2 ... z_n \right]^ { n \over 2 } \nonumber\end{aligned}\ ] ] consequently , in the laughlin case , it is sufficient to have either the recursion relation or the explicit single - determinant quasi - electron form of the wave function to be able to introduce the defects that carry fractional charge . we will present a general single - determinant quasi - electron wave function for the fqhe later in this paper . + _ the pseudopotential : _ to evaluate the two - electron coulomb matrix element it is convenient to transform from state vectors @xmath157 to the basis @xmath158 involving the jacobi coordinates @xmath159 and @xmath160 . the relative wave functions allowed by antisymmetry are @xmath161 the nonzero matrix elements of the coulomb potential , @xmath162 where @xmath163 is the fine structure constant , are plotted in fig . [ fig : figure1 ] as a function of @xmath164 . one consequence of laughlin s construction is the exclusion of the most repulsive @xmath164=1 ( p - wave , @xmath40=3 ) and @xmath165 ( f - wave , @xmath40=5 ) components of the coulomb interaction . the laughlin wave functions can be viewed as a low - momentum restrictions of the true wave functions that capture most of the physics relevant to long - wavelength properties of the system , while excluding high - frequency components . in units of @xmath166 . [ fig : figure1],scaledwidth=46.0% ] the gh operator construction was done in spherical geometry . we retain that geometry here , making the transition to the plane at a later point . the gh wave functions have the form @xmath175\cdot u(j ) \right]^{m-1 } ~\begin{array}{l } s=0 , 1/2 , 1 , ... \\ m=3,5 , ... \end{array } \label{eq : jaingh}\ ] ] where @xmath22 amd @xmath24 are unconstrained apart from @xmath176 . the @xmath177 ( @xmath178 ) operators generate hierarchy ( conjugate ) states . + _ the gh operator for @xmath0=2/5 : _ we use the simplest non - laughlin case of @xmath0=2/5 to illustrate the gh operator construction . the laughlin state at @xmath21 is defined by its two - electron correlation function . as the density is increased above this value , local overdensities arise : quantum mechanics prevents the smooth rescaling of distances that would occur in a degenerate classical system with a scaleless potential . the process begins with the percolation of @xmath11-wave pairs : more precisely , the removal of quanta from a given two - electron correlation creates an overdensity , including a nonzero amplitude for finding the electrons in a relative @xmath11-wave . one seeks a trial variational wave function that keeps local overdensities separated to the extent possible , thus preventing any larger , energetically more costly perturbations in local density . the construction must be possible at all relevant @xmath29 and must produce wave functions that are translational invariant and homogeneous . laughlin addressed the same problem for point electrons as the overdensities " , where the available antisymmetric scalar correlations are odd powers of @xmath179 ^{1 \over 2}\odot [ u(2)]^{1 \over 2 } \equiv u(1 ) \cdot u(2)\ ] ] @xmath180 is a product over all possible pairs of this form . the construction can be extended to any even @xmath29 . we introduce an index @xmath196 to denote a given pair of electrons , in some partition of the electrons , e.g. , @xmath197 , and define @xmath198 @xmath199 \odot u[i=2 ] \equiv [ u(1)u(2)]^1 \odot [ u(3)u(4)]^1\ ] ] @xmath200 \odot u[j ] \right)\left ( \displaystyle\prod_{i=1}^{n/2 } l_d(i ) \right ) \right]\ ] ] again with @xmath201 , @xmath202 when @xmath203 . at large @xmath29 the first term determines the filling , while the second produces the @xmath184 defects that are then arranged in a manner that follows laughlin s construction . antisymmetrization yields a simple determinant @xmath204^{\ell } _ { m_1 } \cdots [ u(n)]^ { \ell}_{m_n } \nonumber \\ & & \times [ d(1)]^{1/2}_{q_1 } \cdots [ d(n)]^{1/2}_{q_n } \label{eq : simpled}\end{aligned}\ ] ] where @xmath205 and @xmath206^ { \ell}_m \equiv \mathcal{d}^\ell_{\ell , m } = \left [ ( 2 \ell ) ! \over ( \ell+m)!(\ell - m ) ! \right]^{1/2 } u_{1/2}^{\ell+m } u_{-1/2}^{\ell - m } \tag{\ref{eq : us}}\ ] ] the @xmath207s can be written to the left or to the right of the @xmath208s , without changing the determinant . this `` @xmath209 '' form of the operator the representation showing the correlation structure can be recast , in the usual way , in an equivalent @xmath210 form , which is the quasi - electron representation that reveals the fll sub - shell structure induced by the correlations . for @xmath211 and thus @xmath203 , @xmath212 , yielding each column of the antisymmetric @xmath214 determinant is a direct sum of the aligned and anti - aligned vectors . the second , more explicit filled sub - shell " form consists of a @xmath215-electron lower ( @xmath216 ) sub - shell and a @xmath217-electron upper ( @xmath218 ) sub - shell . @xmath219 antisymmetrizes over exchange of particles among the two closed shells . this operator generates the @xmath211 state of the @xmath100 hierarchy , when it acts on the symmetric @xmath220 @xmath29-electron state , and the @xmath221 state , when it acts on the @xmath222 symmetric state . + _ @xmath0=2/3 case : _ the @xmath223 conjugate state is the densest of the series .... 4/7,3/5,2/3 and marks the filing where only isolated , two - electron droplets of the @xmath21 remain : we can create these under - dense regions by applying @xmath184 operators of the type @xmath183 to the half - filled shell . the operators that remove flux from the half - filled shell to produce a state with @xmath223 are of the form @xmath224 ^ 1 \cdot [ d(3 ) d(4)]^1 $ ] operating between all of the defects . the arguments are the reverse of those for @xmath211 . we obtain for @xmath29=4 @xmath225 ^ 1 \odot [ d(3)d(4)]^1~ \right]\ ] ] this operator and that for @xmath195 @xmath211 are identical after antisymmetrization ( a pattern that continues for all similar elementary hierarchy / conjugate pairs , and is associated with a symmetry at @xmath6 we discuss later ) . but for @xmath226 @xmath24 is incremented , while @xmath227 is fixed , yielding @xmath228^{1 \over 2 } \otimes [ d(1)]^{s } \right]^{s+{1 \over 2 } } & \left [ [ u(2)]^{1 \over 2 } \otimes [ d(2)]^{s } \right]^{s+{1 \over 2 } } & \cdots & \left [ u(n-1)]^{1 \over 2 } \otimes [ d(n-1)]^{s } \right]^{s+{1 \over 2 } } & \left [ [ u(n)]^{1 \over 2 } \otimes [ d(n)]^s \right]^{s+{1 \over 2 } } \\ ~ \left [ [ u(1)]^{1 \over 2 } \otimes [ d(1)]^{s } \right]^{s-{1 \over 2}}&\left [ [ u(2)]^{1 \over 2 } \otimes [ d(2)]^{s } \right]^{s-{1 \over 2 } } & \cdots & \left [ [ u(n-1)]^{1 \over 2 } \otimes [ d(n-1)]^{s } \right]^{s-{1 \over 2 } } & \left [ [ u(n)]^{1 \over 2 } \otimes [ d(n)]^{s } \right]^{s-{1 \over 2 } } \\ \end{array } \right\| \nonumber \\ \label{eq:52a}\end{aligned}\ ] ] because the gh operators are scalars , the requirements of translational invariance and a homogeneous one - body density are satisfied . the operators must be antisymmetrized among all the relevant partitions of electrons . this yields the generalization of eq . ( [ eq : simpled ] ) @xmath241^{\ell}_{m_1 } \cdots [ u(n)]_{m_n}^{\ell}~ [ d(1)]_{q_1}^{s } \cdots [ d(n)]_{q_n}^{s } , \nonumber \\ ~ \label{eq : gh3}\end{aligned}\ ] ] where @xmath242 is the antisymmetric tensor with @xmath29 indices and @xmath243 , with @xmath244 and @xmath245 . the filling corresponding to eq . ( [ eq : jain ] ) is then @xmath246 each distinct hierarchy state is indexed by some fixed @xmath24 , with @xmath22 running over values @xmath247 . @xmath248 is the large @xmath29 limit , yielding @xmath249 producing the @xmath100 states with @xmath0=1/3 , 2/5 , 3/7 , ... each conjugate state is indexed by some fixed @xmath22 , with @xmath250 . @xmath251 is the large @xmath29 limit , yielding @xmath252 producing the @xmath100 states with @xmath0 = 1 , 2/3 , 3/5 , 4/7 , .... the starting point for the gh@xmath32 wave functions uses the gh operators in their @xmath36 forms . generalizing the previous result for @xmath211 , one can also rewrite this as a product of @xmath85 closed - sub - shell operators , just as was done previously in eqs . ( [ eq:52 ] ) and ( [ eq:52a ] ) . the gh@xmath32 procedures imprint this sub - shell structure on the quasi - electrons , as described in the next section . the resulting wave function is exactly equivalent to that obtained by acting with the laughlin closed - shell operator on the half - filled shell . but these two alternate procedures are not equivalent for @xmath257 gh operators . consequently there are two gh generalizations of laughlin s construction , both of which generate translational invariant , homogeneous states . they differ because the derivatives creating the defects act in slightly different ways in the two procedures . by using the gh operators as in eq . ( [ eq : form ] ) , simpler wave functions revealing more of the underlying physics are obtained , the evolution of the sub - shell structure ( the number of shells , their occupancies ) of the states is quite interesting , as the angular momentum of the quasi - electrons changes as new electrons are added . this is more easily described geometrically . we do so below . the results above show that the quasi - particles that arise in mapping the fqhe hierarchy states of filling @xmath267 into a noninteracting form are simple objects that carry the quantum numbers @xmath268 and @xmath5 , where the allowed @xmath29s are divisible by @xmath11 , @xmath269 , and @xmath270 . in addition to its role in the construction of translationally invariant and homogeneous many - electron states , angular momentum is associated with variational strategies to minimize the coulomb interaction , and thus with the quantum number @xmath5 . the generators of rotations @xcite on the sphere @xmath271 ^ 1_m\ ] ] involve sums over pair - breaking " operators : @xmath272 ^ 1 $ ] , acting on the a scalar pair @xmath236 , produces the aligned pair @xmath273 ^ 1_m$ ] @xcite . that is , the signature of the correlations that arise with increasing density is the generation of angular momentum . constructions like gh@xmath32 , in which regions of overdensity are kept separated to the extent possible by building in multi - electron correlations , will natural lead to a tower of angular momentum sub - shells , and to a noninteracting quasi - electron picture in which the quasi - electrons fill the sub - shells of lowest @xmath274 for @xmath281 ( the anti - aligned laughlin case ) one has @xmath282^{s - { n-1 \over 2 } } \otimes [ z_2 \cdots z_n]^{n-1 \over 2 } \right]^{s - n+1 } \sim \nonumber \\ ~ [ z_1]^{s - n+1 } ~ [ z_1]^{n-1 \over 2 } \odot [ z_2 \cdots z_n]^{n-1 \over 2 } \label{eq : dc0 } \end{aligned}\ ] ] the second term is the product of the @xmath143 for all @xmath283 . thus there are no broken scalar pairs in the first sub - shell . consequent one is guaranteed that the minimum number of quanta in the two - electron relative wave function for electrons 1 and 2 is @xmath284 , with one factor coming from quasi - electron 1 , one from quasi - electron 2 , and one from the determinant . the quasi - electron coupling to its neighbors is always through the symmetric schur - polynomial vector @xmath144^{n-1 \over 2}$ ] . this is the reflection of the scale invariance we have described before : with each successive increment in @xmath5 , an additional quanta is removed from the coupling of electron 1 to its neighbors , without regard for the specific positions of the neighbors . this is the best approximation allowed in quantum mechanics to a uniform rescaling of inter - electron distances . the nondegenerate states corresponding to configurations of quasi - electrons in filled sub - shells are obtained for values of the magnetic field @xmath276 and particle number @xmath29 satisfying @xmath309 with @xmath29 divisible by @xmath310 . for fixed @xmath0 ( fixed @xmath22 ) the number of occupied sub - shells is @xmath311 , but for every increment of the electron number @xmath29 by @xmath311 , the indices of the sub - shells @xmath5 increase by one , indicating the number of electrons @xmath97 lacking a factor @xmath304 has increased by one in each sub - shell . a simple example is the @xmath265 , @xmath141 case , where the quasi - electron for electron one is @xmath144 $ ] : there are no scalars @xmath304 , and thus the number of missing scalars is @xmath14 . the index of the occupied sub - shell is @xmath312 , tracking the electron number . the connections described here between clustering of planar wave functions , their representations in terms of su(2 ) algebra , and the vanishing of the bosonic part of the wave function ( the wave function with one power of the antisymmetric closed shell removed ) when @xmath314 electrons are placed at one point , have been noticed before , most prominently by stone and collaborators @xcite , who also explored the implications of such properties to global topological order and to the nonabelian statistics of the quasi - particles . a specific construction of the sort provided here , however , is new , as is the recognition that these properties were implicit in the gh operators . we will later use the results above to put the wave functions in their generalized pfaffian form , the most symmetric form of the gh@xmath32 wave function in which the composite fermions can be regarded as fermionic excitations of a bosonic vacuum state . + _ relation to the pseudo - potential : _ the ground - state expectation of haldane s pseudopotential provides an order parameter for laughlin s state , and thus that one might expect other hierarchy states to mark densities where similar transitions in this parameter occur . while one can calculate this expectation value from gh@xmath32 wave functions ( this will be done in the future ) , a simpler surrogate is to count the missing scalar pairs @xmath315 , per electron and relative to the laughlin state , at each value of @xmath0 . this yields , in the large @xmath29 limit , 0 ( @xmath21 ) , 1/2 ( @xmath316 ) , 1 ( @xmath298 ) , 3/2 ( @xmath317 ) , .... this implies a change in the slope of the expectation value of the haldane pseudo - potential at the fillings marking the closed - shell states . one can calculate the additional energy associated with the addition or removal of one quasi - electron at these filings . this addition / removal does not create a simple particle or hole state , as changing the electron number alters the angular momentum of the existing quasi - electrons . this process is illustrated in fig . [ fig : fixedmag ] . + _ effective b field : _ the usual argument @xcite leading to the conclusion that a cf is an electron that has absorbed two external magnetic flux units is based on producing the reduced effective b field needed to account for the angular momentum and closed - shell structure of the cfs . two flux units per electron are indeed absorbed into internal ( scalar ) quasi - electron wave functions of gh@xmath32 , although the gh@xmath32 quasi - electron differs from jain s picture , as it involves the coupling of an electron spinor to a single magnetic flux spinor . it may be helpful follow the flux associated with the addition of an electron to the wave function , to explain the bookkeeping . " quasi - electron labels @xmath22 and @xmath24 are related to the electron number @xmath29 and total magnetic flux through the sphere @xmath81 by @xmath318 . consider quasi - electron 1 in the @xmath281 ( laughlin ) shell of a fqhe hierarchy state of filling @xmath0 . in the mapping to quasi - electrons , @xmath14 flux units from electron 1 are used by ( absorbed into ) electrons 2 through @xmath29 , screening them from electron 1 : this is represented mathematically as an anti - aligned coupling of an angular - momentum-1/2 flux unit from electron 1 with the spinors of electrons 2 through @xmath29 , forming a scalar that thus can be considered part of their internal quasi - electron wave functions . conversely , one unit of flux from each neighboring electron is absorbed by electron 1 , incorporated into an anti - aligned couplings with the angular momentum of quasi - electron 1 . this removes another @xmath14 quanta from electron 1 , while produce a scalar that we can regard as the translationally invariant intrinsic wave function of quasi - electron 1 , as in eq . ( [ eq : dc0 ] ) . this leaves an uncanceled angular momentum of @xmath319 in quasi - electron 1 . thus we identify @xmath320 with @xmath321 . as number of occupied sub - shells is @xmath322 and the number of electrons @xmath323 , one can express the resulting @xmath324 ( the reduction in the effective flux ) , for large @xmath29 , as @xmath325 . thus the standard result for the reduced magnetic field @xmath326 is obtained in the present quasi - electron picture , while each quasi - electron involves only a single magnetic flux spinor . + the space of hierarchy and conjugate states is two - dimensional , defined by @xmath232 . thus various one - dimensional trajectories " in the @xmath275 ( or , equivalently , @xmath327 ) parameter space can be followed , to illustrate the rather interesting evolution of the quasi - electron sub - shell structure . we consider three cases : 1 . states of fixed @xmath0 , which for hierarchy states corresponds to fixed @xmath24 , with @xmath29 incremented by running @xmath328 ( @xmath176 ) ; for conjugate states , @xmath22 is fixed , and @xmath29 is incremented by running @xmath250 . half - filled or self - conjugate states : this includes the trajectory where @xmath329 and thus @xmath330 . states of fixed @xmath276 but running @xmath29 , describing the successive additions of electrons while the magnetic field @xmath34 is kept fixed . _ states of fixed @xmath0 , for fixed @xmath24 or @xmath22 : _ the figures group hierarchy states with their conjugates analogs defined by the exchange @xmath331 , where we add bars to distinguish the conjugate state quantum numbers from those of their hierarchy - state analogs . this identification connects the associated fillings according to @xmath332 associating @xmath21 with @xmath141 , @xmath211 with @xmath223 , etc . when described in terms of electrons , the paired states have no obvious common structure and reside in different hilbert spaces ( distinct values of electron angular momenta @xmath276 ) . but in their quasi - electron representations , they have the same sub - shell structure and occupations and a common quasi - electron angular momenta @xmath333 . all states of fixed @xmath0 begin with an elementary state with @xmath329 , the configuration of minimum @xmath29 in which the requisite number of sub - shells is occupied : @xmath85 in the case of hierarchy states , @xmath311 in the case of conjugate states . every choice @xmath232 is associate with a state of definite @xmath0 with the exception of the elementary @xmath329 states , which are the first members of both hierarchy and conjugate series , depending on whether @xmath24 or @xmath22 is kept fixed as @xmath29 is incremented . thus the are members of both the @xmath0 and @xmath334 series . figures [ fig:131 ] and [ fig:2523 ] show the one- and two - sub - shell cases , correspond to the fillings @xmath335 and @xmath336 , respectively . in fig . [ fig:131 ] , as electrons are added , an algebraic correspondence between the sub - hells is maintained : the occupied sub - shells and quasi - electrons have the same rank , at each @xmath29 . the states differ because the shells are labeled by different values of @xmath5 : as electrons are added , a pairing gap opens up between occupied hierarchy and conjugate shells . no scalar pairs are formed in the recursion for @xmath141 , so consequently the occupied shell is that with @xmath312 . in contrast , a scalar pair accompanies the addition of each electron for @xmath21 , so @xmath281 , independent of @xmath29 . figure [ fig:2523 ] shows the two - shell cases of @xmath211 and 2/3 . this structure constrains @xmath29 to increase by two , at each step , as a quasi - electron is added to each sub - shell . the occupied sub - shells for @xmath211 are @xmath281 and 2 : the @xmath337 quasi - electrons in the upper sub - shell are missing a scalar pair with exactly one of their neighbors . the occupied sub - shells for @xmath223 are labeled by @xmath338 and @xmath184 . thus a pair gap again opens up between the hierarchy and conjugates states , but at a rate in @xmath29 half that found for the case illustrated in fig . [ fig:131 ] . the pattern of fig . [ fig:2523 ] continue for the pairs @xmath298 and 3/5 , @xmath339 and 4/7 , etc . , but with 3 , 4 , @xmath340 , shells occupied . + _ the @xmath6 self - conjugate series : @xmath329 : _ the sub - shell structure defined by fixed @xmath24 ( hierarchy states ) and fixed @xmath22 ( conjugate states ) corresponds to paths of fixed @xmath0 in the @xmath341 or @xmath342 plane . another path of fixed @xmath0 , converging to the half - filled shell , is defined by @xmath329 and thus @xmath343=1 , 4 , 9 , @xmath340 . this path is confined to the self - conjugate incompressible states : when @xmath149 , @xmath344 , the @xmath0=1/3 and @xmath0=1 states coincide ; when @xmath195 , @xmath345 , the @xmath211 and @xmath223 states coincide ; when @xmath346 , @xmath347 , the @xmath298 and @xmath348 states coincide ; etc . thus the series can be viewed as converging , in the large @xmath29 limit , to the half - filled shell simultaneously from the high- and low - density sides . this series is illustrated in fig . [ fig : half ] . similar series can be generated for paths defined by @xmath349 , where i is a fixed half integer or integer . all in the large @xmath29 limit correspond to @xmath0=1/2 . this observation is a restatement of the fact that hierarchy and conjugate states are dense near @xmath6 : one can reach these states by trajectories of fixed @xmath350 as well as by dialing @xmath0 from the low - density or high - density sides . below we discuss some of the symmetry properties of wave functions at @xmath6 . + _ states for fixed @xmath34 , variable n : _ if we fix magnetic field @xmath34 ( equivalently @xmath81 ) , @xmath0 will evolve from 0 to 1 as successive electrons are added . in general this path encounters some of the incompressible closed - sub - shell shell states but not all , as @xmath211 closed - shell quasi - electron states are found only if @xmath29 is even , @xmath298 only if @xmath29 is divisible by three , etc . the path runs through quasi - electron states that are not closed shells , and thus in particular produces quasi - electron representations of states involving the addition or subtraction of one electron to the incompressible states , forming particle and hole states . for such states , as the flux through the sphere or its planar analog is fixed , the electrons have a fixed total angular momentum @xmath276 . the quasi - electron for the lowest ( fully anti - aligned ) sub - shell then must be @xmath351^{s - { n-1 \over 2 } } \otimes [ z_2 \cdots z_n]^{n-1 \over 2 } \right]^{s - n+1}\ ] ] as there are @xmath14 other quasi - electrons each containing one factor of @xmath352 . consequently every time @xmath29 is incremented by a unit , the angular momentum of the lowest sub - shell is reduce by a unit , and thus the maximum occupation of that sub - shell is reduced by two . other quasi - electron sub - shells evolve similarly . that is , the addition of an electron alters the sub - shell structure and the pattern of shell occupations : this is not the fixed - shell structure familiar from atomic physics , for instance . figure [ fig : fixedmag ] shows the evolution of states from @xmath353 to @xmath141 , as @xmath29 is incremented , for @xmath354 . the choice of @xmath276 is clearly arbitrary , but @xmath354 supports a self - conjugate state ( @xmath29=16 , @xmath0=4/9 and well as @xmath355 ) as well as other closed - sub - shell quasi - electron states ( @xmath29=11 , @xmath0=1/3 ; @xmath29=15 , @xmath0=3/7 ; @xmath29=20 , @xmath0=2/3 ; @xmath29=31 , @xmath0=1 ) . the sub - shells should be viewed as a procedure for defining a low - momentum hilbert space , within which one can construct zeroth - order wave functions , as one normally does in an effective theory . the figure shows that the hilbert spaces one forms by filling the lowest sub - shells not only describe the incompressible states ( those states where the hilbert space contracts to a single state , and where that state is a scalar ) but also other fillings , where multiple low - energy states can be formed . thus one can use the quasi - electron representation to describe low - lying states of arbitrary filling . a number of the features of fig . [ fig : fixedmag ] are generic , not dependent on the specific choice of @xmath354 . examples include the quasi - electron representations of particle and hole states obtained by adding or subtracting a particle from one of the incompressible states . the simplest cases are the particle and hole excitations of the @xmath0=1/3 states , which one sees from the figure correspond to three - quasi - electron particle and hole states , respectively . as the angular momentum zero state is unique , corresponding to the - quasi - electron coupling @xmath356 , the quasi - electron representation of the low - momentum states predicts that there is a single low - energy translationally invariant , homogeneous particle ( or hole ) state built on the @xmath21 state . + there have been recent discussions about the distinctions between or equivalence of hierarchical descriptions of the fqhe initially proposed by haldane @xcite and by halperin and the cf picture advocated by jain @xcite . given the explicit construction presented here , it is interesting to see how gh@xmath32 description fits into the context of these two views of the fqhe . the gh@xmath32 construction is in good accord with the cf picture of jain , though differences are also apparent : 1 . the gh@xmath32 sub - shell structure is contained within the fll , and thus the incompressible states that corresponding to the filling of these sub - shells also involve only fll degrees of freedom . the angular momentum substructure derives from the fact that the breaking of scalar pairs that must accompany increases in the density generate angular momentum . one minimizes the coulomb by breaking the fewest such scalar pairs , which then identifies the fractional fillings with the nondegenerated states formed from completely filling the lowest sub - shells . the standard description of a cf is an electron coupled to two units of magnetic flux , while the gh@xmath32 quasi - electrons incorporate only a single magnetic flux unit into their intrinsic wave functions . however both descriptions agree that , with the addition of each electron , two units of magnetic flux become intrinsic ( that is , combined into scalars , and thus not contributing to net quasi - electron angular momentum ) . in the gh@xmath32 construction , only one of those flux units is carried by the added electron : the second is divided , one quantum each , among the @xmath14 pre - existing quasi - electrons , absorbed into their internal wave functions . 3 . the incompressible fqhe states are not formed by filling successive sub - shells by equivalent cfs . rather , for a fixed @xmath34 and thus @xmath276 , there is a maximum number of laughlin - like quasi - electrons that can be formed in the hilbert space . when that limit is reached ( at @xmath21 ) , a new flavor " of quasi - electrons forms , involving an additional broken pair , and carrying one additional unit of angular momentum so that a second sub - shell arises .. this ultimately leads to @xmath357 distinct sub - shells , with their respective quasi - electrons co - existing in the hilbert space at @xmath6 , the most complex case . though these differences from the standard cf description are of some consequence , in total the picture that emerges is conceptually compatible with the basic ideas about cfs . there is a mapping of the interacting electron , open - shell fqhe problem into a closed - sub - shell noninteracting quasi - electron one . there is a sub - shell structure though it is a fine - structure , unrelated to the iqhe , with the gap between neighboring sub - shells associated with a single broken pair per electron . there is a reduced magnetic field : the reduction explicitly arises from the anti - alignment of the angular momentum that the electrons would carry in the absence of coulomb interactions , with the angular momentum generated by the successive breaking of scalar pairs , the variational response of the electrons to minimizing their coulomb interactions . aspects of the gh@xmath32 construction are also hierarchical . first , the gh operators are strictly hierarchical , forming a tower where the next object is obtained from the previous one . recall that these operators create fillings @xmath33 by acting on the half - filled shell , both creating the necessary overdensities among @xmath11 electrons ( relative @xmath21 ) that such fillings require , then guaranteeing that these region are separated by a certain minimal number of quanta . denoting the electron labels of the two clusters as @xmath358 and @xmath359 , the series @xmath11=1,2,3 , ... involves the progression @xmath360 where the operator at step @xmath5 is given by the product of the factor in the @xmath5th row above with and those from all previous steps . there is an obvious and simple recursion relation generating the gh operators . however the most common description of a hierarchical structure is at the wave function level , the conjecture that successively more elaborate cfs are created through laughlin - like correlations among the cfs from previous steps in the hierarchy . apart from differences already noted , the sub - shell structure of the gh construction identifies jain s cfs with the quasi - electrons we have defined here , all of which have a common simple structure and are not hierarchical , at least as this term is commonly used . but a hierarchy is an appropriate description of the over - densities that are created , which evolve from single quasi - electrons at @xmath21 to a collection of @xmath357 such objects at @xmath6 . for example , consider the process of adding quasi - electrons to create a state of filling @xmath33 with the next highest electron number . the process is simple one quasi - electron @xmath361 must be added to each of @xmath11 sub - shells and as a series in @xmath11 is hierarchical , @xmath362 this process is accompanied by the addition of magnetic flux : as the starting and new states are both scalars , the added quasi - electrons and the added flux must combine to form a scalar . the steps to create a new state for , say , @xmath363 , are those required to create a new state for @xmath364 , except that a fourth @xmath365 quasi - electron must now be included . the gh@xmath32 construction thus supports the general features of both cf and the hierarchical descriptions , and suggests that some of the debate over the cf / hierarchy issue may derive from assuming the hierarchical objects are the cfs : in the present gh@xmath32 construction , the most natural hierarchical structures are the collections of cfs that are involved in the recursions illustrated in the figures , and that form the regions of over or under density . the states near the half - filled shell have long raised important questions . halperin , lee , and read argued that the ground state at @xmath6 is a fermi liquid @xcite . the sequences of gh@xmath32 states with @xmath366constant converge to @xmath6 in the large @xmath29 limit . the special state with @xmath329 , illustrated in fig . [ fig : half ] , is defined by a vanishing @xmath326 : we previously noted that the effective residual field experienced by the quasi - electrons is determined by @xmath367 . this state can be viewed as the limit of a series of states that can be simultaneously labeled as @xmath368 and @xmath369 . although the limit yields an even - denominator state , the limit is reached through a series of conventional odd - denominator states . the states along this trajectory consist of @xmath357 sub - shells containing an average of @xmath357 particles . thus the states at vanishing @xmath326 correspond to the case where the depth of the quasi - electron fermi sea is maximal , scaling as @xmath357 . in contrast , for any of the conventional fillings characterized by fixed @xmath24 ( hierarchy states ) or @xmath22 ( conjugate states ) , the quasi - electron fermi seas have a finite depth consisting of a few sub - shells , @xmath85 or @xmath370 , in the large @xmath29 limit . because sub - shell splitting are identified with successive removals of one factor of @xmath304 from electron 1 , the gh@xmath32 construction yields at @xmath0=1/2 @xmath371 that is , the energy of the quasi - electron fermi sea is on the order of @xmath357 of the two - electron p - wave energy of eq . ( [ eq : coulomb ] ) , and as the defects involve @xmath357 electrons , the typical area covered by the density perturbations is on the order of @xmath357 times the area available to each electron . + _ @xmath372 symmetry : _ there are some suggestive aspects of this interesting state at @xmath329 that we describe now , that support the notion that the physics near @xmath6 is nontrivial . despite being at @xmath6 , the states with @xmath373 and @xmath374 are not particle - hole symmetric , as @xmath307 and thus @xmath375 . under a particle - hole transformation with respect to the electron or electron - hole vacuums ( the iqhe states at @xmath376 ) , these @xmath29-electron states transform into @xmath14 electron states , and thus not into themselves . however these states have an exact symmetry connected with the interchange @xmath372 . let @xmath377 be the operator that exchanges single - particle spinors with schur polynomial spinors , that is , particle creation with magnetic flux creation @xmath378^{n-1 \over 2 } \leftrightarrow [ z_2 \cdots z_n]^{n-1 \over 2}\ ] ] under this operation the quasi - electrons transform as @xmath379 they behave like majorana states under this operation , transforming to themselves up to a sign , with quasi - electrons in neighboring sub - shells having opposite c. the pattern of neighboring sub - shell of opposite @xmath377 is reminiscent of particle physics scenarios in which two ( degenerate ) majorana neutrinos of opposite cp are patched together to form a dirac neutrino , which then transforms under cp not to itself , but to its anti - particle . a similar transformation can not be performed for a fully spin - polarized system , as that would requires combining states of different angular momentum . but in a two - component system that might arise in the limit of small @xmath380 , it would be possible . one can make this transformation sequentially , beginning with the electron in the @xmath281 sub - shell , and proceeding to the top of the fermi sea . denoting electron spin by @xmath381 and the coupled anwe can form the states @xmath382 which then transform as dirac states , @xmath383 this first step has used the @xmath384 part of the @xmath285 sub - shell but not the @xmath385 components , so we now form @xmath386 one can proceed in such steps , until the fermi surface is reached . for occupied states in a two - component system , a transformation to a new set of basis states has no physical consequence . thus this transformation is of potential interest only at the fermi surface , where for the @xmath329 @xmath6 state , it represents a transformation between occupied ( @xmath387 ) and unoccupied ( @xmath388 ) states . one can compare the ground state of the gh construction - a closed @xmath389 state where the natural basis is majorana and the alternative of a closed @xmath390 sub - shell , where the basis states are dirac , in the case of a two - component system . this second case requires the coupling of two majorana states at the fermi surface with energies , as argued above , @xmath391 , but split by an energy corresponding to a single additional broken pair . thus the dirac states at the fermi sea are pseudo - dirac the two coupled states are nearly degenerate , but not exactly so . one concludes that a qualitatively different fermi surface could arise from relatively modest perturbations . possibly there are connections to recent work by son @xcite , who noted that two conventional jain - like hierarchy / conjugate sequences of different filling , @xmath267 and @xmath392 , could be mapped into the same half - integer filling factor @xmath393 of a dirac composite fermion . + _ particle - hole symmetry : _ consider any hierarchy state corresponding to some choice of @xmath24 , so that @xmath394 . @xmath29 is determined by @xmath328 , @xmath201 , and @xmath395 . now consider a conjugate series . for clarity we label the conjugate series by @xmath396 and @xmath397 the filling is @xmath398 , @xmath399 , and @xmath400 then for any given @xmath401 we find the choice @xmath402 thus our construction produces an explicit conjugate state for every hierarchy state , with the requisite particle number to be the particle - hole ( ph ) conjugate . is the gh@xmath32 conjugate state exactly the ph state ? apart from a few small-@xmath29 cases , the answer is no , though the differences are very small numerically . however this shortcoming is a choice made in our construction : exact ph symmetry can be easily restored . a given gh@xmath32 state , once evaluated , can be readily written in second quantization . doing so for the hierarchy ( @xmath232 ) and conjugate partner ( @xmath403 ) states above yields @xmath404 where @xmath405 is the electron vacuum , @xmath406 is the creation operator for the electron , @xmath407 , and the sum over coefficients @xmath408 yields a scalar contraction of these creation operators . ( this in fact is the procedure followed in the numerical calculations of the next section . ) introducing the destruction operator phased to carry good angular momentum @xmath409 we can form two new scalar states by ph conjugation @xmath410 where @xmath411 is the filled fll , which we can employ in a redefinition of our original states @xmath412 this procedure produces analytic wave functions with exact ph symmetry . the series of states converging to @xmath6 with @xmath413 is defined by the trajectory @xmath414 , and thus lies immediately to the low - density side of the @xmath329 trajectory . the @xmath275 and @xmath415 cases then coincide , unlike all other cases described above . this trajectory runs through the hierarchy states @xmath416 ( @xmath417 ) , @xmath418 ( @xmath419 ) , @xmath420 ( @xmath421 ) , @xmath340 , leading at large @xmath29 to a @xmath6 state distinct from that for @xmath329 . unlike the cases described just above , where the particle and hole states have @xmath422 , this trajectory has @xmath29=@xmath423 . thus there is only a single trajectory of @xmath6 , fully spin - polarized , self - conjugate states defined by @xmath424 leading to a single ph - symmetric state at @xmath6 . a ph - symmetric theory of the fermi liquid ground state at @xmath6 has recently been discussed @xcite . this gh@xmath32 quasi - electrons and wave functions can be written as fermion operators acting on the bosonic @xmath6 state . using eqs . ( [ eq : dc0],[eq : dc1],[eq : dc2 ] ) , we find @xmath425^{l}_m \nonumber \\ \gamma^{n,\mathcal{i}=2}_{l , m}(i ) & = & \sum_{j \ne i } { [ [ z_i]^{\ell - s+(\mathcal{i}-1)/2 } [ z_j]^{(\mathcal{i}-1)/2}]^{l}_m \over z_i - z_j } \nonumber \\ \gamma^{n,\mathcal{i}=3}_{l , m}(i ) & = & \sum _ { j < k;~ j , k \ne i } { [ [ z_i]^{\ell - s+(\mathcal{i}-1)/2 } [ z_j z_k]^{(\mathcal{i}-1)/2}]^{l}_m \over ( z_i - z_j ) ( z_i - z_k ) } \nonumber \\ \gamma^{n,\mathcal{i}=4}_{l , m}(i ) & = & \sum _ { j < k < l;~ j , k , l \ne i } { [ [ z_i]^{\ell - s+(\mathcal{i}-1)/2 } [ z_j z_k z_l]^{(\mathcal{i}-1)/2}]^{l}_m \over ( z_i - z_j ) ( z_i - z_k)(z_i - z_l ) } \nonumber \end{aligned}\ ] ] and so on , where @xmath426 . the numerators are all aligned couplings . the quasi - electrons can be written @xmath427 in the case of the @xmath6 state reached through the series @xmath428 , the above results can be rewritten as @xmath429 ^ 1_m \over z_i - z_j } \nonumber \\ \gamma^{n,\mathcal{i}=3}_{l=2,m}(i ) & = & \sum _ { j < k;~ j , k \ne i } { [ [ z_i z_j]^1 \otimes [ z_i z_k]^1 ] ^{2}_m \over ( z_i - z_j ) ( z_i - z_k ) } \nonumber \\ \gamma^{n,\mathcal{i}=4}_{l=3,m}(i ) & = & \sum _ { j < k < l;~ j , k , l \ne i } { [ [ z_i z_j]^1 \otimes [ z_i z_k]^1 \otimes [ z_i z_l]^1]^3 \over ( z_i - z_j ) ( z_i - z_k)(z_i - z_l ) } \nonumber \end{aligned}\ ] ] these have the form of particle - hole operators on the bosonic @xmath6 state , with the number of particle - hole excitations @xmath303 . @xmath430 is a product of @xmath14 scalar pairs . the denominators destroy @xmath303 of these scalar pairs , defining the holes , while the numerators create their replacements , the corresponding aligned pairs @xmath431 ^ 1 $ ] , the particles . @xmath430 , as a filled sea " of scalar pairs , plays the role of the particle - hole vacuum . related operators have been discussed previously . the moore - read pfaffian @xmath6 state , most often discussed in connection with the @xmath432 state , takes the form @xmath433 the pfaffian operator has the same filling as the gh@xmath32 operator with @xmath434 , the first series on the conjugate side of @xmath329 , just as the ph - symmetric @xmath6 series is the first on the hierarchy side . the @xmath434 gh@xmath32 operator is formed from the antisymmetrized product of the single quasi - electron operators @xmath435^{1/2 } \over z_i - z_j } \nonumber \\ \gamma^{\mathcal{i}=3}(i ) & = & \sum _ { j < k;~ j , k \ne i } { [ z_i z_j z_k]^{3/2 } \over ( z_i - z_j ) ( z_i - z_k ) } \nonumber \\ \gamma^{\mathcal{i}=4}(i ) & = & \sum _ { j < k < l;~ j , k , l \ne i } { [ z_i^2 z_j z_k z_l]^{5/2 } \over ( z_i - z_j ) ( z_i - z_k)(z_i - z_l ) } \nonumber \end{aligned}\ ] ] the gh@xmath32 operator analogous to the pfaffian is a simple determinant , e.g. , taking the form for @xmath436 @xmath437 the wave function generated by the pfaffian at @xmath6 has a poor overlap with that generated by the analogous gh@xmath32 operator , e.g. , @xmath135 0.87 for @xmath438 . the corresponding overlap of the gh@xmath32 wave function with that computed by diagonalizing the coulomb interaction is 0.993 . the spin - paired pfaffian overlaps with numerically generated wave functions for @xmath432 also appear to be rather poor , ranging from 0.69 - 0.87 for @xmath439 , according to scarola , jain , and rezayi @xcite . however the agreement can be improved significantly , if the p - wave contribution to the potential is dialed away from its coulomb value @xcite . the standard pfaffian builds in @xmath184 two - electron correlations similar to those contained in @xmath440 , which generates only a subset of the quasi - electrons contributing at @xmath6 . read and rezayi @xcite generalized this construction to include more complicated correlations that would allow the symmetric part of the wave function to remain nonvanishing when @xmath5 electrons are placed at one point . thus this extension has some common features with @xmath441 , @xmath442 . however , as we have discussed previously , optimal approximate wave functions are unlikely to arise from constructions that consider only short - range behavior , at least in cases where systems are spin polarized and the electrons restricted to fill the lowest lls first . the gh@xmath32 construction is guided by the scale invariance of the potential , and thus effectively produces wave functions that weight in an appropriate way coulomb contributions from all partial waves . + .overlaps of the gh@xmath32 quasi - electron hierarchy wave functions with the exact wave functions computed by direct diagonalization of the coulomb interaction . the corresponding results for the gh / jain wave functions are also given . all results are for the principal hierarchy ( @xmath40=3 ) . the gh / jain wave function overlaps are those available from the original gh paper . [ cols="^,^,^,^,^,^,^",options="header " , ] [ tab : two ] the gh@xmath32 quasi - electron wave functions were obtain by applying the gh operators in an alternative way but a way that retains all of the symmetries of the original construction , including translational invariance , homogeneity , and what we have argued is the best quantum mechanical approximation to the scale - independence of the coulomb potential . are the resulting numerical results comparable ? to test this we used an m - scheme lanczos code to directly diagonalize the coulomb potential on the sphere , then generated the corresponding quasi - particle approximate wave function with a mathematica script , evaluating the overlaps . this was done for various states ranging up to 10 electrons , including laughlin s @xmath100 states , which of course are identical in the gh and gh@xmath32 constructions . the analytic wave functions were generated in the plane , which allowed use of mathematica s polynomial capabilities , then mapped onto the sphere , via the homomorphism we have already described . the comparison was done on the sphere as this was the geometry originally used by gh . from table [ tab : one ] one can see that the original gh wave functions ( an analytic version of those jain constructed ) produce marginally better overlaps but in both cases the overlaps generated are typically as close to unity as those obtained in the @xmath21 laughlin case . gh also generated the higher - density conjugate states , and in this case differences between the gh and gh@xmath32 wave functions are somewhat larger typically comparable to the differences between the gh wave functions and the numerically generated exact wave functions . the advantages of the gh@xmath32 wave function are its exceptional simplicity and its explicit quasi - electron or cf form the ability to express the wave function as a single quasi - electron slater determinant , for both hierarchy and conjugate states , and in both planar and spherical geometry . is this important ? in our view , some discussions about approximate fqhe wave functions in the literature are misguided , treating the approximate wave function as a representation of the true wave function . we do not think this is a sensible viewpoint . first , it is clear that the pursuit of improved wave functions is futile , as all constructions deal with some low - momentum portion of the hilbert space , and thus the resulting wave function always can be improved by mixing in any component not in that hilbert space : this is the variational principle . consequently the overlap of any approximate wave function with the true wave function will deteriorate with increasing @xmath29 : because the approximate wave function resides in a limited low - momentum hilbert space , in any defined area of the plane , it will omit some high - momentum components . if one now doubles the area , and then doubles again , the chances that a high - momentum component will be found somewhere in the extended regions grows combinatorially . consequently the overlap is eventually driven to zero . this trend can be seen in numerical results generically . a more sensible definition of the approximate wave function is as a projection of the true wave function to some low - momentum space an effective wave function . because the wave functions discussed here are filled slater determinants , the prescription for their construction defines a projection operator @xmath443 onto a low - momentum hilbert space that contains only a single state . normally in an effective theory ( et ) , a lo projected wave function is considered a good starting point if it has a strong overlap with @xmath444 , the exact solution projected onto the chosen low - momentum hilbert space . by this standard the gh@xmath32 wave function is a trivially exact lo wave function , as the projection @xmath443 only contains one state . an interesting question as the coulomb interaction provides no scale for use as an expansion parameter is the construction of corrections , to produce an improved nlo wave function . because the gh@xmath32 wave function is based on a simple set of quasi - electron degrees of freedom , one has a starting point that could conceivably could allow nlo corrections . a first step in such a process is suggested by , for example , cases similar to panels a ) and c ) of fig . [ fig : fixedmag ] , open - shell " quasi - electron states . unlike the case of the gh or jain wave functions , the gh@xmath32 construction provides a simple but nontrivial @xmath443 for such cases , consisting of a set of degenerate quasi - electron configurations . we return to this topic below . in this paper we have constructed an explicit quasi - electron representation of the hierarchy and conjugate fqhe states . the quasi - electrons have a generic form , a vector product of an spinor that creates single - electron states and one formed from schur polynomials that adds a unit of magnetic flux to all existing electrons . the quasi - electrons and the sub - shell structure they induced within the fll are quite novel , with both the sub - shell structure and quasi - electrons evolving as particles or magnetic flux are added . effectively the construction explicitly maps the problem of electrons strongly interacting in a partially filled sub - shell , into a noninteracting problem , a single slater determinant of quasi - electrons . the quasi - electrons are fermions that carry good @xmath9 and @xmath10 , and the scalar many - body states they form as slater determinants are translationally invariant and homogeneous ( uniform one - body density ) . the construction is done both on the plane and on the sphere . the hierarchy and conjugate states corresponds to those fillings @xmath0 where the quasi - electron representations of the wave functions are unique , consistent of a set of completely filled shells . the resulting wave function defines a low - momentum hilbert space consisting of one slater determinant that efficiently captures much of the long wave - length behavior of the fqhe . the quasi - electron su(2 ) fine structure that exists in the fll is governed by energy gaps that represent the cost of replacing a favorable scalar pair @xmath445 ^ 0 $ ] by an unfavored pair @xmath445 ^ 1 $ ] . as implemented in the gh@xmath32 wave function , this replacement affects correlations at all distance scales , and thus is a variational ansatz consistent with the scale - invariance of the coulomb potential . ( laughlin s construction has the same property , though it is often misconstrued as a strategy for limiting unfavorable short - range correlations . ) the connection between correlations and angular momentum is natural in the fqhe , as the generator of rotations is the sum over pair - breaking single - electron operators . these observations address the issues that troubled dyakonov and others about the use of multiply - occupied lls in the jain construction : jain borrowed the needed su(2 ) algebra from the iqhe , while in fact the physically relevant su(2 ) algebra comes from correlations within the fll . the construction was extended to the plane in two steps . we first defined a truncated hilbert space in the plane that contains the same number of single - electron degrees of freedom as on the sphere . effectively this defines a planar analog of the angular momentum operator @xmath9 . we then packaged these degrees of freedom in single - particle ( electron ) and schur - polynomial ( magnetic flux ) spinors , so that we could exploit the natural mapping between @xmath446 and the @xmath447 , and thus between rotational invariance on the sphere and translational invariance in the plane . we introduced scalar and tensor products on the plane that preserved the homomorphism between the respective lowering operators . consequently , we were able to construct in a simple procedure @xmath29-electron scalar wave functions in the plane that are translationally invariant and that have a uniform one - body density ( homogeneous ) , where the latter property is defined by the homomorphism with the sphere . while the cf picture of the quantum hall effect has been an important theme of the field for many years , we are not aware of any detailed , explicit construction of the quasi - electrons and their sub - shell structure within the fll , on the plane and on the sphere , and including the full set of both hierarchy and conjugate states . the quasi - electrons that emerge are qualitatively in good accord with the ideas of jain , exhibiting for example the screening of the magnetic flux through the absorption of magnetic flux quanta in the intrinsic ( scalar ) wave functions of the quasi - electrons : screening is manifested in anti - aligned couplings of electrons to magnetic flux . however , there are differences in detail , including the form of the quasi - electrons ( the coupling of an electron spinor to a single magnetic flux spinor , not two ) and the existence of an explicit fll fine structure , reflecting the fact that there are multiple types of quasi - electrons , sharing a common functional form but differing in their electron / magnetic flux angular momentum couplings . there are also hierarchical aspects of the construction , though the notion of hierarchical structures seems most relevant to the regions of over - density or under - density , or equivalently , to the set of @xmath11 distinct quasi - electrons that must be added , one to each occupied quasi - electron shell , to create new wave functions of the same @xmath0 but with @xmath11 additional electrons and @xmath448 new flux quanta . there is a clear hierarchical structure among the @xmath11 added quasi - electrons . consistent with the observation that @xmath6 is the convergence point for the series @xmath267 and @xmath449 , a dense collection of gh@xmath32 states can be defined as the large-@xmath29 limit of the series @xmath320=constant . each corresponds to a deep fermi sea of quasi - electrons , containing @xmath450 occupied shells . we discussed particular symmetries of these states , including the @xmath329 trajectory that is symmetric under interchange of electron and magnetic flux spinors , and associated with an additive charge , and the trajectory @xmath414 , which is the particle - hole symmetric case . finally , motivated by the pfaffian , we described the gh@xmath32 wave functions in perhaps their simplest form , as a operator involving only the coordinates @xmath288 acting on the bosonic half - filled shell . the quasi - electrons are then generated by simple operators @xmath451 than act on the scalars @xmath430 , which can be regarded as the corresponding bosonic quasi - electron . in this language , the @xmath6 @xmath329 state becomes a product of particle - hole excitations of the bosonic vacuum , where the holes correspond to destroyed scalars @xmath292 ^ 0 $ ] , and the particles the @xmath292 ^ 1 $ ] . this form of the gh@xmath32 construction provides a series of operators that can be used as pfaffians are typically used . we argue that they are better motivated physically , as they do not focus exclusively on the short - range properties of the coulomb potential . one of the advantages of a simple representation of wave functions is that it provides an intuitive starting point for understanding the physical mechanisms behind less well understood aspects of the fqhe . we mention two possibilities for extending the gh@xmath32 construction , to make further progress . a set of states of fractional filling 3/8 , 4/11 , 6/17 , ... , residing between @xmath0=1/3 and 1/5 , have been discovered experimentally @xcite . these states nominally correspond to open sub - shell " quasi - particle configurations where the sub - shell that is filled at @xmath211 is instead partially filled , and thus where a set of degenerate many - body quasi - electron states exist ( see fig [ fig : fixedmag ] ) . that is , the hilbert space @xmath443 defined by the quasi - electron construction is no longer trivial , but contains multiple states . the situation is precisely analogous to the original problem that the gh@xmath32 construction addresses , finding the best simple low - momentum representation of interacting electrons that occupy an open shell : quasi - electrons now replace the electrons . in the context of a nonrelativistic et , this problem likely presents the first step toward generalization . any truncation in a hilbert space , such as that implicit in the quasi - electron construction , must be accompanied by the introduction of an effective interaction , to account for the degrees of freedom that have been integrated out . that effective interaction can be derived from the full hamiltonian ; alternatively , it can be parameterized , by fitting the associated low - energy constants to data . we have not needed to address this problem in our construction , because the hierarchy and conjugate states are the unique cases where the projection @xmath443 is onto a single state . in such cases the effective interaction clearly has no role in altering the wave function : it only contributes to ground state energy shift , which we have not addressed in this paper . however , the case of an open quasi - electron sub - shell is different , as the lo hilbert space defined by @xmath443 is no longer trivial , and thus @xmath452 is important , as it breaks the degeneracy within @xmath443 and thus determines the nature of the ground state . progress we have made on this problem , including the analytic wave function that is obtained , will be described elsewhere @xcite . a second open question concerns the nature of the half - filled shell , specifically the observation that at @xmath432 there is a strong fqhe state that may be fully spin polarized , while there is no evidence for such a state at @xmath6 @xcite . descriptions of the @xmath432 state in terms of pfaffians or anti - pfaffians have claimed some success . but in the context of the present construction , the differences between the @xmath6 and @xmath432 states presents a puzzle , as non - degenerate lo candidate states could be constructed for either . the lo wave function would fully populate the lowest two lls in the case of @xmath432 , generating a contribution to the effective interaction for the valence electrons , but such a shift in single electron energies should not affected the correlations . this leads to the speculation that the differences between these two states must arise from corrections to the present construction , that is , from next - to - leading - order ( nlo ) contributions . many ets have an obvious expansion , a small parameter such as the ratio of yukawa masses in a potential with both short- and long - range components . but the coulomb interaction contains no such scales . instead , the natural et expansion for the coulomb interaction is that used in classical electromagnetism , the multipole expansion . the expansion parameter is not intrinsic to the potential , but rather to the geometry : @xmath453 where @xmath89 is the multipole order , @xmath61 is the size of the regions of charge , and @xmath454 their separation . because the fqhe is modeled as a one - component system in a neutralizing but static background , one anticipates that the nlo corrections would be associated with @xmath455 . while in the fll electrons are restricted to the lowest circular cyclotron orbit , this is not the case for the @xmath432 state , where the intrinsic cyclotron motion carries two quanta . the orbit can respond to higher multipoles . this problem has analogs elsewhere in physics , where the symmetry - violating effects connected with quadrupole interactions drive objects to deform , with symmetry then restored by the associated goldstone mode , the correlated low - frequency collective motion of the deformed objects . this physics is frequently most dramatic at half - filled shells , for the reasons apparent in the fqhe : the regions of over - density or under - density involve the correlations among the maximum number of quasi - electrons @xmath450 . that is , @xmath453 is at its most favorable point . the single - particle cyclotron orbits for valence electrons in the @xmath432 state can respond to the interactions beyond the monopole , and presumably do so in the process of minimizing their energy variationally . we are intrigued by the fact that the candidate et is the multipole expansion , that higher lls could respond to terms beyond the monopole , and that the consequence of such corrections are often startling , as they lead to symmetry breaking and restoration through long - wavelength collective modes . these possibilities should be explored . + * acknowledgments : * we thank joe ginocchio for the very enjoyable collaboration that produced the 1996 paper , susanne viefers for suggesting that we supply the missing reference 12 in gh , and byungmin kang and ken mcelvain for help and comments . this work was supported in part by the us doe under de - ac02 - 05ch11231 ( lbl ) de - sc00046548 ( berkeley ) , and de - fg02 - 94er14413 ( colorado ) . 000 d. c. tsui , h. l. stormer , and a. c. gossard , phys . rev . lett . * 48 * ( 1982 ) 1559 . m. i. dyakonov , in recent trends in theory of physical phenomena in high magnetic fields , ed . i. d. vagner et al . ( kluwer academic publishers , 2003 ) , pp . 75 - 88 ( arxiv : cond - mat/0209206 ) . j. k. jain , phys . rev * 63 * ( 1989 ) 199 , phys . b * 40 * ( 1989 ) 8079 , and phys . b * 41 * ( 1990 ) 7653;a . lopez and e. fradkin , phys . b * 44 * ( 1991 ) 5246 and * 47 * ( 1993 ) 7080 ; v. kalmeyer and s .- c . zhang , phys . b * 47 * ( 1993 ) 7312 ; b. i. halperin , p. a. lee , and n. read , phys . rev . b * 47 * ( 1993 ) 7312 . f. d. m. haldane , phys . * 51 * ( 1983 ) 605 . j. k. jain , indian j. phys . * 88 * ( 2014 ) 915 ; t. h. hansson , m. hermans , and s. viefers , phys . rev . b * 80 * ( 2009 ) 165330 ; j. surosa , s. viefers , and t. h. hansson , phys . rev . b * 83 * ( 2011 ) 235130 ; p. bonderson , phys . rev . lett . 108 ( 2012 ) 066806 ; t. kvorning , phys . rev . b * 87 * ( 2013 ) 245315 . d. t. son , phys . x * 5 * ( 2015 ) 031027 . b. i. halperin , p. a. lee , and n. read , phys . rev b * 47 * ( 1993 ) 7312 . g. w. moore and n. read , nucl . b * 360 * ( 1991 ) 362 . r. laughlin , phys . * 50 * ( 1983 ) 1395 . j. n. ginocchio and w. c. haxton , phys . * 77 * ( 1996 ) 1568 . j. k. jain and r. k. kamilla , int . b * 11 * ( 1997 ) 2621 . j. n. ginocchio and w. c. haxton , in _ symmetries in science iv _ , ed . b. gruber ( plenum press , new york , 1993 ) , p. 263 . i. g. macdonald , symmetric functions and hall polynomials , " oxford mathematical monographs ( 2nd edition ) ( oxford university press , 1995 ) . see , e.g. , r. de - picciotto et al . , nature * 389 * ( 1997 ) 162 . r. b. laughlin , rev . * 71 * ( 1998 ) 863 ; c. l. kane and m. p. a. fisher , phys . * 72 * ( 1994 ) 724 ; l. saminadayer , phys . rev . * 79 * ( 1997 ) 2526 . b. halperin , phys . * 52 * ( 1984 ) 1583 ; d. p. arovas , j. r. schrieffer , and f. wilczek , phys . * 53 * ( 1984 ) 722 . m. stone , talk available online at the kitp ( 2006 ) ; e. arddone , r. kedem , and m. stone , j. phys . a * 38 * ( 2004 ) 617 ; m. stone and r. roy , phys . b * 69 * ( 2004 ) 184511 . j. k. jain , http://users.phys.psu.edu/ jain / cftutorial.pdf , based on lecture notes , international summer school on topology in quantum matter , bangalore , india ( 2011 ) v. w. scarola , j. k. jain , and e. h. rezayi , phys . rev . * 88 * ( 2002 ) 216804 . r. f. morf , phys . * 80 * ( 1998 ) 1505 . n. read and r. rezayi , phys . b * 59 * ( 1999 ) 8084 . w. pan , h. l. stormer , d. c. tsui , l. n. pfeiffer , k. w. baldwin , and k. w. west , phys . * 90 * ( 2003 ) 016801 ; w. pan , k. w. baldwin , k. w. west , l. n. pfeiffer , and d. c. tsui , phys . b * 91 * ( 2015 ) 041301(r ) . daniel j. haxton and w. c. haxton , in preparation . r. willet , j. p. eisenstein , h. l. stormer , d. c. tsui , a. c. gossard , and j. h. english , phys . * 59 * ( 1987 ) 1776 . w. pan et al . * 83 * ( 1999 ) 3530 . j. p. eisenstein , k. b. cooper , l. n. pfeiffer , and k. w. west , phys . lett . * 88 * ( 2002 ) 076801 . issues differentiating the gh@xmath32 construction , which was guided by the scale invariance of the coulomb potential , from those emphasizing control of short - range physics , as the coulomb potential is strongest in the p - wave , can be explored by studying the three- and four - electron problems . the hilbert spaces can be defined algebraically for arbitrary @xmath276 in these cases . this appendix derives the most general four - electron wave function and shows its relevance to such correlation questions . the exercise also helps to illustrate that correlations beyond the two - electron ones considered by laughlin are more complicated , because they may not have simple embeddings in many - electron wave functions . the solution of the four - electron problem provided part of the motivation for the gh construction : the gh work is summarized in @xcite . here we extend that discussion to the plane , where the basis states can be enumerated by finding all translationally invariant symmetric polynomials . schur s lemma tells us that the most general few - electron correlation is the product of the antisymmetric closed shell wave function with a symmetric polynomial . thus it is sufficient to identify all such symmetric and translationally invariant polynomials . in the case laughlin studied , two interacting electrons , those polynomials are @xmath456^{(m-1)/2 } \odot [ z_2]^{(m-1)/2}$ ] , @xmath457 . the correlations are scalars , and thus also correspond to two - electron wave functions , making their identification as well as the embedding of this correlation in many - electron systems particularly simple : laughlin simply enforced this correlation on all electron pairs . as he did this for each pair in the most efficient way , the many - body states he constructed are those of maximum density for which the two - electron correlation constraint can be preserved for all pairs . the gh construction began with the observation that other hierarchy wave functions could not be described simply in term of two - electron correlations , as laughlin had exhausted all such possibilities . furthermore laughlin s restriction forbidding p - wave two - electron correlations clearly can not be maintain beyond @xmath458 . it is then natural to turn to the three - electron correlation to limit unfavorable interactions : a three - electron constraint can guarantee that electrons correlated in a p - wave ( iqhe two - electron droplets ) remain separated from any similar overdensity , minimizing the energy , at least until some critical density is reached . such separation can in fact be maintained ( in a defined way ) until @xmath211 . the elementary symmetric polynomials @xmath459,~f_2[z_1,z_2,z_3],~f_3[z_1,z_2,z_3 ] \ } = \{1,~ z_1+z_2+z_3,~ z_1 z_2+z_1 z_3 + z_2 z_3,~ z_1 z_2 z_3 \}$ ] form a basis for the most general symmetric three - electron wave function : as interaction energies only depend on relative coordinates , we are only interested in the subset of polynomials that are translationally invariant . two translationally invariant basis states can be defined , @xmath460 and @xmath461 , leading to the general form for antisymmetric translationally invariant polynomials @xmath462~ f_2^{\prime~p_2}~ f_3^{\prime~p_3}~~~~~~~~~~~~~~~~~~~ \nonumber \\ m \in \ { \mathrm{odd~integers } \}~~~~~p_2 \in \{0,1,2 \}~~~~~p_3 \in \{\mathrm{integers } \ } \nonumber \\ f_2^\prime = f_2-f_1 ^ 2/3~~~f_3^\prime = f_3 - f_1 f_2/3 + 2 f_1 ^ 3/27~~~~~~~\nonumber \\ f_2^\prime[z_1,z_2,z_3 ] \sim \mathcal{s } \left[(z_1-z_2)^2 \right]~~~~~~~~~~~~~~~~~\nonumber \\ ~f_3^\prime[z_1,z_2,z_3 ] \sim \mathcal{s } \left [ ( z_1-z_2)(z_1-z_3)^2 \right]~~~~~~~~~~~ \label{eq:3b } \end{aligned}\ ] ] while @xmath460 and @xmath461 are symmetric , we have broken the symmetry , then introduced the symmetrization operator @xmath463 , to show the underlying two - electron correlations . the restriction on @xmath464 comes from the observation that @xmath465 , so that powers @xmath466 can be eliminated ( while incrementing @xmath40 ) . the structure is that of band - heads labeled by @xmath40 : for fixed m , any pair @xmath467 corresponds to a unique number of quanta , 0,2,3,4,5 , .... thus degeneracies arise only when the @xmath100 band begins ( 9 quanta ) that is , with laughlin s @xmath100 state . thus the correlations corresponding to the higher densities of present interest ( @xmath458 ) are unique . the elementary translationally invariant polynomials @xmath460 and @xmath461 , however , are not scalars and thus are not allowed states . specifically , @xmath468^{3/2 } \otimes [ z_1 z_2 z_3]^{3/2 } \right]^1_{-1 } \nonumber \\ f_3^\prime & \sim & \left [ [ [ z_1]^{1/2}\otimes [ z_2 z_3]^1]^{1/2 } \otimes [ [ z_2]^{1/2 } \otimes [ z_1 z_3]^1]^{1/2 } \otimes [ [ z_3]^{1/2 } \right . \nonumber \\ & & ~~~\left . \otimes [ z_1 z_2]^1]^{1/2 } \right]^{3/2}_{-3/2}. \end{aligned}\ ] ] where , in the expression for @xmath461 , an aligned coupling of three spin-1/2 factors has been formed , with @xmath469^{1/2}\otimes [ z_2 z_3]^1]^{1/2}_{-1/2 } \sim z_2 ( z_1-z_3)+z_3(z_1-z_2)$ ] . these polynomials are translationally invariant because they are the lowest components of rank 1 and 3/2 vectors , not because they are scalars . they do not correspond to homogeneous wave functions . we suspect this result has been implicitly encountered before and may have generated confusion : unlike laughlin s case , one can not identify these correlations by studying three - electron states , nor can one build scalar many - electron wave functions by simply cobbling together products of three - electron scalar building blocks . it is easily proven that the only scalar three - electron states are those identified by laughlin , @xmath470 : on the sphere , haldane recognized this , showing that the only states allowed are @xmath471 where @xmath472 by antisymmetry . thus any attempt to build more general scalar wave functions based on simple products of three - electron factors is bound to fail . however one can construct a four - electron state in which this correlation is embedded this means that three - electron correlations with fewer than six quanta , @xmath473 and @xmath474 do not appear . gh accomplished this as follows , @xmath475_1 \odot [ z_3 z_4]_1 \right ] \nonumber \\ & & = \mathcal{s } \left [ \left\ { ( z_1-z_3)^2 ( z_1-z_4 ) \right\ } \left\{(z_2-z_3 ) ( z_2-z_4)^2 \right\ } \right.+ \nonumber \\ & & ~~~~~~\left . + \left\{(z_1-z_3)(z_1-z_4)^2 \right\ } \left\ { ( z_2-z_3)^2 ( z_2-z_4 ) \right\ } \right ] \nonumber \\ & & \equiv \mathcal{s } \left[g_6[\{z_1,z_2\},\{z_3,z_4\ } ] \right ] \nonumber \sim g_6^\prime[z_1,z_2,z_3,z_4 ] \label{eq : n4}\end{aligned}\ ] ] where @xmath476_1 \odot [ z_3 z_4]_1 = ( z_1-z_3)(z_2-z_4 ) + ( z_1-z_4)(z_2-z_3)$ ] , the spreading operator " of gh that plays a role for pairs of electrons analogous to laughlin s @xmath477 for individual electrons . @xmath478 , before symmetrization , is a simple product of two of the three - electron correlations associated with @xmath461 . the four - electron correlation is a scalar , and thus the antisymmetric state formed from it is a scalar : any three electrons selected from this state will have six quanta . one can contrast this four - electron plaquette with its @xmath100 laughlin counterpart @xmath479 ^ 2 ( z_3-z_4)^2\ ] ] with respect to laughlin s construction , eq . ( [ eq : n4 ] ) replaces separated pairs of electrons in relative @xmath480 waves by their @xmath11 counterparts , and deduced the number of quanta acting between the pairs from 12 to 10 , in the antisymmetric state . the latter determines the filling , changing it from 1/3rd- to 2/5ths - filled . gh , in analogy with laughlin and jain , then wrote this 4-electron wave function as an operator on the half - filled shell . in their spherical notation the four - electron operator is @xmath481_1 \odot [ u(3 ) u(4)]_1 l[n=4,2 ] . \label{eq : op}\ ] ] these arguments can be made more precise by deriving the most general form of the four - electron scalar wave function . the details will not be given here , as the analogous construction on a sphere can be found in ref . . the result has a familiar form , paralleling our three - electron result @xmath482 ~ g_4^{\prime~p_4}~ g_6^{\prime~p_6 } \nonumber \\ & & ~~ m \in \ { \mathrm{odd~integers } \}~~~~p_4 \in \{0,1,2 \}~~~~p_6 \in \{\mathrm{integers } \ } \nonumber \\ & & ~~~~ g_4^\prime = { 1 \over 12 } g_2 ^ 2 -{1 \over 4 } g_1 g_3 + g_4 \nonumber \\ & & ~~~~ g_6^\prime = { 2 \over 27 } g_2 ^ 3 -{1 \over 3 } g_1 g_2 g_3 + g_3 ^ 2 + g_1 ^ 2 g_4 -{8 \over 3 } g_2 g_4 \label{eq:4b } \end{aligned}\ ] ] where @xmath483 are the elementary polynomials for four particles , @xmath484 , ... , @xmath485 . we will ignore for the moment @xmath486 , which we recognize as an @xmath195 scalar in which the @xmath487 correlation @xmath460 is embedded @xmath488 ^ 2 \odot [ z_1 z_2 z_3 z_4]^2 \nonumber \\ & \sim & \left [ [ z_1 z_2 z_3]^{3/2 } \otimes [ z_1 z_2 z_3]^{3/2 } \right]^1 \odot [ z_4 ] ^1 \label{eq : g4 } \end{aligned}\ ] ] but which is rather uninteresting because it acts as a shift " operator . this leaves @xmath478 as the interesting new correlation , which we already observed is built on the three - electron correlation @xmath461 . more explicitly , @xmath489 \odot [ z_2 z_3 ] ~ [ z_2 z_4 ] \odot [ z_1 z_3 ] ~[z_3 z_4 ] \odot [ z_1 z_2 ] \nonumber \\ & \sim & \big [ [ [ z_1]\otimes [ z_2 z_3]^1]^{1/2 } \otimes [ [ z_2 ] \otimes [ z_1 z_3]^1]^{1/2 } \otimes \nonumber \\ & & ~~~ [ [ z_3 ] \otimes [ z_1 z_2]^1]^{1/2 } \big]^{3/2 } \odot [ z_4]^{3/2 } \label{eq : g6 } \end{aligned}\ ] ] eqs . ( [ eq : g4 ] ) and ( [ eq : g6 ] ) demonstrate that there is a one - to - one correspondence between the translationally invariant 3-electron correlations and four - electron scalar wave functions the latter is given by the former dotted into vectors formed from @xmath490 . this is an important result . in the simpler case laughlin investigated , the two - electron correlations were themselves scalars , and once one has scalars , it is easy to build many - electron wave functions from them . here we find that all of the physics contained in three - electron correlations maps uniquely into four - electron scalars . thus in analogy to laughlin s construction , many - electron wave functions respecting any three - electron correlation can be constructed by using the corresponding four - electron scalar building blocks . @xmath211 wave function is unique and has the property that every three - electron correlation contains six quanta . this restriction on the short - range behavior of the wave function can be imposed on configurations of arbitrary @xmath29 : the pattern is shown for @xmath438 , generating a valid scalar wave function when antisymmetrized . the text notes that wave functions optimized for their short range properties are much less successful numerically , illustrating that the scaleless coulomb potential is not a short - range interaction.,scaledwidth=48.0% ] this four - electron wave function is unique and thus corresponds to both the gh / jain and gh@xmath32 @xmath195 @xmath211 results , which are identical . because this state is self - conjugate , it also corresponds to the @xmath195 @xmath223 gh and gh@xmath32 wave functions . as this wave function guarantees that all three - electron correlations contain at least six quanta , it is then tempting to try to extend the construction to any even @xmath29 , preserving this property in the general @xmath211 wave function . this would then yield an approximate @xmath211 fqhe wave function that is the exact eigenfunction of a short - range interaction , effectively generalizing the haldane potential . ( clearly by combining delta functions with gradients , a three - body potential that vanishes among three electrons containing six quanta , but not among electrons correlated at even shorter distances , can be constructed . ) this in fact can be done , using the four - electron plaquettes described previously . as the construction is much easier to describe geometrically than algebraically , it is presented in fig . [ fig : fig25cor ] . this n - electron 2/5-filled state follows the 4-electron form of eq . ( [ eq : n4 ] ) @xmath491 ~ \mathcal{s } \left [ \prod_{i < j=1}^{n/2 } g_6[i_i , i_j ] \right ] \nonumber \\ & \equiv & l[n,1 ] \mathcal{s } \left [ g_6[i_1 , \dots i_{n/2 } ] \right]\end{aligned}\ ] ] where the @xmath29 electrons have been partitioned into @xmath184 pairs labeled by @xmath492 . the wave function is symmetric by construction under the interchanges @xmath493etc . , so it is sufficient to evaluate @xmath463 by symmetrizing the wave function over all electron interchanges that create distinct partitions of the @xmath184 pairs . in addition to being the eigenstate of a short - range effective hamiltonian , this wave function also has a simple recursion relation . despite its nice properties , such a wave function , optimized for its short - range behavior , is not an obvious candidate for a lo wave function . as the coulomb potential has no scale , it would be surprising if its physics could be captured in a short - range interaction . the gh@xmath32 and gh or jain @xmath211 wave functions generate nonzero correlations among three electrons carrying five quanta , for @xmath226 . we verified that these wave functions , for @xmath438 , have significantly better overlaps with the exact ground state wave function , than does the one we constructed via the pattern illustrated in fig . [ fig : fig25cor ] . this test , it seems to us , is definitive , as there is no arbitrariness in the short - range wave function : the four - body correlation one must use in a @xmath211 construction to eliminate configurations with fewer than six quanta is unique . it remains true that wave functions like those of gh / jain and gh@xmath32 , that are not optimized for their short - range behavior , nevertheless can constrain short - range correlations . laughlin s wave function is an example , as one has the haldane pseudopotential . similarly , the gh / jain and gh@xmath32 wave functions contain no three - body correlations containing fewer than five quanta , though obviously they are not uniquely defined by this condition . thus the example worked through here supports the naive observation that because the coulomb potential lacks a short - range scale , good approximate wave functions should not be optimized solely on considerations connected with short - range interactions .	a set of scalar operators , originally introduced in connection with an analytic first - landau - level ( fll ) construction of fractional quantum hall ( fqhe ) wave functions for the sphere , are employed in a somewhat different way to generate explicit representations of both hierarchy states ( e.g. , the series of fillings @xmath0=1/3 , 2/5 , 3/7 , ... ) and their conjugates ( @xmath0 = 1 , 2/3 , 3/5 , ... ) as non - interacting quasi - electrons filling fine - structure sub - shells within the fll . this yields , for planar and spherical geometries , a quasi - electron representation of the incompressible fll state of filling @xmath1 in a magnetic field of strength b that is algebraically identical to the iqhe state of filling @xmath2 in a magnetic field of strength @xmath3 . the construction provides a precise definition of the quasi - electron / composite fermion that differs in some respects from common descriptions : they are eigenstates of @xmath4 ; they and the fll subshells they occupy carry a third index @xmath5 that is associated with breaking of scalar pairs ; they absorb in their internal wave functions one , not two , units of magnetic flux ; and they share a common , simple structure as vector products of a spinor creating an electron and one creating magnetic flux . we argue that these properties are a consequence of the breaking of the degeneracy of noninteracting electrons within the fll by the scale - invariant coulomb potential . we discuss the sense in which the wave function construction supports basic ideas of both composite fermion and hierarchical descriptions of the fqhe . we describe symmetries of the quasi - electrons in the @xmath6 limit , where a deep fermi sea of quasi - electrons forms , and the quasi - electrons take on majorana and pseudo - dirac characters . finally , we show that the wave functions can be viewed as fermionic excitations of the bosonic half - filled shell , producing at @xmath6 an operator that differs from but plays the same role as the pfaffian .
let @xmath0 be a bounded open domain with @xmath14 boundary , @xmath9 and @xmath2 . in this paper we are concerned on study of the dirichlet problem @xmath15 & = & f \quad \mbox{in } \ \omega , \\ \label{exteriordata } u & = & 0 \quad \mbox{in } \ \omega^c,\end{aligned}\ ] ] where @xmath10 is a nonlocal operator approaching the fractional laplacian as @xmath5 approaches @xmath16 . we focus our attention on @xmath10 with the form @xmath17(x ) : = \int_{{\mathbb{r}}^n } [ u(x + z ) - u(x ) ] k_\epsilon(z)dz , \end{aligned}\ ] ] where , for @xmath18 fixed , @xmath19 is defined as @xmath20 notice that for each @xmath2 , @xmath19 is integrable in @xmath21 with @xmath22 norm equal to @xmath23 , where @xmath24 is a constant depending only on @xmath25 and @xmath26 . we point out that operators with kernel in @xmath22 , like @xmath10 , are known in the literature as _ zero order nonlocal operators_. operator @xmath10 is a particular case of a broad class of nonlocal elliptic operators . in fact , given a positive measure @xmath27 satisfying the _ lvy condition _ @xmath28 and , for each @xmath29 and @xmath30 bounded and sufficiently smooth at @xmath31 , the operator @xmath32(x)$ ] defined as @xmath33(x ) = \int_{{\mathbb{r}}^n } [ u(x + z ) - u(x ) - \mathbf{1}_{b_1(0)}(z ) \langle du(x ) , z \rangle]\mu(dz),\ ] ] has been a subject of study in a huge variety of contexts such as potential theory ( @xcite ) , probability ( @xcite ) and analysis ( @xcite ) . an interesting point of view of our problem comes from probability , since represents the infinitesimal generator of a _ jump lvy process _ , see sato @xcite . in our setting , the finiteness of the measure is associated with the so - called _ compound poisson process_. dirichlet problems with the form of - arise in the context of exit time problems with trajectories driven by the jump lvy process defined by @xmath34 , and the solution @xmath7 represents the expected value of the associated cost functional , see @xcite . we may start our discussion with a natural notion of solution to our problem - : we say that a bounded function @xmath30 , continuous in @xmath35 , is a solution of - if it satisfies pointwise in @xmath35 and @xmath36 on @xmath37 . as we see in section 2 , this problem has a unique solution , more interestingly , through an example we will see that such a solution may not be continuous in @xmath21 , since a discontinuity may appear on the boundary of @xmath35 . see remark [ rmku>0borde ] . this situation is in great contrast with the limit case @xmath38 , where the kernel becomes @xmath39 and the associated nonlocal operator is the _ fractional laplacian of order @xmath40 _ , denoted by @xmath41 , see @xcite . in this case , the corresponding dirichlet problem becomes @xmath42 where @xmath43 is a normalizing constant . in the context of the viscosity theory for nonlocal equations ( see @xcite ) , barles , chasseigne and imbert @xcite addressed a large variety of nonlocal elliptic problems including . in that paper , the authors proved the existence and uniqueness of a viscosity solution @xmath44 of satisfying @xmath45 on @xmath13 that is , consequently , continuous when we regard it as a function on @xmath21 . this result is accomplished by the use of a nonlocal version of the notion of viscosity solution with _ generalized boundary conditions _ , see @xcite for an introduction of this notion in the context of second - order equations . additionally , fractional problems like enjoy a _ regularizing effect _ as in the classical second - order case . roughly speaking , for a right - hand side which is merely bounded , the solution @xmath46 of is locally hlder continuous in @xmath35 , see @xcite . in fact , we should mention here that interior hlder regularity for more general fractional problems ( for which is a particular case ) has been addressed by many authors , see for instance @xcite and the classical book of landkof @xcite , for a non - exhaustive list of references . the interior hlder regularity is accomplished by well established elliptic techniques as the harnack s inequality ( @xcite ) and the ishii - lions method ( @xcite ) . in both cases , the nonintegrability of the kernel plays a key role . hlder regularity for problems like can be extended up to the boundary , as it is proved by ros - oton and serra in @xcite , where a boundary harnack s inequality is the key ingredient ( see also @xcite ) . naturally , as a byproduct of these regularity results , compactness properties are available for certain families of solutions of fractional equations . for instance , the family @xmath47 of functions solving @xmath48 satisfies compactness properties when @xmath49 is uniformly bounded in @xmath50 . for zero order problems , regularizing effects as arising in fractional problems are no longer available ( see @xcite ) . in fact , the finiteness of the kernel of zero order operators turns into degenerate ellipticity for which ishii - lions method can not be applied . thus , `` regularity results '' for zero order problems like - are circumscribed to the heritage of the modulus of continuity of the right - hand side @xmath51 to the solution @xmath7 as it can be seen in @xcite . however , the modulus of continuity found in @xcite depends strongly on the size of the @xmath22 norm of @xmath19 , which explodes as @xmath4 . a similar lack of stability as @xmath4 can be observed in the harnack - type inequality results for nonlocal problems found by coville in @xcite . hence , none of the mentioned tools are adequate for getting compactness for the family of solutions @xmath52 of problem - , which is a paradoxical situation since , in the limit case , the solutions actually get higher regularity and stronger compactness control on its behavior . in view of the discussion given above , a natural mathematical question is if there exists a uniform modulus of continuity in @xmath35 , for the family of solutions @xmath53 to - , and consequently compactness properties for it . in this direction , the main result of this paper is the following [ teo1 ] let @xmath0 a bounded domain with @xmath14 boundary and @xmath9 . for @xmath2 , let @xmath7 be a solution to problem - . then , there is a modulus of continuity @xmath54 depending only on @xmath51 , such that latexmath:[\ ] ] it is known that the unique viscosity solution @xmath7 of agrees the prescribed value of the equation on the boundary , and then @xmath471 on @xmath13 for all @xmath2 , see for example @xcite . we have @xmath58 is uniformly bounded in @xmath50 and therefore , the application of half - relaxed limits together with viscosity stability results in @xcite , imply @xmath57 locally uniform in @xmath35 as @xmath4 , where @xmath56 is the unique solution to . since @xmath56 is strictly positive in @xmath13 , the convergence of @xmath7 to @xmath56 can not be uniform in @xmath87 , and therefore the family @xmath58 is not equicontinuous in this case . this example resembles the behavior of the viscosity solutions @xmath7 of the equation @xmath472 which approximate the solution of the equation @xmath473 in this case , the family @xmath474 is not equicontinuous too , see @xcite . * acknowledgements : * the authors want to thank the referees for carefully reading the paper and making suggestions that resulted in a great improvement in the clarity and simplicity of the proofs . was partially supported by fondecyt grant # 1110291 , basal - cmm projects and capde , anillo act-125 . was partially supported by conicyt , grants capital humano avanzado and ayuda realizacin tesis doctoral . barles , g. and burdeau , j. _ the dirichlet problem for semilinear second - order degenerate elliptic equations and applications to stochastic exit time control problems _ comm . in pde . , 20 ( 1 - 2 ) , 129 - 178 ( 1995 ) .	in this paper we consider a smooth bounded domain @xmath0 and a parametric family of radially symmetric kernels @xmath1 such that , for each @xmath2 , its @xmath3norm is finite but it blows up as @xmath4 . our aim is to establish an @xmath5 independent modulus of continuity in @xmath6 , for the solution @xmath7 of the homogeneous dirichlet problem @xmath8 & = & f & \mbox{in } \ \omega . \\ u & = & 0 & \mbox{in } \ \omega^c , \end{array } \right .\ ] ] where @xmath9 and the operator @xmath10 has the form @xmath11(x ) = \frac12\int \limits_{{\mathbb{r}}^n } [ u(x + z ) + u(x - z ) - 2u(x)]k_\epsilon(z)dz\ ] ] and it approaches the fractional laplacian as @xmath12 . the modulus of continuity is obtained combining the comparison principle with the translation invariance of @xmath10 , constructing suitable barriers that allow to manage the discontinuities that the solution @xmath7 may have on @xmath13 . extensions of this result to fully non - linear elliptic and parabolic operators are also discussed .
in spite of progress in the understanding in evolution ever since darwin(1869 ) , the speciation is not yet fully understood . in the recent book , maynard - smith and szathmary(1995 ) wrote that _ we are not aware of any explicit model demonstrating the instability of a sexual continuum_. to discuss the problem of speciation , let us start from reviewing basic standpoints in evolution theory , although it might look too elementary here . * \(i ) existence of genotype and phenotype * \(ii ) fitness for reproduction is given as a function of the phenotype and the environment . the environment " can include interaction with other individuals . in other words , the reproduction rate of an individual is a function of its phenotype , and environment , i.e. , @xmath0phenotype , environment ) . * \(iii ) only the genotype is transferred to the next generation ( weissman doctrine ) * \(iv ) there is flow only from genotype to phenotype ( the central dogma of the molecular biology ) . for example , through the developmental process the phenotype is determined depending on the genotype . now , the process is summarized as _ genotype @xmath1 development @xmath1 phenotype_. here we adopt these standard assumptions . ( although the assumption ( iii ) may not be valid for some cases known as epigenetic inheritance , we accept the assumption here , since the relevance of epigenetic inheritance to evolution is still controversial , and the theory to be proposed is valid in the presence of epigenetic inheritance , but does not require it . ) in the standard evolutionary genetics , the assumption ( iv ) is further replaced by a stronger one , i.e. , ( iv ) `` phenotype is a _ single _ valued function of genotype '' . if this were always true , we could replace @xmath0phenotype , environment ) in ( i ) by @xmath2(genotype),environment ) and then we could discuss the evolutionary process in terms of the population dynamics only of genotypes ( and environment ) . this is the basic standpoint in population genetics . indeed , this reduction to genes is valid for gradual evolution . it is also supported by the following mathematical argument . the change of genotype is slower in time scale than that of phenotype . as is known , variables with slower time scale act as a control parameter to faster ones , if the time scale separation is large enough ( and if the dynamics in the fast time scale do not have such instability that leads to bifurcation ) . still , explanation of the speciation , especially sympatric speciation , is not so easy following this standard evolutionary genetics . if slight genetic change leads to slight phenotype change , then individuals arising from mutation from the same genetic group differ only slightly according to this picture . then , these individuals compete each other for the same niche . unless the phenotype in concern is neutral , it is generally difficult that two ( or more ) groups coexist . those with a higher fitness would survive . one possible way to get out of this difficulty is to assume that two groups are ` effectively ' isolated , so that they do not compete . some candidates for such isolation are searched . the most well - known example is spatial segregation , known as allopatric speciation . since we are here interested in sympatric speciation , this solution can not be adopted here furthermore , there are direct evidences that sympatric speciation really occurred in the evolution , for example in the speciation of cichlid in some lakes(schiliewen et al . 1994 ) . as another candidate for separation mating preference is discussed ( maynard - smith 1966 , felsenstein 1981 , grant 1981 , doebeli 1996 , howard and barlocher eds . recently , there have appeared some models showing the instability of sexual continuum , without assuming the existence of discrete groups in the beginning . probably , the argument based on the runaway is most persuasive ( lande 1981 , turner and burrows 1995,howard and barlocher eds . even though two groups coexist at the same spatial location , they can be genetically separated if two groups do not mate each other . hence , the mating preference is proposed as a mechanism for sympatric speciation . however , in this theory , why there is such mating preference itself is not answered . accordingly , it is not self - contained as a theory . another recent proposal is the introduction of ( almost ) neutral fitness landscape and exclusion of individuals with similar phenotypes ( dieckmann and doebeli , 1999 , kondrashov and kondrashov,1999 kawata and yoshimura 2001 ) . for example , dieckmann and doebeli[1999 ] have succeeded in showing that two groups are formed and coexist , to avoid the competition among organisms with similar phenotypes , assuming a rather flat fitness - landscape . this provides one explanation and can be relevant to some sympatric speciation . however , it is not so clear how the phenotype that is not so important as a fitness works strongly as a factor for exclusion for a closer value . furthermore , we are more interested in the differentiation of phenotypes that are functionally different and not neutral . so far , in these studies , the interaction between individuals lead to competition for their survival . difficulty in stable sympatric speciation without mating preference lies in the lack of a known clear mechanism how two groups , which have just started to be separated , coexist in the presence of mutual interaction . of course , if the two groups were in a symbiotic state , the coexistence could help the survival of each . however , the two groups have little difference in genotype in the beginning of speciation process , according to the assumption ( iv). then , it would be quite difficult to imagine such a ` symbiotic ' mechanism . now , the problem we address here is as follows : if we do not assume ( iv) ( but by assuming ( i)-(iv ) . ) , is there any mechanism that two groups mutually require each other for the survival in the beginning of the separation of the two groups , in the present paper we propose such mechanism , and provide a sympatric speciation scenario robust against fluctuations . note that the above difficulty comes from the assumption that the phenotype is a single - valued function of genotype . is this single - valued - ness always true ? to address this question , we reconsider the g - p relationship . indeed , there are three reasons that we doubt this single - valued - ness . first , yomo and his colleagues have reported that specific mutants of _ e. coli _ show ( at least ) two distinct types of enzyme activity , although they have identical genes(ko et al . , 1994 ) . these different types coexist in an unstructured environment of a chemostat ( ko et al . 1994 ) , and this coexistence is not due to spatial localization . coexistence of each type is supported by each other . indeed , when one type of _ e. coli _ is removed externally , the remained type starts differentiation again to recover the coexistence of the two types . the experiment demonstrates that the enzyme activity of these _ e. coli _ are differentiated into two ( or more ) groups , due to the interaction with each other , even though they have identical genes . here spatial factor is not important , since this experiment is carried out in a well stirred chemostat . second , some organisms are known to show various phenotype from a single genotype . this phenomenon is often related to malfunctions of a mutant ( holmes 1979 ) , and is called as low or incomplete penetrance(opitz 1981 ) . third , a theoretical mechanism of phenotypic diversification has already been proposed as the isologous diversification for cell differentiation(kaneko and yomo , 1994,1997,1999 ; furusawa and kaneko 1998 ) . the theory states that phenotypic diversity will arise from a single genotype and develop dynamically through intracellular complexity and intercellular connection . when organisms with plastic developmental dynamics interact with each other , the dynamics of each unit can be stabilized by forming distinct groups with differentiated states in the pheno - space . here the two differentiated groups are necessary to stabilize each of the dynamics . otherwise , the developmental process is unstable , and through the interaction the two types are formed again , when there is a sufficient number of units . this theoretical mechanism is demonstrated by several models and is shown to be a general consequence of coupled dynamical systems . the isologous diversification theory shows that there can be developmental ` flexibility ' , in which different phenotypes arise from identical gene sets , as in the incomplete penetrance aforementioned . now we have to study how this theory is relevant to evolution . indeed , the question how developmental process and evolution are related has been addressed over decades ( maynard - smith et al . we consider correspondence between genotype and phenotype seriously , by introducing a developmental process with which a given initial condition is lead to some phenotype according to a given genotype . ` development ' here means a dynamic process from an initial state to a matured state through rules associated with genes . ( in this sense , it is not necessarily restricted to multicellular organisms . ) to consider the evolution with developmental dynamics , it is appropriate to represent phenotype by a set of state variables . for example , each individual @xmath3 has variables @xmath4,@xmath5 , @xmath6 , which defines the phenotype . this set of variables can be regarded as concentrations of chemicals , rates of metabolic processes , or some quantity corresponding to a higher function characterizing the behavior of the organism . the state is not fixed in time , but develops from the initial state at birth to a matured state when the organism is ready to produce its offspring . the dynamics of the state variables @xmath4,@xmath5 , @xmath6 is given by a set of equations with some parameters . genes , since they are nothing but information expressed on dna , could in principle be included in the set of variables . however , according to the central dogma of molecular biology ( requisite ( iv ) ) , the gene has a special role among such variables . genes can affect phenotypes , the set of variables , but the phenotypes can not change the code of genes . during the life cycle , changes in genes are negligible compared with those of the phenotypic variables they control . in terms of dynamical systems , the set corresponding to genes can be represented by parameters @xmath7 that govern the dynamics of phenotypes , since the parameters in an equation are not changed through the developmental process , while the parameters control the dynamics of phenotypic variables . accordingly , we represent the genotype by a set of parameters . only when an individual organism is reproduced , this set of parameters changes slightly by mutation . for example , when @xmath8 represents the concentrations of metabolic chemicals , @xmath7 is the catalytic activity of enzymes that controls the corresponding chemical reaction . now , our model is set up as follows : \(1 ) dynamical change of states giving a phenotype : the temporal evolution of the state variables @xmath4,@xmath5 , @xmath6 is given by a set of deterministic equations , which are described by the state of the individual , and parameters @xmath7 ( genotype ) , and the interaction with other individuals . this temporal evolution of the state consists of internal dynamics and interaction . ( 1 - 1)the internal dynamics ( say metabolic process in an organism ) are represented by the equation governed only of @xmath4,@xmath5 , @xmath6 ( without dependence on @xmath9 ( @xmath10 ) ) , and are controlled by the parameter sets @xmath7 . ( 1 - 2 ) interaction between the individuals : the interaction is given through the set of variables @xmath11 . for example , we consider such interaction form that the individuals interact with all others through competition for some ` resources ' . the resources are taken by all the individuals , giving competition among all the individuals . since we are interested in sympatric speciation , we take this extreme all - to - all interaction , by taking a well stirred soup of resources , without including any spatially localized interaction . \(2 ) reproduction and death : each individual gives offspring ( or splits into two ) when a given ` maturity condition ' for growth is satisfied . this condition is given by a set of variables @xmath11 . for example , if @xmath11 represents cyclic process corresponds to a metabolic , genetic or other process that is required for reproduction , we assume that the unit replicates when the accumulated number of cyclic processes goes beyond some threshold . \(3 ) mutation : when each organism reproduces , the set of parameters @xmath12 changes slightly by mutation , by adding a random number with a small amplitude @xmath13 , corresponding to the mutation rate . the values of variables @xmath11 are not transferred but are reset to initial conditions . ( if one wants to include some factor of epigenetic inheritance one could assumed that some of the values of state variables are transferred . indeed we have carried out this simulation also , but the results to be discussed are not altered ( or confirmed more strongly ) . \(4 ) competition : to introduce competition for survival , death is included both by random removal of organisms at some rate as well as by a given death condition based on their state . for a specific example , see appendix . from several simulations satisfying the condition of the model in 2 , we have obtained a scenario for a sympatric speciation process(kaneko and yomo 2000,2002 ) the speciation process we observed is schematically shown in fig.1 , where the change of the correspondence between a phenotypic variable ( p " ) and a genotypic parameter ( g " ) is plotted at every reproduction event . this scenario is summarized as follows . in the beginning , there is a single species , with one - to - one correspondence between phenotype and genotype . here , there are little genetic and phenotypic diversity that are continuously distributed.(see fig.1a ) . we assume that the isologous diversification starts due to developmental plasticity with interaction , when the number of these organisms increase . indeed , the existence of such phenotypic differentiation is supported by isologous diversification , , and is supported by several numerical experiments . this gives the following stage - i . * stage - i : interaction - induced phenotypic differentiation * when there are many individuals interacting for finite resources , the phenotypic dynamics start to be differentiated even though the genotypes are identical or differ only slightly . phenotypic variables split into two ( or more ) types ( see fig.1b ) . this interaction - induced differentiation is an outcome of the mechanism aforementioned . slight phenotypic difference between individuals is amplified by the internal dynamics , while through the interaction between organisms , the difference of phenotypic dynamics tends to be clustered into two ( or more ) types . here the two distinct phenotype groups ( brought about by interaction ) are called ` upper ' and ` lower ' groups , tentatively . this differentiation is brought about , since the population consisting of individuals taking identical phenotypes is destabilized by the interaction . such instability is for example , caused by the increase of population or decrease of resources , leading to strong competition . of course , if the phenotype @xmath14 at a matured state is rigidly determined by developmental dynamics , such differentiation does not occur . only the assumption we make in the present theory is that there exists such developmental plasticity in the internal dynamics , when the interaction is strong . recall again that this assumption is theoretically supported . note that the difference is fixed at this stage neither at the genetic nor phenotypic level . after reproduction , an individual s phenotype can switch to another type . * stage - ii : amplification of the difference through genotype - phenotype relationship * at the second stage the difference between the two groups is amplified both at the genotypic and at the phenotypic level . this is realized by a kind of positive feedback process between the change of geno- and pheno - types . first the genetic parameter(s ) separate as a result of the phenotypic change . this occurs if the parameter dependence of the growth rate is different between the two phenotypes . generally , there are one or several parameter(s ) @xmath15 , such that the growth rate increases with @xmath15 for the upper group and decreases for the lower group ( or the other way round ) ( see fig.1c and fig.2 ) . certainly , such a parameter dependence is not exceptional . as a simple illustration , assume that the use of metabolic processes is different between the two phenotypic groups . if the upper group uses one metabolic cycle more , then the mutational change of the parameter @xmath15 to enhance the cycle is in favor for the upper group , while the change to reduce it may be in favor for the lower group . indeed all numerical results support the existence of such parameters . this dependence of growth rate on the genotypes leads to the genetic separation of the two groups , as long as there is competition for survival , to keep the population numbers limited . the genetic separation is often accompanied by a second process , the amplification of the phenotypic difference by the genotypic difference . in the situation of fig.1c , as a parameter @xmath16 increases , a phenotype @xmath17 ( i.e. , a characteristic quantity for the phenotype ) tends to increase for the upper group , and to decrease ( or to remain the same ) for the lower group . it should be noted that this second stage is _ always _ observed in our model simulation when the phenotypic differentiation at the first stage occurred . as a simple illustration , assume that the use of metabolic processes is different between the two groups . if the upper group uses one metabolic cycle more , then the mutational change of the parameter @xmath18 ( e.g. , enzyme catalytic activity ) to enhance the cycle is in favor for the upper group , while the change to reduce it may be in favor for the lower group . indeed , all the numerical results carried out so far support that there always exist such parameters . this dependence of growth on genotypes leads to genetic separation of the two groups . * stage - iii genetic fixation * after the separation of two groups has progressed , each phenotype ( and genotype ) starts to be preserved by the offspring , in contrast to the situation at the first stage . however , up to the second stage , the two groups with different phenotypes can not exist in isolation by itself . when isolated , offspring with the phenotype of the other group starts to appear . the two groups coexist depending on each other ( see fig.1d ) . only at this third stage , each group starts to exist by its own . even if one group of units is isolated , no offspring with the phenotype of the other group appears . now the two groups exist on their own . such a fixation of phenotypes is possible through the evolution of genotypes ( parameters ) . in other words , the differentiation is fixed into the genes ( parameters ) . now each group exists as an independent ` species ' , separated both genetically and phenotypically . the initial phenotypic change introduced by interaction is fixed to genes , and the ` speciation process ' is completed . at the second stage , the separation is not fixed rigidly . units selected from one group at this earlier stage again start to show phenotypic differentiation , followed by genotypic separation , as demonstrated by several simulations . after some generations , one of the differentiated groups recovers the geno- and phenotype that had existed before the transplant experiment . this is in strong contrast with the third stage . at the third stage , two groups with distinct genotypes and phenotypes are formed , each of which has one - to - one mapping from genotype to phenotype . this stage now is regarded as speciation ( in the next section we will show that this separation satisfies hybrid sterility in sexual reproduction , and is appropriate to be called speciation ) . when we look at the present process only by observing initial population distribution ( in fig.1a ) and the final population distribution ( in fig.1d ) , without information on the intermediate stages given by fig . 1b ) and 1c ) , one might think that the genes split into two groups by mutations and as a result , two phenotype groups are formed , since there is only a flow from genotype to phenotype . as we know the intermediate stages , however , we can conclude that this simple picture does not hold here . here phenotype differentiation drives the genetic separation , in spite of the flow only from genotype to phenotype . phenotype differentiation is consolidated to genotype , and then the offspring take the same phenotype as their ancestor . note that the speciation process of ours occurs under strong interaction . at the second stage , these two groups form symbiotic relationship . as a result , the speciation is robust in the following sense . if one group is eliminated externally , or extinct accidentally at the first or second stage , the remaining group forms the other phenotype group again , and then the genetic differentiation is started again . the speciation process here is robust against perturbations . note that each of the two groups forms a niche for the other group s survival mutually , and each of the groups is specialized in this created niche . for example , some chemicals secreted out by one group are used as resources for the other , and vice versa . hence the evolution of two groups are mutually related . at the first and second stages of the evolution , the speed for reproduction is not so much different between the two groups . indeed , at these stages , the reproduction of each group is strongly dependent on the other group , and the ` fitness ' as a reproduction speed of each group by itself alone can not be defined . at the stage - ii , the reproduction of each group is balanced through the interaction , so that one group can not dominate in the population ( see fig.2 ) . if phenotypic differentiation at the stage 1 occurs in our model , then the genetic differentiation of the later stages _ always _ follows , in spite of the random mutation process included . how long it takes to reach the third stage can depend on the mutation rate , but the speciation process itself does not depend on the mutation rate . however small the mutation rate may be , the speciation ( genetic fixation ) always occurs . once the initial parameters of the model are chosen , it is already determined whether the interaction - induced phenotype differentiation will occur or not . if it occurs , then always the genetic differentiation follows . on the other hand , in our setting , if the interaction - induced differentiation does not exist initially , there is no later genetic diversification process . if the initial parameters characterizing nonlinear internal dynamics or the coupling parameters characterizing interaction are small , no phenotypic differentiation occurs . also , the larger the resource per individual is , the smaller the effective interaction is . then , phenotypic differentiation does not occur . in these cases , even if we take a large mutation rate , there does not appear differentiation into distinct genetic groups , although the distribution of genes ( parameters ) is broader . or , we have also made several simulations starting from a population of units with widely distributed parameters ( i.e. , genotypes ) . however , unless the phenotypic separation into distinct groups is formed , the genetic differentiation does not follow . ( fig.4a and b ) only if the phenotype differentiation occurs , the genetic differentiation follows(fig.4c ) . for some other models with many variables and parameters , the phenotypes are often distributed broadly , but continuously without making distinct groups . in this case again , there does not appear distinct genetic groups , through the mutations , although the genotypes are broadly distributed . ( see fig.5 ) . the speciation process is defined both by genetic differentiation and by reproductive isolation ( dobzhansky 1937 ) . although the evolution through the stages i - iii leads to genetically isolated reproductive units , one might still say that it should not be called ` speciation ' unless the process shows isolated reproductive groups under the sexual recombination . in fact , it is not trivial if the present process works with sexual recombination , since the genotypes from parents are mixed by each recombination . to check this problem , we have considered some models so that the sexual recombination occurs to mix genes . to be specific , the reproduction occurs when two individuals @xmath19 and @xmath20 satisfy the maturity condition , and then the two genotypes are mixed . as an example we have produced two offspring @xmath21 and @xmath22 , from the individuals @xmath19 and @xmath20 as @xmath23 with a random number @xmath24 to mix the parents genotypes ( see also appendix 11.2 ) . in spite of this strong mixing of genotype parameters , the two distinct groups are again formed . of course , the mating between the two groups can produce an individual with the parameters in the middle of the two groups . when parameters of an individual take intermediate values between those of the two groups , at whatever phenotypes it can take , the reproduction takes much longer time than those of the two groups . before the reproduction condition is satisfied , the individual has a higher probability to be removed by death . as the separation process to the two groups further progresses , an individual with intermediate parameter values never reaches the condition for the reproduction before it dies . this post - mating isolation process is demonstrated clearly by measuring the average offspring number of individuals over given parameter ( genotype ) ranges and over some time span . an example of this average offspring number is plotted in fig.3 , with the progress of the speciation process . as the two groups with distinct values of parameters are formed , the average offspring number of an individual having the parameter between those of the two groups starts to decrease . soon the number goes to zero , implying that the hybrid between the two groups is sterile . in this sense , sterility ( or low reproduction ) of the hybrid appears as a * result , without any assumption on mating preference. now genetic differentiation and reproductive isolation are satisfied . hence it is proper to call the process through the stages i - iii as speciation . so far we have not assumed any preference in mating choice . hence , a sterile hybrid continues to be born . then it is natural to expect that some kind of mating preference evolves to reduce the probability to produce a sterile hybrid . here we study how mating preference evolves as a result of post - mating isolation . as a simple example , it is straightforward to include loci for mating preference parameters . we assume another set of genetic parameters that controls the mating behavior . for example , each individual @xmath3 has a set of mating threshold parameters @xmath25 , corresponding to the phenotype @xmath26 . if @xmath27 for some @xmath28 , the individual @xmath19 denies the mating with @xmath20 even if @xmath19 and @xmath20 satisfy the maturity condition . in simulation with a model with @xmath29 , we choose a pair of individuals that salsify the maturity condition , and check if one does not deny the other . only if neither denies the mating with the other , the mating occurs to produce offspring , when the genes from parents are mixed in the same way as as in the previous section . if these conditions are not satisfied , the individuals @xmath19 and @xmath20 wait for the next step to find a partner again ( see also appendix 11.3 ) . here the set of @xmath30 is regarded as a set of ( genetic ) parameters , and changes by mutation and recombination . the mutation is given by addition of a random value to @xmath31 . initially all of @xmath32 ( for @xmath33 ) are smaller than the minimal value of @xmath26 , so that any mating preference doe not exist . if some @xmath34 gets larger than some of @xmath35 , there appears mating preference . an example of numerical results is given in fig.5 , where the change of phenotype @xmath36 and some of the parameters @xmath37 , are plotted . here , by the phenotype differentiation , one group ( to be called ` up ' group ) has a large @xmath38 value for some @xmath39 and almost null values for some other @xmath40 . hence , sufficiently large positive @xmath41 gives a candidate for mating preference . right after the formation of two genetically distinct groups that follows the phenotype separation , one of the mating threshold parameters ( @xmath42 ) starts to increase for one group . in the example of the figure , ` up ' group has phenotype with ( large @xmath43 , small @xmath44 ) and the other ( ` down ' ) group with ( small @xmath43 , large @xmath44 ) . there the ` up ' group starts to increase @xmath45 , and @xmath46 is satisfied for an individual @xmath47 of the ` down ' group . now the mating between the two groups is no more allowed , and the mating occurs only within each group . the mating preference thus evolved prohibits inter - species mating producing sterile hybrid . note that the two groups do not simultaneously establish the mating preference . in some case , only one group has positive @xmath48 , which is enough for the establishment of mating preference , while in some other cases one group has positive @xmath49 , and the other has positive @xmath50 , where the mating preference is more rigidly established . although the evolution of mating preference here is a direct consequence of the post - mating isolation , it is interesting to note that the coexistence of the two species is further stabilized with the establishment of mating preference . without this establishment , there are some cases that one of the species disappears due to the fluctuation after very long time in the simulation . with the establishment , the two species coexist much longer ( at least within our time of numerical simulation ) . in diploid , there are two alleles , and two alleles do not equally contribute to the phenotype . for example , often only one allele contributes the control of the phenotype . if by recombination , the loci from two alleles are randomly mixed , then the correlation between genotype and phenotype achieved by the mechanism so far discussed might be destroyed . indeed , this problem was pointed out by felsenstein ( 1981 ) as one difficulty for sympatric speciation . of course , this problem is resolved , if genotypes from two alleles establish high correlation . to check if this correlation is generated , we have extended our model to have two alleles , and examined if the two alleles become correlated . here , we adopted the model studied so far , and added two alleles further ( see also appendix 11.4 ) . in mating , the alleles from the parents are randomly shuffled for each locus . in other words , each organism @xmath3 has two sets of parameters @xmath51 and @xmath52 . each @xmath53 is inherited from either @xmath54 or @xmath55 of one of the parents , and the other @xmath56 is inherited from either @xmath54 of @xmath55 of the other parents . here parameters at only one of the alleles work as a control parameter for the developmental dynamics of phenotype . we have carried out some simulations of this version of our model ( kaneko , unpublished ) . here again , the speciation proceeds in the same way , through the stages i , ii , and iii . hence our speciation scenario works well in the presence of alleles . in this model , the genotype - phenotype correspondence achieved at the stage iii , could be destroyed if there were no correlation between two alleles . hence we have plotted the correlation between two alleles by showing two - dimensional pattern @xmath57 in fig.7 . initially there was no correlation , but through temporal evolution , the correlation is established . in other words , the speciation in phenotype is consolidated to genes , and later is consolidated to the correlation between two alleles . as already discussed , our speciation proceeds , starting from phenotypic differentiation , then to genetic differentiation , and then to post - mating isolation , and finally to pre - mating isolation ( mating preference ) . this ordering might sound strange from commonly adopted viewpoint , but we have shown that this ordering is a natural and general consequence of a system with developmental dynamics with potential plasticity by interaction . in a biological system , we often tend to assume causal relationship * between two factors , from the observation of just * correlation * of the two factors . for example , when the resident area of two species , which share a common ancestor species , is spatially separated , we often guess that the spatial separation is a cause for the speciation . indeed , allopatric speciation is often adopted for the explanation of speciation in nature . however , in many cases , what we observed in field is just correlation between spatial separation and speciation . which is the cause is not necessarily proved . rather , spatial segregation can be a result of ( sympatric ) speciation . by extending our theory so far , we can show that spatial separation of two species is resulted from the sympatric speciation discussed here . to study this problem , we have extended our model by allocating to each organism a resident position in a two - dimensional space . each organism can move around the space randomly but slowly , while resources leading to the competitive interaction diffuse throughout space much faster . if the two organisms that satisfy the maturation condition meet in the space ( i.e , they are located within a given distance ) , then they mate each other to produce offspring . in this model , we have confirmed that the sympatric speciation first occurs through the stages i - iii in 3 . later these two differentiated groups start to be spatially segregated , as shown in fig.8 . now sympatric speciation is shown to be consolidated to spatial segregation ( kaneko , in preparation ) . the spatial segregation here is observed when the range of interaction is larger than the typical range of mating . for example , if mobility of resources causing competitive interaction is larger than the mobility of organisms , spatial segregation of symptarically formed species is resulted . instead of spatially local mating process , one can assume slight gradient of environmental condition , for example , as gradient in resources . in this case again , sympatric speciation is expected to be later fixed to spatial separation . to sum up , we have pointed out here the possibility that some of speciation that are considered to be allopatric can be a result of sympatric speciation of our mechanism . the sympatric speciation is later consolidated to spatial segregation of organisms . our theory reviewed here is related with several earlier theories , but is conceptually different . here we will briefly discuss these points . since our mechanism crucially depends on the interaction , one might think that it is a variant of frequency - dependent selection . the important difference here is that phenotype may not be uniquely determined by the genotype , even though the environment ( including population of organisms ) is given . in the frequency - dependent selection , genetically ( and accordingly phenotypically ) different groups interact with each other , and the fitness depends on the population of each group ( futsuyma , 1986 ) . at the third stage of our theory , the condition for this frequency - dependent selection is satisfied , and the evolution progresses with the frequency - dependent selection . however , the important point in our theory lies in the earlier stages where a single genotype leads to different phenotypes . indeed this intrinsic nature of differentiation is the reason why the speciation process here works at any ( small ) mutation rate and also under sexual recombination , without any other ad hoc assumptions . in our theory phenotype change is later consolidated to genotype . indeed , genetic ` takeover ' of phenotype change was also discussed as baldwin s effect , where the displacement of phenotypic character is fixed to genes . in the discussion of baldwin s effect , the phenotype character is given by epigenetic landscape ( waddington,1957 ) . in our case , the phenotype differentiation is formed through developmental process to generate different characters due to the interaction . distinct characters are stabilized each other through the interaction . with this interaction dependence , the two groups are necessary with each other , and robust speciation process is resulted . hence , the fixation to genotype in our theory is related with baldwin s effect , but the two are conceptually different . since the separation of two groups with distinct phenotypes is supported by the interaction , the present speciation mechanism is possible without supposing any mating preference . in fact , the hybrid becomes inferior in the reproduction rate , and the mating preference based on the discrimination in phenotype is shown to evolve . indeed , a mechanism to amplify the differentiation by mating preference was searched for as reinforcement since dobzhansky[1951 ] . our theory also gives a plausible basis for the evolution of mating preference without assuming ad - hoc reinforcement , or without any presumption on the inferiority in hybrid . note that our phenotypic differentiation through development is different from the so called ` phenotypic plasticity ' , in which a single genotype produces * alternative phenotypes * in * alternative environments(callahan , pigliucci and schlichting 1987 ; spitze and sadler 1996 ; weinig 2000 ) . in contrast , in our case , distinct phenotypes from a single genotype are formed under the same environment. in fact , in our model , this phenotypic differentiation is necessary to show the later genetic differentiation . without this differentiation , even if distinct phenotypes appear for different environments as in ` phenotypic plasticity ' , genetic differentiation does not follow . in spite of this difference , it is true that both are concerned with flexibility in phenotypes . some of phenotypic plasticity so far studied may bring about developmental flexibility of ours , under a different environmental condition . in our case , competitive interaction is relevant to speciation . indeed , coexistence of two ( or more ) species after the completion of the speciation is discussed as the resource competition by tilman[1976,1981 ] . although his theory gives an explanation for the coexistence , the speciation process is not discussed , because two individuals with a slight genotypic difference can have only a slight difference there . in our theory , even if the genotypes of two individuals are the same or only slightly different , their phenotypes can be of quite different types . accordingly , our theory provides a basis for resource competition also . general conclusion of our theory is that sympatric speciation can generally occur under strong interaction , if the condition for interaction - induced phenotype differentiation is satisfied . we briefly discuss relevance of our theory to biological evolution . since the present speciation is triggered by interaction , the process is not so much random as deterministic . once the interaction among individuals brings about phenotypic diversification , speciation always proceeds directionally without waiting for a rare , specific mutation . the evolution in our scenario has a ` deterministic ' nature and a fast tempo for speciation , which is different from a typical ` stochastic ' view of mutation - driven evolution . some of the phenotypic explosions in the history of evolution have been recorded as having occurred within short geologic periods . following these observations , punctuated equilibrium was proposed [ gould and eldegridge 1977 ] . our speciation scenario possibly gives an interpretation of this punctuated equilibrium . it may have followed the deterministic and fast way of interaction - induced speciation . in the process of speciation , the potentiality of a single genotype to produce several phenotypes is consumed and may decline . after the phenotypic diversification of a single genotype , each genotype newly appears by mutation and takes one of the diversified phenotypes in the population . thus , the one - to - many correspondence between the original genotype and phenotypes is consumed . through the present process of speciation , the potentiality of single genotypes to produce various phenotypes decreases unless the new genotypes introduce another positive feedback process to amplify the small difference . as a result , one may see single genotypes expressing only one ( or a small number of ) phenotypes in nature . since most organisms at the present time have gone through several speciation processes , they may have reduced their potentiality to produce various phenotypes . according to our theory , if the organisms have a high potentiality , they will undergo a speciation process before long and the potentiality will decrease . in other words , natural organisms tend to lose the potentiality to produce various phenotypes in the course of evolution . as a reflection on the evolutionary decline of the potentiality , one can expect that mutant genotypes tend to have a higher potentiality than the wild - type genotype . as mentioned in 1 , the low or incomplete penetrance(opitz 1981 ) is known to often occur in mutants , compared with higher penetrance in a wild type . our result is consistent with these observation , since wild types are in most cases , a consequence of evolution , where the one - to - one correspondence is recovered , while the mutants can have higher potentiality to have a loose correspondence . relationship between development and evolution has been discussed extensively . our theory states the relevance of developmental plasticity to speciation . taking our results and experimental facts into account , one can predict that organisms emerging as a new species have a high potentiality to produce a variety in phenotypes . it is interesting to discuss why insects , for example , have higher potentiality to speciation from this viewpoint . also examining if living fossils , such as _ latimeria chalumnae _ , _ limulus _ and so forth , have a stable expression of a small number of phenotypes . in our speciation theory , plasticity is declined through the evolution . of course , there should be some occasions when the potentiality is regained , so that the evolution continues . for example , change of environment may influence the developmental dynamics to regain loose correspondence , or introduction of novel degrees of freedom or genes may provide such looseness . endosymbiosis can be one of such causes . also , change of the interaction through spatial factor may introduce novel instability in dynamics , resulting in the loose correspondence . one important point in our theory is that the speciation in asexual and sexual organisms are explained within the same theory . of course , the standard definition of species using hybrid sterility is applied only for sexual organisms . however , it is true that the asexual organisms , or even bacteria , exhibit discrete geno- and pheno - types . it is suggested that ` species ' , i.e. , discrete types with reproductive isolation , may exist in asexual organisms ( roberts and cohan 1995 , holman 1987 ) . there are also discussions that the potentiality of speciation in asexual organisms is not lower than the sexual organisms . in this sense , the present theory sheds a new light to the problem of speciation in asexual organisms as well . according to our theory , sympatric speciation under sexual reproduction starts first from phenotypic differentiation , and then genetic diversification takes place , leading to hybrid sterility , and finally the speciation is fixed by mating preference . this order may be different from studies most commonly adopted . hence , our theory will be verified by confirming this chronic order in the field . one difficulty here , however , lies in that the process from phenotypic differentiation to the last stage is rather fast according to our theory . still , it may be possible to find this order in the field , by first searching for phenotypic differentiation of organisms with identical genotype and under the identical environment . in this respect , the data of cichlid of nicaraguan lake may be promising ( wilson , noack - kunnmann , and meyer 2000 ) , since phenotypic difference corresponding to different ecological niche is observed even though clear genetic difference is not observed yet . discussion on the mechanism of evolution using past data , however , often remains anyone s guess . most important in our scenario , in contrast , is its experimental verifiability , since the process of speciation is rather fast . for example , the evolution of _ e. coli _ is observed in the laboratory , as has been demonstrated by kashiwagi et al.(1998 , 2001 ) and w .- z . xu et al.(1996 ) . as mentioned in 1 , phenotypic differentiation of _ e. coli _ is experimentally observed when their introduction is strong . since the strength of interaction can be controlled by the resources and the population density , one can check whether or not the evolution in genetic level is accelerated through interaction - induced phenotypic diversification ( kashiwagi et al . , examination of the validity of our speciation scenario will give a first step to such study . to sum up , we have shown that developmental plasticity induced by interaction leads to phenotypic differentiation , which is consolidated to genes . thus , distinct species with distinct geno- and pheno types are formed . this leads to hybrid sterility , and later mating preference evolves . further later , this differentiation can be fixed to correlation in alleles or to spatial segregation . how the original differentiation in phenotypes can be understood as symmetry breaking from a homogeneous state , in the term of physics , while successive consolidation of the broken symmetry to different properties observed at later stages is more important for biological evolution . this dynamic process of consolidation is a key issue in development and evolution ( see also ( newman , 2002 ) ) . acknowledgment * the author would like to thank tetsuya yomo for collaboration in studies on which the present paper is based . he would also like to thank hiroaki takagi and chikara furusawa for useful discussions , and masakazu shimada , jin yoshimura , masakado kawata , and stuart newman for illuminating suggestions . the present study is supported by grants - in - aid for scientific research from the ministry of education , culture , sports , science and technology of japan ( 11ce2006 ) . we study a simple abstract model of evolution with an internal dynamical process for development . in the model , each individual @xmath3 has several ( metabolic or other ) cyclic processes , and the state of the @xmath58-th process at time @xmath59 is given by @xmath60 . with @xmath61 such processes , the state of an individual is given by the set @xmath62,@xmath63 , @xmath64 , which defines the phenotype . this set of variables can be regarded as concentrations of chemicals , rates of metabolic processes , or some quantity corresponding to a higher function . the state changes temporally according to a set of deterministic equations with some parameters . to be specific , our toy model consists of the following dynamics : \(1 ) dynamics of the state : here , we split @xmath65 into its integer part @xmath66 and the fractional part @xmath67 $ ] . the integer part @xmath68 is assumed to give the number of times the cyclic process has occurred since the individual s birth , while the fractional part @xmath69 gives the phase of oscillation in the process . the dynamics of the variables @xmath65 consist of a mutual influence of cyclic processes and interaction with other organisms . as a simple example , the former is assumed to be given by @xmath70 , while the latter is given by the competition for resources among the @xmath71 organisms existing at the moment , given by @xmath72 . ( the second term comes from the constraint @xmath73 , i.e. , the condition that @xmath74 individuals compete for a given resource @xmath75 at each time step . the first term represents the ability to secure the resource , depending on the state . ) our toy model is given by \(2 ) growth and death : each individual splits into two when a given condition for growth is satisfied . taking into account that the cyclic process corresponds to a metabolic , genetic or other process that is required for reproduction , we assume that the unit replicates when the accumulated number of cyclic processes goes beyond some threshold . thus , the condition is given by @xmath76 ( the maturity condition ) . the state @xmath77 is reset to a random value between 0 and 1 , when the corresponding individual splits . to introduce competition , individuals are eliminated by a given death condition , as well as by random removal with a given rate . as for the former condition , an individual with @xmath78 ( i.e. , with a reverse process ) is removed . \(3 ) genetic parameter and mutation : following the discussion in the text , genes are represented as parameters in the model , since the control parameters affect the dynamics of phenotypic variables , but no direct reverse process exists , as dictated by the central dogma of molecular biology . here , genotypes are given by a set of parameters @xmath79 , representing the relationship between the two cyclic processes @xmath28 and @xmath80 ( @xmath81 ) . this set of parameters changes slightly through mutation when offspring is reproduced . with each division , the parameters @xmath82 are changed to @xmath83 with @xmath13 , a random number over @xmath84 $ ] , with small @xmath85 , corresponding to the mutation rate . in the present model , due to the nonlinear nature of the dynamics , @xmath86 often oscillates in time chaotically or periodically . hence it is natural to use @xmath87 including its integer part , as a representation of the phenotype , since its integer part represents the number of cyclic process used for reproduction . we have also carried out some simulations of a model with reaction network , where @xmath88 represents the @xmath80th chemical concentration of an individual @xmath3 . each individual gets resources depending on its internal state . through the above catalytic reaction process , some products are synthesized from the resources . when they are beyond a given threshold , they split to two , as given in the model for isologous diversification ( kaneko and yomo , 1994,1997,1999 , furusawa and kaneko , 1998 ) . with the increase of the number of individuals , they compete for resources , while they are removed randomly to include competition . since genes code the catalytic activity of enzymes , the rate of each reaction in the catalytic network is controlled by a gene . hence , as a genetic parameter @xmath15 , the parameter for each reaction rate is adopted . through the mutation to this reaction rate , the speciation process discussed throughout the paper is also observed ( takagi , kaneko , yomo 2000 ) . to include sexual recombination , we have extended our model so that organisms satisfying the threshold condition mate to reproduce two offspring . when they mate , the offspring have parameter values that are intermediate of those of the parents . here , the offspring @xmath21 and @xmath22 are produced from the individuals @xmath19 and @xmath20 as here the set of @xmath30 is introduced as a set of ( genetic ) parameters , and changes by mutation and recombination . the mutation is given by addition of a random value over @xmath90 $ ] . initially @xmath91 ( for @xmath33 is set smaller than the minimal value of @xmath26 , so that any mating preference does not exist . if @xmath92 gets larger than some of @xmath93 there appears mating preference . here we assume that each individual has two sets of parameters @xmath94 , given by @xmath95 and @xmath96 . in mating , the alleles from the parents are randomly shuffled for each locus @xmath97 . each @xmath98 is inherited from either @xmath99 or @xmath100 of one of the parents @xmath101 , and the other @xmath102 is inherited from either @xmath103 or @xmath104 of the other parents @xmath105 . for the dynamics for @xmath106 , only @xmath107 is used . the other parameter @xmath108 is not used , but can be used as @xmath109 of the offspring after the shuffling . to an individual @xmath3 in the model of 11.2 ( with sexual reproduction ) , spatial position @xmath110 is assigned . the individual shows brownian motion in the 2-dimensional space , by adding random number over @xmath111 $ ] to @xmath110 per each step . they move within a square of a given suze with a periodic boundary condition . if two individuals @xmath3 and @xmath58 that satisfy the maturity condition ( @xmath112 ) are within a given distance @xmath113 , they can reproduce two offspring , which are located between @xmath110 and @xmath114 . 1 . c. a. beam , r. m. preparata , m. himes , d. l. nanney , ribosomal rna sequencing of members of the crypthecodinium cohnii ( dinophyceae ) species complex ; comparison with soluble enzyme studies . journal of eukaryotic microbiology . 40(5):660 - 667 , ( 1993 ) . 2 . callahan h.s . , pigliucci m. , and schlichting c.d . 1997 , developmental phenotypic plasticity : where ecology and evolution meet molecular biology , bioessays * 19 * 519 - 525 3 . coyne j.a . , & orr h.a . , the evolutionary genetics of speciation " , phil . trans . london * b 353 * 287 - 305 ( 1998 ) 4 . darwin c. _ on the origin of species by means of natural selection or the preservation of favored races in the struggle for life _ ( murray , london,1859 ) . dieckmann u. & doebeli m. , on the origin of species by sympatric speciation " , _ nature _ * 400 * 354 - 357 ( 1999 ) 6 . doebeli m. a quantitative genetic competition model for sympatric speciation " j. evol . * 9 * 893 - 909 ( 1996 ) 7 . dobzhansky t. , _ genetics and the origin of species _ ( columbia univ . press . ( 1937,1951 ) 8 . felsenstein j. 1981 , skepticism towards santa rosalia , or why are there so few kinds of animals ? , evolution * 35 * 124 - 138 9 . furusawa c. & kaneko k. , emergence of rules in cell society : differentiation , hierarchy , and stability " bull.math.biol . 60 ; 659 - 687 ( 1998 ) 10 . d. j. futsuyma , _ evolutionary biology second edition _ , sinauer associates inc . , sunderland , mass ( 1986 ) . 11 . gilbert s.f . , opitz j.m . , and raff r.a . 1996 , resynthesizing evolutionary and developmental biology , developmental biol . * 173 * 357 - 372 12 . gould s.j . , and eldredge n. punctuated equilibria : the tempo and mode of evolution reconsidered " , _ paleobiology _ * 3 * , 115 - 151 ( 1977 ) 13 . holman e. , 1987 recognizability of sexual and asexual species of rotifers , syst . zool . * 36 * 381 - 386 14 . holmes l.b . , penetrance and expressivity of limb malformations " _ birth defects . ser . _ * 15 * , 321 - 327 ( 1979 ) . 15 . howard and s.h . berlocher ( eds . ) _ endless form : species and speciation _ , oxford univ . press . ( 1998 ) 16 . kaneko k. clustering , coding , switching , hierarchical ordering , and control in network of chaotic elements " physica 41 d , 137 - 172 ( 1990 ) 17 . kaneko k. relevance of clustering to biological networks " , physica 75d , 55 - 73 ( 1994 ) 18 . kaneko k. , coupled maps with growth and death : an approach to cell differentiation " , physica 103 d ; 505 - 527 ( 1998 ) 19 . kaneko k. & yomo t. , cell division , differentiation , and dynamic clustering " , physica 75 d , 89 - 102 ( 1994 ) . kaneko k. & yomo t , isologous diversification : a theory of cell differentiation " , bull.math.biol . 59 , 139 - 196 ( 1997 ) 21 . kaneko k. & yomo t , isologous diversification for robust development of cell society " , j. theor . biol . , 199 243 - 256 ( 1999 ) 22 . kaneko k. & yomo t , symbiotic speciation from a single genotype " , proc . b , 267 , 2367 - 2373 ( 2000 ) 23 . kaneko k. & yomo t , symbiotic sympatric speciation through interaction - driven phenotype differentiation " , evol . ( 2002 ) , in press 24 . kashiwagi a. , kanaya t. , yomo t. , urabe i. , how small can the difference among competitors be for coexistence to occur " , _ researches on population ecology _ * 40 * , 223 ( 1998 ) . kashiwagi a. , noumachi w. , katsuno m. , alam m.t . , urabe i. , and yomo t. plasticity of fitness and diversification process during an experimental molecular evolution " , j. mol . evol . , ( 2001 ) in press 26 . kawata m. & yoshimura j. , _ speciation by sexual selection in hybridizing populations without viability selection _ , ev.ec . res . 2 : ( 2000 ) 897 - 909 27 . ko e. , yomo t. , & urabe i. , dynamic clustering of bacterial population " physica 75d , 81 - 88 ( 1994 ) 28 . kobayashi c. , suga y. , yamamoto k. , yomo t. , ogasahara k. , yutani k. , and urabe i j. biol . 272 , 23011 - 6 , thermal conversion from low- to high - activity forms of catalase i from bacillus stearothermophilus " ( 1997 ) 29 . kondrashov a.s . & kondrashov a.f , interactions among quantitative traits in the course of sympatric speciation " , _ nature _ * 400 * 351 - 354 ( 1999 ) 30 . lande r. , models of speciation by sexual selection on phylogenic traits " , proc . natl . usa * 78 * 3721 - 3725 ( 1981 ) 31 . maynard - smith j. , sympatric speciation " , the american naturalist * 100 * 637 - 650 ( 1966 ) 32 . maynard - smith j. and szathmary e. , _ the major transitions in evolution _ ( w.h.freeman , 1995 ) 33 . maynard - smith j. , burian r. , kauffman s. , alberch p. , campbell j. , goodwin b. , lande r. , raup d. , and wolpert l. , 1985 developmental constraints and evolution , q. rev . 60 * 265 - 287 34 . newman , _ from physics to development : the evolution of morphogenetic mechanism _ ( to appear in origins of organismal form " , eds . mller and s.a . newman , mit press , cambridge , 2002 ) 35 . j.m.opitz , some comments on penetrance and related subjects " , _ am - j - med - genet . _ * 8 * 265 - 274 ( 1981 ) . 36 . roberts m.s . , and cohan f.m . , 1995 recombination and migration rates in natural populations of _ bacillus subtilis _ and _ bacillus mojavensis _ , evolution * 49 * : 1081 - 1094 37 . rosenzweig , competitive speciation " , biol . j. of linnean soc . * 10 * , 275 - 289 , ( 1978 ) 38 . schiliewen , u.k . , tautz , d. & pbo s. , sympatric speciation suggested by monophly of crater lake cichilids , nature * 368 * , 629 - 632 ( 1994 ) 39 . spitze k. and sadler t.d . , 1996 , evolution of a generalist genotype : multivariate analysis of the adaptiveness of phenotypic plasticity , am . nat . * 148 * , 108 - 123 40 . takagi , h. , kaneko k. , & yomo t. evolution of genetic code through isologous diversification of cellular states " , artificial life , 6 ( 2000 ) 283 - 305 . tilman , d. ecological competition between algae : experimental confirmation of resource - based competition theory " , _ science _ * 192 * 463 - 465 ( 1976 ) 42 . tilman , d. test of resource competition theory using four species of lake michigan algae " , _ ecology _ * 62 * , 802 - 815 ( 1981 ) 43 . turner g.f . , & burrows m.t . a model for sympatric speciation by sexual selection " , london * b 260 * 287 - 292 ( 1995 ) 44 . waddington c.h . , _ the strategy of the genes _ , ( george allen & unwin ltd . , bristol , 1957 ) 45 weinig c. , 2000 , plasticity versus canalization : population differences in the timing of shade - avoidance responses evolution * 54 * 441 - 451 46 . wilson a.b . , noack - kunnmann k. , and meyer a. `` incipient speciation in sympatric nicaraguan crater lake cichlid fishes : sexual selection versus ecological diversification '' proc . london * b 267 * 2133 - 2141 ( 2000 ) 47 . xu , a. kashiwagi , t. yomo , i. urabe , fate of a mutant emerging at the initial stage of evolution " , _ researches on population ecology _ * 38 * , 231 - 237 ( 1996 ) .	a mechanism of sympatric speciation is presented based on the interaction - induced developmental plasticity of phenotypes . first , phenotypes of individuals with identical genotypes split into a few groups , according to instability in the developmental dynamics that are triggered with the competitive interaction among individuals . then , through mutational change of genes , the phenotypic differences are fixed to genes , until the groups are completely separated in genes as well as phenotypes . it is also demonstrated that the proposed theory leads to hybrid sterility under sexual recombination , and thus speciation is completed in the sense of reproductive isolation . as a result of this post - mating isolation , the mating preference evolves later . when there are two alleles , the correlation between alleles is formed , to consolidate the speciation . when individuals are located in space , different species are later segregated spatially , implying that the speciation so far regarded to be allopatric may be a result of sympatric speciation . relationships with previous theories , frequency - dependent selection , reinforcement , baldwin s effect , phenotypic plasticity , and resource competition are briefly discussed . relevance of the results to natural evolution are discussed , including punctuated equilibrium , incomplete penetrance in mutants , and the change in flexibility in genotype - phenotype correspondence . finally , it is discussed how our theory is confirmed both in field and in laboratory ( in an experiment with the use of _ e coli . _ ) . key words : dynamical systems , development , phenotypic plasticity , post - mating isolation , mating preference , genotype - phenotype mapping e - mail : [email protected]
-5 mm a locally free sheaf @xmath0 of @xmath1-modules is called _ a vector bundle _ on an algebraic variety @xmath2 . as a natural generalization of line bundles vector bundles have two important roles in algebraic geometry . one is the linear system . if @xmath0 is generated by its global sections @xmath3 , then it gives rise to a morphism to a grassmann variety , which we denote by @xmath4 , where @xmath5 is the rank of @xmath0 . this morphism is related with the classical linear system by the following diagram : @xmath6 where @xmath7 is the determinant line bundle of @xmath0 and @xmath8 is the morphism associated to it . the other role is the moduli . the moduli space of line bundles relates a ( smooth complete ) algebraic curve with an abelian variety called the _ jacobian variety _ , which is crucial in the classical theory of algebraic functions in one variable . the moduli of vector bundles also gives connections among different types varieties , and often yields new varieties that are difficult to describe by other means . in higher rank case it is natural to consider the moduli problem of @xmath0 under the restriction that @xmath9 is unchanged . in view of the above diagram , vector bundles and their moduli reflect the geometry of the morphism @xmath10 via grassmannians and plcker relations . in this article we consider the case where @xmath2 is a k3 surface , which is one of two 2-dimensional analogues of an elliptic curve and seems an ideal place to see such reflection . -5 mm the moduli space of line bundles on an algebraic variety is called the _ picard variety_. the picard variety @xmath11 of an algebraic curve @xmath12 is decomposed into the disjoint union @xmath13 by the degree @xmath14 of line bundles . here we consider the case of genus 1 . all components @xmath15 are isomorphic to @xmath12 if the ground field is algebraically closed . has a rational point . ] but this is no more true otherwise . for example the jacobian @xmath16 has always a rational point but @xmath12 itself does not . and @xmath17 deserve the name _ jacobian_. they coincide in our case @xmath18 . ] we give other examples : * example 1 * let @xmath19 be an intersection of two quadrics @xmath20 in the projective space @xmath21 and @xmath22 the pencil of defining quadrics . then the picard variety @xmath23 is the double cover of @xmath24 and the branch locus consists of 4 singular quadrics in @xmath22 . precisely speaking , its equation is given by @xmath25 . let @xmath26 be the 6-dimensional grassmann variety embedded into @xmath27 by the plcker coordinate . its projective dual is the dual grassmannian @xmath28 , where @xmath29 parameterizes 2-dimensional subspaces and @xmath30 quotient spaces . * example 2 * a transversal linear section @xmath31 is a curve genus 1 and of degree 5 . its picard variety @xmath32 is isomorphic to the dual linear section @xmath33 , the intersection with the linear subspace spanned by 5 points @xmath34 . -5 mm a compact complex 2-dimensional manifold @xmath35 is _ a k3 surface _ if the canonical bundle is trivial and the irregularity vanishes , that is , @xmath36 . a smooth quartic surface @xmath37 is the most familiar example . let us first look at the 2-dimensional generalization of example 1 : * example 3 * let @xmath38 be an intersection of three general quadrics in @xmath39 and @xmath40 the net of defining quadrics . then the moduli space @xmath41 is a double cover of @xmath42 and the branch locus , which is of degree 6 , consists of singular quadrics in @xmath40 . here @xmath43 , @xmath7 being a line bundle , is the moduli space of stable sheaves @xmath0 on a k3 surface @xmath35 with rank @xmath5 , @xmath44 and @xmath45 . surprisingly two surfaces @xmath46 and @xmath47 in this example are both k3 surfaces . this is not an accident . in respect of moduli space , vector bundles a k3 surface look like picard varieties in the preceding section . ( @xcite,@xcite ) the moduli space @xmath43 is smooth of dimension @xmath48 . @xmath43 is again a k3 surface if it is compact and and of dimension @xmath49 . a k3 surface @xmath35 and a moduli k3 surface appearing as @xmath50 are not isomorphic in general but their polarized hodge structures , or periods , are isomorphic to each other over @xmath51 ( @xcite ) . the moduli is not always fine but there always exists a universal @xmath52-bundle over the product @xmath53 . let @xmath54 be the associated sheaf of azumaya algebras , which is of rank @xmath55 and locally isomorphic to the matrix algebra @xmath56 . @xmath54 is isomorphic to @xmath57 if a universal family @xmath58 exists . the hodge isometry between @xmath59 and @xmath60 is given by @xmath61 . example 2 has also a k3 analogue . let @xmath62 be a 10-dimensional spinor variety @xmath63 , that is , the orbit of a highest weight vector in the projectivization of the 16-dimensional spinor representation . the anti - canonical class is 8 times the hyperplane section and a transversal linear section @xmath64 is a k3 surface ( of degree 12 ) . as is similar to @xmath29 the projective dual @xmath65 of @xmath66 is again a 10-dimensional spinor variety . * example 4 * the moduli space @xmath67 is isomorphic to the dual linear section @xmath68 . in this case , moduli is fine and the relation between @xmath35 and the moduli k3 are deeper . the universal vector bundle @xmath58 on the product gives an equivalence between the derived categories @xmath69 and @xmath70 of coherent sheaves , the duality @xmath71 holds ( cf . @xcite ) and moreover the hilbert schemes @xmath72 and @xmath73 are isomorphic to each other . * remark * ( 1 ) theorem 1 is generalized to the non - compact case by abe @xcite . \(2 ) if a universal family @xmath58 exists , the derived functor with kernel @xmath58 gives an equivalence of derived categories of coherent sheaves on @xmath35 and the moduli k3 ( bridgeland @xcite ) . in even non - fine case the derived category @xmath69 is equivalent to that of the moduli k3 @xmath74 twisted by a certain element @xmath75 ( cldraru @xcite ) . -5 mm let @xmath35 and @xmath76 be algebraic k3 surfaces and @xmath77 a hodge isometry between @xmath78 and @xmath79 . then the associated cycle @xmath80 on the product @xmath81 is a hodge cycle . this is algebraic by virtue of the torelli type theorem of shafarevich and piatetskij - shapiro . shafarevich conjectured in @xcite a generalization to hodge isometries over @xmath51 . our moduli theory is able to answer this affirmatively . let @xmath82 be a hodge isometry . then the associated ( hodge ) cycle @xmath83 is algebraic . in @xcite , we already proved this partially using theorem 1 ( cf . @xcite also ) . what we need further is the moduli space of projective bundles . let @xmath84 be a @xmath52-bundle over @xmath35 . the cohomology class @xmath85 \in h^1(s , pgl(r , { \cal o}_s))$ ] and the natural exact sequence ( in the classical topology ) @xmath86 define an element of @xmath87 , which we denote by @xmath88 . fix @xmath89 and @xmath5 , we consider the moduli of @xmath90-bundles @xmath22 over @xmath35 with @xmath91 mod @xmath5 which are stable in a certain sense . if the self intersection number @xmath92 is divisible by @xmath93 , then the moduli space contains a 2-dimensional component , which we denote by @xmath94 . the following , a honest generalization of computations in @xcite , is the key of our proof : assume that @xmath92 is divisible by @xmath95 . then @xmath96 is isomorphic to @xmath97 as polarized hodge structure , where @xmath98 is the submodule of @xmath78 consisting of @xmath99 such that the intersection number @xmath100 is divisible by @xmath5 . for example let @xmath101 be a double cover of @xmath102 with branch sextic . if @xmath89 satisfies @xmath103 and @xmath104 , then @xmath105 is a k3 surface of degree 8 . this is the inverse correspondence of example 1 ( cf . @xcite , @xcite ) . details will be published elsewhere . -5 mm let @xmath12 be a smooth complete algebraic curve . as a set a brill - noether locus of @xmath12 is a stratum of the picard variety @xmath11 defined by @xmath106 , the number of global sections of a line bundle @xmath7 . the standard notation is @xmath107 \,\|\ , h^0(l ) \ge r + 1\ } \subset { \rm pic}_d\ , c,\ ] ] for which we refer @xcite . non - abelian analogues are defined in the moduli space @xmath108 of stable 2-bundles on @xmath12 similarly . the _ non - abelian brill - noether locus of type iii _ is @xmath109 for a non - positive integer @xmath110 , and _ type ii _ is @xmath111 for a vector bundle @xmath112 of rank 2 and @xmath113 . by virtue of the ( serre ) duality , these have very special determinantal descriptions . we give them scheme structures using these descriptions ( @xcite ) . assume that @xmath12 lies on a k3 surface @xmath35 . if @xmath0 belongs to @xmath43 , then the restriction @xmath114 is of canonical determinant and we have @xmath115 . so @xmath114 belongs to @xmath116 if it is stable . this is one motivation of the above definition . the case of genus 11 , the gap value of genera where fano 3-folds of the next section do not exist , is the molst interesting . ( @xcite ) if @xmath12 is a general curve of genus @xmath117 , then the brill - noether locus @xmath118 of type iii is a k3 surface and the restriction @xmath7 of the determinant line bundle is of degree @xmath119 . there exists a universal family @xmath58 on @xmath120 . we moreover have the following : * the restriction @xmath121 is is stable and belongs to @xmath122 , for every @xmath123 , and * the classification morphism @xmath124 is an embedding . this embedding is a non - abelian analogue of the albanese morphism @xmath125 @xmath126 and we have the following : * corollary * _ a general curve of genus eleven has a unique embedding to a k3 surface . _ in @xcite , we studied the forgetful map @xmath127 from the moduli space @xmath128 of pairs of a curve @xmath12 of genus @xmath129 and a k3 surface @xmath35 with @xmath130 to the moduli space @xmath131 of curves of genus @xmath129 and the generically finiteness of @xmath132 . the above correspondence @xmath133 gives the inverse rational map of @xmath132 . we recall the fact that @xmath134 is not dominant in spite of the inequality @xmath135 ( @xcite ) . -5 mm a smooth 3-dimensional projective variety is called a _ fano 3-fold _ if the anti - canonical class @xmath136 is ample . in this section we assume that the picard group @xmath137 is generated by @xmath136 . the self intersection number @xmath138 is always even and the integer @xmath139 is called the _ genus _ , by which the fano 3-folds are classified into 10 deformation types . the values of @xmath129 is equal to @xmath140 and @xmath141 . a fano 3-folds of genus @xmath142 is a complete intersection of hypersurfaces in a suitable weighted projective space . by shokurov @xcite , the anticanonical linear system @xmath143 always contain a smooth member @xmath35 , which is a k3 surface . in @xcite we classified fano 3-folds @xmath2 of picard number one using _ rigid bundles _ , that is , @xmath144 with @xmath145 . for example @xmath2 is isomorphic to a linear section of the 10-dimensional spinor variety , that is , @xmath146 in the case of genus 7 and a linear section @xmath147 of the 6-dimensional symplectic , or lagrangian , grassmann variety @xmath148 @xmath149 in the case of genus 9 . the non - abelian brill - noether loci shed new light on this classification . a fano 3-fold @xmath2 of genus @xmath150 is isomorphic to the brill noether locus @xmath151 of type iii for a smooth curve @xmath12 of genus @xmath150 . this description is dual to the description ( [ linsec ] ) in the following sense . first two ambient spaces of @xmath2 , the moduli @xmath108 and the grassmannian @xmath152 are of the same dimension . secondly let @xmath153 and @xmath154 be the normal bundles of @xmath2 in these ambient spaces . then we have @xmath155 , that is , two normal bundles are twisted dual to each other . a fano @xmath156-fold of genus @xmath157 is isomorphic to the brill noether locus @xmath158 of type ii for a nonsingular plane quartic curve @xmath12 and @xmath112 a rank @xmath49 stable vector bundle over @xmath12 of odd degree . this descriptions is also dual to ( [ linsec2 ] ) in the above sense : the moduli @xmath108 and the grassmannian @xmath159 are of the same dimension and the two normal bundles of @xmath2 are twisted dual to each other . each fano 3-fold of genus 8 , 12 and conjecturally 10 has also such a pair of descriptions . -5 mm the four examples in sections 1 and 2 are very simple and invite us to a simplification of moduli construction . let @xmath19 be as in example 1 and @xmath160 the affine space associated to the 16-dimensional vector space @xmath161 , where @xmath162 is the homogeneous coordinate of @xmath21 . let @xmath163 be the closed subscheme defined by the condition that and @xmath166 its coordinat ring . then the picard variety @xmath167 is the projective spectrum @xmath168 of the invariant ring by construction . ( see @xcite for details . ) the above condition is equivalent to that @xmath169 is a linear combination of @xmath170 and @xmath171 . the invariant ring is generated by three elements by theorem 2.9.a of weyl @xcite . two of them , say @xmath172 , are of degree 2 and correspond to @xmath170 and @xmath171 , respectively . the rest , say @xmath76 of degree 4 , is the determinant of 4 by 4 matrix obtained from the four coefficients @xmath173 of @xmath174 . there is one relation @xmath175 . hence @xmath176 is a double cover of @xmath177 as desired . the moduli space @xmath41 in example 3 is constructed similarly . let @xmath178 be the affine space associated to the vector space @xmath179 and @xmath180 the subscheme defined by the condition that @xmath181 is of rank @xmath182 everywhere on @xmath38 . then the invariant ring of the action of @xmath183 on @xmath180 is generated by four elements @xmath184 of degree 2 , 2 , 2 , 6 . there is one relation @xmath185 and @xmath186 , the projective spectrum @xmath187 , is a double cover of @xmath102 as described . the moduli space of vector bundles on a surface was first constructed by gieseker @xcite . he took the mumford s git quotient @xcite of grothendieck s quot scheme @xcite by @xmath188 and used the gieseker matrix to measure the stability of the action . in the above construction , we take the quotient of @xmath189 , which is nothing but the affine variety of gieseker matrices of suitable rank 2 vector bundles , by a general linear group @xmath190 . the jacobian , or the picard variety , of a curve is more fundamental . weil @xcite constructed @xmath191 as an algebraic variety using the symmetric product @xmath192 and showed its projectivity by lefschetz @xmath193 theorem . later seshadri and oda @xcite constructed @xmath15 for arbitray @xmath14 ( over the same ground field as @xmath12 ) by also taking the git quotient of quot schemes . the above constructions eliminate quot schemes and the concept of linearization from those of gieseker , seshadri and oda . s. mori and s. mukai : the uniruledness of the moduli space of curves of genus 11 , in algebraic geometry , proceedings , tokyo / kyoto 1982 , series : lecture notes in mathematics , vol . 1016 , ( m , raynaud and t. shioda eds . ) , springer verlag , 1983 , 334353 . : curves , @xmath195 surfaces and fano manifolds of genus @xmath196 10 , in algebraic geometry and commutative algebra in honor of masayoshi nagata , ( h. hijikata and h. hironaka et al eds . ) , kinokuniya , tokyo , 1988 , 367377 . : biregular classification of fano threefolds and fano manifolds of coindex 3 , proc . , usa , * 86 * ( 1989 ) , 30003002 . : new developments in the theory of fano 3-folds : vector bundle method and moduli problem , sugaku , * 47*(1995 ) , 125144 . : english translation , sugaku expositions , to appear .	a k3 surface is a quaternionic analogue of an elliptic curve from a view point of moduli of vector bundles . we can prove the algebraicity of certain hodge cycles and a rigidity of curve of genus eleven and gives two kind of descriptions of fano threefolds as applications . in the final section we discuss a simplified construction of moduli spaces . 4.5 mm * 2000 mathematics subject classification : * 14j10 , 14j28 , 14j60 .
the mathematical study of enzyme kinetics ( or michaelis - menten kinetics ) is mainly based on the standard quasi - steady state approximation ( sqssa ) - which has been used in biochemistry , since the pioneering papers by bodenstein @xcite and chapman and underhill @xcite - which starts from the observation that enzyme reactions are characterized by a first , short transient phase , where the intermediate complex rapidly grows , and a second , longer quasi - equilibrium phase , where the complex slowly decays in the product , in general the activated substrate . each phase of the reaction has a time scale ( @xmath0 and @xmath1 , respectively ) . the quasi - steady state approximation is a very efficient way of simplification for the description of a typical saturation phenomenon , occurring in enzyme kinetics , but present in several other biological systems . let us cite , just as non exhaustive examples , the monod - wyman - changeux molecular model of cooperativity in allosteric reactions @xcite , or , more recently , the model of lekszycki and coworkers concerning the bone regeneration @xcite and the model of infarcted cardiac tissue regeneration by means of stem cells @xcite , where saturation phenomena are observed . the reaction can be described as follows . let us consider an enzyme , @xmath2 , which reacts with a protein , @xmath3 , resulting in an intermediate complex , @xmath4 . in turn , this complex can break down into a product , @xmath5 , and the enzyme @xmath2 . it is frequently assumed that the formation of @xmath4 is reversible while its breakup is not . the process is represented by the following sequence of reactions @xmath6 where @xmath7 are the reaction rates . for notational convenience we will use variable names to denote both a chemical species and its concentration . for example , @xmath2 denotes both an enzyme and its concentration . reaction ( [ eq : a1 ] ) obeys two natural constrains : the total amounts of protein and enzyme remain constant . therefore , @xmath8 for positive constants @xmath9 and @xmath10 . in conjunction with the constraints ( [ eq : a2 ] ) , the following cauchy problem for a system of two ordinary differential equations can be used to model reaction ( [ eq : a1 ] ) : @xmath11 \notag \\ & x(0)=x_t , \quad c(0)=0.\end{aligned}\ ] ] where @xmath12 and where @xmath10 , @xmath13 , @xmath14 , @xmath15 , @xmath16 are viewed as fixed positive constants and @xmath17 is the michaelis affinity constant . similarly , we can define the dissociation constant @xmath18 and the van slyke - cullen constant @xmath19 . since , after a short transient , where the complex @xmath4 rapidly grows , reaching its maximal concentration , it slowly decays , the sqssa consists in supposing that , after the transient phase , the complex can be considered in a quasi - equilibrium , i.e. , posing @xmath20 . with this approximation , the system becomes the differential - algebraic system @xmath21 ( @xmath22 ) where only the initial condition @xmath23 can be imposed , because the sqssa describes only the slow phase , where the initial value of @xmath24 is its maximal value , instead of @xmath25 . in the sixties of the last century mathematicians ( see , in particular , @xcite ) interpreted the sqssa in terms of leading order term of asymptotic expansions with respect to a perturbation parameter @xmath26 , which must be supposed small . heineken et al . used @xmath27 , because in literature it is widely used to impose that the initial concentration of the enzyme @xmath2 is much less than the concentration of the substrate @xmath3 . the parameter can also arise by virtue of a biochemical condition imposing the separation between the two timescales @xmath0 and @xmath1 characterizing the reaction ( see also @xcite ) . in this way , segel - slemrod @xcite showed that the sqssa can be obtained also as the leading order of an asymptotic expansion in terms of @xmath28 , enlarging the parameter range of validity of the sqssa . inspired by the papers by laidler @xcite , swoboda @xcite , and schauer and heinrich @xcite , borghans et al . @xcite introduced a different approximation , called total quasi - steady state approximation ( tqssa ) which uses the new variable @xmath29 , called total substrate . formally , introducing the `` lumped '' variable @xmath30 , problem ( [ eq : a3 ] ) can be rewritten as @xmath31 , \notag \\ & \overline{x}(0)=x_t , \quad c(0)=0.\end{aligned}\ ] ] also the tqssa posits that @xmath4 equilibrates quickly compared to @xmath32 . imposing also in this case a quasi - steady state approximation ( @xmath20 ) , we obtain @xmath33 where @xmath34 is the only biologically allowed solution of @xmath35 . let us remark that since , thanks to the conservations laws , @xmath36 , the tqssa can be viewed as the other side of the coin of laidler s theory , though the approach followed in @xcite implicitly contains much more information about the reliability of the approximation , as shown in @xcite . also the tqssa can be seen as the leading order term of an asymptotic expansion in terms of a suitable parameter , @xmath37 , producing a new approximation , which is valid in a much wider parameter range . the parameter , introduced in @xcite , appears already in a paper by palsson @xcite , where the author determines sufficient conditions for the application of any quasi - steady state approximation , based again on the time scale separation . taking into account that the perturbation parameter is always less than @xmath38 , its introduction in terms of time scale separation appears much more natural than the previous parameters . this result gives a theoretical mathematical foundation of the choice of the parameter in the tqssa . moreover , several authors ( see , for example , @xcite ) study the transient phase of the reaction supposing that in this phase @xmath3 does not change substantially . this hypothesis is not realistic , while , using the total substrate @xmath39 , we observe that at time 0 , we have @xmath40 , which addresses much more naturally the request of small changes of the total substrate in the initial time of the reaction . in figure [ figura1 ] we show the different efficiency of the two quasi - steady state approximations , when the parameters are stressed in such a way that the sqssa is no more valid . ) , with their sqssa ( [ eq : a3bis ] ) and tqssa ( [ eq : tqssa_single ] ) . the parameter set is the following : @xmath41 . the inadequacy of the sqssa , mainly in the first part of the reaction , is evident , while the tqssa in indistinguishable from the numerical solution of the system.,title="fig:",scaledwidth=50.0% ] ) , with their sqssa ( [ eq : a3bis ] ) and tqssa ( [ eq : tqssa_single ] ) . the parameter set is the following : @xmath41 . the inadequacy of the sqssa , mainly in the first part of the reaction , is evident , while the tqssa in indistinguishable from the numerical solution of the system.,title="fig:",scaledwidth=50.0% ] in previous literature the different qssas are approached by means of two different tools : tihonov s theorem @xcite , which studies the asymptotic stability of systems of differential equations characterized by the presence of small perturbative parameters and center manifold theory , which is one of the most powerful tools to study the dimensional reduction of differential systems . for example , on the one hand , heineken et al . @xcite and dvok and ika @xcite quote tihonov s theorem in order to justify the sqssa , while khho and hegland @xcite refer to this theorem to apply the tqssa ; on the other hand , other authors @xcite interpret the sqssa and the tqssa , respectively , as the slow manifold of the michaelis - menten kinetics . however , at the best of our knowledge , the two approaches are not yet compared , in order to check whether there exist any equivalences between the so - called singular points , introduced by tihonov , and the center manifolds , as studied , for example , by carr @xcite . applying the techniques exposed in @xcite , we show that the two approximations are asymptotically equivalent , concluding that the sqssa and the tqssa can be interpreted both as the singular point of the michaelis - menten kinetics and its center manifold . this means that tihonov s theorem implies that any qssa can be mathematically interpreted as the study of the reduced system of the original system setting the perturbative parameter @xmath42 , instead of setting the derivative of the complex @xmath24 equal to zero . this fact formally allows the application of a `` mechanistic '' passage , consisting in equating to zero the derivatives of the complexes , in the single reaction scheme , as in more complex reactions , because this is the simplest way to reach ( an approximate ) expression of the center manifold . for example , wang and sontag @xcite apply this technique for the study of the sqssa of the double phosphorylation - dephosphorylation cycle . as shown in @xcite , however , these approximation are no more applicable to mechanisms where oscillations can appear and an a priori analysis of the application should be performed every time we have to deal with any qssa . the paper is organized as follows in section 2 we recall all the main definitions and properties of tihonov s theory and of the center manifold theory ; in section 3 we show the equivalence of the two approaches in the case of the sqssa , of the tqssa and for a class of more general systems of differential equations characterized by the presence of a small , perturbative parameter ; in section 4 we discuss some future applications of this results to more complex enzymatic reactions . in this section we introduce the notations and summarize the results we need for the formulation of the problem investigated here . for convenience of the reader we closely follow the notations of the fundamental book by wiggins @xcite . we will investigate systems in the class of general autonomous vector fields @xmath43 it is natural to consider the linearized system @xmath44 associated to the vector field ( [ eq:1 ] ) , if @xmath45 is one of its fixed points , and the constant jacobian @xmath46 matrix @xmath47 . then @xmath48 can be represented as the direct sum of three subspaces denoted @xmath49 , @xmath50 , and @xmath51 , which are defined as follows : @xmath52 where @xmath53 , @xmath54 , @xmath55 are the ( generalized ) eigenvectors of @xmath56 corresponding to the eigenvalues having negative real part , positive real part and zero real part , respectively . @xmath49 , @xmath50 , and @xmath51 are referred to as the stable , unstable , and center subspaces , respectively . they are invariant subspaces ( or manifolds ) since solutions of ( [ eq:7 ] ) with initial conditions entirely contained in either @xmath49 , @xmath50 , or @xmath51 must forever remain in that particular subspace for all time . it is well known that there exists a linear transformation @xmath57 which transforms the linear equation ( [ eq:7 ] ) into block diagonal form @xmath58 where @xmath59 , @xmath60 , @xmath61 is an @xmath62 matrix having eigenvalues with negative real part , @xmath63 is an @xmath64 matrix having eigenvalues with positive real part , and @xmath65 is an @xmath66 matrix having eigenvalues with zero real part . the @xmath25 in the block diagonal form ( [ eq:14 ] ) indicate appropriately sized block consisting of all zero s . using this same linear transformation to transform the coordinates of the nonlinear vector field ( [ eq:1 ] ) gives the equation @xmath67 where @xmath68 , @xmath69 , and @xmath70 are the first @xmath71 , @xmath72 , and @xmath73 components , respectively , of the vector @xmath74 . the following theorem shows how this structure changes when the nonlinear vector field ( [ eq:13 ] ) is considered . it is stated without proof ( see @xcite for details ) . [ th:2 ] suppose ( [ eq:13 ] ) is @xmath75 , @xmath76 . then the fixed point @xmath77 of ( [ eq:13 ] ) possesses a @xmath75 s - dimensional local , invariant stable manifold , @xmath78 , a @xmath75 u - dimensional local , invariant unstable manifold , @xmath79 , and a @xmath75 c - dimensional local , invariant center manifold , @xmath80 , all intersecting in @xmath77 . these manifolds are all tangent to the respective invariant subspaces of the linear vector field ( [ eq:14 ] ) at the origin and , hence , are locally representable as graphs . in particular , we have @xmath81 @xmath82 @xmath83 @xmath84 @xmath85 @xmath86 where @xmath87 , @xmath88 , @xmath89 , @xmath90 , @xmath91 , and @xmath92 are @xmath75 functions . moreover , trajectories in @xmath78 and @xmath79 have the same asymptotic properties as trajectories in @xmath49 and @xmath50 , respectively . namely , trajectories of ( [ eq:13 ] ) with initial conditions in @xmath78 ( resp . , @xmath79 ) approach the origin at an exponential rate asymptotically as @xmath93 ( resp . , @xmath94 ) . if the eigenvalues of the center subspace are all precisely zero - rather than having just real part zero - then a center manifold is called a _ slow manifold_. if @xmath95 , then any orbit will rapidly decay to @xmath51 . thus , in order to investigate the long - time behavior ( i.e. , stability ) we need only to investigate the system restricted to @xmath51 . this simple reasoning is the foundation of the `` reduction principle '' applied to the study of the stability of nonhyperbolic fixed points of nonlinear vector fields . for our purposes , let us consider vector fields of the following form @xmath96 where @xmath97 in the above , @xmath56 is a @xmath66 matrix having eigenvalues with zero real parts , @xmath98 is an @xmath99 matrix having eigenvalues with negative real parts , and @xmath100 and @xmath101 are @xmath75 functions ( @xmath102 ) . for the sake of notation simplicity , let us write the center manifold in the following way : @xmath103 with @xmath104 sufficiently small . [ r:1 ] we remark that the conditions @xmath105 and @xmath106 imply that @xmath107 is tangent to @xmath51 at @xmath108 . the following three theorems are taken from the seminal book @xcite , as reported in @xcite . [ th:3 ] there exists a @xmath75 center manifold for ( [ eq:34 ] ) . the dynamics of ( [ eq:34 ] ) restricted to the center manifold is , for @xmath72 sufficiently small , given by the following c - dimensional vector field @xmath109 the next result implies that the dynamics of ( [ eq:35 ] ) near @xmath110 determines the dynamics of ( [ eq:34 ] ) near @xmath108 . [ th:4 ] i ) let the zero solution of ( [ eq:35 ] ) be stable ( asymptotically stable ) ( unstable ) ; then the zero solution of ( [ eq:34 ] ) is also stable ( asymptotically stable ) ( unstable ) . ii ) let the zero solution of ( [ eq:35 ] ) be stable . then if @xmath111 is a solution of ( [ eq:34 ] ) with @xmath112 sufficiently small , there is a solution @xmath113 of ( [ eq:35 ] ) such that as @xmath114 @xmath115 where @xmath116 is a constant . it is possible to compute the center manifold so that we can reap the benefits of theorem [ th:4 ] . using invariance of @xmath107 under the dynamics of ( [ eq:34 ] ) , we derive a quasi - linear partial differential equation that @xmath117 must satisfy , in order for its graph to be a center manifold for ( [ eq:34 ] ) . this is done as follows : 1 . the @xmath118 coordinates of any point on @xmath107 must satisfy @xmath119 2 . differentiating ( [ eq:36 ] ) with respect to time implies that the @xmath120 coordinates of any point on @xmath107 must satisfy @xmath121 3 . any point on @xmath107 obeys the dynamics generated by ( [ eq:34 ] ) . therefore , substituting @xmath122 into ( [ eq:37 ] ) gives @xmath123 or @xmath124 then , to find a center manifold , all we need to do is to solve ( [ eq:40 ] ) . [ th:5 ] let @xmath125 be a @xmath126 mapping with @xmath127 such that @xmath128 as @xmath129 for some @xmath130 . then @xmath131 the theorem gives us a method for computing an approximate solution of ( [ eq:40 ] ) to any desired degree of accuracy . so , for this task , we will employ power series expansions ( note that by remark [ r:1 ] power series expansions start from second order ) . suppose now that ( [ eq:34 ] ) depends on a vector of parameters @xmath132 : @xmath133 where @xmath134 with the same assumptions as in ( [ eq:34 ] ) . following wiggins @xcite , we will handle parameterized systems including the parameter @xmath26 as a new dependent variable as follows @xmath135 this system has a fixed point at @xmath136 . the matrix associated with the linearization of ( [ eq:44 ] ) around this fixed point has @xmath71 eigenvalues with negative real part and @xmath137 eigenvalues with zero real part . let us now apply center manifold theory . modifying definition given in formula ( [ eq : center ] ) , a center manifold will be represented as the graph of @xmath138 for @xmath139 and @xmath26 sufficiently small . theorem [ th:3 ] still applies , with the vector field reduced to the center manifold given by @xmath140 let us calculate the center manifold . using invariance of the graph of @xmath141 under the dynamics generated by ( [ eq:44 ] ) , we have @xmath142 substituting ( [ eq:44 ] ) into ( [ eq:46 ] ) results in the following quasi - linear partial differential equation that @xmath141 must satisfy in order for its graph to be a center manifold . @xmath143+\notag \\ & -bh(x,\varepsilon)-g\left(x , h(x,\varepsilon),\varepsilon\right)=0\end{aligned}\ ] ] although center manifolds exist , they do not need to be unique . this can be seen from a well - known example due to anosov ( see @xcite , @xcite ) . it can be proven ( see , among others , @xcite as reported in @xcite ) that any couple of center manifolds of a given fixed point differ by ( at most ) exponentially small terms . thus , the taylor series expansions of any two center manifolds agree to all orders . moreover , it can be shown that , due to the attractive nature of the center manifold , certain orbits ( for example , fixed points , periodic orbits , homoclinic orbits , and heteroclinic orbits ) that remain close to the origin for all the time must be on every center manifold of a given fixed point . for this subsection we will refer to the widespread book by w. wasow @xcite , and - in particular - to its relevant section on singular perturbations . a systematic study of the qualitative aspects of such singular perturbation problems can be found in a series of papers by tihonov ( @xcite , @xcite and @xcite ) . we consider differential systems of the form @xmath144 where @xmath139 is @xmath73-dimensional vector and @xmath145 an @xmath71-dimensional vector . all variables are real , and @xmath26 is positive . we assume that : * the functions @xmath100 and @xmath101 in ( [ eq : c1 ] ) are continuous in an open region @xmath146 of the @xmath147-space . * there is an @xmath71-dimensional vector function @xmath148 continuous in @xmath149 such that the points @xmath150 , for all @xmath149 , are in @xmath146 and @xmath151 * there exists a number @xmath152 , independent of @xmath139 , such that the relations @xmath153 imply @xmath154 the function @xmath148 will be referred to as a root of the equation @xmath155 . it is not excluded that @xmath155 may have other roots besides @xmath148 . a root @xmath148 that satisfies condition c will be called _ isolated _ in @xmath149 . the system of differential equations @xmath156 in which @xmath139 is a parameter , will be called the boundary layer equation belonging to the system ( [ eq : c1 ] ) . to ( [ eq : c1 ] ) there corresponds the reduced ( or degenerate ) system @xmath157 the solutions of ( [ eq : c1 ] ) and ( [ eq : c1bis ] ) define trajectories @xmath158 and @xmath159 in the @xmath147-space . we also assume : * the singular point @xmath160 of the boundary layer equation ( [ eq : c2 ] ) is asymptotically stable for all @xmath149 . the root @xmath148 will be called , briefly , a stable root in @xmath149 , if assumption ( d ) is satisfied . in accordance with our previous terminology we refer to the problem consisting of equations ( [ eq : c1 ] ) together with the initial condition @xmath161 as the full problem . the reduced problem is here defined by @xmath162 @xmath163 the differential equation ( [ eq : c4 ] ) is , of course , obtained by setting @xmath164 in ( [ eq : c1 ] ) and determining the root @xmath165 of the equation @xmath155 . moreover , we assume : * the full problem , as well as the reduced one , has a unique solution in an interval @xmath166 . * the asymptotic stability of the singular point @xmath165 is uniform with respect to @xmath139 in @xmath149 . let @xmath167 . the set of points in the @xmath147-space for which the inequalities @xmath168 hold will be called a `` @xmath169-tube '' . the set @xmath170 constitutes the `` lateral boundary '' of the @xmath169-tube . [ l:1 ] suppose assumptions ( a ) to ( f ) are satisfied . let @xmath167 be arbitrary but so small that the closure of the @xmath169-tube lies in @xmath146 . there exist then two numbers @xmath171 and @xmath172 such that for @xmath173 the following is true : any solution of the full equation that is in the interior of the @xmath169-tube for some value @xmath174 of @xmath175 , @xmath176 , and in the closure of the @xmath169-tube for all @xmath175 in @xmath177 , does not meet the lateral surface of the @xmath169-tube for @xmath177 . the lemma states that , for small @xmath26 , any solution that comes close to the curve @xmath165 in @xmath149 remains close to it , as long as @xmath149 . for a convenient formulation of tihonov s theorem , according to @xcite , we introduce one more term . [ def:9 ] a point @xmath178 , @xmath179 is said to lie in the domain of influence of the stable root @xmath165 if the solution of the problem @xmath180 exists and remains in @xmath146 for all @xmath181 , and if it tends to @xmath182 , as @xmath183 . [ th:6 ] let assumptions ( a ) to ( f ) be satisfied and let @xmath184 be a point in the domain of influence of the root @xmath165 . then the solution @xmath185 , @xmath186 of the full initial value problem ( [ eq : c1 ] ) , ( [ eq : c3 ] ) is linked with the solution @xmath187 , @xmath188 of the reduced problem ( [ eq : c4 ] ) , ( [ eq : c5 ] ) by the limiting relations @xmath189 here @xmath190 is any number such that @xmath191 is an isolated stable root of @xmath192 for @xmath193 . the convergence is uniform in @xmath193 , for @xmath185 , and in any interval @xmath194 for @xmath186 . tihonov s theorem [ th:6 ] is only the first step in the asymptotic solution of initial value problems of the singular perturbation type . the most natural approach to this problems is to attempt a solution ( _ outer solution _ ) in the form of a series in powers of @xmath26 : @xmath195 and to determine the coefficients @xmath196 , @xmath197 by means of formal substitution and comparison of coefficients . it is clear that we have to relate the series ( [ eq : c6 ] ) to the behavior of the solution of ( [ eq : c1 ] ) in the boundary layer , as shown , for example , in @xcite . for values of @xmath175 that are of order @xmath198 the solution to our perturbation problem can be found starting from the stretching transformation @xmath199 . hence , the stretched form of the original problem is @xmath200 also in this case we determine the solution of the transient phase ( _ inner solution _ ) in terms of a series in powers of @xmath26 . the developments of the passages is beyond the scope of this paper . for the other accounts and a more detailed discussion , see @xcite . let us consider the enzymatic reaction described in ( [ eq : a1 ] ) and ( [ eq : a3 ] ) . clearly @xmath201 is a fixed point of ( [ eq : a3 ] ) . following @xcite , let us first adimensionalize equations ( [ eq : a3 ] ) . let us observe that we could use different adimensionalization procedures , in particular using the parameter @xmath202 , as proposed in @xcite . however , we follow the simpler scheme shown in @xcite , just in order to test our theoretical results and compare them with the results shown in @xcite : @xmath203 where @xmath204 and @xmath205 this is the heineken - tsuchiya - aris system @xcite . carr ( @xcite , p.8 , example 3 ) uses equations @xmath206 to obtain ( [ eq : a9s ] ) from ( [ eq : a10carr ] ) just impose @xmath207 and @xmath208 . we will start from ( [ eq : a9s ] ) for having a more realistic system . by applying the sqssa ( which corresponds to impose @xmath164 ) , we have the reduced system ( _ outer solution _ ) of ( [ eq : a9s ] ) @xmath209 as above remarked , heineken et al . @xcite and dvok and ika @xcite quote tihonov s theorem in order to justify the sqssa . the aim of this subsection is now to determine the center manifold for ( [ eq : a9s ] ) , using the techniques described in @xcite and to show that it is asymptotically equivalent to the singular points related to tihonov theory . to this aim , let us now set @xmath210 . equations ( [ eq : a9s ] ) can be rewritten in the equivalent form ( _ inner solution _ ) @xmath211 where @xmath212 . in order to obtain for ( [ eq : a13 ] ) a block form , of the type ( [ eq:34 ] ) , where the submatrix having eigenvalues with zero real parts is separated from the submatrix having eigenvalues with negative real parts , we operate the substitution @xmath213 , i.e. , @xmath214 . hence , @xmath215 and @xmath216 following @xcite , the way we will handle parametrized systems consists of including the parameter @xmath26 as a new dependent variable as in ( [ eq : a14 ] ) , which merely acts to augment the matrix @xmath56 by adding a new center direction that has no dynamics . in this way , system ( [ eq : a13 ] ) becomes @xmath217 where the parameter @xmath26 is a new variable and the system could have also other fixed points . the associated linearized system has a diagonal form , where the eigenvalues are given by @xmath25 ( multiplicity @xmath218 ) and @xmath219 . to find a center manifold , all we need to do is to solve ( [ eq:40 ] ) for system ( [ eq : a14 ] ) , employing theorem [ th:5 ] , which gives us a method for computing an approximate solution of ( [ eq:40 ] ) to any desired degree of accuracy . referring to ( [ eq:40 ] ) and ( [ eq:34 ] ) , @xmath220 , @xmath221 , so we search for a function @xmath222 such that @xmath223 where @xmath224 using theorem [ th:5 ] we assume @xmath225 substituting ( [ eq : a16 ] ) into ( [ eq : a15 ] ) , one has : @xmath226 where @xmath227 accordingly , substituting ( [ eq : a18 ] ) into ( [ eq : a17 ] ) , one has @xmath228+\kappa \left(a_1 u^2+a_2 u \varepsilon+a_3 \varepsilon^2+\dots\right)+ \notag \\ & -u\left(u - a_1 u^2-a_2 u \varepsilon - a_3 \varepsilon^2 + \dots\right)-\varepsilon \biggl[-\frac{\lambda}{\kappa } u + \left(-a_1+\frac{1}{\kappa}+\frac{\lambda}{\kappa}a_1\right ) u^2+\notag \\ & + \left(-a_2+\frac{\lambda}{\kappa}a_2\right)u\varepsilon+ \left(-a_3+\frac{\lambda}{\kappa}a_3\right)\varepsilon^2+\dots\biggr ] = 0\end{aligned}\ ] ] truncating at second order terms , we obtain : @xmath229 equating terms of the same power to zero gives @xmath230 , @xmath231 and @xmath232 . hence , the center manifold for system ( [ eq : a14 ] ) is : @xmath233 which , for @xmath208 , gives the result shown in @xcite . finally , substituting ( [ eq : a20 ] ) into ( [ eq : a14 ] ) , we obtain the vector field reduced to the center manifold , according to equation ( [ eq:35 ] ) of theorem [ th:3 ] . in fact , if @xmath230 , @xmath231 and @xmath232 , formula ( [ eq : a18 ] ) becomes : @xmath234 thus : @xmath235 , \notag \\ \frac{d \varepsilon}{ds}&=0\end{aligned}\ ] ] or , in terms of the original time scale , @xmath236 , \notag \\ \dot\varepsilon&=0\end{aligned}\ ] ] let us now conclude showing that the center manifold obtained following this method is asymptotically sufficiently close to ( [ eq : a11 ] ) . we can obtain back the equation in @xmath237 . in fact , since @xmath214 , from ( [ eq : a20 ] ) and @xmath238 we have @xmath239 considering @xmath240 , one has @xmath241 which is the second equation of ( [ eq : a11 ] ) . we can conclude that , supposing @xmath242 , the center manifold determined in this way approximates the solution given by the sqssa , which coincides with the root related to the application of tihonov s theorem . in figure ( [ figura2 ] ) we compare the sqssa of system ( [ eq : a9s ] ) , obtained from ( [ eq : a10carr ] ) , with the center manifold ( [ eq : aaa20 ] ) , at the zeroth order and at the first order in @xmath26 , respectively . obviously , the latter gives a better approximation of the numerical solution of ( [ eq : a9s ] ) , while the former well approximates the sqssa curve , which in fact can be considered the zeroth order term of an asymptotic expansion of the solution in terms of @xmath26 . of the numerical solution of the system ( [ eq : a9s ] ) ( blue solid line ) with its sqssa ( [ eq : a10carr ] ) ( black solid line ) and its zeroth order ( dashed line ) and first order ( dashed / dotted line ) center manifold ( [ eq : aaa20 ] ) . the parameter sets are the following . left : @xmath243 . the set was taken from @xcite . right : @xmath244 . the set was taken ( and modified ) from @xcite . since in both cases the value of @xmath245 is sufficiently small , the different approximations converge to the graph of the numerical solution . in the plot on the right it is possible to appreciate the different behavior of the zero - th order and the first order center manifolds . while the first order manifold approximates in a better way the numerical solution , the zero - th order converges to the sqssa , that does not approximate sufficiently well the numerical solution , since it is the zero - th order term of the singular perturbation of the solution in terms of the parameter @xmath245.,title="fig:",scaledwidth=50.0% ] of the numerical solution of the system ( [ eq : a9s ] ) ( blue solid line ) with its sqssa ( [ eq : a10carr ] ) ( black solid line ) and its zeroth order ( dashed line ) and first order ( dashed / dotted line ) center manifold ( [ eq : aaa20 ] ) . the parameter sets are the following . left : @xmath243 . the set was taken from @xcite . right : @xmath244 . the set was taken ( and modified ) from @xcite . since in both cases the value of @xmath245 is sufficiently small , the different approximations converge to the graph of the numerical solution . in the plot on the right it is possible to appreciate the different behavior of the zero - th order and the first order center manifolds . while the first order manifold approximates in a better way the numerical solution , the zero - th order converges to the sqssa , that does not approximate sufficiently well the numerical solution , since it is the zero - th order term of the singular perturbation of the solution in terms of the parameter @xmath245.,title="fig:",scaledwidth=50.0% ] let us now consider system ( [ eq : a4 ] ) . let us adimensionalize the system , as in @xcite @xmath246 where @xmath247 and @xmath248 with system parameters @xmath249 such that @xmath250 . by applying the tqssa ( which corresponds to imposing @xmath164 ) to ( [ eq : aaa12 ] ) , we have @xmath251 and , solving in @xmath237 : @xmath252 which represents the singular point ( or _ outer solution _ ) of ( [ eq : aaa12 ] ) , where @xmath253 , @xmath254 , @xmath255 are viewed as fixed positive constants and @xmath26 is the parameter . its fixed point is @xmath256 . the new parameter @xmath26 appears already in @xcite and was used in @xcite to determine the asymptotic expansions whose leading term is just the tqssa . in 2008 khoo and hegland @xcite applied tihonov s theorem @xcite in order to study the tqssa , obtaining similar results as in @xcite . the aim of this subsection is now to determine the center manifold for ( [ eq : aaa12 ] ) , using the techniques described in @xcite and to show that they are asymptotically equivalent to the singular points related to tihonov theory . to this aim , let us now set @xmath210 , system ( [ eq : aaa12 ] ) can be rewritten in the form ( _ inner solution _ ) @xmath257 in order to obtain for ( [ eq : a22 ] ) a block form , of the type ( [ eq:34 ] ) , we make the substitution @xmath258 , i.e. , @xmath259 . hence , @xmath260 @xmath261 @xmath262 doing so , and introducing the new variable @xmath26 , system ( [ eq : a22 ] ) becomes @xmath263 the associated linearized system has a diagonal form and , in fact , the eigenvalues are given by @xmath25 ( with multiplicity @xmath218 ) and @xmath264 . also in this case the eigenvalue @xmath25 has multiplicity @xmath218 . we solve ( [ eq:40 ] ) for system ( [ eq : a23 ] ) , employing theorem [ th:5 ] and determine the center manifold . referring to ( [ eq:40 ] ) and ( [ eq:34 ] ) , we have that @xmath220 , @xmath265 . accordingly , we search a function @xmath266 such that @xmath267 using theorem [ th:5 ] we assume @xmath268 substituting ( [ eq : a24 ] ) into ( [ eq : a25 ] ) , one has : @xmath269 truncating at second order terms , we obtain : @xmath270 so , @xmath271 u^2+\ ] ] @xmath272 u \varepsilon+ ( \eta + \kappa_m)a_3 \varepsilon^2 = 0\ ] ] from which : @xmath273 hence , the center manifold for system ( [ eq : a23 ] ) is : @xmath274 finally , substituting ( [ eq : a27 ] ) into ( [ eq : a23 ] ) we obtain the vector field reduced to the center manifold , according to equation ( [ eq:35 ] ) of theorem [ th:3 ] . then : @xmath275 , \\ \dot\varepsilon&=0 \end{cases}\ ] ] or , in terms of the original time scale , @xmath276 , \notag \\ \dot\varepsilon&=0\end{aligned}\ ] ] let us show that the center manifold obtained in ( [ eq : a27 ] ) is asymptotically close to the root given by ( [ eq : a29 ] ) , in terms of tihonov s theorem . from ( [ eq : a27 ] ) , and since @xmath259 , with @xmath277 we have @xmath278 since equation ( [ eq : a29 ] ) is obtained putting @xmath240 , one has @xmath279 while , approximating the square root in ( [ eq : a29 ] ) , one has @xmath280 \sim \frac{\eta + \kappa_m+\sigma u}{\eta \sigma } \varepsilon = \frac{u}{\eta + \kappa_m+\sigma u } , \ \ \ \text{for } \ u\rightarrow0\end{aligned}\ ] ] it follows that both ( [ eq : a30 ] ) and ( [ eq : a31 ] ) are asymptotic to @xmath281 when @xmath282 . this means that both the expressions can be intended as two different approximations of the center manifold . in figure ( [ figura3 ] ) we compare the tqssa of system ( [ eq : aaa12 ] ) , obtained from ( [ eq : a29 ] ) , with the center manifold ( [ eq : aaa30 ] ) , at the zeroth order and at the first order in @xmath26 , respectively . obviously , the latter gives a better approximation of the numerical solution of ( [ eq : aaa12 ] ) , while the former well approximates the tqssa curve , which in fact can be considered the zeroth order term of an asymptotic expansion of the solution in terms of @xmath26 . of the numerical solution of the system ( [ eq : aaa12 ] ) ( blue solid line ) with its tqssa ( [ eq : a29 ] ) ( black solid line ) and its zeroth order ( dashed line ) and first order ( dashed / dotted line ) center manifold ( [ eq : aaa30 ] ) . the parameter sets are the following . left : @xmath283 . the set was taken from @xcite . right : @xmath284 . the set was taken from @xcite . in the plot on the left , since the value of @xmath26 is sufficiently small , the different approximations converge to the graph of the numerical solution . in the plot on the right it is possible to appreciate the different behavior of the zero - th order and the first order center manifolds . while the first order manifold approximates in a better way the numerical solution , the zero - th order converges to the tqssa , that does not approximate sufficiently well the numerical solution , since it is the zero - th order term of the singular perturbation of the solution in terms of the parameter @xmath26.,title="fig:",scaledwidth=50.0% ] of the numerical solution of the system ( [ eq : aaa12 ] ) ( blue solid line ) with its tqssa ( [ eq : a29 ] ) ( black solid line ) and its zeroth order ( dashed line ) and first order ( dashed / dotted line ) center manifold ( [ eq : aaa30 ] ) . the parameter sets are the following . left : @xmath283 . the set was taken from @xcite . right : @xmath284 . the set was taken from @xcite . in the plot on the left , since the value of @xmath26 is sufficiently small , the different approximations converge to the graph of the numerical solution . in the plot on the right it is possible to appreciate the different behavior of the zero - th order and the first order center manifolds . while the first order manifold approximates in a better way the numerical solution , the zero - th order converges to the tqssa , that does not approximate sufficiently well the numerical solution , since it is the zero - th order term of the singular perturbation of the solution in terms of the parameter @xmath26.,title="fig:",scaledwidth=50.0% ] let us consider now a more general system of the following form ( _ outer solution _ ) @xmath285 and the corresponding _ inner solution _ @xmath286 ( with @xmath287 ) where @xmath288 the origin is a fixed point for ( [ eq : b1 ] ) . heineken - tsuchiya - aris system ( [ eq : a13 ] ) and the system obtained by the tqssa approximation ( [ eq : a22 ] ) , are particular cases of the system ( [ eq : b1])-([eq : b2 ] ) . we are able to state a more general theorem concerning the center manifold , which is the main result of our paper . let @xmath289 ; hence : @xmath290+b\left[au+bv+\psi\left(u , v\right)\right]\ ] ] @xmath291 doing so , system ( [ eq : b1 ] ) becomes , for @xmath292 and @xmath293 , @xmath294 the associated linearized system has a block form of type ( [ eq:34 ] ) and , in fact , the eigenvalues are given by @xmath25 ( with multiplicity @xmath218 ) and @xmath293 . thus in every system of the form ( [ eq : b3 ] ) we are in presence of a center manifold . we write equation ( [ eq:40 ] ) for system ( [ eq : b3 ] ) , employing theorem [ th:5 ] . referring to ( [ eq:40 ] ) and ( [ eq:34 ] ) , we have that @xmath220 , @xmath295 . accordingly , we search for a function @xmath296 such that @xmath297 @xmath298 from which , since @xmath299 is a function at least of third order in @xmath26 and @xmath72 , while we are interested in a second order expression of function @xmath266 , we can neglect this term and focus on @xmath300 the center manifold of ( [ eq : b1 ] ) and the isolated point of ( [ eq : b1 ] ) are asymptotically equivalent . * step 1 . * using theorem [ th:5 ] we assume @xmath301 and it is trivial to prove that @xmath266 satisfies ( [ eq : b4 ] ) for @xmath302 . moreover , from ( [ eq : b2 ] ) , @xmath303+\dots\ ] ] where @xmath304 contains the quadratic terms in @xmath72 and @xmath305 . since the terms in @xmath306 and @xmath307 , with @xmath308 , are at least of third order in @xmath26 and @xmath72 , we consider only term in @xmath309 . therefore , @xmath310u^2+\dots\ ] ] while for @xmath311 it is sufficient to consider the first order expansion in @xmath72 because , otherwise , in ( [ eq : b4 ] ) we would have third order terms for @xmath312 in @xmath26 and @xmath72 . thus , @xmath313u+\dots\ ] ] where we recall that @xmath314 . accordingly , equation ( [ eq : b4 ] ) becomes : @xmath315 \notag \\ & + \frac{b}{2 } \left[\psi_{uu}(0,0)-2\frac{a}{b } \psi_{u , v}(0,0)+\left(\frac{a}{b}\right)^2 \psi_{v , v}(0,0)\right]u^2+\dots = 0\end{aligned}\ ] ] equating to zero terms of the same power gives @xmath316 \notag \\ \lambda_2&=-\frac{a}{b}\left[\varphi_u(0,0)-\frac{a}{b } \varphi_v(0,0)\right ] \notag \\ \lambda_3&=0\end{aligned}\ ] ] hence , the center manifold for system ( [ eq : b1 ] ) is @xmath317 u^2 \notag \\ & -\frac{a}{b}\left[\varphi_u(0,0)-\frac{a}{b } \varphi_v(0,0)\right ] u \varepsilon+\dots\end{aligned}\ ] ] setting in the rhs @xmath42 , we obtain the center manifold @xmath318 of ( [ eq : b1 ] ) . singular point technique @xcite * on the other hand , @xmath319 since , setting @xmath164 , we have that @xmath320 , equation ( [ eq : b4 ] ) becomes @xmath321 which gives an identity up to @xmath322 , if we substitute @xmath323 as above and if we operate a taylor expansion around @xmath256 . the vector field reduced to the center manifold , from equation ( [ eq:35 ] ) of theorem [ th:3 ] , is : @xmath324 or , in terms of the original time scale , @xmath325 moreover : @xmath326 where @xmath314 and all the derivatives are calculated in @xmath327 . hence , @xmath328u+w\frac{\varphi_v(0,0)}{b}+\dots\ ] ] and , since @xmath329 $ ] , we have : @xmath330 for @xmath314 . from ( [ eq : b5 ] ) , the vector field reduced to the center manifold , in terms of the original time scale , near the origin , becomes : @xmath331 for values of @xmath323 and @xmath332 as above . summarizing , we have obtained two relations : * a ) from the center manifold theory * : considering ( [ eq : b4 ] ) , for @xmath296 , and setting @xmath42 , we have : @xmath333 * b ) from singular perturbation techniques * : by assumption b of section [ sec : singpert ] , and since @xmath334 in ( [ eq : b3 ] ) , we have , putting @xmath164 : @xmath335 comparing ( [ eq : center1 ] ) and ( [ eq : center2 ] ) , we observe a relation between @xmath336 and @xmath337 but we can not infer that @xmath338 , due to the non - uniqueness of center manifold . however , in the above steps we have proven that @xmath339 q.e.d . this theorem means that the center manifolds obtained by means of ( [ eq : center1 ] ) and ( [ eq : center2 ] ) are asymptotically equivalent . this allows us to interpret any qssa , obtained imposing @xmath340 , as a manifold which is asymptotically equivalent to the center manifold , as confirmed by equation ( [ eq : b10 ] ) for heineken - tsuchiya - aris system , and by equations ( [ eq : a30])-([eq : a31 ] ) for the tqssa . this explains why , in order to achieve the center manifold of ( [ eq : a9s ] ) and ( [ eq : aaa12 ] ) , it is sufficient to consider - for @xmath282 - the expression obtained equating to zero the second equation of these systems , ( i.e. for @xmath164 ) . we recall that in many papers ( see , for example , @xcite , who refer to tihonov s theorem , and @xcite , according to singular perturbation theory ) , the center manifold is obtained equating to zero the right hand side of the equation of the form : @xmath341 the quasi - steady state approximation has been a challenge for applied mathematicians , who had to explain the feasibility of an approximation which imposes to the complex @xmath4 both to be constant and to depend on @xmath3 . some biochemistry texts ( see , for example , @xcite ) mislead the reader , interpreting the qssa as a true equality , which brings to assert that the ratio @xmath342 is constant during all the quasi - steady state phase . this is obviously not true . in @xcite the authors solve the apparent incongruence , determining the asymptotic value of @xmath342 , showing that , for every choice of the kinetic parameters and of the initial conditions , @xmath343 ( where @xmath344 $ ] ) , differently from what is wrongly stated . heineken et al . @xcite and successively other authors @xcite interpreted sqssa and tqssa as leading order expansions of the solutions in terms of a suitable parameter , which has to be considered small . this interpretation allows us to embed the qssa theory in a framework which is related to tihonov s theorem @xcite , where the parameter multiplies the derivative of @xmath4 and the qssa can be obtained as the singular point of the original system , setting @xmath42 . in this paper we have shown that , at least in the classical simple scheme ( [ eq : a3 ] ) , the approximation obtained applying tihonov s theorem is asymptotically equivalent to the center manifold of the system , which means that reduced system and center manifold are two sides of the same coin . once again , the total qssa has shown to be much more efficient and natural than the standard one , mainly thanks to the fact that the parameter used for the expansions in the total framework is always less than @xmath345 . in our actual researches we are applying the techniques shown in this paper to more complex enzyme reactions , as the fully competitive inhibition @xcite , the phosphorylation - dephosphorylation cycle ( or goldbeter - koshland switch @xcite ) , the linear double phosphorylation reaction , the double phosphorylation - dephosphorylation cycle @xcite and , more in general , futile cycles @xcite . these mechanisms were already studied in terms of tqssa in previous papers @xcite . the techniques here shown will allow to read the tqssa as the leading term of an asymptotic expansion in terms of a suitable perturbation parameter , in these more complex cases , too . the authors are deeply grateful to prof . enzo orsingher , from sapienza university ( rome , italy ) and prof . jan andres , from palacky university ( olomouc , czech republic ) for their translations of the papers @xcite and some precious clarifications concerning some passages of the papers . d. andreucci , a. m. bersani , g. dellacqua , e. bersani , c. de lazzari , m. ledda , a. lisi , g. pontrelli , a reaction - diffusion numerical model to predict cardiac tissues regeneration via stem cell therapy , mascot11 proceedings , r. m. spitaleri ( ed . ) , imacs series in computational and applied mathematics 17 ( 2013 ) , imacs , rome , pp . a.m. bersani , e. bersani , g. dellacqua , m.g . pedersen , new trends and perspectives in nonlinear intracellular dynamics : one century from michaelis menten paper , continuum mechanics and thermodynamics , 1 - 26 , ( 2014 ) . a.m. bersani , e. bersani , l. mastroeni , deterministic and stochastic models of enzymatic networks - applications to pharmaceutical research , comp . , special issue : r. tadei and n. bellomo ( editors ) , `` modeling and computational methods in genomic sciences '' , 55 , pp . 879 - 888 ( 2008 ) a. m. bersani , a. borri , a. milanesi , p. vellucci , tihonov theory and center manifolds for inhibitory mechanisms in enzyme kinetics , submitted to comm . ind . math . a.m. bersani , g. dellacqua , is there anything left to say on enzyme kinetic constants and quasi - steady state approximation ? , j. math . , 50 , 335 - 344 ( 2012 ) a.m. bersani , g. dellacqua , g. tomassetti , on stationary states in the double phosphorylation - dephosphorylation cycle , aip conf . 1389 , numerical analysis and applied mathematics icnaam , halkidiki ( greece ) , 19 - 25 september 2011 , pp . 1208 - 1211 ( 2011 ) m.z . bodenstein , eine theorie der photochemischen reaktionsgeschwindigkeiten . 85 , 329397 ( 1913 ) j. borghans , r. de boer and l. segel , extending the quasi - steady state approximation by changing variables , bull . biol . , 58 , pp . 4363 ( 1996 ) j. carr , applications of center manifold theory . springer - verlag : new york , heidelberg , berlin , ( 1981 ) . chapman , l.k . underhill , the interaction of chlorine and hydrogen . the influence of mass , j. chem . , 103 , pp . 496508 ( 1913 ) g. dellacqua and a.m. bersani , a perturbation solution of michaelis - menten kinetics in a `` total '' framework , j. math . , 50 , pp . 11361148 ( 2012 ) g. dellacqua , a.m. bersani , quasi - steady state approximations and multistability in the double phosphorylation - dephosphorylation cycle , communications in computer and information science 273 , pp . 155 - 173 ( 2012 ) g. dellacqua , a.m. bersani , bistability and the complex depletion paradox in the double phosphorylation - dephosphorylation cycle , proceedings bioinformatics 2011 , pp . 55 - 65 ( 2012 ) . s. dimitrov , s. markov , metabolic rate constants : some computational aspects , mathematics and computers in simulation , 133 , pp . 91110 ( 2017 ) j. w. dingee and a. b. anton , a new perturbation solution to the michaelis - menten problem , aiche j. , 54 , pp . 13441357 ( 2008 ) i. dvok , j. ika , analysis of metabolic systems with complex slow and fast dynamics , bull . biol . , 51 , pp . 255 - 274 ( 1989 ) i. giorgio , u. andreaus , d. scerrato , f. dellisola , a visco - poroelastic model of functional adaptation in bones reconstructed with bio - resorbable materials . model mechanobiol . ( 2016 ) 15(5 ) : 1325 - 1343 . a. goldbeter , d.e . koshland , an amplified sensitivity arising from covalent modification in biological system , proc . , 78 , pp . 6840 - 6844 ( 1981 ) g. g. hammes , thermodynamics and kinetics for the biological sciences , wiley - interscience , new york ( 2000 ) . f. g. heineken , h. m. tsuchiya , r. aris , on the mathematical status of the pseudo - steady state hypothesis of biochemical kinetics , mathematical biosciences 1.1 , 95 - 113 , ( 1967 ) . khoo , m. hegland , the total quasi - steady state assumption : its justification by singular perturbation and its application to the chemical master equation , anziam j. , 50 , pp . c429c443 ( 2008 ) a. kumar , k. josic , reduced models of networks of coupled enzymatic reactions . journal of theoretical biology 278.1 , 87 - 106 , ( 2011 ) . k. j. laidler , theory of the transient phase in kinetics , with special reference to enzyme systems , can . 33 , pp . 16141624 ( 1955 ) a. l. lehninger , principles of biochemistry , w.h . freeman & company , new york ( 2008 ) lin , l.a . segel , mathematics applied to deterministic problems in the natural sciences , society for industrial and applied mathematics ( siam ) , philadelphia ( 1988 ) y. lu , t. lekszycki , a novel coupled system of non - local integro - differential equations modelling young s modulus evolution , nutrients supply and consumption during bone fracture healing . . ( 2016 ) 67 : 111 . j. monod , j. wyman , j .- changeux , on the nature of allosteric transitions : a plausible model . j. mol 12 , pp . 88118 ( 1965 ) a.h . nguyen and s.j . fraser , geometrical picture of reaction in enzyme kinetics , j. chem . 91 , pp . 186193 ( 1989 ) b.o . palsson , e.n . lightfoot , mathematical modelling of dynamics and control in metabolic networks . i. michaelis - menten kinetics , j. theor . biol . , 111 , pp . 273 - 302 ( 1984 ) b.o . palsson , e.n . lightfoot , mathematical modelling of dynamics and control in metabolic networks . simple dimeric enzymes , j. theor . , 111 , pp . 303 321 ( 1984 ) b.o . palsson , h. palsson , e.n . lightfoot , mathematical modelling of dynamics and control in metabolic networks . iii . linear reaction sequences , j. theor . biol . , 113 , pp . 231 259 ( 1985 ) b.o . palsson , on the dynamics of the irreversible michaelis - menten reaction mechanism , chem . , 42 , pp . 447 - 458 ( 1987 ) m. g. pedersen , a. m. bersani and e. bersani , the total quasi steady - state approximation for fully competitive enzyme reactions , bull . biol . , 69 , pp . 433457 ( 2005 ) m. g. pedersen , a. m. bersani , e. bersani and g. cortese , the total quasi - steady state approximation for complex enzyme reactions , mathematics and computers in simulation ( matcom ) , 79 , pp . 10101019 ( 2008 ) m. g. pedersen , a. m. bersani and e. bersani , quasi steady - state approximations in intracellular signal transduction a word of caution , j. math . , 43 , pp . 13181344 ( 2008 ) m. g. pedersen and a. m. bersani , introducing total substrates simplifies theoretical analysis at non - negligible enzyme concentrations : pseudo first - order kinetics and the loss of zero - order ultrasensitivity , j. math . biol . , 60 , pp . 267 - 283 ( 2010 ) m. schauer , r. heinrich , analysis of the quasi - steady - state approximation for one - substrate reaction , j. theor . biol . , 79 , pp . 425 - 442 ( 1979 ) s. schnell , p.k . maini , enzyme kinetics far from the standard quasi - steady state and equilibrium approximation , math . comput . model . , 35 , pp . 137 - 144 ( 2002 ) l. a. segel , modeling dynamic phenomena in molecular and cellular biology , cambridge univ . press , cambridge ( 1984 ) l. a. segel , on the validity of the steady state assumption of enzyme kinetics , bull . math . biol . , 50 , pp . 579593 ( 1988 ) l. a. segel and m. slemrod , the quasi steady - state assumption : a case study in pertubation , siam rev . , 31 , pp . 446477 ( 1989 ) j. sijbrand , properties of center manifolds , trans . 289 , 431 - 469 , ( 1985 ) . swoboda , the kinetics of enzyme action , biochim . acta , 23 , pp . 7080 ( 1957 ) swoboda , the kinetics of enzyme action , ii . the terminal phase of the reaction , biochim . biophys . acta , 25 , pp . 132135 ( 1957 ) a.n . tikhonov , on the dependence of the solutions of differential equations on a small parameter , matematicheskii sbornik 64.2 , 193 - 204 , ( 1948 ) . tikhonov , on a system of differential equations containing parameters . , 27 , 147156 , ( 1950 ) . tikhonov , systems of differential equations containing small parameters in the derivatives . matematicheskii sbornik 73.3 , 575 - 586 , ( 1952 ) . l. wang , e.d . sontag , on the number of steady states in a multiple futile cycle , j. math . , 57 , pp . 29 - 52 ( 2008 ) l. wang , e.d . sontag , singularly perturbed monotone systems and an application to double phosphorylation cycles , j. nonlinear sci . , 18 , pp . 527550 ( 2008 ) w. wasow , asymptotic expansions for ordinary differential equations . courier corporation , ( 2002 ) . s. wiggins , normally hyperbolic invariant manifolds in dynamical systems . springer - verlag : new york , heidelberg , berlin , ( 1994 ) . s. wiggins , introduction to applied nonlinear dynamical systems and chaos . 2 . springer science @xmath346 business media , ( 2003 ) . e. n. yeremin , the foundations of chemical kinetics , mir pub . , moscow ( 1979 )	in this paper we prove that the well - known quasi - steady state approximations , commonly used in enzyme kinetics , which can be interpreted as the reduced system of a differential system depending on a perturbative parameter , according to tihonov theory , are asymptotically equivalent to the center manifold of the system . this allows to give a mathematical foundation for the application of a mechanistic method to determine the center manifold of ( at this moment , still simple ) enzyme reactions .
this section is devoted to the computation of the reynold s stress @xmath90 using the stochastic equation for the fluctuations @xmath91 which is equation ( 5 ) in the article . the noise @xmath87 is white in time and has spatial correlation function @xmath92 . using the ito convention for stochastic calculus , it is straightforward to obtain the enstrophy balance for the fluctuations @xmath93 and thus in statistical stationary state we get the very useful relation @xmath94.\ ] ] thanks to the incompressibility condition , @xmath95 is related to the reynolds stress through @xmath96 . first we can take advantage of the invariance along the @xmath55 direction by taking the fourier transform of ( [ eq : fluctuations ] ) in @xmath55 . the fourier transform in @xmath77 does not provide an obvious simplification as the profile @xmath3 depends on @xmath77 . however , we can use the linearity to express the solution as the sum of particular solutions for independent stochastic forcings @xmath97 . each of these forcings has a correlation function @xmath98 , this means that we have the relation * _ @xmath99=e^{il(y - y')}\delta(t - t')$]_. we take the fourier transform in @xmath55 defined by @xmath100 with @xmath54 taking the values @xmath101 , @xmath102 is an integer . @xmath103 is then defined as the function @xmath104 that is solution of ( [ eq : fluctuations ] ) with a stochastic forcing with only one fourier component @xmath105 . we then obtain @xmath106,\label{eq : tenseur decompose}\ ] ] where the positive constants @xmath107 are the fourier coefficients of the spatial correlation function of the noise @xmath87 , @xmath108=c(x - x',y - y')\delta(t - t').$ ] be careful that * in this formula the bracket @xmath109 denotes a stochastic averaging , because the zonal average is already taken into account by the sum over all vector @xmath54 . the vorticity @xmath103 is the solution of the stochastic partial differential equation @xmath110 @xmath111 is the stream function defined through @xmath112 . as the reader would have notice , we try to reduce the problem by expressing the solution as the sum of particular problems that we hope to be much simpler . now we have to find an expression for @xmath113 instead of the full solution @xmath114 . we will go one step further and show that the stochastic problem described by the two equations ( [ eq : tenseur decompose]-[eq : quasilinearcomplex ] ) reduces in fact to a _ deterministic one , _ following _ _ @xcite__. _ _ equation ( [ eq : quasilinearcomplex ] ) can be formally written as @xmath115=-\alpha\omega_{k , l}+\eta_{l},\ ] ] where @xmath116=iku\omega_{k , l}+ik(\beta - u")\psi_{k , l}\label{eq : lk}\ ] ] is a linear operator for a given @xmath3 . then we use the fact that the noise @xmath117 is white in time and has an exponential correlation function * @xmath98 * to express the quantity @xmath109 as @xmath118\right\|^{2}.\label{eq : omegacarre}\ ] ] this formula should be understood as follows : @xmath119 $ ] is the solution at time @xmath120 of the _ deterministic equation _ @xmath121=0 $ ] with initial condition @xmath122 . the subscript @xmath123 will mean that we are dealing with the solution of a deterministic equation . the exponential @xmath124 ensures the convergence of this integral . the great advantage to have reduced the stochastic problem to a deterministic one is that we now have to solve an hydrodynamic problem , the propagation of a vorticity fluctuation in a shear flow , a problem for which much has already been done in the literature . we thus have to solve the deterministic equation @xmath125 and we will take advantage of the small scale forcing limit where @xmath32 . we will consider the case of an infinite space in the @xmath77 direction , but we expect the result to be qualitatively similar for a finite space because it is sufficient that the scale of the forcing @xmath49 is small compared to the domain size , @xmath126 please be careful here that @xmath17 and @xmath54 are different , @xmath127 . we consider the limit @xmath17 infinite with both @xmath54 and @xmath128 going to infinity * * such that @xmath129 with @xmath72 a fixed constant . we will check that the limit @xmath32 and @xmath130 are not at all equivalent . as the leading order term in equation ( [ eq : deterministic ] ) is the free transport term * @xmath131 , * we will use the natural ansatz @xmath132 in equation ( [ eq : deterministic ] ) . we obtain for @xmath133 the equation @xmath134 where @xmath135 is the green function of the laplacian operator @xmath136 solving @xmath137 . the green function for the full laplacian @xmath138 writes @xmath139 . for infinite space , we have thus @xmath140 . this expression is very convenient to do asymptotic calculations because , as the reader can see , @xmath141 is of order @xmath142 and thus goes to zero for large @xmath54 . at zero order , we have only free transport of the perturbation , and the expression of @xmath114 is given by @xmath143 the computation of the next order is more technical . from equation ( [ eq : omegacarre ] ) we have the basic equality @xmath144 again , the bracket @xmath23 is simply a stochastic averaging . an integration by parts gives @xmath145 because @xmath146 , it compensates exactly the @xmath147 coming from the enstrophy injection in expression ( [ eq : tenseur decompose ] ) . replacing @xmath141 using ( [ eq : amplitude ] ) and equation ( [ eq : tenseur decompose ] ) that gives the expression of @xmath148 , we get the _ exact _ expression for the reynolds stress divergence @xmath149 now we will have to use our hypothesis @xmath32 to go on in the computation . here comes a small subtlety , the @xmath128 component does not appears explicitly , it is hidden in the initial condition @xmath150 therefore we can not just expand @xmath151 in power of @xmath142 as one * * may guess at first sight . but expression ( [ eq : amplitude ] ) tells us that @xmath152 as @xmath17 goes to infinity . the right way to do the asymptotic expansion consists in an expansion of @xmath153 wrt time and we use that each temporal derivation of @xmath133 is smaller of order @xmath49 @xmath154_{t=0}+o\left(\frac{1}{k^{2}}\right),\end{aligned}\ ] ] and we expand also @xmath3 using that @xmath54 is large @xmath155 the last expression defines @xmath156 and @xmath157 . using that @xmath158 , @xmath159 for infinite space , the first nonzero contribution is given by @xmath160 with @xmath161 . one could check easily that the expression ( [ eq : machinappendiceb ] ) is zero when @xmath162 , that s why we have to keep the @xmath157 term to get the leading order term in the expansion in powers of @xmath163 . we now prove * * that the second term in the expansion of @xmath164 is zero . with expression ( [ eq : amplitude ] ) , we compute @xmath165 and then @xmath166 the sum of both terms is zero . the next contribution then implies two derivatives and is of order @xmath167 . the presence of both exponentials in ( [ eq : machinappendiceb ] ) allows us to invert the order of integration and integrate the time first . we get @xmath168 now we have to use that @xmath169 is small and develop the denominator . this gives @xmath170 now comes a trick : @xmath171 and , recalling that * @xmath172 , * we can put the derivative in @xmath77 in front of the integral . this is a method to recognize that our expression is a derivative in @xmath77 and get the analytic expression of the reynolds stress . the idea to recognize a derivative seems to fall from nowhere , but it comes in fact from the work @xcite and the expression they derived for the reynolds stress . in fourier space , the incompressibility condition writes * @xmath173 , * we then get @xmath174 and finally @xmath175 the integration procedure defines @xmath90 up to a constant . this constant has no influence on the solution of the equation for the mean velocity profile , equation ( 4 ) in the article . in general , the expression of the reynolds stress will depend on the boundary conditions . let us comment on this result . the parameter @xmath1 is small because we have done the quasilinear approximation , but we keep it finite . it has a regularizing effect , and ensures that the integral remains well defined for every profile @xmath3 even if @xmath79 somewhere in the flow . the result above does not really makes sense for @xmath72 close to @xmath176 but the value @xmath176 is excluded because we assumed that the stochastic forcing does not inject energy in the mean flow . in numerical simulations , the spectrum of the stochastic forcing is often an annulus with a weight function to make it more or less anisotropic . the fourier component @xmath177 is excluded because it corresponds to a direct stochastic forcing on the zonal flow ( see figure ( [ fig : typical - spectrum ] ) ) . to get results we can compare to simulations , we have to integrate the contributions to the reynolds stress over the whole spectrum . ] we will thus define a weight function @xmath178 such that the correlation function fourier component is @xmath179*_. _ * on figure ( [ fig : typical - spectrum ] ) for example , we have a constant function @xmath180 except around the excluded values @xmath176 and @xmath181 . we approximate the discrete sum in ( [ eq : tenseur decompose ] ) by an integral assuming @xmath182 because we have fixed the mean kinetic energy to @xmath183 , the function @xmath180 must satisfy @xmath184 in the last integral , we have done the change of variable @xmath185 using this function to characterize the spectrum , we obtain the simple expression for the reynold s stress @xmath186 where @xmath187 is the inverse fourier transform of @xmath180 , @xmath188 . now that we have an expression for the reynolds stress in terms of the mean profile @xmath3 we can solve the equation for the mean velocity profile and find the stationary solution , if any . we let @xmath36 be the function defined through @xmath189 , and the stationary profile is defined by the set @xmath190 the beautiful fact is that this system is integrable . if we replace @xmath90 in the first equation , it comes @xmath191 so we can now multiply the equality by @xmath5 and integrate . let @xmath38 be a primitive of @xmath192 and we have @xmath193 this is equation [ eq : dynamics ] in the article . the aim of this section is to explain how we can compute the reynolds stress in the inertial limit @xmath31 starting from the relation ( [ eq : tenseur decompose ] ) . we have to compute @xmath194\right\|^{2},\label{eq : formuleannexea}\ ] ] where @xmath119:=\omega_{d}$ ] is the solution to the deterministic equation @xmath195 with initial condition @xmath98 . we will first assume there are no neutral modes solutions of ( [ eq : perturbationannexea ] ) . first , we do the change of time scale @xmath196 in the integral of ( [ eq : formuleannexea ] ) . it gives us @xmath197\right\|^{2}.\ ] ] when @xmath1 goes to zero , the term @xmath198 $ ] is the long time limit of the solution of ( [ eq : perturbationannexea ] ) . we use the nontrivial result for the case of non monotonous flows , of @xcite , that there exists a function @xmath199 such that @xmath200 when there are no neutral modes . hence @xmath201\right\|\rightarrow\|\omega_{d}^{\infty}(y)\|$ ] , and the presence of the exponential in the integral ensures the convergence of the whole . this proves that without neutral modes @xmath202 how then can we compute the function @xmath203 ? we start from equation ( [ eq : perturbationannexea ] ) that describes the linear evolution of a perturbation @xmath204 of meridional wave number @xmath54 , and with streamfunction * @xmath205 . * the idea is to transform equation ( [ eq : perturbationannexea ] ) into an inhomogeneous rayleigh equation , as classically done , and then to study its asymptotic solutions close to the real axis , which is the limit @xmath206 , with the notations below . we introduce the function @xmath207 which is the laplace transform of the stream function @xmath208 i.e @xmath53 . the equation for @xmath58 is @xmath209 with the boundary conditions that @xmath58 vanishes at infinity . we do not have a flow infinite in the @xmath77 direction , but as was already stated , the properties of the flow become local for large @xmath17 . the choice to take vanishing boundary conditions at infinity is done for convenience and it is expected that this particular choice does not modify the physical behavior of the perturbation . for all @xmath210 the function @xmath58 is well defined . the inhomogeneous rayleigh equation ( [ eq : inhomogeneous_rayleigh ] ) is singular for @xmath59 and for any critical point ( or critical layer ) @xmath211 such that the zonal flow velocity is equal to the phase speed : @xmath212*. * one can show that @xmath58 has a limit denoted @xmath213 when @xmath214 goes to zero . the function @xmath203 is then given by @xmath215 see @xcite for more details . boris galperin , roland mb young , semion sukoriansky , nadejda dikovskaya , peter l read , andrew j lancaster , and david armstrong . cassini observations reveal a regime of zonostrophic macroturbulence on jupiter . , 229:295320 , 2014 . carolyn c porco , robert a west , alfred mcewen , anthony d del genio , andrew p ingersoll , peter thomas , steve squyres , luke dones , carl d murray , torrence v johnson , et al . cassini imaging of jupiter s atmosphere , satellites , and rings . , 299(5612):15411547 , 2003 . colette salyk , andrew p ingersoll , jean lorre , ashwin vasavada , and anthony d del genio . interaction between eddies and mean flow in jupiter s atmosphere : analysis of cassini imaging data . , 185(2):430442 , 2006 .	it is extremely uncommon to be able to predict analytically , from first principles , the velocity profile of a turbulent flow . in two - dimensional flows , atmosphere dynamics , and plasma physics , large scale coherent jets are created through inverse energy transfers from small scales to the largest scales of the flow . we prove that in the limits of vanishing energy injection , vanishing friction , and small scale forcing , the velocity profile of a jet obeys a universal equation independent of the details of the forcing . we find an other universal relation for the maximal curvature of a jet and we give strong arguments to support the existence of an hydrodynamic instability at the point with minimal jet velocity . those results are the first computations of reynolds stresses from first principle in a genuine turbulent flow , and the first consistent analytic theory of zonal jets in barotropic turbulence . theoretical prediction of velocity profiles of inhomogeneous turbulent flows is a long standing challenge , since the nineteenth century . it involves closing hierarchy for the velocity moments , and for instance obtaining a relation between the reynolds stress and the velocity profile . since boussinesq in the nineteenth century , most of the approaches so far have been either empirical or phenomenological . even for the simple case of a three dimensional turbulent boundary layer , plausible but so far unjustified similarity arguments may be used to derive von krmn logarithmic law for the turbulent boundary layer ( see for instance @xcite ) , but the related von krmn constant @xcite has never been computed theoretically . still this problem is a crucial one and has some implications in most of scientific fields , in physics , astrophysics , climate dynamics , and engineering . in this paper we find a way to close the hierarchy of the velocity moments , from first principles , for the equation of barotropic flows with or without @xmath0 effect . this two dimensional model is relevant for laboratory experiments of fluid turbulence @xcite , liquid metals @xcite , plasma @xcite , and is one of the key toy model for understanding planetary jet formation @xcite and basics aspects of plasma dynamics on tokamaks in relation with drift waves and zonal flow formation @xcite . it is also a relevant model for jupiter troposphere organization @xcite . moreover , our approach should have future implications for more complex turbulent boundary layers , which are crucial in climate dynamics in order to quantify momentum and energy transfers between the atmosphere and the ocean . an interesting property of two dimensional turbulent flows is their inverse energy transfer from small to large scales , sometimes through a cascade among scales , but much more often through a direct transfer from small to largest scales mediated by the average flow . this inverse energy transfer is responsible for the self organization of the flow into coherent structures , that are much more energetic and evolve much slower than the eddies . it has been realized since the sixties and seventies in the atmosphere dynamics and plasma communities that in some regimes turbulent flows are strongly dominated by large scale coherent structures , in which cases a quasilinear approach may be relevant . such a quasilinear approach requires solving a coupled equation for the mean flow and the lyapunov equation that describes the fluctuations with a gaussian approximation , just like the lenard balescu equation in plasma kinetic theory . numerical approaches and theoretical analysis have been systematically developed for fifteen years in order to solve and study such quasilinear or related approximations @xcite . in a recent theoretical paper , the range of validity of such an approach has been established by proving the self consistency of the approximations in weak forcing and dissipation limit ( small @xmath1 , see below ) @xcite . while this work gave theoretical ground to the approach , explicit formula for the reynolds stress can not be expected in general . however , in a recent work @xcite , an expression for the reynolds stress has been derived from the momentum and energy balance equations by neglecting the perturbation cubic terms in the energy balance ( this follows from the quasilinear approach justification @xcite ) , but also neglecting pressure terms ( not justified so far , see @xcite ) . this approach surprisingly predicts a universal velocity profile for the outer region of a large scale vortex in two dimensions , which means that the profile does not depend on the detailed characteristics of the stochastic forcing but only on the total energy injection rate @xmath2 expressed in . another analytic expression for the reynolds stress , has also been derived in the particular case of a linear velocity profile @xmath3 in @xcite . what are the criteria for the validity of these results ? can we reconcile the different results giving a full theoretical justification and extend these for more general cases ? using the same hypothesis as in @xcite , it is straightforward to generalize the expression obtained for a vortex to the case of a jet with mean velocity @xmath3 . the reynolds stress writes @xmath4 where @xmath5 is the spatial derivative of @xmath3 . to fully derive expression ( [ eq : reynolds stress ] ) , the key idea is to use the already justified quasilinear approximation in the limit of small friction @xcite and to further consider the limit of small scale forcing . the limit of small scale forcing and vanishing friction is relevant for giant gaseous planets like jupiter , for which a number of observations are available @xcite . we start from the equations for a barotropic flow on a periodic beta plane with stochastic forces @xmath6 where @xmath7 is the two dimensional velocity field with @xmath8 . @xmath9 models a linear friction , and f is a stochastic force white in time . we recall here that @xmath10 is the coriolis parameter that depends on the north - south coordinate y. following @xcite we choose temporal and spatial units such that the mean kinetic energy is 1 , and @xmath11 . the nondimensional equations are @xmath12 where @xmath13 . now @xmath14 is a nondimensional parameter although we will often refer to it as the `` friction '' . @xmath15 is the new nondimensional coriolis parameter while @xmath10 is the dimensional one . eq . ( [ eq : navier - stokes adim ] ) still has three nondimensional parameters , @xmath16 and @xmath17 . @xmath17 is the typical wavenumber where energy is injected in fourier space . we consider the asymptotics of small friction and forcing ( inertial regime , @xmath18 and small scale forcing @xmath19 , with fixed @xmath0 . in the inertial and small scale forcing limits , energy is injected at small scale and is dissipated at the largest scale of the flow . the main idea is then to separate the flow @xmath20 in two parts , @xmath21 . the mean velocity @xmath22 $ ] called indifferently _ mean flow _ or _ zonal flow _ , is defined as both the zonal and stochastic average of the velocity field . in the following , the bracket @xmath23 will be used for this zonal and stochastic average . we are left with two coupled equations , one governing the dynamics of the mean flow , the other one describing the evolution of eddies . in the limit where @xmath1 is small , it has been proven that fluctuations are of order @xmath24 and thus it is selfconsistent to neglect nonlinear terms in the equation for fluctuations @xcite . using also the incompressibility condition , we obtain the quasilinear model @xmath25\label{eq : quasilinear}\\ & \partial_{t}\omega+u\partial_{x}\omega+(\beta - u'')v=-\alpha\omega+\eta\label{eq : quasilinerar2}\end{aligned}\ ] ] where we have introduced @xmath26 , the vorticity of the fluctuations . eq . ( [ eq : quasilinear ] ) shows that the typical time scale for the evolution of the mean flow @xmath3 is @xmath27 which is , following our assumption @xmath28 , much larger than the time scale for the evolution of eddies . using this time scale separation , we will consider that @xmath3 is a constant field in the second eq . ( [ eq : quasilinerar2 ] ) , and we will always solve @xmath29 for a given @xmath3 . we now sketch the strategy we will follow : * first we solve the linear eq . ( [ eq : quasilinerar2 ] ) and compute the stationary distribution _ @xmath30 as a function of @xmath3_. * the enstrophy balance for the fluctuations allows us to relate @xmath30 to the divergence of the reynolds stress tensor ( see appendix [ sec : appendixa ] ) . * last we can use this expression to close the first eq . ( [ eq : quasilinear ] ) , and discuss possible stationary profiles @xmath3 . to reach the first objective , we take advantage of the asymptotic regimes @xmath31 and @xmath32 . when we take those two limits , it is natural to ask whether they commute or not , and which nondimensional parameter will govern the difference between @xmath31 first or @xmath32 first . our asymptotic calculations show that the key parameter is the ratio between @xmath33 and @xmath1 . taking the limit @xmath31 first amounts to saying that @xmath34 is very large . taking the small scale forcing limit @xmath32 first is technically much easier . the idea is to write eq . ( [ eq : quasilinerar2 ] ) in an integral form using the green function of the laplacian , and use the fact that the green function decreases very fast when @xmath17 is large , which implies that the evolution of the flow is local in space . at this stage of the calculation , @xmath1 is small but finite , and the expression of the reynolds stress depends both on @xmath1 and on the properties of the stochastic forcing . the complete calculation is reported in appendix [ sec : appendixa ] . we emphasize that as long as @xmath1 is kept finite , the reynolds stress is not at all `` universal '' . the result shows that the reynolds stress can be expressed analytically as @xmath35 where the explicit expression of @xmath36 is a parametric integral given in appendix [ sec : appendixa ] . the close equation describing the stationary profile @xmath3 in this limit is thus @xmath37 , which can be integrated using a primitive @xmath38 of the function @xmath39 in @xmath40 it is in itself remarkable that for some range of the parameters , the flow of the barotropic quasilinear model can be computed from a newtonian equation like ( [ eq : dynamics ] ) . in eq . ( [ eq : dynamics ] ) , @xmath38 plays the role of a potential as if the equation would describe a particle moving in a one - dimensional potential . the constant in the right - hand side is set by @xmath41 and depending on the value of this constant , there can be one , two or three solutions as shown on fig . ( [ fig : petitschema ] ) . if @xmath42 , there is one solution for which @xmath3 never vanishes . if @xmath43 , there are three possible solutions , one is periodic , the other two diverge . the periodic solution corresponds to @xmath44 confined in the well of @xmath38 . in that case the flow is periodic and the solution exchanges kinetic energy in the term @xmath45 with potential energy @xmath46 just like a particle moving in a confining potential . outside the well , the solutions are diverging , one corresponds to an increasing @xmath3 and the other to a decreasing @xmath3 . ) . we show here that the situation is analogous to a particle moving in the one - dimensional potential displayed by the blue curve . depending on the value of the constant of motion , the solution can be regular and periodic , or diverges . [ fig : petitschema ] ] if we now take the limit of vanishing @xmath1 in expression ( [ eq : reynolds stress small scale ] ) , a straightforward calculation using the explicit expression of @xmath36 ( given in appendix [ sec : appendixa ] ) allows us to recover the `` universal '' expression for the reynolds stress because @xmath47 which is eq . ( [ eq : reynolds stress ] ) with dimensional units . the physical interpretation of the limiting case ( [ eq : reynolds stress ] ) is very enlightening . the term @xmath48 can be interpreted as the rate of energy transferred from small scale to large scale , therefore expression ( [ eq : reynolds stress ] ) is consistent : with the limit of large @xmath17 , the evolution of eddies becomes local as if the perturbation only sees a region of width @xmath49 around itself , and thus the different parts of the flow are decoupled . the other limit of small @xmath1 forces the energy to go to the largest scale to be dissipated because the dissipation at small scales becomes negligible . let us study the other limit where @xmath1 goes to zero first . the techniques used in this second case are very different than the previous one . the calculation involves laplace transform tools that were first used in @xcite to study the asymptotic stability of the linearized euler equations . in the limit @xmath1 goes to zero , we show in appendix [ sec : appendixb ] how we can derive the relation between the reynolds stress divergence to the long time behavior of a disturbance @xmath50 carried by a mean flow @xmath51 . this is an old problem in hydrodynamics , one has to solve the celebrated rayleigh equation for the stream function @xmath52where @xmath53 is the laplace transform of the stream function , and @xmath54 is the @xmath55-component of the wavevector . involved computations are then required to give the explicit expression of the reynolds stress . let us just mention that the difficulty comes from the fact that we have to take the limit @xmath56 first in eq . ( [ eq : inhomogeneous rayleigh ] ) before @xmath32 . then we can relate @xmath57 to the laplace transform @xmath58 taken at @xmath59 . again , the asymptotic expansion for @xmath60 requires the condition that @xmath61 is large . therefore , there exists a small region of size @xmath49 around the maximum where the asymptotic expansion breaks down . in the outer region away from the extrema , we recover the expression ( [ eq : reynolds stress ] ) in the inertial limit @xmath31 first . for strictly monotonic velocity profiles @xmath3 , we conclude that the inertial limit and the small scale forcing limit commute and the universal expression ( [ eq : reynolds stress ] ) is expected to be valid . using expression ( [ eq : reynolds stress ] ) for the reynolds stress , we can solve eq . ( [ eq : quasilinear ] ) for the stationary profile . it writes @xmath62 whatever the value of the free parameter @xmath41 , all profiles @xmath3 are diverging in finite length . an example of such a profile @xmath3 is given on fig . ( [ fig : divergence ] ) . in red , we have plotted different solutions of eq . ( [ eq : diverging solution ] ) together such that the mean velocity profile @xmath3 is composed of many diverging jets . in blue , we have drawn at hand what we expect qualitatively from a real velocity profile . an example of a real profile obtained by numerical simulations in @xcite is displayed on fig.([fig : numerics ] ) . we observe that both divergences are regularized , by a cusp at the eastward jet maximum , and by a parabolic profile at the westward jet minimum . the second aim of this paper is to explain the asymmetric regularization of the eastward and westward jets respectively . 0.25 ) ( red curve ) together with a qualitative real velocity profile ( blue curve ) . ( b ) real velocity profile obtained in numerical simulations @xcite . the analytic solution for @xmath3 has divergences at the extrema that are regularized by a cusp at the eastward jet and a parabola at the westward jet.,title="fig : " ] 0.25 ) ( red curve ) together with a qualitative real velocity profile ( blue curve ) . ( b ) real velocity profile obtained in numerical simulations @xcite . the analytic solution for @xmath3 has divergences at the extrema that are regularized by a cusp at the eastward jet and a parabola at the westward jet.,title="fig : " ] numerical simulations like the one performed on fig . ( [ fig : numerics ] ) show that the mean velocity profile is regularized at a very small scale at its maximum . as we explained already previously , there exists a region of typical size @xmath49 around the maximum where the asymptotic expansion for the reynolds stress breaks down . it is thus natural to choose the ansatz @xmath63 the scaling in @xmath49 implies that the ratio @xmath64 is very large at the cusp because @xmath65 . the relevant equation we have to use to describe the cusp is the inhomogeneous rayleigh eq . ( [ eq : inhomogeneous rayleigh ] ) . if we put the ansatz @xmath66 in ( [ eq : inhomogeneous rayleigh ] ) and consider the limit of large @xmath17 , we get @xmath67where @xmath68 . the solution of this equation with @xmath56 gives us the laplace transform of the stream function with which we can express the reynolds stress . the @xmath0-effect disappears completely from the equation of the cusp because in the region of the cusp , the curvature is so large that it overcomes completely the @xmath0-effect , as can be seen on the green curve of fig . ( [ fig : numerics ] ) . eq . ( [ eq : rayleigh cusp ] ) together with the equations linking the reynolds stress with @xmath60 have a numerical solution , which proves that our scaling @xmath69 is self - consistent . this solution can not be expressed analytically and depends on the fourier spectrum of the stochastic forcing . an example of a numerical integration of the reynolds stress divergence @xmath70 for the cusp is displayed on fig . ( [ fig : tenseur cusp ] ) for @xmath71 and a stochastic forcing with a semi annular fourier spectrum where @xmath72 ranges between @xmath73 and @xmath74 . ( blue curve ) computed for a parabolic profile @xmath71 around its maximum @xmath75 . it is satisfying to observe that the red curve @xmath76 matches the blue curve for large @xmath77 according to the universal expression ( [ eq : reynolds stress]).[fig : tenseur cusp ] ] on fig . ( [ fig : tenseur cusp ] ) , the red curve is the asymptote @xmath76 obtained from formula ( [ eq : reynolds stress ] ) with @xmath78 . when a jet is in a stationary state , the cusp profile joins smoothly the outer region of the jet where the universal result ( [ eq : reynolds stress ] ) is valid . another physical phenomenon at the maximum of the jet is called `` depletion at the stationary streamlines '' and has first been observed in @xcite . it means that at critical latitudes where @xmath79 , any vorticity perturbation of the flow @xmath80 has to asymptotically vanish with time . one main consequence of this phenomenon is the relation @xmath81 where @xmath82 is the latitude of the extremum . from ( [ eq : curvature ] ) we learn that , even if the velocity profile of the cusp is not universal , the maximal curvature of the profile satisfies a universal relation , and it would therefore be very interesting to check it in full numerical simulations with different types of forcing spectra . the previous discussion successfully explained the jet regularization of the eastward jet cusp . as clearly observed on fig . ( [ fig : numerics ] ) , and on jupiter data , westward jets do not produce cusps . at first sight , it may seem that all the theoretical arguments used so far , the energy balance , the asymptotic expansions , and the results ( [ eq : reynolds stress ] ) and ( [ eq : curvature ] ) , do not break the symmetry between eastward and westward jets , as @xmath0 disappears from all these computations . however , as clearly stressed in @xcite , the whole theoretical approach relies on an assumption of hydrodynamic stability for @xmath3 . the asymmetry is clearly visible in the rayleigh - kuo criterion , that states that when @xmath83 change sign , an instability may develop . we will now argue that the turbulent flow is constantly oscillating between a stable and unstable solution , in order to control the westward jet behavior . to check this marginal stability hypothesis , we solved numerically the homogeneous pde for a perturbation carried by a mean flow @xmath84 we chose a parabolic mean velocity with a small perturbation at its minimum @xmath85 the main curvature @xmath86 is chosen to be slightly smaller than @xmath0 , and we performed simulations using different values of @xmath87 and @xmath88 . we check that the growth rate of the perturbation is indeed exponential with time with a complex rate @xmath89 . in this paper we gave a global consistent picture of how a zonal jet is sustained in a steady state through continuous energy transfer from small scale to large scale , from first principles . extension and generalizations of our approach can be foreseen for other geometries ( on the sphere ) , for more comprehensive quasi - geostrophic models of atmosphere jets , and for classes of flows dominated by a strong mean jet , for instance in some instances of boundary layer theory .
the `` bullet '' cluster 1e0657 - 56 represents one of the most complex and unusual large - scale structures ever observed . located at a redshift @xmath5 it has the highest x - ray luminosity and temperature of all known clusters as a result of overheating due to a recent supersonic mach @xmath6 @xcite central encounter of a sub - cluster ( the bullet ) with its main cluster . the 500 ks chandra acis - i image of 1e0657 - 56 ( fig . 1 in markevitch 2006 ) shows two plasma concentrations with the bullet sub - cluster on the right of the image being deformed in a classical bow shock on the western side as a result of its motion through the hot gas of the main cluster . the analysis of the shock structure leads to the conclusion that the bullet is now moving away from the main cluster with a velocity of @xmath7 km s@xmath8 @xcite . the line - of - sight velocity difference between the two systems is only 600 km s@xmath8 suggesting that the encounter is seen nearly in the plane of the sky @xcite . as the core passage must have occurred @xmath9 gyr ago we have the unique opportunity to study this interaction in a very special short - lived stage , far away from thermal and dynamical equilibrium . as a result of the encounter , the collision - dominated hot plasma and the collisionless stellar and dark matter components have been separated . the galaxy components of both clusters are clearly offset from the associated x - ray emitting cluster gas @xcite . in addition , weak and strong lensing maps @xcite show that the gravitational potential does not trace the distribution of the hot cluster gas that dominates the baryonic mass but follows approximately the galaxy distribution as expected for a collisionless dark matter component . the likelihood to find such a high velocity cluster encounter in a @xmath10cdm cosmology has recently been investigated by @xcite using the millenium run simulation . according to the newest estimates from x - ray and gravitational lensing results the @xcite likelihood becomes @xmath11 @xcite which means that 1e0657 - 56 represents an extremely rare system in a @xmath10cdm universe . recent numerical works @xcite have demonstrated that the relative velocity of the dark matter components associated to the main and the sub - cluster are not necessarily coincident with the bullet velocity inferred from the shock analysis . in details , @xcite using a 2-d eulerian code well reproduced the observed increase in temperature across the shock front with a dark halo velocity @xmath12 lower than that of the shock , while @xcite found even a larger difference between the shock velocity ( @xmath13 km s@xmath8 in their best model ) and the speed of the halo ( only @xmath14 km s@xmath8 ) . moreover , according to @xcite , due to a drop in ram - pressure after the cores interaction the gas component of the sub - cluster can eventually be larger than that of its dark matter counterpart . [ cols="<,^,^,^,^,^,^,^,^,^ " , ] [ runsresults ] the main cluster is initially at rest and the sub - cluster moves in the @xmath15 direction with a velocity which ranges from 2000 to 5000 km s@xmath8 . the initial conditions of the different runs are summarized in table [ runs ] . the velocities of the sub - cluster relative to the center of mass of the system are listed in the last but one column of the table . all the simulations were carried out using gasoline , a parallel sph tree - code with multi - stepping @xcite . most of the runs are adiabatic , with @xmath16 . radiative cooling for a primordial mixture of hydrogen and helium in collisional equilibrium is implemented in 1:6c and 1:3lfgc . the consequences of assuming a lower concentration for the main dark halo are investigated in run 1:6c4 , where @xmath17 . the main cluster is modeled with @xmath18 particles , @xmath19 sph and @xmath20 collisionless . the sub - cluster , with the exception of run 1:8 ( where the number of gas particles in the bullet is @xmath21 ) , has @xmath22 particles , @xmath23 collisional and the remainder dark matter particles . the gravitational spline softening is set equal to 5 kpc for the gaseous and dark component . each numerical work which aims to simulate the bullet cluster should be able to reproduce simultaneously the main features observed in x - ray maps ( the bow shock , the relative surface brightness of the bullet and the main cluster ) , and the observed surface brightness and temperature profiles across the shock discontinuity . an additional constraint is provided by the observed displacement between x - ray and lensing maps , which is not negligible in both the main and the sub - cluster ( @xmath24 and @xmath25 kpc respectively , according to clowe et al . 2006 ) . a first indication about the validity of a model arises from the qualitative comparison of our simulated x - ray surface brightness maps with the x - ray 500 ks _ acis - i images provided by @xcite . the impact parameter @xmath26 is not strictly constrained by observations . nevertheless a head - on merger , with @xmath27 , seems to be excluded by comparing deep x - ray observations and weak lensing maps . in particular in the top panel of fig . 1 which is fig . 1b of @xcite the brightest gas associated with the main cluster is not located along the line which connects the centers of the two total mass distributions . moreover , the x - ray emission from the main cluster is asymmetric , with a peak in the north of the image and an extended tail of less bright material pointing south . these features are hardly associable with a zero impact parameter interaction , as shown in the middle and bottom panels of fig . [ b=0 ] which illustrate two simulated 1:6 head - on encounters where the sub - cluster moves from the left to the right of the image ( @xmath15 axis of the simulation ) with decreasing initial velocities ( the middle panel corresponds to run 1:6vb0 of table [ runs ] , the bottom one to 1:6v3000b0 ) . images are projected along an axis perpendicular to the collision plane ( the encounter is seen face - on ) and the selected snapshot is the one which most closely matches the observed distance between the centers of the total mass distributions , associated with the two clusters about 720 kpc from @xcite once the sub - cluster has passed through the core of the main system ( hereafter the present time ) . in the lower two panels of fig . [ b=0 ] colours represent x - ray maps in the _ chandra _ energy band ( 0.8 - 4 kev ) generated using the theoretical image processing system ( tipsy ) , which produces projected x - ray surface brightness maps with the appropriate variable sph kernel applied individually to the flux represented by each particle . assuming complete ionization and zero metallicity ( metal lines are expected to provide a significant contribution to emissivity only at relatively low temperatures , less than 2 kev ) the x - ray luminosity in a given energy band is defined as @xcite @xmath28 where @xmath29 is the cooling function in the specific band , @xmath30 , @xmath31 and @xmath32 are temperature , density and mass associated with the @xmath33 hot ( @xmath34k ) gas particle , respectively , @xmath35 is the proton mass and @xmath36= 0.6 the mean molecular weight . the sum runs over all the @xmath37 particles within an oblong of base equal to the pixel size and major axis oriented along the line of sight . when the x - ray luminosity of the entire cluster is calculated @xmath37 is the number of hot particles within the virial radius @xmath38 . the cooling function is computed using a raymond - smith code @xcite for a gas of primordial composition . the energy band ( 0.8 - 4 kev ) is chosen in such a way to reproduce the 500 ks _ acis - i image of the bullet cluster @xcite . the entire energy band , used to calculate bolometric x - ray luminosities in the following of the paper , goes from 5 ev to @xmath39 kev . the lower two panels of fig . [ b=0 ] show that after an encounter with zero impact parameter the displacement of the main cluster s x - ray peak is aligned with the @xmath15 axis . a large relative velocity ( central panel ) induces a significant offset between the dark and baryonic component of the main cluster ( see table [ runsresults ] for details ) but it also leads to substantial disruption of the main cluster gaseous core . moreover , the displacement of the bullet from its dark halo ( 278 kpc ) is much larger than observed . decreasing the relative velocity between the two clusters ( bottom panel ) two x - ray peaks are clearly visible but the displacement of the main cluster gas is now negligible due to the lower ram - pressure experienced by the main cluster core . @xcite ( fig . 7 in their article ) provide further examples of head - on encounters with even lower mass ratios and relative velocities ( @xmath40 km s@xmath8 in the center of mass rest frame ) . even assuming extremely low concentrations ( @xmath41 ) for the main halo , the authors never reproduce the displacement observed in the two systems . increasing the concentration strongly increases the luminosity of the main cluster , which appears much brighter than the bullet , contrary to what is observed . a bow shock is clearly visible on the right of each image . the shape of the shock front is only marginally dependent on the kinematics of the model while the distance between the edge of the bullet ( the so called contact discontinuity ) and the shock front becomes larger for decreasing bullet velocities . the contact discontinuity itself is much flatter in the case of 1:6b0 than in the low velocity encounter 1:6v3000b0 and clearly not comparable with observations , which show a more narrow structure . in general , a more efficient ram - pressure during the phase of core - core interaction is associated with a larger opening angle of the contact discontinuity at the present time @xcite . the rest of the runs listed in table [ runs ] have an impact parameter @xmath26 equal to 150 kpc , comparable with the core radius @xmath42 of the main cluster gas distribution for most of the models . for a plasma distributed according to a @xmath2 profile like the one adopted in this paper the assumption @xmath43 kpc implies that the maximum external density crossed by the core of the sub - cluster is 1.5 times smaller than the one it would pass through for @xmath27 kpc . for the same initial velocity , the sub - cluster also sees a weaker potential with respect to the case of a head - on encounter and experiences a smaller maximum orbital velocity . a choice of a much larger value of @xmath26 would further decrease the mutual ram - pressure between the two systems and require much higher relative velocities in order to explain the observed offset between the gas and dark matter . in fig . [ b>0 ] and [ b>02 ] we illustrate the projected x - ray surface brightness maps of some interesting models with impact parameter @xmath44 kpc . the encounters are shown face - on . the box size and the surface mass density contours are the same as in the last two panels of fig . [ b=0 ] . in order to underline the morphological details of the high emission regions the upper limit of the surface brightness scale varies in the different images . individual values are indicated in the captions . among this sub - sample of runs , 1:3 produces the largest displacement of the x - ray peak associated with the main cluster , but the x - ray map differs from the observations . in particular a large strip of strongly emitting gas still connects the two x - ray peaks while the morphology of the main cluster peak is much more elongated than observed . decreasing the mass ratio between the two interacting systems ( runs 1:6 and 1:8 ) the displacement in the bullet becomes larger than that in the main cluster . the run 1:6 is characterized by an initial sub - cluster velocity of 5000 km s@xmath8 ( as well as runs 1:3 and 1:8 ) in the system of reference where the main cluster is at rest , that corresponds to a present time velocity of @xmath45 km s@xmath8 in the center of mass rest frame . the same model is simulated assuming lower relative velocities ( 1:6v3000 and 1:6v2000 ) . with decreasing velocities the offset in the sub - cluster becomes smaller and the x - ray emission from the bullet less pronounced with respect to the bright x - ray emitting region at the center of the main cluster . at the same time the shape of the contact discontinuity changes , getting progressively more narrow while the distance between the contact discontinuity and the shock front increases as will be shown more quantitatively in the next section . the displacement associated with the main - cluster is determined by the distance of closest approach between the centers of the two clusters , which becomes smaller assuming the same initial impact parameter @xmath26 with decreasing bullet velocities . indeed , the separation between the main cluster x - ray emission peak and its dark matter counterpart is maximum in the case of the low velocity run 1:6v2000 . [ future ] a not negligible fraction of the x - ray emission visible at the present time near the center of the main - cluster is actually associated with hot gas stripped from the external regions of the bullet . fig . [ mixing ] refers to run 1:6v3000 . it shows the individual distribution of gas originating from the main ( top ) and sub - cluster ( bottom ) and lying at the present time within 1 mpc from the center of the system . comparing these images with the middle right panel of fig . [ b>0 ] it appears evident that the bright elongated x - ray feature crossing the second innermost isodensity contour is associated with the displaced gaseous center of the main cluster while the surrounding more diffuse region hosts a significant amount of sub - cluster gas . indeed , as it shown in the upper panel of fig . [ mixing ] , the motion of the bullet across the inner regions of the main system creates a low density `` tunnel '' in the main cluster gas distribution . at the same time the sub - cluster looses a large amount of baryonic material during the phase of core - core interaction . this material , which fills the tunnel , falls back into the gravitational center of the main cluster and resides now at more than 500 kpc distance from the x - ray bullet . as we will show in the next section , the amount of gas deposited by the sub - cluster in the central regions of the main system increases with decreasing relative velocities . this trend explains the relative increase in luminosity of the diffuse strongly emitting component if compared to the peak associated to the main cluster core gas ( always in the upper right region with respect to the center of the mass density distribution ) , when we compare run 1:6 with the low velocity run 1:6v2000 where it becomes the primary peak of x - ray emission . as shown in the bottom right panel of fig . [ b>0 ] a main halo with low concentration ( @xmath17 ) does not survive a 1:6 sub - cluster encounter with velocity @xmath46 km s@xmath8 and its x - ray peak is destroyed . as previously noticed by @xcite the choice of a lower gas fraction ( @xmath47 in three of the last four runs of table [ runs ] ) does not affect significantly the x - ray map morphology although the displacement of the two luminosity peaks with respect to their dark matter counterparts slightly changes . finally , we tested the consequences caused by including radiative cooling and choosing a larger main halo model ( fig . [ b>02 ] ) . cooling makes the contact discontinuity more narrow and the amount of diffuse x - ray gas around the peak associated with the main cluster smaller . the offset in the main and sub - cluster remains unaltered . run 1:6v3000big is characterized by a main cluster total mass of @xmath48 , which is closer to the value adopted by @xcite and predicted by fitting the large field weak lensing data with extremely low concentrated ( @xmath49 ) nfw halos . although the initial relative velocity is only 3000 km s@xmath8 , due to the large mass of the host halo the present time velocity of the bullet in the center of mass rest frame is much higher than in the corresponding 1:6v3000 run . consequently the offset of the x - ray peak associated with the main cluster is less than 100 kpc and the amount of gas lost by the bullet in the core of the main system closer to that observed in 1:6 than in 1:6v3000 . summarizing , run 1:6v3000 , with mass ratio 1:6 , initial relative velocity @xmath50 km s@xmath8 and present time sub - cluster velocity @xmath51 km s@xmath8 in the center of mass rest frame , best reproduces the main features observed in x - ray maps , in particular the peculiar morphology of the x - ray emission associated with the main cluster , the relative surface brightness between the main and the sub - cluster , the shape of the shock front and of the contact discontinuity . although the low velocity run 1:6v2000 leads to a x - ray displacement closer to that observed by @xcite both in the main and in the sub - cluster , this model seems to be excluded on the basis of a pure morphological comparison with the observational data . indeed , the bullet seems to be much less bright than the center of the main cluster . [ masses ] compares the integrated mass profiles of the main and the sub - cluster for different mass models with the gravitational lensing results from @xcite and previous simulations by @xcite . the profiles inferred from observations are obtained by measuring the enclosed mass in cylinders centered on the southern cd of the main cluster and on the bcg of the sub - cluster . here we calculated the present time projected mass within cylindrical bins centered on the center of mass of the two clusters . the massive cluster run 1:6v3000big ( green lines ) , overestimates the cumulative profile of the main cluster , while an initial main cluster mass of @xmath52 ( 1:6v3000 , in red ) fits the observational points better than the large - mass - low - concentration model of @xcite in the central regions and underestimates the projected mass by less than @xmath53 between 250 and 450 kpc from the center . blue lines refer to run 1:3lfg where the main halo has the same dark matter mass than 1:6v3000 but lower baryonic fraction . it is interesting to notice that using almost the same initial sub - cluster mass ( the sub - cluster model in run 1:6v3000big and 1:3lfg only differs in the baryonic fraction ) we can nearly reproduce the observed projected mass in the bullet only in the case of a very massive host halo ( green dashed line , 1:6v3000big run ) which overpredicts the main cluster profile ( green solid line ) . on the other hand a smaller main halo which better fits the observational data ( blue solid line ) associated with a similar sub - cluster mass ( 1:3lfg ) underpredicts the lensing results ( blue dashed line ) . our favourite run 1:6v3000 gives even lower projected sub - cluster mass estimates ( red dashed line ) and lies close to the model proposed by @xcite ( their sub - cluster mass is similar to the one used in our 1:6 models ) . these results seem to indicate that the mass of the sub - cluster is a significant part of the main cluster mass , with initial mass ratio between the sub and the main cluster much larger than 1:3 . @xcite successfully reproduces the projected mass associated with the sub - cluster only assuming very massive halos and high mass ratios . in particular the smallest mass ratio allowed by their models is 1:2.7 , with a main cluster and bullet mass of @xmath54 and @xmath55 , respectively . such high masses are however clearly incompatible with the values inferred by galaxy kinematics . in particular the virial mass derived by @xcite for the main cluster is @xmath56 ( comparable with the value obtained by girardi & mezzetti 2001 ) . they also measure the main cluster total r - band luminosity and find @xmath57 using @xmath58 to estimate the cluster mass @xcite we obtain @xmath59 . the bullet mass determination of @xcite is less reliable since it is based on only seven galaxies and on the assumption of equilibrium while the sub - cluster seems to be tidally perturbed beyond @xmath60 kpc from its center . they find @xmath61 . from the total r - band luminosity ( @xmath62 ) we get a value almost five times larger ( @xmath63 ) . the bullet mass proposed by @xcite is even one order of magnitude larger than that obtained from weak lensing analysis ( @xmath64 ) of a new larger lensing field which covers most of the area occupied by the system ( clowe et al . in prep . ) . with such low mass ratios we can not reproduce the morphology of the x - ray maps ( upper left panel of fig . [ b>0 ] ) since the luminosity peak associated with the main cluster core is almost completely destroyed by the interaction . actually any attempt to derive the mass of the main and sub - cluster simply by projecting the mass of two isolated nfw halos and fitting the observed lensing data like the analysis of @xcite is going to provide wrong results . indeed , as we will show in section 4 , the dark matter profiles of the main and sub - cluster change significantly during the central phases of the interaction , with a density increase in the inner regions and a decrement beyond 0.3 @xmath38 . this effect , which characterizes both the main and the sub - cluster , is not related to stripping processes but can be explained by the fact that the two halos are strongly perturbed and can not be idealized as the sum of two isolated equilibrium models . fitting the projected cumulative mass around the centers of the two interacting clusters with nfw models we would overestimate either the virial mass or the concentration of the original clusters . = 8truecm according to @xcite the line of sight velocity difference between the main and the sub - cluster is relatively small , about 600 km s@xmath8 , which implies that the encounter must be occurring nearly in the plane of the sky . in fig . [ inclination ] we show the x - ray surface brightness map of run 1:6v3000 when the orbital plane is viewed inclined by 45 degrees left ( top panel ) and down ( bottom panel ) . in the upper image the bullet ( which is closer to the observer ) appears to be much brighter than the main cluster while the contact discontinuity and the shock front are much rounder than in the case where the encounter is seen perpendicular to the orbital plane . such a large inclination has to be excluded by a comparison with observations . moreover , the line of sight velocity difference between the two clusters ( we assume the line of sight kinematics of the dark matter to be coincident with that of the stellar component observed by barrena et al . 2002 ) is much larger than the observed one ( @xmath65 km s@xmath8 ) due to the fact that the main component ( @xmath66 km s@xmath8 in the center of mass rest frame ) of the sub - cluster velocity is oriented in the direction of the positive @xmath15 axis in the original face - on encounter . a velocity difference comparable to the observed one is reached for a right to left inclination of @xmath67 degrees . on the opposite , a top - down inclination of the plane of the encounter with respect to the line of sight ( bottom panel ) does not produce drastic changes in the morphology of the bow shock and of the edge of the bullet . indeed the main and the sub - cluster still lie close to the plane of the sky , but the relative surface brightness between the dense and the diffuse strongly emitting components at the center of the main system changes . in particular the peak of x - ray emission due to the presence of hot gas deposited by the sub - cluster within the core of the main cluster is now comparable with the main cluster x - ray peak . the difference in line of sight velocity in this case is only @xmath68 km s@xmath8 and would probably not be distinguishable with respect to a pure face - on encounter . [ future ] = 8truecm fig . [ sb ] represents the x - ray brightness profile measured in a narrow slit ( with thickness 20 kpc ) parallel to the @xmath15-axis across the shock front . the bullet is located at @xmath69 with @xmath15 increasing towards the pre - shock region . the general trend does not depend on the model and is comparable with the profile suggested by observations @xcite : all the simulations show an inner bump associated with the bullet and an outer one ( between @xmath70 and @xmath71 kpc ) caused by the shock . the abrupt jump at @xmath72 kpc is the shock front while the pre - shock region is well fitted by a two dimensional @xmath2-profile @xmath73^{-3\beta + 1/2}$ ] . in the case of run 1:6 @xmath74 kpc , @xmath75 and @xmath76 erg s@xmath8 kpc@xmath77 . a different orientation ( @xmath78 , bottom - up inclination of the plane of the encounter ) of the line of sight does not affect substantially the surface brightness profile . if cooling is activated the surface brightness of the pre and post - shock region is smaller , but the thickness of the shock - front and the jump in surface brightness are similar . in order to calculate projected temperatures we need to define a weighting function . the emission weighted temperature @xmath79 was originally introduced to provide a better comparison between simulations and observations with respect to a simple mass weighted temperature definition and has been commonly used in the analysis of simulations @xcite . it assumes a weighting function proportional to the emissivity of each hot gas particle and is defined as @xmath80 recently @xcite have demonstrated that for clusters with a complex thermal structure the emission weighted temperature always overestimates the spectroscopic temperature obtained from x - ray observations due to fact that the source is not a single temperature plasma . for clusters with a temperature @xmath81 kev the discrepancy is @xmath82 . this difference becomes particularly large in the presence of strong temperature gradients like shocks which appear to be much weaker in observations than what is predicted by emission - weighted temperature maps ( mathiesen & evrard 2001 , gardini et al . 2004 , rasia et al . 2005 ) . @xcite proposed a new definition of temperature , the spectroscopic - like temperature @xmath83 : @xmath84 when applied to clusters hotter than 2 - 3 kev this equation , with @xmath85 , gives a good approximation ( within few percent ) of the spectroscopic temperature obtained from data analysis of chandra . the bullet cluster 1e0657 - 56 is characterized by the highest luminosity , temperature and the strongest bow shock of all known clusters @xcite and seems indeed to be the ideal candidate for adopting the spectroscopic - like temperature definition . therefore in the remainder of this paper @xmath83 will be analyzed and @xmath79 only indicated for a comparison . [ temppro1 ] illustrates the projected spectroscopic - like temperature profiles across the shock for the different runs of table [ runs ] . all the values refer to the present time . in particular , the upper left panel shows the temperature jumps associated with different relative velocities of the two clusters . decreasing the initial relative velocity from @xmath86 to @xmath87 km s@xmath8 ( in the system of reference where the main cluster is at rest ) reduces the temperature peak by @xmath88 kev while the peak itself becomes broader since the thickness of the shock increases by almost a factor of two due to the lower pressure exercised by the pre - shock gas . both the 1:6v3000 and 1:6v2000 models seem to fit quite nicely the observed height ( @xmath89 kev ) and thickness ( 150 - 200 kpc ) of the shock front @xcite while the 1:6 run produces a peak which is too narrow ( @xmath68 kpc ) and pronounced ( @xmath90 kev ) . the upper right panel refers to different mass ratios . if the encounter is characterized by the same initial relative velocity and gas fraction the strongest shock is associated with the most massive sub - cluster . a 1:3 adiabatic encounter produces a @xmath91 kev temperature peak with thickness @xmath92 kpc while the 1:8 temperature profile is not substantially different from the one of the 1:6 run . in the same plot we also show the shock created in the massive run 1:6v3000big . clearly the maximum temperature is much higher than the observed one . including radiative cooling ( bottom left panel ) has the effect of reducing the peak in temperature but does not influence the thickness of the shock region . the adoption of a simple cooling model is actually questionable . indeed , although the estimate of the temperature jump is reasonable , the entire temperature profile drops by 5 kev due to the fact that once cooling is activated the main cluster gas component becomes thermally unstable in the early phases of the interaction and overcools in the central regions . if cooling is important , models with higher mass ratios and relative velocities ( like the 1:3lfg run in plot ) have still to be taken in account and can not be excluded as the high temperature peaks could actually cool significantly . the choice of a different line of sight ( bottom right plot ) does not affect significantly the temperature profile along the shock . even decreasing the baryonic fraction in the clusters and assuming a much less concentrated main halo , the height and thickness of the temperature peak do not change . = 8truecm in fig . [ temppro2 ] we compare the spectroscopic - like temperature profile across the shock region with the emission weighted one for our favourite run 1:6v3000 . while the projected temperature profile calculated according to the two definitions is similar ( @xmath79 is only slightly higher than @xmath83 ) in the pre and post - shock regions , the emission weighted temperature @xmath79 in the region @xmath93 kpc is @xmath94 higher than @xmath83 . @xmath95 represents the actual temperature along the@xmath15-axis through the shock . indeed the blue curve in fig . [ temppro2 ] gives the exact temperature jump across the shock ( the `` true temperature '' of the shock ) which is characterized by an even higher peak with respect to the projected ones . this deprojected temperature profile is actually the one shown by @xcite in their fig . 9 where they compare their model with observations which on the other hand refer to projected quantities @xcite . as the calculated local temperatures of @xcite fit the projected observed temperature profile very well we conclude that their projected temperature profiles are actually inconsistent with the observations by @xcite . in fig . [ sequencexray ] , [ sequencetemp ] , [ sequenceden ] we show for our selected adiabatic run 1:6v3000 the evolution with time of the @xmath96 kev x - ray surface brightness , spectroscopic - like temperature and gas surface density during the central phases of the interaction . all the quantities are projected along an axis perpendicular to the plane of the collision . on top of the color maps we draw in black the projected isodensity contours of the total mass distribution , which is dominated by the dark matter component . time increases from the left to the right and from the top to the bottom . the sequence of six panels covers an interval in time of 400 myr , between 1.3 and 1.7 gyr from the beginning of the simulation . the snapshot corresponding to the present time is the one on the bottom left . the bullet approaches the main cluster from the left , with an initial velocity of 3000 km s@xmath8 in the main cluster rest - frame . the shock front has an arc - like shape @xcite and becomes progressively more asymmetric as the bullet moves closer to the core of the main cluster . when the centers of the two clusters are less than 250 kpc separated , ram - pressure becomes effective in producing a displacement ( visible in the projected density maps as well as in surface brightness ) between the gaseous core of the bullet and the peak of its associated mass distribution . the offset in the main cluster is evident only when the bow shock passes through its core . in the central panels of fig . [ sequencexray ] and [ sequenceden ] the x - ray luminosity and surface density saturate in order to distinguish features in the maps at later times . nevertheless , it is visible how the main cluster core gets compressed and displaced from the center of the potential towards the top right of the image and appears in the x - ray maps ( last two panels ) as an elongated structure , characterized by high surface brightness . temperature maps better describe the evolution of the shock region , which gets compressed and hotter during the core passage while in a later phase it cools down and becomes thicker due to lower pressure of the pre - shock gas . the bullet itself expands as it leaves the central regions of the main cluster . as observed by @xcite and previously noticed by @xcite , despite of its strong x - ray emission ( it is the brightest feature in the post core - core interaction x - ray maps ) the bullet remains relatively cold . even if radiative cooling is not activated , the core of the sub - cluster is heated only to a maximum temperature of @xmath97 k , while the shock front is much hotter ( @xmath98 k ) . the projected temperature associated with the bullet is higher when it passes through the core of the main system ( central panels of fig . [ sequencetemp ] ) due to the line of sight overlap with the hot gas from the main cluster and the shock heated material stripped from the bullet itself which surrounds it . as soon as the sub - cluster moves out into the cool external regions of the main cluster and loses part of the hot envelope of stripped gas , its projected temperature decreases to values comparable to the observed ones . another peculiar feature in the temperature maps is the high temperature region next to the innermost total density contour of the main cluster and visible in the middle right and bottom left panels of fig . [ sequencetemp ] . this area could be associated with the southeastern high temperature region observed by @xcite ( regions @xmath99 , @xmath100 and @xmath101 in their fig . 2 ) in the main cluster x - ray map and assumed to be coincident with the main merger site . in our simulations the high temperature region is filled with hot gas stripped from the external regions of the sub - cluster and deposited within the core of the main system ( compare with fig . [ mixing ] ) . this high temperature material , combined with the main cluster gas which lies in the same projected region , produces the diffuse x - ray high emission feature visible at the present time in the main cluster below the primary peak ( bottom left panel of fig . [ sequencexray ] or middle right panel of fig . [ b>0 ] for a better colour contrast ) . the eastern side of the high temperature region is less bright in x - ray emission and lies beyond the luminosity peak associated with the main cluster . as the sub - cluster moves to larger radii , the high temperature region expands and cools . shortly ( @xmath102 myr ) after the present time this gas is not anymore clearly distinguishable in the x - ray and density maps . in this section we will focus on a sample of runs from table [ runs ] namely 1:6 , 1:6v3000 , 1:6v2000 and 1:3clfg and investigate in details their three - dimensional characteristics . three of these simulations are adiabatic models characterized by different initial relative velocities and will permit us to study the effects of the sub - cluster speed on the present and future state of the encounter . the remaining run 1:3clfg is the one which better reproduces together with 1:6v3000 and 1:6v2000 the observed jump in temperature across the shock front . [ shock ] illustrates the behavior of hydrodynamical quantities across the shock discontinuity for a snapshot which corresponds to the present time . the horizontal axis is centered on the gaseous bullet and oriented perpendicular to the bow shock nose . the bow shock location is indicated by a vertical line while the edge of the bullet corresponds to the peak in density . the sub - cluster is moving outward from the main cluster core towards positive values of @xmath15 . all the physical quantities are calculated and mass averaged on a one dimensional grid where each grid element has a volume of @xmath103 kpc@xmath104 . @xmath105 is the component of the velocity perpendicular to the shock front and is calculated with respect to the system center of mass . the mach number @xmath106 is determined from the temperature jump which shows a better defined discontinuity compared with the density jump using the rankine - hugoniot conditions . we find values in good agreement with observations @xcite for 1:6v3000 and the cooling run 1:3clfg ( both with @xmath107 ) . 1:6v2000 is characterized by a slightly lower value of @xmath106 ( @xmath108 ) while the high velocity run 1:6 has a stronger shock , with @xmath109 . as seen in the previous section , the pre - shock temperature ( @xmath11011 - 12 kev for the adiabatic runs ) is slightly higher than the one obtained by projection along the line of sight ( fig . [ temppro2 ] ) . assuming @xmath111=11 kev and @xmath112 we predict a pre - shock sound speed @xmath113 km s@xmath8 and a shock velocity @xmath114 km s@xmath8 . this value reduces to the observed shock velocity @xmath115 km s@xmath8 if we use the projected average pre - shock temperature ( @xmath116 kev ) adopted by @xcite . as previously noticed by other authors @xcite the velocity jump shown in the last panel of each plot is much smaller than the theoretically inferred shock velocity . actually the pre - shock gas is not at rest but shows a negative velocity along the @xmath15-axis which however can only be partially explained by the fact that the center of mass of the system is moving in the positive direction of the @xmath15-axis following the bullet . indeed the upstream velocity maintains a negative sign even with respect to the rest frame of the parent cluster , indicating a pre - shock infall towards the bullet . this effect is explained in @xcite by studying the dynamical evolution of the system s global potential , which becomes deeper after the core - core interaction and induces infall of material from the region ahead of the shock . nevertheless the infall velocity which characterizes our models is a significantly smaller than the sub - cluster velocity in contrast to the values found by @xcite and ranges only between 500 and 900 km s@xmath8 in the different runs . = 8truecm = 8truecm figs . [ vel3000 ] and [ vel5000_2000 ] illustrate the characteristic velocities of the sub - cluster in the orbital plane for the three runs 1:6v3000 , 1:6 and 1:6v2000 . all the velocities are calculated in the center of mass rest frame and the time corresponding to the present position is indicated by a vertical line . the velocity of the dark matter component peaks at the moment of closest approach between the two cores and then decreases faster than for a ballistic orbit as a result of dynamical friction . the escape velocity at a given sub - cluster position is calculated assuming a spherical unperturbed host potential and is indicated by a black dotted curve . all the runs have initially unbound sub - clusters . due to the effects of dynamical friction after the phase of core - core interaction the 1:6v2000 sub - cluster is actually bound to the main system , while sub - clusters with initial velocities @xmath50 km s@xmath8 and 5000 km s@xmath8 have velocities slightly or much larger than the escape velocity from the main cluster . the gaseous bullet initially follows its dark matter counterpart but before the point of closest approach it is slowed down by ram - pressure . at the same time the morphology of the sub - cluster gas distribution changes . the contact discontinuity assumes an arc - like shape ( partially reducing the effect of ram - pressure ) and the bow shock forms . it is interesting to notice that after the point of closest approach the evolution of the relative velocity between gas and dark matter in the sub - cluster strongly depends on the intensity of the ram - pressure force . in particular for relatively low ram - pressure values ( 1:6v2000 and 1:6v3000 ) the gaseous bullet is accelerated towards its dark matter counterpart as soon as it leaves the core of the host cluster where it experienced the largest external densities and ram - pressure . as a result the gravitational acceleration relative velocity between the gaseous and dark component of the sub - cluster ( whose two components in the orbital plane are represented for 1:6v3000 by the blue solid and dashed curves in fig . [ vel3000 ] ) is larger than zero . at the time corresponding to the present position the two velocities look comparable in the case of 1:6v3000 while in 1:6v2000 , where the ram - pressure acting on the bullet is lower , the acceleration starts earlier and the gaseous bullet is already @xmath117 km s@xmath8 faster than its dark counterpart . for larger impact velocities as in the case of 1:6 ram - pressure is effective in slowing down the gaseous bullet even at large distances from the center of the main - cluster . the velocity of the gaseous bullet is therefore smaller than that of its dark counterpart until when the sub - cluster is well outside of the virial radius of the main system . the green solid curve in each plot represents the velocity of the front shock obtained by differentiating the positions of the shock front at increasing times . the shock velocity rapidly increases after the core - core interaction and at the present time it is always larger than the velocity of the gaseous bullet . in order to calculate the bullet and shock velocities in the system of reference of the pre - shock gas ( red and green dashed curve , respectively ) the infall velocity of the upstream gas is calculated at different times before and after the present one . the shock velocity @xmath118 at the present position is 4100 , 4500 and 5100 km s@xmath8 in the case of 1:6v2000 , 1:6v3000 and 1:6 , respectively , consistently with the shock velocity @xmath119 inferred from the rankine - hugoniot jump conditions and with the value provided by observations . only 1:6v2000 shows a two - dimensional shock velocity well below the observational uncertainties . as found by @xcite , after the point of closest interaction the shock velocity is always larger than the velocity of the sub - cluster mass centroid , but the amount of the difference strongly depends on the model . in particular , in the case of the two low velocity runs 1:6v2000 and 1:6v3000 @xmath118 is @xmath120 and @xmath121 larger than the velocity of the dark matter component , while in 1:6 the difference is almost negligible ( only @xmath122 ) . + the cluster collision produces a drastic increase in luminosity and temperature . [ lvst ] and [ tvst ] show the bolometric x - ray luminosity @xmath123 and average spectroscopic - like temperature @xmath83 as functions of time . both quantities are calculated for the entire simulated box and scaled to their initial values . = 8truecm = 8truecm the first phase of the interaction which involves only the external low density regions of the two clusters is characterized by an identical slow increment of luminosity and temperature in all the adiabatic runs . the jump is associated with the phase of core - core interaction and peaks right after the time of closest approach . the present time is indicated with an empty squared and sits on the downturning curve . the high velocity run 1:6 is associated with the largest increase in temperature ( @xmath124 3.5 ) and with the smallest jump in luminosity ( @xmath1254 ) . for decreasing sub - cluster velocities the peak in temperature becomes slightly less pronounced while the luminosity jump rises by a factor of 1.5 . the loss of baryonic material from the sub - cluster within the high density core of the main system is indeed larger for low velocity encounters and leads to higher luminosities even if the increment in temperature is smaller with respect to the high velocity runs . excluding the bound run 1:6v2000 , the amount of gas stripped from the bullet and lying within the virial radius of the main cluster is @xmath126 in 1:6 and @xmath127 in 1:6v3000 , which in both cases corresponds to almost @xmath128 of the initial baryonic content of the sub - cluster . the difference becomes more pronounced if we consider only the core ( @xmath129 ) of the main cluster , where the mass of sub - cluster gas is @xmath130 in the 1:6v3000 run and one order of magnitude less in the case of 1:6 . at 4 gyrs after the beginning of the simulation the total luminosity of the system is similar to the initial one in the case of the bound system 1:6v2000 where the center of mass of the bullet does not move out to distances beyond the virial radius of the main system . the luminosity drops to @xmath131 or even less of the initial luminosity for the runs 1:6v3000 and 1:6 respectively . this decrease in luminosity is motivated by the fact that at the final stages of the simulations a significant fraction of the gas is unbound and very extended . due to its low density it does not contribute to the luminosity of the system despite the high temperature . in particular , for the same intracluster distance , the high velocity encounter 1:6 is associated to the highest fraction of unbound material , as will be shown later in this section . on the other hand , the final temperature of the system is higher than the one associated to the initially isolated clusters and converges to a value @xmath132 almost independently of the sub - cluster velocity . the cooling run 1:3clfg shows a somehow different behaviour : both , luminosity and temperature profiles have an initial decrement due to the cooling of the central regions of the two approaching clusters . already during the early phases of the interaction the cooling run moves out of the equilibrium . the peaks in luminosity and temperature are much smaller than the corresponding adiabatic ones ( not illustrated ) . the final luminosity approaches zero . = 8truecm = 15truecm @xcite found that the bullet cluster lies exactly on the @xmath133 relation for nearby clusters @xcite but its temperature is much higher than the one expected according to weak lensing mass estimates . [ tl ] illustrates the drift of the simulated systems along the @xmath133 diagram . the position of 1e0657 - 560 is indicated by a star . @xmath123 and @xmath83 are calculated in a cylindrical region centered on the center of mass of the main cluster and radius equal to its initial virial radius . the luminosity of each model is normalized in such a way that the initial main cluster lies on the @xmath133 relation for local clusters despite of the different initial gas fractions . the starting time of the simulations is indicated with a black solid square . all the adiabatic runs present a similar evolution and move roughly parallel to the @xmath133 relation , as previously noticed by @xcite for major mergers in cosmological simulations . during the early stages of the encounter the cluster moves along a curve which is flatter than the observed @xmath133 relation : the compression of the low density gas at the outskirts of the cluster produces an increase in temperature which is only marginally accompanied by a luminosity growth . the time when the core of the sub - cluster enters the virial radius of the main system represents an inversion point in the diagram : despite the formation of a bow - shock the temperature decreases due to the expansion of the main - cluster gas and the presence of low temperature baryons belonging to the sub - cluster within the virial radius of the main system . at the same time the luminosity rises as a result of the shock and the cluster moves perpendicularly towards the @xmath133 relation . this phase is actually quite short ( on average @xmath134 gyr ) but it characterizes all the adiabatic runs . later on the cluster moves almost parallel to the @xmath133 relation toward larger values of @xmath83 , with the peak in temperature being reached at the point of closest approach . most of the runs show a small delay ( @xmath135 myr ) between the time characterized by the highest temperature and the time with highest luminosity with the curve making a knot in the diagram . the branch of the curve associated with the post core - core interaction is parallel to the increasing one but shifted to smaller luminosities : during the strongest phase of the interaction some of the hot material is lost beyond the virial radius of the main cluster and indeed the largest shift in luminosity is observable in the high velocity run 1:6 . both , luminosity and temperature now decrease until they reach a second inversion point in the curve ( in the case of 1:6 it is only a change in slope ) , associated with the egress of the bullet from the virial radius of the main system . the cooling run 1:3clfg is characterized by a first decrease in temperature which corresponds to the initial phase of thermal instability and central cooling of the main cluster . later on it moves in the @xmath133 diagram similarly to the adiabatic runs although the peaks in luminosity and temperature are much less prominent . nevertheless in a pure cooling model the main cluster does not return to a state of thermal equilibrium at the end of the interaction since nothing prevents the central regions from cooling and the system moves toward extremely low values of luminosity and temperature . [ profiles ] illustrates the final structure of the remnants . in the case of the two highest velocities the final time is chosen in such a way that the distance between the centers of the two systems is about @xmath136 kpc larger than the sum of the virial radii at @xmath137 . this occurs at @xmath138 and @xmath139 gyr in the case of 1:6 and 1:6v3000 , respectively . the low velocity run 1:6v2000 is analyzed at the time corresponding to the first apocenter when the core of the sub - cluster is close to the virial radius of the main cluster . the collisionless component of the main cluster ( top left panel ) in all cases is not substantially affected by the interaction while the sub - cluster dark matter halo appears to be strongly perturbed and retains the original spherical symmetry only within its scale radius . the sub - cluster central density profile ( top right ) is vertically shifted downward without a significant change of slope while for @xmath140 the loss of material becomes more significant and in the case of low velocity encounters ( @xmath141 and @xmath142 km s@xmath8 ) with mass ratio 1:6 the profile shows a large jump of more than one order of magnitude between @xmath143 and @xmath144 . beyond the scale radius the isodensity contours appear elongated and show a large plateau ( see also fig . [ future ] ) associated to tidally stripped material . the evolution of gas density profiles is represented by the two panels on the bottom row of fig . [ profiles ] . in the 1:6 runs the interaction affects the central slope of the main cluster which becomes shallower while the 1:3 encounter produces the largest deviation from the initial values , with the final density profile shifted down by @xmath145 . as mentioned earlier , a not negligible part of the gas within the virial radius of the main cluster originally belonged to the sub - cluster and was subsequently stripped by ram - pressure during the central phases of the interaction . the density profiles of the stripped gas are drawn in blue for the different runs . in the case of 1:6v2000 only the gas outside the virial radius of the sub - cluster is considered . both 1:6v3000 and 1:6v2000 are characterized by a large fraction of sub - cluster gas lost to the core of the main system , with a flat ( 1:6v2000 ) or even positive ( 1:6v3000 ) central slope , while the high velocity run 1:6 , despite the larger ram - pressure values , has less time to deposit gas in the central regions and shows a clear cut - off for @xmath146 . at larger radii , baryonic material is still accreting onto the remnant . indeed , part of the main cluster gas has been pushed out by the bow shock and is now falling back into the cluster potential together with a fraction of the material lost by the sub - cluster . curves in the bottom right panel of fig . [ profiles ] represent the sub - cluster gas density profiles . the low velocity sub - clusters 1:6v3000 and 1:6v2000 retain less than one tenth of their initial gas within @xmath147 while the density profiles in the external regions drop by a factor @xmath148 . in general , the encounter flattens the gas density profile of the bullet core and this effect is evident in the cooling simulation as well . the 1:6 bullet at @xmath138 gyr has lost almost all the gas within @xmath38 . half of this baryonic material was stripped from its dark halo by ram - pressure during the central phases of the interaction and now accelerates towards its dark matter counterpart . it will be accreted by the sub - cluster halo at later times . [ profiles ] compares dark matter and gas density profiles at times when the main and sub - cluster are close to a state of virial equilibrium . despite the changes observable in the gas density profiles and the tidal stripping affecting the bullet beyond its core radius , the central slope of the collisionless component seems to be not strongly perturbed by the interaction @xcite the situation changes comparing the density profiles of the initial systems with those of the main and sub - cluster at the present time , when the bullet is located at almost one third of the host virial radius . as illustrated in the two upper panels of fig . [ profiles2 ] , the density profile of the interacting dark matter halos increases in the inner regions and shows a decrement beyond @xmath149 . this behavior is similar for different models and therefore independent of the orbital details , of the mass ratios and is seen for both , the main and the sub - cluster . the two - body relaxation time scale beyond the softening radius is longer than @xmath150 gyr @xcite , thus implying that dark matter density profiles are not affected by numerical relaxation . as noticed in section 3 , fitting the projected masses around the centers of main and sub - cluster at the present time we would get more concentrated or more massive systems with respect to the real ones . the last two panels of fig . [ profiles2 ] represent gas density profiles at the present time . the loss of gas in the central regions of the sub - cluster is expected as a result of ram - pressure stripping . but interestingly also the inner 200 kpc of the main cluster are completely devoid of gas originally belonging to the main cluster . later on , after the displacement of the main cluster gaseous core , the baryonic material stripped from the bullet replenishes the central regions of the host system . as soon as the bullet moves toward the outskirts of the main system , the main cluster gas has time to collapse again into the center of the potential ( bottom left panel of fig . [ profiles ] ) . + according to @xcite the probability to find a cosmological configuration where the most massive sub - halo has a velocity larger than @xmath151 is well fitted by the function : @xmath152 where @xmath153 is the virial velocity of the main cluster . adopting for @xmath151 the present time velocity of the sub - cluster dark matter component ( in table [ runs ] ) we find a probability of @xmath154 , @xmath155 and @xmath156 for runs 1:6 , 1:6v3000 and 1:6v2000 respectively . we used high resolution n - body / sph simulations to perform an extensive parameter study of the `` bullet '' cluster system 1e0657 - 56 . collisions of two nfw halos with hot isothermal gas components in hydrostatic equilibrium were studied , adopting initial relative velocities of 5000 , 3000 and 2000 km s@xmath8 , which is 3.8 , 2.3 and 1.5 times the main cluster s virial velocity , respectively . we varied masses , mass ratios , impact parameter , baryonic fraction and concentrations . most of the runs are adiabatic . radiative cooling is included in two cases , adopting a standard cooling function for primordial gas . we analyzed the projected properties of the system at a time where the distance between the centers of the mass distribution associated with the main and sub - cluster is comparable to the value provided by lensing observations . we have shown that : * most of the main features in the observed x - ray maps are not well reproduced by encounters with zero impact parameter . indeed , depending on the relative velocity of the bullet , a perfectly head - on encounter either destroys the x - ray peak associated with the main cluster or does not produce a significant displacement between gas and dark matter in the main system . moreover , in a pure head - on collision , the gaseous core of the main cluster is displaced along the line which connects the centers of the two total mass distributions , contrary to observations . an impact parameter corresponding to the core radius of the main cluster gas distribution provides enough ram - pressure to produce a displacement comparable to observations and introduces asymmetries in the main cluster emissivity map , similar to those detected in x - ray . * a low concentrated ( @xmath17 ) main cluster does not survive the collision and its x - ray emissivity peak is destroyed . * encounters with mass ratios as large as 1:3 do not match the observed x - ray morphology and the size of the projected temperature jump across the shock discontinuity . introducing cooling in the simulations , the temperature peak is cooled to a value comparable with the observed one . on the other hand , lensing mass reconstructions seem to suggest even larger mass ratios . this remains an open question . gravitational lensing analysis of the simulations presented in this paper will be part of a follow - up paper ( mastropietro , maccio & burkert in prep ) . * a pure cooling model neglecting energetic feedback still does not give a realistic description of the interacting system . what we are witnessing is overcooling in the central regions following a thermal instability in the early phases of the interaction . nevertheless , cooling simulations provide a significant reduction of the temperature peak across the shock indicating that a more realistic treatment of the gas physics make models with higher mass ratios more realistic . * the choice of a different gas fraction does not affect significantly the results . * the morphology of x - ray maps is best simulated by adiabatic 1:6 encounters . in particular the run 1:6v3000 , with initial velocity @xmath46 km s@xmath8 ( @xmath157 km s@xmath8 in the center of mass system of reference ) and present time dark matter velocity @xmath158 km s@xmath8 ( again in the center of mass rest - frame ) , reproduces most of the main x - ray features : the peculiar morphology of the x - ray emission associated with the main cluster , the relative surface brightness between the main and the sub - cluster , the shape of the bow shock and of the contact discontinuity . decreasing the relative velocity ( 1:6v2000 has a present time dark matter velocity which is almost 300 km s@xmath8 smaller than that used by springel & farrar 2007 ) the bullet becomes much less bright with respect to the center of the host system in disagreement with observations , although the displacement in the two clusters is better reproduced . the high velocity run 1:6 produces a contact discontinuity much broader than observed . * a significant fraction of the x - ray emission next to the center of the main cluster is associated with gas stripped from the external regions of the bullet . in the 1:6 runs the main cluster x - ray peak presents two distinct components : a compact strongly emitting region associated with the displaced core of the host system and a more diffuse component spatially coincident with gas stripped from the sub - cluster during the central phases of the interaction . the relative luminosity of this secondary component increases with decreasing bullet velocities . * the observed difference in line of sight velocity is compatible with an inclination of about 10 degrees of the plane of the encounter . a bottom - down inclination of our simulations keeps both the clusters close to the plane of the sky and does not produce essential changes in the the x - ray maps . but it affects the morphology of the main cluster peak and the luminosity of the secondary peak with respect to that associated with the main cluster core . * the projected temperature jump across the shock discontinuity gives important indications about the nature of the encounter . indeed , both , the height and the thickness of the peak change , with the peak becoming broader for decreasing bullet velocities due to the lower pressure , exercised by the pre - shock gas after the central phases of the encounter . assuming an adiabatic equation of state , both runs 1:6v3000 and 1:6v2000 , match quite well the observed temperature jump , while for larger velocities ( 1:6 ) the peak becomes too narrow and pronounced compared to observations . * temperature maps reveal some interesting features . in particular , as already found by @xcite , the bullet remains relatively cold despite of its dominant x - ray emission . for a short time , a high temperature region appears next to the center of the main cluster mass distribution immediately after the central phases of the interaction . this feature still visible at the present time is not spatially coincident with the main cluster gaseous core . it is associated with hot gas stripped from the sub - cluster within the core of the host system . its location partially corresponds to the x - ray diffuse peak , while its eastern component could be related to the high temperature region observed by @xcite southeast of the main - cluster peak . for a selected sub - sample of runs we performed a detailed three dimensional analysis following their past , present and future evolution for four gyr . * only the low velocity bullet 1:6v2000 is actually bound to the main cluster at the end of the simulation while in 1:6v3000 and 1:6 the sub - cluster has a velocity larger than the escape velocity from the host system . * the velocity of the gaseous component of the bullet starts diverging from that of its dark matter counterpart before the point of closest approach between the two clusters , when the gas is slowed down by ram - pressure . as predicted by @xcite , due to the drop in ram - pressure after the central phases of the interaction and gravitational acceleration by its dark halo counterpart , at the present time the gaseous bullet is not necessarily slower than the dark matter halo at the present time . we explored this question in greater details and found that the behavior of the sub - cluster gas after the point of closest approach strongly depends on the initial velocity of the bullet . for relatively low velocities ( and ram - pressure values , run 1:6v3000 and 1:6v2000 ) the gaseous bullet is accelerated towards its dark matter counterpart as it leaves the core of the main cluster . at the present time the gaseous bullet moves as fast as its dark matter halo in 1:6v3000 and even 500 km s@xmath8 faster in 1:6v2000 . for larger encounter velocities ( 1:6 ) ram - pressure is more effective in slowing down the bullet even beyond the core radius of host cluster and the gaseous bullet is always slower than dark matter . * it has been already noticed by @xcite and @xcite that the sub - cluster velocities do not coincide with the shock velocity @xmath159 as measured by observers . in the case of the three runs ( 1:6v3000 , 1:6v2000 and 1:6clfg ) which best reproduce the projected temperature jump , the mach number and shock velocity determined using the rankine - hugoniot conditions across the shock discontinuity show a good agreement with the values provided by @xcite . the two dimensional shock velocity calculated by tracking the shock position as function of time is consistent with these values . we find that , although after the point of closest approach @xmath159 is always larger than the velocity of the sub - cluster dark matter halo , the difference depends on the model and becomes less significant for higher velocities . in particular in the high velocity run 1:6 the present time shock velocity is only @xmath122 higher than the dark matter bullet . * the collision produces a drastic increase in luminosity and temperature . the highest velocity impacts are associated with the largest increase in temperature ( and shocks ) and the smallest peaks in luminosity due to the fact that in high velocity encounters a significant fraction of the baryons stripped from the bullet are deposited at large radii within the main cluster . @xcite found that the bullet cluster lies on the @xmath4 relation for local clusters @xcite at very high temperature . we followed the evolution of the main cluster in the @xmath4 diagram during the different phases of the interaction and find that it moves roughly parallel to the @xmath4 relation for nearby clusters . the maximum temperature is reached at the point of closest approach and the peak in luminosity immediately afterwards , with a small delay which produces a knot in the diagram . the branch of the curve associated with the post core - core interaction is still parallel to the @xmath4 relation but shifted towards smaller luminosities with respect to the early increasing branch , due to the loss of gas beyond the virial radius of the main cluster . * after the encounter , as soon as the bullet is close or beyond the virial radius of its host system , the dark matter density profile of the main cluster does not deviate anymore significantly from the original one , while the gas profile becomes shallower in the central regions , with a large decrement in density observed only in run 1:3clfg . the situation changes drastically if we compare the density profiles of the initial systems with those at the present time , when the bullet is still located well within the virial radius of the main cluster . in this case the interacting systems are not in virial equilibrium and the dark matter densities of both , the main and sub - cluster , increase in the inner regions and show a decrement beyond 0.4 @xmath38 . comparing these present time profiles with nfw halos would give wrong estimates of the halo parameters . at a time corresponding to the present configuration the center of the host system is completely devoid of main cluster gas as a result of the displacement of the main cluster gaseous core by the bullet . on the other hand , the baryons stripped from the bullet have replenished the central regions of the host . this is only a temporary situation however since a few gyrs later the main cluster gas falls back and becomes again the dominant component in the central regions , although a significant fraction of gas stripped from the bullet is still present . the final sub - cluster dark matter density profiles seem significantly affected by the interaction beyond their scale radius , where the isodensity contours are elongated and show a plateau . we would like to thank r. piffaretti , s. borgani , a. biviano , t. naab , a. maccio and m. girardi for useful discussions . we acknowledge t. quinn for support with the tipsy x - ray package and d. clowe for providing the upper panel of fig . the work was partly supported by the dfg sonderforschungsbereich 375 `` astro - teilchenphysik '' . the numerical simulations were performed on a local sgi - altix 3700 bx2 which was partly funded by the cluster of excellence `` origin and structure of the universe '' .	we present high resolution n - body / sph simulations of the interacting cluster 1e0657@xmath056 . the main and the sub - cluster are modeled using extended cuspy @xmath1 dark matter halos and isothermal @xmath2-profiles for the collisional component . the hot gas is initially in hydrostatic equilibrium inside the global potential of the clusters . we investigate the x - ray morphology and derive the most likely impact parameters , mass ratios and initial relative velocities . we find that the observed displacement between the x - ray peaks and the associated mass distribution , the morphology of the bow shock , the surface brightness and projected temperature profiles across the shock discontinuity can be well reproduced by offset 1:6 encounters where the sub - cluster has initial velocity ( in the rest frame of the main cluster ) close to 2 times the virial velocity of the main cluster dark matter halo . a model with the same mass ratio and lower velocity ( @xmath3 times the main cluster virial velocity ) matches quite well most of the observations . however , it does not reproduce the morphology of the main cluster peak . dynamical friction strongly affects the kinematics of the sub - cluster so that the low velocity bullet is actually bound to the main system at the end of the simulation . we find that a relatively high concentration ( c=6 ) of the main cluster dark matter halo is necessary in order to prevent the disruption of the associated x - ray peak . for a selected sub - sample of runs we perform a detailed three dimensional analysis following the past , present and future evolution of the interacting systems . in particular , we investigate the kinematics of the gas and dark matter components as well as the changes in the density profiles and the motion of the system in the @xmath4 diagram . [ firstpage ] methods : n - body simulations galaxies : clusters : individual : 1e0657 - 56 dark matter x - rays : galaxies : clusters
this proceedings contribution is based on @xcite . while an exact analytic formula for gluon production in @xmath2 collisions in the color glass condensate ( cgc ) framework is still not known , we do know that in @xmath3 and @xmath4 collisions gluon production at the level of classical gluon fields and leading-@xmath5 nonlinear quantum evolution is given by the @xmath1-factorization formula @xcite : @xmath6 here @xmath7 is the total rapidity interval of the collision , @xmath8 , boldface variables denote two - component transverse plane vectors @xmath9 , and @xmath10 , @xmath11 are the unintegrated gluon distributions in the proton and the nucleus , respectively , which are defined by @xcite @xmath12 and @xmath13 in the quantity @xmath14 denotes the forward scattering amplitude for a gluon dipole of transverse size @xmath15 with its center located at the impact parameter @xmath16 scattering on a target nucleus with total rapidity interval @xmath17 . @xmath14 can , in general , be found from the jimwlk evolution equation . in the large-@xmath18 limit it is related to the quark dipole forward scattering amplitude on the same nucleus @xmath19 by @xmath20 where @xmath21 can be found from the bk evolution equation . the quantity @xmath22 from is also a gluon dipole amplitude , but taken in a dilute regime , where it is found by solving the linear balitsky - fadin - kuraev - lipatov ( bfkl ) evolution equation . for the gluon production was derived in the fixed coupling approximation . however , the dipole amplitudes @xmath23 , @xmath14 and @xmath22 are now known for the running coupling case due to the completion of the running coupling calculations for the bfkl / bk / jimwlk evolution equations @xcite . using the running - coupling bk ( rcbk ) equation , the calculational framework presented above , when applied to heavy ion collisions by replacing @xmath24 in , and implemented with a careful inclusion of the nuclear geometry fluctuations , led to the prediction made in @xcite of the charged particle multiplicity as a function of the collision centrality for the lhc . this prediction was confirmed by the alice data in @xcite . however , at the time the prediction @xcite was made , the scales of the couplings explicitly shown in eqs . , , and were not known and had to be modeled . our goal here is to fix the scales of those couplings for the lowest - order gluon production . to include running coupling corrections we follow the blm scale - setting procedure @xcite , which is known to be correct at least at the leading order in @xmath25 . one first needs to resum the contribution of all quark bubble corrections giving powers of @xmath26 , with @xmath27 the number of quark flavors and @xmath28 the physical coupling at some arbitrary renormalization scale @xmath29 . we then complete @xmath27 to the full beta - function by replacing @xmath30 in the obtained expression . here @xmath31 is the one - loop qcd beta - function . after this , the powers of @xmath32 should combine into physical running couplings @xmath33 at various momentum scales @xmath34 which would follow from this calculation . we use the @xmath35 renormalization scheme . as was originally argued in @xcite , including running coupling corrections into the diagrams of below assuming that only a gluon can be produced in the final state would leave one factor of the coupling at an arbitrary renormalization scale @xmath29 . following @xcite we rectify the problem by redefining the gluon production cross - section to include production of collinear gluon gluon and quark anti - quark pairs with the invariant mass lower than some collinear ir cutoff @xmath36 . this new observable is completely @xmath29-independent and expressible in terms of the running coupling constants . gluon production at the lowest order in the coupling is shown in in the @xmath37 light - cone gauge . at this order the unintegrated gluon distribution is @xmath38 such that reduces to @xmath39 our goal is to set the scales for the three couplings in . light - cone gauge . , width=566 ] the final result for the lowest - order gluon production cross section with the running coupling corrections included is @xcite @xmath40 with the momentum scale @xmath34 being a complicated function of @xmath41 and @xmath42 given in @xcite . an interesting feature is that the scale @xmath43 is complex - valued ! the cross - section is , of course , real , as it contains a complex - valued coupling constant multiplied by its conjugate , @xmath44 . clearly looks like the fixed - coupling cross - section with three factors of fixed - coupling replaced by the seven running couplings : we chose to refer to this structure as _ the septumvirate _ of couplings @xcite . an important feature of the scale @xmath34 is that in the @xmath45 ( or @xmath46 ) limit @xmath47 reducing to @xmath48 using the lowest - order expression we would like to conjecture the following running - coupling generalization of : @xmath49 with the new ( rescaled ) distribution functions defined by ( cf . eqs . and ) @xmath50 and @xmath51 one may use this ansatz , along with @xmath52 obtained from the rcbk evolution to further improve the existing cgc phenomenology for particle multiplicity along with its centrality and rapidity dependence at rhic and lhc . this research is sponsored in part by the u.s . department of energy under grant no . de - sc0004286 . 1 w. a. horowitz and yuri v. kovchegov . . , a849:7297 , 2011 . yuri v. kovchegov and kirill tuchin . inclusive gluon production in dis at high parton density . , d65:074026 , 2002 . m. a. braun . . , b483:105114 , 2000 . y. kovchegov and h. weigert . . , 784:188226 , 2007 . i. i. balitsky . . , 75:014001 , 2007 . javier l. albacete and adrian dumitru . . k. aamodt et al . . , 106:032301 , 2011 . stanley j. brodsky , g. peter lepage , and paul b. mackenzie . on the elimination of scale ambiguities in perturbative quantum chromodynamics . , d28:228 , 1983 . yuri v. kovchegov and heribert weigert . . , a807:158189 , 2008 .	we calculate running coupling corrections for the lowest - order gluon production cross section in high energy hadronic and nuclear scattering using the blm scale - setting prescription . at leading order there are three powers of fixed coupling ; in our final answer , these three couplings are replaced by seven factors of running coupling : five in the numerator and two in the denominator , forming a ` septumvirate ' of running couplings , analogous to the ` triumvirate ' of running couplings found earlier for the small-@xmath0 bfkl / bk / jimwlk evolution equations . it is interesting to note that the two running couplings in the denominator of the ` septumvirate ' run with complex - valued momentum scales , which are complex conjugates of each other , such that the production cross section is indeed real . we use our lowest - order result to conjecture how running coupling corrections may enter the full fixed - coupling @xmath1-factorization formula for gluon production which includes non - linear small-@xmath0 evolution .
in this section we discuss experimental and theoretical issues about strain in manganite films . in most cases tensile strain suppresses ferromagnetism and , consequently , critical temperature @xmath13 . on the contrary , compressive strain should reduce the resistivity and shifts @xmath13 toward higher temperatures with respect to the transition temperature of the bulk . the observed strain effect is usually interpreted within the double - exchange model , since the hopping matrix element @xmath20 can be altered by epitaxial strain through the change of the @xmath21 bond length @xmath22 and the bond angle @xmath12 . however , recent studies show that compressive strain does not always lead to enhancement of @xmath13 , @xcite and some anomalous results have been also reported for tensile strain . @xcite in order to explain unusual results in manganite films , an extra mechanism mediated by the orbital state has been proposed . @xcite furthermore the theoretical work concerning manganite films has distinguished between uniform bulk strain and biaxial strain effects on the curie point . @xcite the biaxial strain increases the jahn - teller splitting and favors electron localization in @xmath10 levels , causing @xmath13 to decrease . the agreement with experimental data is in sign and order of magnitude pointing out that a jahn - teller coupling is a crucial variable . @xcite however , this model does not account for the observed decrease of @xmath13 upon thickness reduction mainly in films of @xmath19 on @xmath23 ( @xmath24 ) and @xmath25 ( @xmath26 ) . @xcite when the properties of epitaxial films are studied as a function of thickness , different and complex features are usually observed . first , a reduction of the curie point @xmath13 and the @xmath6 transition temperature @xmath27 of @xmath17 and @xmath19 films has been reported when thickness decreases . this occurs both for films with a gradually relaxed structure @xcite and for fully strained films @xcite showing that the strain can not be the only factor responsible for the reduction of @xmath13 in very thin films . second , when thickness is reduced , the resistivity increases and the magnetic moment often decreases . this has been usually interpreted as due to the presence of a dead layer located at interfaces . @xcite the thickness of these dead layers is of the order of a few @xmath28 depending on the substrate . the decrease of the curie temperature and the increase of resistivity are gradual suggesting that regions with higher resistivity could be present at distances from the interface larger than size of the dead layer . third , changes in the phase coexistence linked to the thickness dependence can lead to charge trapping in thin films . @xcite thus the thickness dependence and the strain effect are far from being fully understood and challenging . the adopted model takes into account the double - exchange mechanism , the coupling of the @xmath10 electrons to lattice distortions and the super - exchange interaction between neighboring localized @xmath9 electrons . @xcite within the single orbital approximation ( reasonable in the doping regime where @xmath5 occurs ) , the hamiltonian reads @xmath29 the quantities in eq.([1r ] ) have been discussed in previous papers . @xcite in films the electron transfer element is assumed to be depending on the direction of hopping and open boundary conditions are imposed along the @xmath30 axis assumed as the out - of - plane axis of growth . as in the bulk , the dimensionless parameter @xmath31 is introduced to measure the strength of the @xmath14 interaction in the adiabatic regime . two canonical transformations are performed in order to treat the @xmath14 interaction variationally . then the bogoliubov inequality is employed in order to derive the variational free energy of the system using a test hamiltonian characterized by free electron , phonon and spin degrees of freedom . @xcite the electron free energy per site reads @xmath32+\mu \rho , \label{fro}\ ] ] where @xmath33 is the temperature , @xmath34 the inverse of @xmath33 , @xmath35 the number of sites along the @xmath36 axis , and @xmath37 the electron density . in eq.([fro ] ) we have @xmath38 , where @xmath39 is the renormalized electronic band . the band dispersion @xmath40 is @xmath41 -2 t_{eff}^z cos(k_z ) , \label{11ur}\ ] ] where the effective transfer integrals @xmath42 and @xmath43 take into account the polaronic and double - exchange effects , and the quantity @xmath44 measures the electronic band shift due to the @xmath14 interaction . due to the finite size along the @xmath30 axis , the values of @xmath45 are given by @xmath46 with @xmath47 . the thermodynamic limit is performed along @xmath48 and @xmath49 directions , using a constant electronic density of states @xmath50 , that represents a simple approximate expression for the exact density of states used for a @xmath51 lattice . within the variational approach , in a way analogous to the bulk case , the linear response to an external electromagnetic field of frequency @xmath52 and , consequently , the transport properties in the limit @xmath53 can be calculated . @xcite the focus is on the real part of the conductivity tensor @xmath54 given by @xmath55 where the @xmath56 is the retarded current - current correlation function . it can be deduced making the analytic continuation of the correlation function defined in matsubara frequencies as @xmath57 where the current operator @xmath58 suitable for the film geometry is @xmath59 with @xmath60 electron charge . we perform the two canonical transformations and make the decoupling of the correlation function in electron , phonon and spin terms through the introduction of the test mean - field hamiltonian . in manganite films the electron correlation function is @xmath61 where the function @xmath62 is @xmath63 with @xmath64 given by @xmath65/sin(k_z ) } } , \ ] ] and @xmath66 free electron green s function . to derive the optical properties , second order fluctuations on the mean - field approach are fundamental since they introduce scattering between charge carriers . @xcite the effect of the damping can enter the calculation substituting @xmath67 for @xmath68 that in matsubara frequencies is expressed as @xmath69 where the spectral function @xmath70 is assumed to be @xmath71 ^ 2/4 + ( \omega - \xi_{{\bf k } } ) ^2 } , \label{43r}\ ] ] with @xmath72 rate of the scattering with optical phonons and spin fluctuations evaluated on the energy shell . @xcite the conductivity tensor @xmath54 is deduced following the lines of our previous works . @xcite in the film geometry this tensor is diagonal with @xmath73 , thus , in the limit @xmath53 , two different resistivities are obtained : in - plane resistivity @xmath74 and out - of - plane resistivity @xmath75 . in this section we analyze the effects of the thickness on the phase diagram , the magnetization and the transport properties of the film . for the moment we do not introduce hopping anisotropies being @xmath76 , with @xmath77 . first the role of the size is investigated in the regime of weak @xmath14 coupling appropriate for @xmath17 films . in fig . 1 the phase diagram of the film is reported for the bulk ( solid line ) , for a film made of 10 planes ( dashed line ) and 5 planes ( dotted line ) . for these values of the parameters the system exhibits a continuous transition from a double - exchange ferromagnet to a metallic paramagnet . a decrease of size leads to a reduction of the ferromagnetic phase that seems to be a generic feature of magnetic films . @xcite actually the critical temperature shifts to lower temperatures than that of the bulk when the spin - spin correlation exceeds the film thickness . @xcite in the inset of fig.1 the variation of the curie temperature as function of the number of the planes is reported . the critical temperature shows a large decrease as the number of planes is reduced . this effect is very pronounced when the number of planes is smaller than @xmath78 : in fact in this case there is a shift of the critical temperature larger than @xmath79 of the corresponding temperature of the bulk . we have investigated the thickness dependence also for a manganite film in the regime of intermediate @xmath14 coupling and bandwidth suitable for @xmath19 films . in this case near the @xmath6 transition the system segregates in ferromagnetic and antiferromagnetic or paramagnetic domains of itinerant and localized carriers , respectively . @xcite in fig . 2 the phase diagram for these systems is shown in correspondence with different sizes of the system . we find that the reduction of the ferromagnetism does not depend on the order of the transition . although the transition lines are different from the phase diagram of the film in the weak - coupling regime , the decrease of the ferromagnetic phase is of the same order . this is confirmed also by the behavior of the critical temperature as a function of the number of planes , as reported in the inset of fig . even if the transition temperatures are lower , the shape of the curve bears a strong similarity with the corresponding plot for systems in weak - coupling regime : in both cases the shift of curie temperature @xmath80 follows a power law @xmath81 . @xcite we note that , in manganite films grown without detectable lattice defects , the decrease of the critical temperature occurs for a thickness smaller than @xmath82 . @xcite therefore for thin films characterized by thickness larger than the possible dead layer , the size effect can be relevant in the interpretation of experimental behaviors allowing to explain the gradual decrease of the critical temperature with decreasing film size . actually , the tendency towards the localization in the thinnest films can not be entirely ascribed to the presence of a dead layer . due to the thickness effect , the ground state presents a weaker order of spins , that can be easier destroyed with increasing temperature . indeed , as the size of the film is reduced , the magnetization assumes smaller values at low temperatures . @xcite in a double - exchange model a weakening of the ferromagnetism is expected to reduce the mobility of the carriers . this is confirmed by the calculated in - plane resistivity @xmath74 that , as reported in fig . 3 , at @xmath83 , exhibits a significant increase as the number of planes decreases ( the out - of - plane resistivity @xmath75 shows a similar behavior ) . like the critical temperature , large variations occur when the film is very thin suggesting that size effect can be important to explain the gradual increase of the resistivity as the thickness decreases . however , the behavior of resistivity with respect to the number of planes is different as the temperature increases . in fact , as shown in the inset of fig . 4 , at higher temperatures the resistivity of the film is smaller for thinner films . this can be understood considering that the differences in the spin order due to size effects diminish with increasing temperature . so the bare behavior of the scattering rate with optical phonons emerges : this type of scattering decreases when the size of the system is reduced in agreement with what happens in quantum wells . @xcite at high temperatures , in addition to size effect , other effects such as the strain have to be considered in order to get a better agreement with experimental data . in this section we deal with strain effects on the properties of the film . the strain in the film is introduced through the modifications of in - plane and out - of - plane hoppings , that are linked to lattice parameters measured in these systems . it can be assumed @xmath84 , where @xmath85 is the in - plane bond length and @xmath86 , with @xmath87 in - plane bond angle . @xcite in an analogous manner , we can have @xmath88 , where @xmath89 is the out - of - plane bond length and @xmath90 , with @xmath91 out - of - plane bond angle . if one neglects the angular dependence , the ratio of the hopping amplitudes of the film with respect to the bulk values are given by @xmath92 and @xmath93 , where @xmath94 is the lattice parameter of bulk , @xmath95 and @xmath96 are in - plane and out - of - plane lattice parameters of the film measured by x - ray diffraction experiments . @xcite these ratios simulate the strain of the film within the variational approach , so its effects on the phase diagram and the transport properties can be estimated . the variations of the phase diagram are controlled mainly by the changes of in - plane hopping , so that compressive strain ( @xmath97 ) leads to an enhancement of the critical temperature , while tensile strain ( @xmath98 ) to a reduction . the increase of the compressive ( tensile ) strain is able to perturb the system also at low temperatures where the ferromagnetic alignment of the spins is reinforced ( reduced ) . deriving in - plane and out - of - plane hoppings from the lattice parameters of a compressively strained @xmath17 film,@xcite it is found that the strain introduces anisotropies in the transport properties of the films in the weak coupling regime . @xcite indeed , as shown in fig . 4 , the in - plane resistivity @xmath74 decreases in temperature with increasing compressive strain , while the out - of - plane resistivity @xmath75 increases . this is explained by the fact that , increasing the strain , the in - plane hopping is enhanced , while the out - of - plane hopping is reduced . in any case , at low temperatures , due to the enhancement of the ferromagnetic order induced by the compressive strain , both the resistivities decrease with increasing strain . next we have investigated the effects of compressive and tensile strain on the phase diagram of a manganite film in the intermediate coupling regime . the in - plane and out - of - plane hopping amplitudes have been deduced considering the lattice parameters of @xmath19 films grown on different substrates . @xcite the film on @xmath26 corresponds to the ratio @xmath99 , that on @xmath100 to @xmath101 , finally that grown on @xmath24 to @xmath102 . as shown in fig . 5 , in the first case , the compressive strain ( dotted line with diamonds ) leads to an enhancement of the critical temperature with respect to the bulk value ( solid line with circles ) unlike experimental results . @xcite in the second case , even if field strain is near the unity ( dash - dotted line with down triangles ) , the critical temperature shifts at lower temperatures in agreement with experimental data . actually in this case both the lattice parameters exceed the bulk value , determining a reduction of the in - plane hopping . in the third case , for tensile strain ( dashed line with squares ) , the shift of the curie temperature in comparison with the bulk value is consistent with experimental measurements . @xcite we stress that the results discussed above have been obtained neglecting the dependence of hopping amplitudes on the @xmath1 bond angle @xmath12 . however , in bulks of @xmath0 perovskites , it has been shown that changes in @xmath2 site ionic size affect only @xmath12 , while the bond length @xmath22 seems to be unchanged ( @xmath103 ) suggesting as a maximum lattice parameter @xmath104 . @xcite recent measurements in @xmath19 thin films on @xmath26 and @xmath24 substrates @xcite have showed that @xmath85 is fixed to a value that does not depend on the type of strain , while the in - plane bond angle is changed . the angle @xmath87 becomes larger ( smaller ) under tensile ( compressive ) strain , showing consistency with the elongation ( contraction ) of @xmath95 . similar results have been observed also in @xmath17 films on @xmath26 . @xcite hence , it is reasonable to assume that values of the lattice parameter larger than those of the bulk imply stretching of @xmath22 ( @xmath12 near @xmath105 ) and smaller lattice parameters induce a contraction of @xmath12 ( @xmath22 near @xmath106 ) . thus , for @xmath19 films grown on @xmath26 the angular dependence is not negligible . @xcite actually , deriving @xmath87 from experimental data @xcite ( @xmath107 ) and considering only the changes of in - plane parameters ( that are the most important ) , we can deduce the variation of the in - plane hopping from the knowledge of the @xmath1 angle of the bulk @xcite ( @xmath108 ) . as reported in fig . 5 ( dash - dotted line with up triangles ) , even if the strain is compressive , the critical temperature is reduced compared with bulk value in agreement with experimental data . in this section we focus our attention on the variations of the @xmath14 coupling linked to the presence of strain that can be included in order to improve the description of manganite films . both uniform bulk strain and biaxial strain can affect the @xmath14 coupling strength of the carriers in manganite films . @xcite while the biaxial distortions tend to increase the localization of the carriers , the bulk strain can lead to an increase or a decrease of @xmath14 coupling depending on the sign of the strain . actually , a uniform compression tends to increase the electron hopping amplitudes reducing the importance of the @xmath14 coupling . so , unlike tensile strain , for compressive strain two opposite behaviors influence the variations of the @xmath14 coupling . in any case , for very thin films , the tendency towards localization can be simulated in our scheme by an enhancement of the @xmath14 coupling . for @xmath17 films grown on @xmath24 the size and strain dependence of the transition temperature is strongly influenced by the strain - induced @xmath14 coupling . @xcite actually we have verified that a small reduction ( increase ) in @xmath14 coupling can induce a large enhancement ( decrease ) of the critical temperature in the @xmath5 regime . for fully strained @xmath19 films grown on different substrates the delicate balance of segregating phases is strongly influenced by finite size and strain effects . deducing sizes and strain fields of the @xmath19 films grown on @xmath24 , @xcite in fig . 6 we show the corresponding phase diagrams calculated supposing a likely distribution of the quantity @xmath109 , that represents the ratio between the @xmath14 coupling @xmath110 of the film and that of the thickest film . the concomitant effects induce a strong lowering of the transition temperature and at low @xmath33 an increase in the weight of antiferromagnetic insulating phases . in fact , for a film characterized by a thickness of @xmath111 planes and a ratio @xmath112 , an antiferromagnetic insulating phase of localized charges is stabilized at low temperatures in agreement with experiment . @xcite therefore , the insulating dead layer could originate as natural consequence of reduced size of the film , strain effect and increased @xmath14 coupling . in fig . 7 the magnetization ( upper panel ) and the in - plane resistivity ( lower panel ) are reported as function of the temperature for three different values of the strain and @xmath14 coupling . as result of the competition between ferromagnetic metallic and antiferromagnetic or paramagnetic insulating phases , @xcite the magnetization is strongly reduced , while the resistivity is largely enhanced showing consistency with experiments @xcite . finally we stress that the nearly constant behavior of the resistivity at intermediate temperatures is characteristic of the antiferromagnetic phase of localized carriers . only at higher temperatures , in the paramagnetic phase , the resistivity strongly diminishes with increasing temperature . in this paper we have analyzed the thickness dependence and strain effects in films of @xmath0 perovskites within the @xmath5 regime . a variational approach previously proposed for manganite bulk has been generalized in order to consider the film geometry . a reduction in the thickness of the film causes a decrease of the transition temperature and the magnetization , and increase of the resistivity especially at low temperatures . if the film is very thin but has thickness larger than the possible dead layer , the calculated results show good agreement with experimental behaviors . the strain is associated with the changes of in - plane and out - of - plane electron hopping amplitudes induced by the substrate . the variations of the phase diagram with respect to the bulk are controlled mainly by the changes of in - plane hopping , so that the compressive strain leads to an enhancement of the critical temperature , while tensile strain weakens the ferromagnetic phase . if the dependence of the hopping matrix elements on the @xmath1 bond angle is properly considered , the strain effects on the transition temperature and transport properties are consistent with experimental data . finally it is shown that substrate - induced variations of the @xmath14 coupling , together with finite size and strain effects , can strongly influence the subtle balance of coexisting phases favoring insulating phases when the thickness of the film decreases . although the results presented in this paper are quite satisfactory , we wish to mention other mechanisms that could influence the thermodynamic and transport properties , at least qualitatively . first a strain - induced modification of @xmath10 electron orbital stability can affect the phase diagram of the system . @xcite changes in oxygen content resulting in cationic vacancies due to the annealing may shift the critical temperature at values much higher than any bulk values in the series compounds . @xcite the loss of magnetic moment and the increase of resistivity could be caused by the domain - type disorder . @xcite even if in manganite films spatial inhomogeneities can be intrinsic , @xcite the disorder induced by the substrate could be responsible also for phase separation . @xcite as stressed in the previous section , the tendencies toward phase separation can be important to explain the presence of the insulating dead layer in manganite films . the variational approach used in this paper can be further generalized to consider insulating planes separated through an interface from metallic planes . strain effects and thickness dependence could strongly perturb the size of insulating and metallic planes giving rise to the dead layer . work in this direction is in progress . v. cataudella , g. de filippis , g. iadonisi , c. a. perroni , and f. ventriglia acknowledge financial support by prin `` electronic and optical properties in manganite films '' . f1 phase diagrams corresponding to @xmath113 , @xmath114 and @xmath115 for different numbers of film s planes . @xmath116 means paramagnetic metallic , @xmath117 ferromagnetic metal . in the inset the corresponding variation of the critical temperature @xmath13 as a function of the number of the film s planes at @xmath118 . the temperatures are expressed in units of @xmath119 . f2 phase diagrams corresponding to @xmath120 , @xmath121 and @xmath122 for different numbers of film s planes . @xmath123 stands for paramagnetic insulating . in the inset the resulting variation of the critical temperature @xmath13 as a function of the number of the film s planes at @xmath118 . the temperatures are expressed in units of @xmath119 . f3 the resistivity @xmath74 as a function of the number of planes for @xmath83 at @xmath118 and @xmath124 . in the inset the variation in temperature ( in units of @xmath119 ) of the resistivity @xmath74 for different numbers of planes ( the arrows indicate the transition temperatures ) . the resistivity is in units of @xmath125 , where @xmath96 is the hole concentration and @xmath126 . f6 phase diagrams corresponding to @xmath120 , @xmath129 and @xmath130 for different values of size , strain and @xmath14 coupling . @xmath123 means paramagnetic insulating . the temperatures are expressed in units of @xmath119 . f7 magnetization ( in units of the saturation magnetization ) and resistivity ( in units of @xmath125 , wit @xmath96 hole concentration and @xmath126 ) as a function of the temperature ( in units of @xmath119 ) at @xmath120 and @xmath118 for different values of size , strain and @xmath14 coupling . r. von helmolt , j. wecker , b. holzapfel , l. schultz , and k. samwer , phys . lett . * 71 * , 2331 ( 1993 ) ; s. jin , t. h. tiefel , m. mccormack , r.a . fastnacht , r. ramesh , and l.h . chen , science * 264 * , 413 ( 1994 ) . r. shreekala , m. rajeswari , r. c. srivastava , k. ghosh , a. goyal , v. v. srinivasu , s. e. lofland , s. m. bhagat , m. downes , r. p. sharma , s. b. ogale , r. l. greene , r. ramesh , t. venkatesan , r. a. rao , and c. b. eom , appl . 74 * , 1886 ( 1999 ) . a. biswas , m. rajeswari , r. c. srivastava , y. h. li , t. venkatesan , r. l. greene , and a. j. millis , phys . b * 61 * , 9665 ( 2000 ) ; l. zhang , c. israel , a. biswas , r. l. greene , and a. de lozanne , science * 298 * , 805 ( 2002 ) . m. bibes , s. valencia , ll . balcells , b. martinez , j. fontcuberta , m. wojcik , and s. nadolski , phys . lett . * 87 * , 67210 ( 2001 ) ; m. bibes , s. valencia , ll . balcells , b. martinez , j. fontcuberta , m. wojcik , s. nadolski , and e. jedryka , phys . b * 66 * , 134416 ( 2002 ) .	thickness dependence and strain effects in films of @xmath0 perovskites are analyzed in the colossal magnetoresistance regime . the calculations are based on a generalization of a variational approach previously proposed for the study of manganite bulk . it is found that a reduction in the thickness of the film causes a decrease of critical temperature and magnetization , and an increase of resistivity at low temperatures . the strain is introduced through the modifications of in - plane and out - of - plane electron hopping amplitudes due to substrate - induced distortions of the film unit cell . the strain effects on the transition temperature and transport properties are in good agreement with experimental data only if the dependence of the hopping matrix elements on the @xmath1 bond angle is properly taken into account . finally variations of the electron - phonon coupling linked to the presence of strain turn out important in influencing the balance of coexisting phases in the film . the perovskite oxides @xmath0 ( @xmath2 stands for a divalent alkali element such as @xmath3 or @xmath4 ) have been studied intensively since the discovery of `` colossal '' magnetoresistance ( @xmath5 ) in thin films . @xcite these dramatic changes in electron and magnetic properties are found at temperatures around the combined ferromagnetic - paramagnetic and metallic - insulating ( @xmath6 ) transitions . the ferromagnetic phase was usually explained by introducing the double exchange mechanism , @xcite in which hopping of an outer shell electron from a @xmath7 to a @xmath8 site is favored by a parallel alignment of the core spins . in addition to the double - exchange term that promotes hopping of the carriers , a strong interaction between electrons and lattice distortions has been proposed to play a non negligible role in these compounds @xcite and confirmed by many experimental measurements . @xcite actually for the @xmath7 site , with three electrons in the energetically lower spin triplet state @xmath9 and the mobile electron in the higher doublet @xmath10 , a jahn - teller distortion of the oxygen octahedron can lead to splitting of the doublet and the trapping of the charge carriers in a polaronic state . a second important connection between crystal structure and @xmath6 transition lies in the dependence of the @xmath11 electron transfer on the @xmath1 bond angle @xmath12 , that is on the orientation of the oxygen octahedra with respect to the main crystal axes . this implies a strong effect of either external pressure or mean @xmath2-site ionic radius on the ferromagnetic critical temperature @xmath13 . @xcite finally , due to the complex interplay of electron , orbital , spin and lattice degrees of freedom , strong tendencies toward phase separation are present in these materials . @xcite in recent papers , @xcite some of us have shown on the basis of a variational approach that the interplay of the electron - phonon ( @xmath14 ) interaction and the double- and super - exchange magnetic effects can be important to explain the experimentally observed tendency of manganite bulk to form inhomogeneous magnetic structures near the phase boundaries . employing this scheme , spectral and optical properties in absence and in presence of magnetic field have been derived in the regime of @xmath5 finding good agreement with experimental data . the situation for manganite films is more complicated . maximum @xmath15 values in films are usually larger and at lower temperatures than in equivalent bulk materials . @xcite this enhanced change of the resistivity of the films has immediately suggested device applications based on the sensitivity to magnetic fields . @xcite an essential issue for manganite films is to understand the role of the strain due to lattice mismatch between the substrate and the film . indeed it has been found that properties such as magnetoresistance , magnitude of the temperature @xmath13 , resistivity , magnetization , @xcite transport and magnetic anisotropies , @xcite and spin and orbital order structure @xcite are sensitive to the epitaxial strain . these properties are different from the changes induced by hydrostatic or chemical pressure , since in - plane strain generally leads to an out - of - plane strain of different sign . moreover the effects induced by the substrate are able to influence the tendency toward phase separation , @xcite induce inhomogeneities @xcite in films , and cause new electronic behavior not found in bulk materials of the same composition . @xcite actually , the strain affects so many quantities that it could be used to control the properties of interest by depositing films on various substrates , changing the deposition conditions and the post - annealing procedure , and varying the thickness . @xcite in this paper we extend the variational approach of the bulk in order to deal with manganite films . the size of the film is taken into account considering the system made of a finite number of planes and imposing open boundary conditions along the out - of - plane direction of growth . the thickness dependence has been studied finding similar results in the weak @xmath14 coupling regime appropriate for large bandwidth systems such as @xmath16 ( @xmath17 ) films and in the intermediate regime suitable for @xmath18 ( @xmath19 ) films : reduction of the critical temperature , decrease of the magnetization and increase of the resistivity especially at low temperatures . for thin films characterized by a thickness larger than the possible dead layer , the calculated results are in good agreement with experimental behaviors . the strain in the film is simulated through modifications of in - plane and out - of - plane hoppings , that are related to lattice parameters measured in these systems . it is found that the compressive strain leads to an enhancement of the critical temperature , while tensile strain weakens the ferromagnetic phase . when the dependence on the bond angle @xmath12 is properly taken into account , it is shown that also for compressive strain a reduction of the transition temperature can occur . the strain introduces also anisotropies in the transport properties showing the consistency of the theory with experimental data . finally substrate - induced variations of the @xmath14 coupling are considered pointing out that , together with finite size and strain effects , they can strongly influence the subtle balance of coexisting phases favoring insulating phases when the thickness of the film decreases . in section i experimental and theoretical issues about strain in manganite films are discussed ; in section ii the variational approach and the transport properties of the bulk are generalized to deal with manganite films ; in section iii the thickness dependence is analyzed ; in section iv the strain effects on the phase diagram and transport properties are calculated ; in section v changes of the @xmath14 coupling induced by the substrate are considered .
the sorting of the suffixes of a text plays a fundamental role in text algorithms with several applications in many areas of computer science and bioinformatics . for instance , it is a fundamental step , in implicit or explicit way , for the construction of the suffix array ( @xmath0 ) and the burrows - wheeler transform ( @xmath1 ) . the @xmath0 , introduced in 1990 ( cf . @xcite ) , is a sorted array of all suffixes of a string , where the suffixes are identify by using their positions in the string . several strategies that privilege the efficiency of the running time or the low memory consumption have been widely investigated ( cf . @xcite ) . the @xmath1 , introduced in @xmath2 ( cf . @xcite ) , permutes the letters of a text according to the sorting of its cyclic rotations , making the text more compressible ( cf . @xcite ) . a recent survey on the combinatorial properties that guarantee such a compressibility after the application of @xmath1 can be found in @xcite ( cf . also @xcite ) . moreover , in the last years the @xmath0 and the @xmath1 , besides being important tools in data compression , have found many applications well beyond its original purpose ( cf . @xcite ) . the goal of this paper is to introduce a new strategy for the sorting of the suffixes of a word that opens new scenarios of the computation of the @xmath0 and the @xmath1 . our strategy uses a well known factorization of a word @xmath3 called the _ lyndon factorization _ and is based on a combinatorial property proved in this paper , that allows to sort the suffixes of @xmath3 ( `` global suffixes '' ) by using the sorting of the suffixes inside each block of the decomposition ( `` local suffixes '' ) . the lyndon factorization is based on the fact that any word @xmath3 can be written uniquely as @xmath4 , where * the sequence @xmath5 is non - increasing with respect to lexicographic order ; * each @xmath6 is strictly less than any of its proper cyclic shift ( lyndon words ) . this factorization was introduced in @xcite and a linear time algorithm is due to duval @xcite . the intuition that the knowledge of lyndon factorization of a text can be used for the computation of the suffix array of the text itself has been introduced in @xcite . conversely , a way to find the lyndon factorization from the suffix array can be found in @xcite . if @xmath7 is a factor of a word @xmath3 we say that the sorting of the local suffixes of @xmath7 is _ compatible _ with the sorting of the global suffixes of @xmath3 if the mutual order of two local suffixes in @xmath7 is kept when they are extended as global suffixes . the main theorem in this paper states that if @xmath7 is a concatenation of consecutive lyndon factors , then the local suffixes in @xmath7 are compatible with the global suffixes . this suggests some new algorithmic scenarios for the constructions of the @xmath0 and the @xmath1 . in fact , by performing the lyndon factorization of a word @xmath3 by duval s algorithm , one does not need to get to the end of the whole word in order to start the decomposition into lyndon factors . since our result allow to start the sorting of the local suffixes ( compatible with the sorting of the global suffixes ) as soon as the first lyndon word is discovered , this may suggest an online algorithm , that do not require to read the entire word to start sorting . moreover , the independence of the sorting of the local suffixes inside the different lyndon factors of a text suggests also a possible parallel strategy to sort the global suffixes of the text itself . in section [ sec : prel ] we give the fundamental notions and results concerning combinatorics on words , the lyndon factorization , the burrows - wheeler transform and the suffix array . in section [ sec : method ] we first introduce the notion of global suffix on a text and local suffix inside a factor of the text . then we prove the compatibility between the ordering of local suffixes and the ordering of global suffixes . in section [ sec : algo ] we describe an algorithm that uses the above result to incrementally construct the @xmath1 of a text . such a method can be also used to explicitly construct the @xmath0 of the text . in section [ sec : conclusion ] we discuss about some possible improvements and developments of our method , including implementations in external memory or in place constructions . finally , we compare our strategy for sorting suffixes with the method proposed in @xcite in which a lightweight computation of the @xmath1 of a text is performed by partitioning it into factors having the same length . let @xmath8 be a finite alphabet with @xmath9 . given a finite word @xmath10 , @xmath11 for @xmath12 , a _ factor _ of @xmath3 is written as @xmath13 = a_i \cdots a_j$ ] . a factor @xmath14 $ ] is called a _ prefix _ , while a factor @xmath15 $ ] is called a _ suffix_. in this paper , we also denote by @xmath16 as the suffix of @xmath3 starting from position @xmath17 . we omit @xmath3 when there is no danger of ambiguity . we say that @xmath18 are _ conjugate _ ( or _ cyclic shift _ ) or _ @xmath19 is a conjugate of @xmath20 _ if @xmath21 and @xmath22 for some @xmath23 . recall that conjugacy is an equivalent relation . a _ lyndon _ word is a primitive word which is also the minimum in its conjugacy class , with respect to the lexicographic order relation . in @xcite , one can find a linear algorithm that for any word @xmath24 computes the lyndon word of its conjugacy class . we call it the lyndon word of @xmath3 . lyndon words are involved in a nice and important factorization property of words . @xcite every word @xmath25 has a unique factorization @xmath4 such that @xmath26 is a non - increasing sequence of lyndon words . we call this factorization the _ lyndon factorization _ of a word and it can be computed in linear time ( see for instance @xcite ) . duval in @xcite presents two variants of an algorithm of factorization of a word into lyndon words in time linear in the length of the word . the first variant of the algorithm uses only three variables for a complete computation and it requires no more than @xmath27 comparisons between two letters . the second one is slightly faster in that sense that it requires no more than @xmath28 comparisons but it uses an auxiliary storage of size @xmath29 . the basis idea for both these variants is finding each factor of the decomposition of the word @xmath3 from left to right by eventually reading a long enough prefix of the next lyndon factor . lyndon factorization has been realized also in parallel ( cf . @xcite ) and in external memory ( cf . @xcite ) . one way to define the burrows - wheeler transform ( @xmath1 ) @xcite of a string @xmath3 of length @xmath30 ( although not the most efficient way to compute it ) is to construct all @xmath30 cyclic shifts of @xmath3 and sort them lexicographically . the output of @xmath1 consists of the pair ( @xmath31 , @xmath32 ) , where @xmath31 is the sequence of the last character of each rotation in the sorted list and @xmath32 is an integer denoting the position of the original word in the list . + another more efficient way consists in the concatenating at the the input string @xmath3 a symbol @xmath33 that is smaller than any other letter . in this case , the @xmath1 is intuitively described as follows : given a word @xmath34 , @xmath35 is a word obtained by sorting the list of the suffixes of @xmath36 and by concatenating the symbols preceding in @xmath3 each suffix in the sorted list . in both the cases , it is an invertible transform , i.e. , one can recover the original text from its @xmath1 . note that , in general , the sorting of the conjugates of a word @xmath3 and the sorting of the suffixes of a word @xmath36 is different , but , as consequence of the properties of lyndon words , when the word @xmath3 is the lyndon word , then the two sorting coincide ( cf . * lemma 12 ) ) . a study of the combinatorial aspects that connect these two sorting can be found in @xcite . in this study an important role is played by the notion of lyndon word . given a text @xmath3 of length @xmath30 , the suffix array ( sa ) for @xmath3 is an array of integers of range @xmath37 to @xmath38 specifying the lexicographic ordering of the suffixes of the string @xmath3 . it will be convenient to assume that @xmath39 = \$$ ] , where @xmath33 is smaller than any other letter . that is , @xmath40 = i$ ] if and only if @xmath41 $ ] is the @xmath42-th suffix of @xmath3 in ascending lexicographical order . @xmath43 for instance , if @xmath44 then @xmath45 and @xmath46 $ ] . the table obtained by lexicographically sorting all the suffixes of @xmath36 is depicted in figure [ fig : bwt ] . let @xmath24 and let @xmath47 be its lyndon factorization . for each factor @xmath48 , we denote by @xmath49 and @xmath50 the position of the first and the last character , respectively , of the factor @xmath48 in @xmath3 . let @xmath51 be a factor of @xmath3 . we denote by @xmath52 $ ] and we call it _ local suffix _ at the position @xmath17 with respect to @xmath51 . note that @xmath53 $ ] and we call it _ global suffix _ of @xmath3 at the position @xmath17 . we write @xmath54 instead of @xmath16 when there is no danger of ambiguity . let @xmath3 be a word and let @xmath51 be a factor of @xmath3 . we say that the sorting of suffixes of @xmath51 is _ compatible _ with the sorting of suffixes of @xmath3 if for all @xmath55 with @xmath56 , @xmath57 notice that in general taken an arbitrary factor of a word @xmath3 , the sorting of its suffixes is not compatible with the sorting of the suffixes of @xmath3 . consider for instance the word @xmath58 and its factor @xmath59 . then @xmath60 whereas @xmath61 . [ th : suforder ] let @xmath34 and let @xmath4 be its lyndon factorization . let @xmath62 . then the sorting of the suffixes of @xmath51 is compatible with the sorting of the suffixes of @xmath3 . let @xmath17 and @xmath42 be two indexes with @xmath63 both contained in @xmath51 . we just need to prove that @xmath64 . let @xmath65 $ ] and @xmath66 $ ] . suppose that @xmath67 . then @xmath68 by the definition of lexicographic order . if @xmath69 there is nothing to prove . if @xmath70 , then @xmath71 is prefix of @xmath72 , so by the definition of lexicographic order @xmath73 suppose now that @xmath73 . this means that @xmath68 . if @xmath69 there is nothing to prove . if @xmath70 , the index @xmath74 is in some lyndon factor @xmath75 with @xmath76 , then @xmath77 . we denote @xmath78 $ ] . then @xmath79 , since @xmath80 ( because @xmath75 is a lyndon word ) and @xmath81 ( since the factorization is a sequence of non increasing factors ) . the above theorem states , in other words , that mutual order of the suffixes of @xmath3 starting in two positions @xmath17 and @xmath42 is the same as the mutual order of the `` local '' suffixes starting in @xmath17 and @xmath42 inside the block obtained as concatenation of the consecutive lyndon factors including @xmath17 and @xmath42 . as particular case , the theorem is also true when the two suffixes start in the same lyndon factor . we recall that , if @xmath82 and @xmath83 denote two sorted lists of elements taken from any well ordered set , the operation @xmath84 consists in obtaining the sorted list of elements in @xmath82 and @xmath83 a consequence of previous theorem is stated in the following proposition . let @xmath85 and @xmath86 denote the sorted lists of the suffixes of @xmath87 and the suffixes @xmath88 , respectively . then @xmath89 . this proposition suggests a possible strategy for sorting the list of the suffixes of some word @xmath3 : * find the lyndon decomposition of @xmath3 , @xmath90 ; * find the sorted list of the suffixes of @xmath91 and , separately , the sorted list of the suffixes of @xmath92 ; * merge the sorted lists in order to obtain the sorted lists of the suffixes of @xmath93 ; * find the sorted list of the suffixes of @xmath94 and merge it to the previous sorted list ; * keep on this way until all the lyndon factors are processed ; this kind of strategy could have several advantages : first of all , one can work online , i.e. one can start sorting suffixes as soon as the first lyndon factor is individuated . this also allow to integrate the sorting process with the duval s algorithm for lyndon decomposition that outputs lyndon factors online as well . the second advantage is that this kind of strategy allows parallelization , since every lyndon factor can be processed separately for sorting its suffixes . these kind of application would require an efficient algorithm to perform the merging of two sorted lists . a detailed algorithmic description of this method in order to obtain the @xmath1 of a text is given in next section . in this section we propose an algorithm that incrementally constructs the suffix array @xmath0 and the burrows - wheeler transform @xmath1 of the text @xmath3 by using its lyndon factorization . in particular , here we detail the construction of the @xmath1 but an analogous reasoning can be done in order to obtain the suffix array . we assume that @xmath90 is the lyndon factorization of the word @xmath95 $ ] . so @xmath96 . such an hypothesis , although strong , is not restrictive because one can obtain the lyndon factorization of any word in linear time ( cf . as shown in previous section , the hypothesis that @xmath3 is factorized in lyndon words suggests to connect the problem to the sorting of the local suffixes of @xmath3 to the lexicographic sorting of the global suffixes of @xmath3 . our algorithm , called bwt_lynd , considers the input text @xmath95 $ ] as logically partitioned into @xmath97 blocks , where each block corresponds to a lyndon word , and computes incrementally the @xmath98 via @xmath97 iterations , one per block of @xmath3 . each block is examined from right to left so that at iteration @xmath17 we compute @xmath99 given @xmath100 , @xmath101 and @xmath102 . remark that the positions in @xmath102 range in @xmath103 $ ] . this means that we sum the amount @xmath104 to the values of the usual suffix array of @xmath105 . the key point of the algorithm comes from theorem [ th : suforder ] , because the construction of @xmath99 from @xmath100 requires only the insertion of the characters of @xmath106 in @xmath100 in the same mutual order as they appear in @xmath101 . note that the character @xmath33 that follows @xmath6 is not considered in this operation . moreover , such an operation does not modify the mutual order of the characters already lying in @xmath100 . for each block @xmath6 with @xmath17 ranging from @xmath37 to @xmath97 , the algorithm bwt_lynd executes the following steps : 1 . [ it : bwtsa ] compute the @xmath107 and @xmath108 . [ it : a ] compute the counter array @xmath109 $ ] which stores in @xmath110 $ ] the number of suffixes of the string @xmath111 which are lexicographically smaller than the @xmath42-th suffix of @xmath112 . [ it : merge ] merge @xmath100 and @xmath107 in order to obtain @xmath113 . [ ex : merge ] let @xmath114 . the lyndon factorization of @xmath3 is @xmath115 , where @xmath116 . figure [ fig:1 ] illustrates how step [ it : merge ] of the algorithm works . note that the positions of the suffixes in @xmath117 ( i.e. in @xmath118 ) are shifted of @xmath119 positions . notice that in the algorithm bwt_lynd we do not actually compute the sorted list of suffixes , but we show it in figure [ fig:1 ] to ease the comprehension of the algorithm . moreover , the algorithm can be simply adapt to compute the suffix array of @xmath3 , so in figure [ fig:1 ] the suffix arrays are also shown . @xmath120 step [ it : bwtsa ] can be executed in linear time @xmath121 , if @xmath107 and @xmath108 are stored in internal memory ( see @xcite ) . during step [ it : a ] , the algorithm uses the functions @xmath122 and @xmath123 described as follows . for any character @xmath124 , let @xmath125 denote the number of characters in @xmath51 that are smaller than @xmath20 , and let @xmath126 denote the number of occurrences of @xmath20 in @xmath127 $ ] . such functions have been introduced in @xcite for the fm - index . for sake of simplicity we can firstly construct the array @xmath128 $ ] which stores in @xmath129 $ ] the number of suffixes of the string @xmath111 which are lexicographically smaller than the suffix of @xmath112 starting at the position @xmath42 . remark that we set @xmath130=0 $ ] because @xmath131\$$ ] has the same rank of @xmath33 between the suffixes of @xmath111 and it is preceded by the same symbol @xmath132 in @xmath133 . consequently , in our algorithm considers the suffixes @xmath131\$$ ] and the suffix @xmath33 ( of the string @xmath111 ) as the same suffix . it is easy to prove that the value @xmath134 $ ] is @xmath135 . the array @xmath136 is computed from the position @xmath137 to @xmath138 by using proposition [ prop : arraya ] . [ prop : arraya ] let @xmath42 be a integer ranging from @xmath137 to @xmath138 and let @xmath139 $ ] be the number of suffixes of @xmath111 lexicographically smaller than @xmath140 $ ] . let @xmath141 be the first symbol of the suffix @xmath142 $ ] . then , @xmath143=c(\lbwt(l_1 \cdots l_{i-1}\$),c)+rank(\lbwt(l_1 \cdots l_{i-1}\$),c , a[j+1]).\ ] ] since @xmath141 is the first symbol of the suffix @xmath142 $ ] , then @xmath142=cl_i[j+1,\|l_i\|+1]$ ] . all the suffixes of @xmath111 starting with a symbol smaller than @xmath141 are lexicographically smaller than @xmath142 $ ] . the number of such suffixes is given by @xmath144 . let us count now the number of suffixes that starting with @xmath141 and are smaller than @xmath142 $ ] . this is equivalent to counting how many @xmath141 s occur in @xmath145 $ ] . such a value is given by @xmath146)$ ] . it is easy to verify that we can obtain the array @xmath147 by using the array @xmath136 and the suffix array @xmath102 , i.e. @xmath148 = a[sa(l_i\$)[i]]$ ] . note that the array @xmath147 contains the partial sums of the values of the @xmath149 array used in @xcite . however , we could directly compute the array @xmath147 by using the notion of inverse suffix array @xmath150 of a word @xmath36 is the inverse permutation of @xmath0 , i.e. , @xmath151 = i$ ] for all @xmath152 $ ] . the value @xmath153 $ ] is the lexicographical rank of the suffix starting at the position @xmath42 . ] . step [ it : a ] could be realized in @xmath154 time because we can build a data structure supporting @xmath155 time rank queries over @xmath100 . the same time complexity is obtained if the rank queries are executed over @xmath101 . step [ it : merge ] uses @xmath147 to create the new array @xmath99 by merging @xmath107 with the @xmath100 computed at the previous iteration . such a step implicitly constructs the lexicographically sorted list of suffixes starting in @xmath156 and extending up to end of @xmath6 together with the suffixes of @xmath6 . in order to do this we keep the mutual order between the suffixes of @xmath111 and @xmath105 thanks to theorem [ th : suforder ] . from the definition of the array @xmath147 , it follows that the first two positions of the array @xmath99 are the first symbol of @xmath107 and the first symbol of @xmath100 , respectively . for @xmath157 we copy @xmath110 $ ] values from @xmath100 followed by the value @xmath158 $ ] . it is easy to see that the time complexity of step [ it : merge ] is @xmath154 , too . from the description of the algorithm and by proceeding by induction , one can prove the following proposition . at the end of the iteration @xmath97 , algorithm bwt_lynd correctly computes @xmath159 . each iteration @xmath17 runs in @xmath154 time . the overall time complexity is @xmath160 , where @xmath161 . the goal of this paper is to propose a new strategy to compute the @xmath1 and the @xmath0 of a text by decomposing it into lyndon factors and by using the compatibility relation between the sorting of its local and global suffixes . at the moment , the quadratic cost of the algorithm could make it impractical . however , from one hand , in order to improve our algorithm , efficient dynamic data structure for the rank operations and for the insertion operations could be used . navarro and nekrich s recent result @xcite on optimal representations of dynamic sequences shows that one can insert symbols at arbitrary positions and compute the rank function in the optimal time @xmath162 within essentially @xmath163 bits of space , for a sequence @xmath164 of length @xmath30 . on the other hand , our technique , differently from other approaches in which partitions of the text are performed , is quite versatile so that it easily can be adapted to different implementative scenarios . for instance , in @xcite the authors describe an algorithm , called bwte , that logically partitions the input text @xmath3 of length @xmath30 into blocks of the same length @xmath165 , i.e. @xmath166 and computes incrementally the @xmath1 of @xmath3 via @xmath167 iterations , one per block of @xmath3 . text blocks are examined from right to left so that at iteration @xmath168 , they compute and store on disk @xmath169 given @xmath170 . in this case the mutual order of the suffixes in each block depends on the order of the suffixes of the next block . our algorithm bwt_lynd builds the @xmath171 of a text or its @xmath0 by scanning the text _ from left to right _ and it could run online , i.e. while the lyndon factorization is realized . one of the advantages is that adding new text to the end does not imply to compute again the mutual order of the suffixes of the text analyzed before , unless for the suffixes of the last lyndon word that could change by adding characters on the right . moreover , as described in the previous section , the text could be partitioned into several sequences of consecutive blocks of lyndon words , and the algorithm can be applied _ in parallel _ to each of those sequences . furthermore , also the lyndon factorization can be performed in parallel , as shown in @xcite . alternatively , since we read each symbol only once , also an in - place computation could be suggested by the strategy proposed in @xcite , in which the space occupied by text @xmath3 is used to store the @xmath172 . finally , in the description of the algorithm we did not mention the used workspace . in fact , it could depend on the time - space trade - off that one should reach . for instance , the methodologies used in @xcite where disk data access are executed only via sequential scans could be adapted in order to obtain a lightweight version of the algorithm . an external memory algorithm for the lyndon factorization can be found in @xcite . we remark that that the method proposed in @xcite could be integrated into bwt_lynd in the sense that one can apply bwte to compute at each iteration the @xmath1 and the @xmath0 of each block of the lyndon partition . in conclusion , our method seems lay out the path towards a new approach to the problem of sorting the suffixes of a text in which partitioning the text by using its combinatorial properties allows it to tackle the problem in local portions of the text in order to extend efficiently solutions to a global dimension .	the process of sorting the suffixes of a text plays a fundamental role in text algorithms . they are used for instance in the constructions of the burrows - wheeler transform and the suffix array , widely used in several fields of computer science . for this reason , several recent researches have been devoted to finding new strategies to obtain effective methods for such a sorting . in this paper we introduce a new methodology in which an important role is played by the lyndon factorization , so that the local suffixes inside factors detected by this factorization keep their mutual order when extended to the suffixes of the whole word . this property suggests a versatile technique that easily can be adapted to different implementative scenarios . ( # 1)bwt(#1 ) # 1_#1 ( ) # 1_#1 ( ) # 1_#1(w ) # 1_#1(w ) # 1(#1 ) sorting suffixes , bwt , suffix array , lyndon words , lyndon factorization
many main - sequence stars that are old enough for their primordial disks to have dissipated still emit significant excess infrared flux . this emission is usually attributed to starlight - heated dust located in a debris disk where the dust population is continually resupplied by processes such as collisions between planetesimals or sublimation of cometary material ( e.g. aumann et al . 1984 , backman et al . 1992 ; wyatt et al . 1999 ) . mid - ir ( @xmath18 - 25 @xmath0 m ) imaging is a useful tool for investigating these disks . photospheric emission at these wavelengths is often small compared to the disk emission , and so fruitful imaging of these disks , especially the innermost regions , does not require the use of a coronagraph , which blocks the starlight but also hides the key regions of interest near the star . large ( 8 - 10 m ) ground - based telescopes can directly probe these inner subarcsecond regions of the disk at mid - ir wavelengths . this emission can also sample a different particle population than those seen in scattered optical or near - ir light , thereby making it a valuable complementary approach ; submicrometer - to - micrometer - sized particles are particularly evident in the mid - ir . here we report the first multi - wavelength mid - ir images that resolve the debris disk of hd 32297 , and we note the independent discovery of resolved thermal emission by @xcite . these observations were motivated by high - contrast coronagraphic imaging of scattered light in the extended dust disk around hd 32297 in near - ir and optical bandpasses @xcite . the high fractional infrared luminosity ( 0.0027 ; schneider et al . 2005 ) of this a0 star indicates a considerable contribution from thermal emission that , it seemed to us , would likely be resolvable at a distance of 112 pc ( perryman et al . 1997 , hereafter p97 ) at mid - ir wavelengths . over a hundred debris disk candidates have been identified photometrically ( e.g. , rieke et al . 2005 ) , but only about a dozen have been spatially resolved at any wavelength . of the few resolved disks , seven belong to a - type stars like hd 32297 , with two of these ( vega and fomalhaut [ su et al . 2005 ; stapelfeldt et al . 2004 ] ) having surface brightnesses too low to be detected in the infrared with reasonable integration times from the ground . these and most other resolved debris disks around a stars ( @xmath2 pic , hr 4796a , 49 ceti , and hd 141569 ) extend more than 100 au in radius @xcite ; @xmath3 lep remains the exception among resolved disks , being comparable in size to our asteroid belt @xcite . hd 32297 is therefore an important addition to the small but growing sample of debris disks resolved at mid - ir wavelengths . _ data._ we obtained mid - ir images of hd 32297 on four nights in 2006 ( february 3 , march 4 , 9 , and 10 [ ut ] ) at gemini south ( program i d : gs-2005b - dd-8 ) with t - recs ( thermal region camera and spectrograph ) , using the narrowband si-5 ( @xmath4 = 11.66 @xmath0 m , @xmath5 = 1.13 @xmath0 m ) and narrowband qa ( @xmath4 = 18.30 @xmath0 m , @xmath5 = 1.51 @xmath0 m ) filters . t - recs utilizes a raytheon si : as bib detector with 320 x 240 pixels . each pixel subtends 0.09 `` , and the total field of view is 28.8 '' x 21.6 `` . the observations used the standard mid - ir technique of chopping and nodding to remove time - variable sky background , telescope thermal emission , and low - frequency detector noise ; the chop throw was 15 '' . the data were reduced with the gemini iraf package . to account for different exposure times , sky transparency , and sky background , the images from the four nights were weighted by the spatially integrated signal - to - noise ratio to produce the final images from each filter . the flux standard hd 37160 was observed before and after the group of target observations ; this source also served as a point - spread - function ( psf ) comparison star . hd 37160 is a g star located 8.7@xmath6 away from hd 32297 and is approximately 125 times brighter than hd 32297 at 11.7 @xmath0 m and 25 times brighter at 18.3 @xmath0 m . for the four nights of observations , the psf fwhm ranged from 0.370.46 at 11.7 @xmath0 m and from 0.530.60 at 18.3 @xmath0 m . by quadratic subtraction , we infer that the greatest deviation from the @xmath7/d diffraction limit ( 0.31 at 11.7 @xmath0 m and 0.49 at 18.3 @xmath0 m ) occurred with seeing disks of 0.33 at 11.7 @xmath0 m and 0.34 at 18.3 @xmath0 m . the final psf images were combined with the same weights as the source images and indicated a resolution ( fwhm ) of 0.40 at 11.7 @xmath0 m and 0.54 at 18.3 @xmath0 m . the total ( disk + star ) flux densities of hd 32297 are 53 mjy at 11.7 @xmath0 m and 90 mjy at 18.3 @xmath0 m . observations of the flux standards were not repeated throughout the night , so we adopt nominal calibration uncertainties of 10% at 11.7 @xmath0 m and 15% at 18.3 @xmath0 m , which are typical for photometric variations in the mid - ir ( e.g. de buizer et al . 2005 , packham et al . 2005 ) . our 11.7 @xmath0 m flux density is consistent with the @xmath8 12 @xmath0 m upper limit . @xmath8 detected hd 32297 ( iras 04597 + 0723 ) at 25 @xmath0 m ( moderate quality ) and 60 @xmath0 m ( high quality ) . _ disk geometry and morphology._ we observe an apparent nearly edge - on disk ( figure [ fig : images ] ) with a total extent ( at the 3-@xmath9 level ) of 291 au ( 2.58 ) and 341 au ( 3.02 ) at 11.7 and 18.3 @xmath0 m , respectively . with ellipse fitting of isophotes in the 11.7 @xmath0 m image , we find a pa for the major axis of 51.1@xmath6 @xmath10 2.4@xmath6 ( east of north ) , consistent to within 2@xmath6 with the value measured for the entire disk in near - ir images @xcite and for the southwest side of the disk in optical images @xcite . the mid - ir spatial extent is only half that measured at near - ir wavelengths @xcite . at 11.7 @xmath0 m , the integrated flux of the ne side of the disk is 1.34 @xmath10 0.09 times as bright as the sw side ; the statistical significance of the difference in the brightness between the two sides is 6.1 @xmath9 . the fluxes used in this calculation measure the total flux on each side exterior to the 0.75 region corresponding to the first airy ring , in order to exclude most of the photospheric flux . this asymmetry is opposite to that determined by @xcite , where the scattered light in the coronagraphic image is brighter in the sw by a difference of 2.6 @xmath9 . at 18.3 @xmath0 m , the sw side is 1.28 @xmath10 0.29 times as bright as the nw side ( opposite to that at 11.7 @xmath0 m ) , but this difference is not significant ( 1.5 @xmath9 ) . the fluxes for each side at 18.3 @xmath0 m are measured from the central star position outward , since the photospheric contribution at this wavelength is small ( see table [ tab : fluxes ] ) . other than the modest difference in brightness between sides , there are no obvious morphological features in the disk at either wavelength . asymmetries are seen in many disks , and they may have diverse origins such as resonant trapping or cataclysmic collisions @xcite . however , until deeper multi - wavelength images are available , detailed considerations of any asymmetries in hd 32297 seem to us to be unwarranted . _ comparison with key debris disk archetypes._ we compare the global mid - ir characteristics of hd 32297 to those of two important archetypes for disks around a stars that are probably of comparable ( @xmath11 myr ) age : @xmath2 pic and hr 4796a . hd 32297 s age is currently not well constrained , but a possible association of the source with the gould belt or taurus - aurigae supports an age estimate of @xmath130 myr . however , @xcite notes that the optically observed morphology could be interpreted as the result of outflows suggestive of a much younger age . the fractional ir luminosity of hd 32297 ( 0.0027 ; schneider et al . 2005 ) is similar to those of @xmath2 pic and hr 4796a : 0.0015 @xcite , and 0.005 @xcite , respectively . hd 32297 is much farther away ( 112 pc [ p97 ] ) than @xmath2 pic ( 19 pc [ p97 ] ) and hr 4796a ( 67 pc [ p97 ] ) , which results in a lower total signal - to - noise ratio and poorer resolution in its images . therefore , it is not possible with our current images to study the inner region of the disk closest to hd 32297 ( @xmath1230 au ) in as great detail as these other two sources . we can , however , usefully compare for the first time its mid - ir colors to those of the two disks . we estimate the photospheric flux densities for hd 32297 at 11.7 @xmath0 m and 18.3 @xmath0 m by extrapolating the k - band magnitude of 7.59 @xcite to 10 @xmath0 m . we assume that the flux density varies as @xmath13 over this wavelength range , as is estimated by @xcite to be appropriate for an a0 star ( e.g. jura et al . 1998 ) . beyond 10 @xmath0 m , we adopted the standard rayleigh - jeans ( @xmath14 ) relation to estimate photospheric flux densities and resulting excess flux densities as given in table 1 . the 24.5 @xmath0 m flux density of 198 @xmath10 32 mjy was interpolated from our 11.7 @xmath0 m measurement and the @xmath8 detection at 25 @xmath0 m . the color - color plot ( figure [ fig : colors ] ) of hd 32297 , @xmath2 pic and hr 4796a shows that all three sources possess similar @xmath15(24.5 @xmath0m)/@xmath15(18.3 @xmath0 m ) colors , but the @xmath15(18.3 @xmath0m)/@xmath15(11.7 @xmath0 m ) color of hr 4796a is twice that of both @xmath2 pic and hd 32297 @xcite . the high value for this latter color for hr 4796a is likely due to a deficiency of hot 11.7 @xmath0m - emitting dust very close to the star , as is indeed seen in mid - ir images of the source ( e.g. jayawardhana et al . 1998 ; koerner et al . 1998 ; telesco et al . 2000 ) where the region interior to 70 au is largely devoid of dust . the @xmath15(18.3 @xmath0m)/@xmath15(11.7 @xmath0 m ) color of hd 32297 appears more consistent with its innermost particles having smaller average radial distances , more comparable to the case of @xmath2 pic . in @xmath163.2 , we provide additional evidence and discussion regarding the detailed geometry of the central clearing . _ characteristic particle temperatures._ the mid - ir color temperature computed from the two excess flux densities is 186 @xmath10 15 k. the uncertainties in the temperature reflect those from measurement and photometric calibration . if the dust particles were blackbody emitters , the color temperature would imply a distance of @xmath17 au for the majority of the particles . however , since we observe the disk to extend at least ten times farther , most of the constituent particles must be small , inefficient emitters that are less than a few microns in radius for them to be so warm @xcite . for comparison , we also compute the temperature of the dust particles assuming an emission efficiency proportional to frequency , which yields 156 @xmath10 10 k. we measured the radial brightness profiles of the photosphere - subtracted disk by summing over a 0.8 swath along the disk axis and then summing the two sides of the disk . the radial bin widths are @xmath120 au , with the exception of the outermost data point that reflects the mean value for a @xmath140 au bin to increase s / n . for each filter , the psf was scaled to the level of the photosphere ( see table [ tab : fluxes ] ) and then subtracted from the disk image . to achieve the same resolution in the two images , each image was then convolved with a gaussian profile having a width corresponding to the psf fwhm in the other filter . we used these brightness profiles to generate a color temperature profile of the disk , shown in figure [ fig : temps ] . we see that the color temperature remains constant to within a few kelvins from the center of the disk out to @xmath180 au . particle temperature should decrease as distance from the star increases . therefore , this zone of constant temperature implies that the particles dominating the emission are only viewed in projection and are , in fact , at approximately the same physical distance from the star . based on the location where the temperature begins to decline , that distance is 80 au , which must correspond to the approximate outer boundary of a central region that is highly deficient in dust , as suggested by the color - color plot in @xmath163.1 . however , in apparent contradiction to figure [ fig : colors ] , in which the colors of hd 32297 are nearly the same as those of @xmath2 pic , this inferred central clearing size is almost identical to that determined for hr 4796a ( e.g. jayawardhana et al . 1998 ; koerner et al . 1998 ; telesco et al . 2000 ) . we suggest that both findings can be reconciled in a picture where the zone markedly deficient in dust is approximately the same size as the one in hr 4796a , but that this zone still contains significantly more dust than the cleared region of hr 4796a , which is almost completely free of debris . thus we may be observing a debris disk with a morphology intermediate between @xmath2 pic and hr 4796a . we do not mean to suggest that these stars correspond to a temporal sequence , since that is inconsistent with the known ages of @xmath2 pic and hr 4796a . however , the disk in hd 32297 appears to broaden our appreciation of the range of morphologies that can occur during disk evolution . we plot temperature curves in figure [ fig : temps ] for a blackbody and for particles of three characteristic radii ( 0.05 @xmath0 m , 0.075 @xmath0 m , and 0.2 @xmath0 m ) . these latter curves were calculated using equations presented in @xcite , which enable a particle temperature estimate at a given distance from the star for particles with a range of characteristics , described by a critical wavelength parameter @xmath17 . this parameter is related to radiation efficiency , such that the efficiency is assumed to be nearly unity for wavelengths shorter than @xmath17 . the value of @xmath17 depends on composition , and we adopt the case where @xmath17 is approximately equal to the particle size , as expected for particles that absorb efficiently but emit inefficiently , such as graphite and amorphous silicate @xcite . we find that hd 32297 s color temperature profile does not coincide with that expected for particles of only one characteristic size . the profile is well approximated by the curve for 0.075-@xmath0 m particles between @xmath180120 au , but at larger distances the color temperature falls below this curve . since @xmath17 reflects both particle size and composition , a change in either ( or both ) of these characteristics of the dominant population at @xmath1120 au could produce the deviation we observe from the temperature curve for a single @xmath17 value . such a radial transition might occur at the snow line . the snow line in protoplanetary disks is expected to lie where temperatures fall to @xmath1145170 k ( e.g. hayashi 1981 ; sasselov & lecar 2000 ) . we estimate color temperatures below this level for the two outermost points in the disk , beyond @xmath1140 au . an increase in particle surface density beyond the putative snow line would likely result in the growth of larger icy particles than would be found in the inner region of the disk . such particles would show different absorption and emission behavior compared to non - ice - enhanced particles closer to the star , not only because of their larger size , but also because of the significant change in composition , which dictates the particle heating and cooling processes ( e.g. podolak & zucker 2004 ) . this idea is purely speculative and must be investigated further with spatially resolved spectroscopic observations and subsequent modeling . while the disk of hd 32297 must be explored with deeper direct imaging and spectroscopy to clarify the distribution and composition of dust , we have established the key conclusion that hd 32297 is yet another debris disk with a central clearing . the number of disks currently resolved with imaging is still small , but most of these resolved sources share the common feature of a cleared central area . by studying these debris disks further , we can better understand how their inner regions are shaped by planets or other processes ( e.g. , klahr & lin 2001 ; kalas et al . 2005 ) . we wish to acknowledge very useful comments from the referee , karl stapelfeldt . mm gratefully acknowledges fellowship support from the michelson science center . observations were obtained at the gemini observatory , operated by aura , inc . , under agreement with the nsf on behalf of the gemini partnership . m flux density measurements along the disk . large error bars reflect the combined photometric and measurement uncertainties and small error bars reflect only the measurement uncertainties . the dotted line indicates the expected blackbody temperature change with distance from the star and the three solid lines indicate predicted temperatures ( as discussed in @xcite and 3.2 ) for characteristic grain sizes : 0.25 @xmath0 m ( left ) , 0.075 @xmath0 m ( center ) , and 0.05 @xmath0 m ( right ) . [ fig : temps ] ]	we present the first subarcsecond - resolution images at multiple mid - ir wavelengths of the thermally - emitting dust around the a0 star hd 32297 . our observations with t - recs at gemini south reveal a nearly edge - on resolved disk at both 11.7 @xmath0 m and 18.3 @xmath0 m that extends @xmath1150 au in radius . the mid - ir is the third wavelength region in which this disk has been resolved , following coronagraphic observations by others of the source at optical and near - ir wavelengths . the global mid - ir colors and detailed consideration of the radial color - temperature distribution imply that the central part of the disk out to @xmath180 au is relatively deficient in dust .
the list of materials in which carriers are strongly interacting with each other and scatter by random impurities is quite long . it comprises _ inter alia _ high temperature superconducting oxides @xcite and various heavy fermion alloys @xcite . two models are usually used to study such system . it is either hubbard or anderson model suitably extended to allow for the description of real materials . here the single band hubbard model has been used , as it is the simplest model of correlated system . the disorder is introduced into the model by allowing for fluctuations of the local on - site energies @xmath0 . the main purpose of this work is to study the interplay of disorder and correlations in the hubbard model . i shall present analytical and numerical calculations indicating that correlations induce additional disorder in the system . _ mutatis mutandis _ disorder in the system affects the parameters of the mott - hubbard metal - insulator transition . in the recent studies of weakly interacting disordered systems it has been found that due to disorder the interactions scale to strong coupling limit @xcite . here to treat many particle aspect of the problem we use the slave boson technique , which is known to be qualitatively valid for all strength of interaction . in particular this method reproduces the results of the gutzwiller approximation at the saddle - point level @xcite . we start here with random version of the hubbard model @xmath1 the meaning of symbols is standard . the first term describes the hopping of carriers through the crystal specified by lattice sites @xmath2 , @xmath3 . @xmath4 denotes the fluctuating site energy - it introduces disorder into the model . @xmath5 is the repulsion between two opposite spin electrons occupying the same site . in the slave boson technique@xcite one introduces 4 auxiliary boson fields : @xmath6 , @xmath7 , @xmath8 such that @xmath9 , @xmath10 , @xmath11 project onto the empty , singly occupied by @xmath12 and doubly occupied site @xmath2 . the hamiltonian ( 1 ) is in this enlarged hilbert space written as @xmath13 where @xmath14 . equation ( 2 ) is strictly equivalent to ( 1 ) , when the constraints @xmath15 are fulfilled . in disordered system and at @xmath16 = 0 k , the mean field approach to slave bosons can be formulated by suitably generalizing the clean limit . one introduces the constraints into hamiltonian with help of langrange multipliers @xmath17 and @xmath18 and replaces boson operators @xmath19 , @xmath20 , @xmath21 by classical , site dependent , amplitudes @xmath22 , @xmath23 , @xmath24 . these amplitudes are calculated from the , configuration dependent , ground state energy @xmath25 by minimizing @xmath26 with respect to all seven parameters @xmath22 , @xmath23 , @xmath24 , @xmath27 , @xmath18 . as a result one gets three constraints : @xmath28 , @xmath29 and four additional equations which read @xmath30 where @xmath31 . the important next step is connected with calculation of @xmath32 and @xmath33 in the presence of disorder . in principle these quantities do depend on the particular distribution of all impurities ( i.e. distribution of site energies @xmath4 ) . we shall calculate them from the corresponding green s function . we use a version@xcite of the coherent potential approximation ( cpa ) to calculate ( averaged and conditionally averaged ) green s functions . to this end we rewrite the mean field hamiltonian in the form @xmath34 note , that hamiltonian ( 5 ) contains not only the site dependent parameters @xmath35 but due to correlations there appear also @xmath36 which are the additional source of , diagonal and spin dependent , disorder . besides that , the parameters @xmath37 , @xmath38 which vary from site to site make effective hopings @xmath39 random quantities . this off - diagonal ( in wannier space ) disorder is particularly important . fortunately multiplicative dependence of effective hopping on the kind of atoms at sites @xmath2 and @xmath3 is easy to handle within cpa . the procedure is standard@xcite and one gets the following equation for the density of states in paramagnetic phase @xmath40}{1-\left [ \sigma ( \varepsilon ^{+})-(\varepsilon -\varepsilon _ { i}+\mu -\lambda _ { i}^{(2)})/\xi _ { i}^{2 } \right ] \,f[\sigma ( \varepsilon ) ] } } \rangle _ { { \rm imp}}\ ] ] and the coherent potential @xmath41 is determined as a solution of equations @xmath42 \,f[\sigma ( z)]}}\rangle _ { { \rm imp}}=0 \nonumber \\ & & f[\sigma ( z)]={\frac{1}{n}}\sum_{\vec{k}}{\frac{1}{\sigma ( z)-\varepsilon ( \vec{k})}}={\frac{1}{n}}\sum_{\vec{k}}\,\bar{g}_{\vec{k}}(z ) \end{aligned}\ ] ] here @xmath43 means averaging over disorder and @xmath44 is electron spectrum in host ( clean ) material . to solve equations ( 4 ) one still has to calculate the quantities @xmath45 , which in general depend on the configuration of all impurities in the system . they can be calculated from the knowledge of the cpa green s functions in the following way . we assume that site @xmath2 is described by actual parameters _ i.e. _ @xmath4 , @xmath46 , @xmath37 , while all other sites in the system are replaced by effective ones described by the coherent potential @xmath47 and the green s functions @xmath48 . then it is easy to find that _ e.g_. @xmath49 and @xcite @xmath50 here @xmath51 , and by definition @xmath52 $ ] and @xmath53 this closes the system of equations to be solved in a self - consistent manner . in the next section we show numerical calculations of the effect of correlations on the density of states in disordered alloy . for the purpose of numerical illustration of the general approach sketched in previous section we calculate here the density of states ( dos ) of perfect interacting , random noninteracting and random interacting systems . for the purpose of numerical analysis we shall assume @xmath54 limit and _ bcc _ crystal structure , which leads to the following canonical tight - binding spectrum of noninteracting carriers in clean material using t=0.0625@xmath56 leads to the noninteracting system bandwidth @xmath57 . in the following all energies and frequencies are expressed in @xmath56 . the density of states corresponding to this spectrum is well known . it possesses van hove singularity in the middle of the band , which extends from @xmath58 to @xmath59 . the spectrum @xmath60 of a clean but interacting system in the @xmath54 limit is replaced by @xmath61 . the corresponding density of states is thus of the same form but placed in the energy window @xmath62 $ ] and scaled by the factor @xmath63 . note that in this @xmath54 limit the upper hubbard band is pushed to infinity and not visible . the changes of the spectrum due to correlations in disordered @xmath64 alloy are illustrated in figure ( 1 ) where we show the averaged @xmath65 and conditionally averaged densities of states @xmath66 and @xmath67 ( @xmath68 , @xmath69 ) without ( fig . 2a ) and with ( fig . 2b ) electron correlations . the carrier concentration @xmath70 . noninteracting alloy dos is symmetric . we observe strong asymmetry , the opening of the real gap in the spectrum of interacting carriers and the appreciable increase of the density of states at the fermi level ( taken as e=0 in the figure ) . in conclusion , we have shown that interplay of disorder and correlations leads to strong renormalisation of the electron spectrum . let us stress that all other properties of such systems will also be strongly affected by interaction induced disorder . the calculations of _ dc _ and _ ac _ transport properties are in progress .	i address here the question of the mutual interplay of strong correlations and disorder in the system . i consider random version of the hubbard model . diagonal randomness is introduced _ via _ random on - site energies and treated by the coherent potential approximation . strong , short ranged , electron - electron interactions are described by the slave boson technique and found to induce additional disorder in the system . as an example i calculate the density of states of the random interacting binary alloy and compare it with that for non - interacting system . pacs numbers : 71.10.fd , 71.23.-k , 71.55.ak .
it is generally believed that @xmath5 ( @xmath6 ) pairing is important in nuclei with roughly equal numbers of neutrons and protons @xcite . the standard technique for treating these correlations is through the bardeen cooper schrieffer ( bcs ) or hartree fock bogolyubov ( hfb ) approximation , generalized to include the @xmath6 pairing field in addition to the @xmath7 and @xmath8 pairing fields @xcite . questions arise , however , as to whether these methods can adequately represent the physics of the competing modes of pair correlations , without full restoration of symmetries @xcite . important insight into this issue has been achieved recently in the context of exactly - solvable models that include these different pairing modes . analysis of the so(8 ) model @xcite , in which isoscalar and isovector pairing act in either a single active orbital or a series of degenerate orbitals , suggests that isospin restoration or equivalently quartet correlations are extremely important , especially near @xmath9 @xcite . more recent studies , carried out for models involving non - degenerate orbitals @xcite , reinforce earlier conclusions as to where isoscalar pairing correlations should be most important @xcite,@xcite . furthermore , they make possible the description of deformation , as is critical for systems with @xmath10 , by treating the non - degenerate orbitals as nilsson - like . however , it is still not possible to restore certain symmetries within these models , for example rotational symmetry . as a consequence , there still remain many open issues concerning the role of the different possible modes of pairing in @xmath11 nuclei . in this work , we report a systematic study of pairing correlations in the context of the nuclear shell model , whereby deformation can be readily included and symmetries maintained throughout . in this way , we are able to address many of the open issues on the role of the various pairing modes in the presence of nuclear deformation . an outline of the paper is as follows . in section ii , we briefly describe our model and then in section iii describe selected results . finally , in section iv we summarize the key conclusions of the paper . to address in a systematic way the role of pairing correlations in the presence of nuclear deformation , we consider neutrons and protons restricted to the orbitals of the @xmath0 shell outside a doubly - magic @xmath12 core and interacting via a schematic hamiltonian @xmath13 where @xmath14 is the mass quadrupole operator , @xmath15 creates a correlated @xmath16 , @xmath17 , @xmath18 , @xmath19 pair and @xmath20 creates a correlated @xmath16 , @xmath21 , @xmath22 , @xmath23 pair . the first term in the hamiltonian produces rotational collective motion , whereas the second and third term are the isoscalar and isovector pairing interactions , respectively , whose matrix elements can be found in ref . ( @xcite ) . the last term is the one - body part of the spin - orbit interaction , which splits the @xmath24 levels with a given @xmath25 . we carry out shell - model calculations systematically as a function of the various strength parameters . we begin with pure su(3 ) rotational motion @xcite associated with the @xmath26 interaction and then gradually ramp up the various su(3)-breaking terms to assess how they affect the rotational properties . this includes the isocalar and isovector pairing interactions and the spin - orbit term . we first consider the nucleus @xmath27 , with @xmath28 , and then systematically increase @xmath29 and @xmath30 to study the role of the number of active neutrons and protons , e.g. whether there is an excess of one over the other and whether the nucleus is even - even , odd - mass or odd - odd . the nuclei we have treated are @xmath27 ( @xmath31 ) , @xmath2ti ( @xmath32 , @xmath33 ) , @xmath3ti ( @xmath34 , @xmath35 ) , @xmath3v ( @xmath36 , @xmath33 ) , and @xmath4cr ( @xmath37 , @xmath35 ) . some of the observables we have studied are ( 1 ) the energies and associated be(2 ) values of the lowest rotational band , ( 2 ) the number of isovector @xmath20 pairs and ( 3 ) the number of isoscalar @xmath15 pairs . in the following , we present selected results that derive from these calculations . before turning to our results for specific nuclei , we first ask whether the hamiltonian ( [ hamiltonian ] ) has sufficient flexibility to realistically describe the nuclei under investigation . without making an effort towards an absolute fit , we note that the choice @xmath38 , and @xmath39 gives an acceptable fit to the spectra of all the nuclei we have considered . comparison of experimental spectra for @xmath40 with the calculated spectra obtained using the _ optimal _ hamiltonian described in the text . all energies are in @xmath41.,width=453 ] we first show in figure 1 its prediction for @xmath42sc , in comparison with the experimental spectrum . other than its inability to reproduce the low - lying @xmath43 , @xmath19 state , the optimal hamiltonian does an acceptable job in qualitatively describing the features of the low - energy spectrum . the lack of an acceptable description of the @xmath44 state reflects the fact that our optimal hamiltonian is missing the strong attraction between @xmath45 nucleons in the stretch configuration . comparison of experimental spectra for @xmath1ti , @xmath3ti and @xmath4cr with the calculated spectra obtained using the _ optimal _ hamiltonian described in the text . all energies are in @xmath41.,width=529 ] in figure 2 , we show its predictions for @xmath1ti , @xmath3ti and @xmath4cr . as can be seen , the non - rotational character of @xmath1ti at low angular momenta is reproduced by our calculations , as are the highly rotational patterns seen experimentally for the heavier nuclei . as we will see later , even the experimentally observed backbend in @xmath4cr is acceptably reproduced with this hamiltonian . we refer to the choice @xmath46 in the _ optimal _ hamiltonian as the su(4 ) choice , from the dynamical symmetry that derives from this choice of parameters in the so(8 ) model . calculated energy splittings @xmath47 ( in @xmath41 ) in the ground band of @xmath1ti as a function of the strength of the ( a ) isoscalar pairing interaction and the ( b ) isovector pairing interaction , with no spin - orbit splitting . the strengths of the respective pairing interactions are shown at the ends of the lines , as they are elsewhere in the manuscript.,width=680 ] the first nucleus we discuss is @xmath1ti , with two active neutrons and two active protons . in figure 3 , we show the calculated energy splittings @xmath47 associated with the ground - state band as a function of the strength parameters @xmath48 and @xmath49 that define the isoscalar and isovector pairing interactions , respectively . for these calculations we assumed a quadrupole strength of @xmath50 _ and no spin - orbit interaction_. as expected , in the absence of a spin - orbit splitting the isoscalar and isovector pairing interactions have precisely the same effect on the properties of the ground state rotational band . the same conclusion derives when we consider the effect of isoscalar and isovector pairing on other observable properties when no spin - orbit splitting is included . as an example , we show in figure 4 results for the be(2 ) values connecting the states in the ground band , again as a function separately of the isovector and isoscalar pairing strengths . calculated e2 transition matrix elements @xmath51 in the ground band of @xmath1ti as a function of the strength of the ( a ) isoscalar pairing interaction and of the ( b ) isovector pairing interaction , with no spin - orbit splitting . the angular momentum @xmath52 of the initial state appears to the left of each line . , width=604 ] we next show in figure 5 the same results as in figure 2 , but now in the presence of our realistic spin - orbit strength , @xmath53 . here we see that the effects of isoscalar and isovector pairing are very different . in the presence of realistic single - particle energies , isovector pairing produces a much more rapid loss of rotational collectivity than isoscalar pairing . calculated energy splittings @xmath47 in @xmath41 in the ground band of @xmath1ti as a function of the strength of the ( a ) isoscalar pairing interaction and of the ( b ) isovector pairing interaction , with the optimal spin - orbit term present . , width=680 ] next we consider the relative effect of the spin - orbit interaction on isovector and isoscalar pairing properties in the optimal su(4 ) limit , where both modes of pairing contribute with the same strength . this is addressed in figure 6 , where we show the average number of @xmath20 and @xmath15 pairs in the ground band as a function of the strength of the spin - orbit force . these are determined by considering @xmath54 and @xmath55 and scaling them with respect to the results that would derive from pure @xmath19 and @xmath23 pairing hamiltonians ( for a system of two pairs in an @xmath56 shell ) , respectively . while the number of isovector pairs does not change substantially with increasing spin - orbit strength , the isoscalar pair number is reduced dramatically , especially for the lower angular momentum states of the band . we conclude , therefore , that the spin - orbit interaction suppresses isoscalar pairing , already at @xmath9 . the mechanism whereby this suppression takes place was discussed recently in ref . @xcite . the number of ( a ) isocalar @xmath57 pairs and ( b ) isovector @xmath20 pairs in @xmath1ti as a function of the strength of the spin - orbit interaction @xmath58 . all other hamiltonian parameters are the optimal values . , width=680 ] comparison of the experimental spectrum of @xmath1ti with those obtained using a pure @xmath26 interaction and both a @xmath26 interaction and a spin - orbit term . , width=453 ] finally , in figure 7 we show the spectrum of @xmath1ti that derives _ solely _ from turning on a strong spin - orbit force , i.e. with no pairing present . we see that the spectrum is still highly rotational , despite the fact that the resulting single - particle energies are no longer su(3)-like . to obtain the physical spectrum with a non - rotational character , it is thus essential to have pairing . it has been traditionally thought that it is the non - su(3 ) order of the single - particle levels that is responsible for the non - rotational character seen in the experimental spectrum @xcite , a conclusion that is not supported by our results . it is pairing that is responsible for the non - rotational character of @xmath1ti . this point was already suggested in fig . 3 , where we we saw that for the physical pairing strengths a highly non - rotational spectrum near the ground band emerged even in the absence of a spin - orbit splitting . further understanding of this conclusion follows from the important role of quasi - su(3 ) @xcite in producing deformation . even though the spin - orbit interaction breaks the su(3 ) symmetry , it leaves quasi - su(3 ) symmetry approximately preserved . calculated energy splittings @xmath59 in @xmath41 within the odd group of levels ( see text ) of the ground state band of @xmath2ti as a function of the strength of the ( a ) isoscalar pairing interaction and of the ( b ) isovector pairing interaction , with no spin - orbit term present . , width=604 ] calculated energy splittings @xmath59 within the even group of levels ( see text ) of the ground state band of @xmath2ti as a function of the strength of the isoscalar pairing interaction ( left panel ) and of the isovector pairing interaction ( right panel ) , with no spin - orbit term present . , width=604 ] next we consider the odd - mass nucleus @xmath2ti , with one additional neutron relative to @xmath1ti . for notational purposes , we divide the results according to whether @xmath60 is odd ( referring to this as the odd group ) or whether it is even ( the even group ) . this reflects the fact that states within the ground state band decay by strong e2 transitions within their own groups . in figures 8 and 9 , we present results for the calculated energy splittings of the ground band in the odd and even groups , respectively , as a function of pure isoscalar ( panel a ) and pure isovector ( panel b ) pairing , in both cases with no spin - orbit force present . the results suggest that blocking due to an odd neutron does not affect the symmetry between isoscalar and isovector pairing when there is no spin - orbit force . in figures 10 and 11 , we show the corresponding results in the presence of a spin - orbit force , with the optimal strength @xmath53 . now isoscalar pairing is suppressed and there is a much more dramatic effect of isovector pairing . calculated energy splittings @xmath47 in @xmath41 within the odd group of levels ( see text ) of the ground state band of @xmath2ti as a function of the strength of the ( a ) isoscalar pairing interaction and of the ( b ) isovector pairing interaction , with the optimal spin - orbit term present . , width=604 ] calculated energy splittings @xmath47 in @xmath41 within the even group of levels ( see text ) of the ground state band of @xmath2ti as a function of the strength of the ( a ) isoscalar pairing interaction and of the ( b ) isovector pairing interaction , with the optimal spin - orbit term present . , width=604 ] calculated energy splittings @xmath47 in @xmath41 in the ground band of @xmath3ti as a function of the strength of the ( a ) isoscalar pairing interaction and of the ( b ) isovector pairing interaction , with no spin - orbit term present . , width=604 ] next we turn to @xmath3ti with two excess neutrons present . here too we compare the effect of the isoscalar and isovector pairing interactions on deformation , showing the results in figure 12 with no spin - orbit term present . here the effect of isoscalar pairing is strongly suppressed relative to isovector pairing , suggesting that even without a spin - orbit term isoscalar pairing is very strongly focused on those nuclei with @xmath9 with a slight excess being sufficient to suppress this pairing mode . we next consider @xmath3v , an odd - odd @xmath9 nucleus . the @xmath23 states of @xmath3v are of course precisely the same as those already considered in @xmath3ti . we first address in figure 13 the manner whereby the symmetry between isocalar and isovector pairing in the absence of a spin - orbit force is reflected in @xmath3v . in the absence of isoscalar and isovector pairing , the @xmath61 state and the @xmath62 state form a degenerate ground state doublet . when only isoscalar pairing is turned on ( panel a ) , the @xmath61 state is pushed down below the @xmath62 state . when only isovector pairing is turned on ( panel b ) the reverse happens and the @xmath62 is pushed down and becomes the ground state . in the su(4 ) limit ( panel c ) with equal isovector and isocalar pairing strengths , the degeneracy reappears . in figure 14 , we show what happens in the presence of the physical spin - orbit interaction , for equal isovector and isoscalar pairing . now the degeneracy is broken and the @xmath63 state emerges as the ground state , as in experiment . the experimental splitting is 1.23 @xmath41 , whereas our optimal hamiltonian produces a splitting of 1.05 @xmath41 . we should note that the first excited state in @xmath3v experimentally is found to be a @xmath64 state , at 801 @xmath65 . in our calculations the lowest @xmath64 state occurs at significantly higher energy , at 1.89 @xmath41 . this may be related to the schematic nature of our hamiltonian . calculated energies in @xmath41 of the lowest @xmath66 and @xmath67 states of @xmath3v with no spin - orbit term present . panel ( a ) shows the results of pure isoscalar pairing , panel ( b ) shows the results of pure isovector pairing and panel ( c ) shows the results of su(4 ) pairing . , width=642 ] calculated energies in @xmath41 of the lowest @xmath66 and @xmath67 states of @xmath3v as a function of the equal strength of isoscalar and isovector pairing , with the optimal spin - orbit term ( @xmath53 ) present . , width=415 ] lastly , we turn to @xmath4cr , which again has @xmath9 , but now with two quartet structures present . here we assume as our starting point both the optimal quadrupole - quadrupole force and the optimal one - body spin - orbit force and then ramp up the two pairing strengths from zero to their optimal values . the results are illustrated in figure 15 , for scenarios in which we separately include isoscalar pairing , isovector pairing and su(4 ) pairing with equal strengths . as a reminder , the experimental spectrum for @xmath4cr shows a backbend near @xmath68 , which as noted earlier is reproduced by our _ optimal _ hamiltonian . the results of figure 15 make clear that ( a ) the backbend can not be reproduced with pure isoscalar pairing , but requires isovector pairing as well , and ( b ) there is no significant difference between the results obtained with pure isovector pairing and su(4 ) pairing . calculated energy splittings @xmath47 in @xmath41 in the @xmath4cr ground band , for ( a ) isovector , ( b ) isoscalar , and ( c ) su(4 ) pairing , respectively , as described in the text . the optimal spin - orbit term is included.,width=680 ] the backbend in @xmath4cr was discussed extensively in the context of a shell - model study with a fully realistic hamiltonian in @xcite , where it was first shown to derive from isovector pairing . our results are in agreement with that earlier conclusion . to see these points more clearly , we show in figure 16 the numbers of isovector @xmath20 and isoscalar @xmath15 pairs as a function of angular momentum for the optimal hamiltonian . as in our treatment of @xmath1ti ( see fig . 6 ) , the pair numbers are obtained by evaluating @xmath54 and @xmath69 and scaling them with respect to the results that would derive from pure @xmath23 and @xmath19 pairing hamiltonians , respectively . [ now , however , the analysis is carried out for a system of four pairs in an @xmath56 shell . ] as in ref . @xcite , the contribution of isovector pairing in the @xmath62 ground state in much larger than the contribution of isoscalar pairing . as the system cranks to higher angular momenta , the isovector pairing contribution falls off with angular momentum very rapidly eventually arriving at a magnitude roughly comparable with the isoscalar pairing contribution at roughly @xmath70 . as the angular momentum increases even further we see a fairly substantial increase in the isovector pairing contribution at @xmath71 , which according to figure 15 is where the backbend becomes prominent . after the backbend , both isoscalar and isovector pairing contributions decrease to near zero as alignment is achieved . calculated numbers of isovector @xmath20 pairs and isoscalar @xmath15 pairs in the ground ( yrast ) band of @xmath4cr for the optimal values of the hamiltonian parameters.,width=604 ] we have also studied the properties of the lowest excited ( yrare ) band that emerges from the same calculation , a k=2@xmath72 band . the energies of this band are illustrated in figure 17 , together with those of the ground ( yrast ) band . from this figure , we conclude that the backbend in @xmath73cr does not derive from level crossing . calculated excitation energies of the ground ( yrast ) band and the first excited ( yrare ) band in @xmath4cr for the optimal values of the hamiltonian parameters . , width=604 ] in figure 18 , we show the number of isocalar and isovector pairs in the excited yrare band , to be compared with the results for the yrast band of figure 16 . in the backbend region , the numbers of isoscalar and isovector pairs are found to behave differently in the two bands . whereas the numbers of isoscalar and isovector pairs are roughly the same in the yrast band ( fig . 16 ) , there are substantially fewer isoscalar pairs than isovector pairs in the yrare band ( fig . we believe that this is an interesting observation worthy of further study . calculated numbers of isovector @xmath20 pairs and isoscalar @xmath15 pairs in the first excited ( yrare ) band of @xmath4cr for the optimal values of the hamiltonian parameters.,width=604 ] in this work , we have reported a systematic shell - model study of proton - neutron pairing in @xmath0 shell nuclei using a parametrized hamiltonian that includes deformation and spin - orbit effects as well as both isoscalar and isovector pairing . by working in a shell - model framework we are able to assess the role of the various pairing modes in the presence of nuclear deformation without violating symmetries . we first showed that our parametrized hamiltonian has enough flexibility to provide a reasonable description of the evolution of nuclear structure properties in this region . we then probed the role of the various modes of pairing on deformation with and without a spin - orbit term . we did this as a function of the number of neutrons and protons , so as to assess the role both of a neutron excess and of the number of active particles . some of the conclusions that emerged are : ( 1 ) in the absence of a spin - orbit term , isoscalar and isovector pairing have identical effects at @xmath9 , but isoscalar pairing ceases to have an appreciable effect for nuclei with just two excess neutrons ; ( 2 ) the non - rotational character of @xmath1ti can not be explained solely in terms of spin - orbit effects , but requires pairing for its understanding ; ( 3 ) in the presence of a spin - orbit interaction , isoscalar pairing is suppressed even at @xmath9 , ( 4 ) the fact that ground state of @xmath3v has @xmath66 @xmath23 derives primarily from the spin - orbit interaction and its effect of suppressing isoscalar pairing , ( 5 ) the known backbend in @xmath4cr has its origin in isovector pairing and does not derive from level crossing , and ( 6 ) in the region of the @xmath4cr backbend , the numbers of isoscalar and isovector pairs behave quite differently in the yrast and yrare bands . * acknowledgements : * the work reported herein began while two of the authors , s.p . and n.s . , were visiting the consejo superior de investigaciones cientficas in madrid , whose hospitality is gratefully acknowledged . much of it was carried out while y.l . was visiting the bartol research institute of the university of delaware , whose hospitality is likewise acknowledged . the work of s.p . , b.t . and y.l . was supported in part by the national science foundation under grant # phy-0854873 , that of n.s . by the romanian ministry of education and research through cncsis grant idei nr . 1975 , that of a.p . by the projects fpa2009 - 13377 micinn(spain ) and hephacos s2009/esp-1473 comunidad de madrid(spain ) , and that of y.m.z . and y.l . by the national science foundation of china under grant # 10975096 and by the chinese major state basic research developing program under grant # 2007cb815000 8 a. l. goodman , adv . nucl . phys . * 11 * , 263 ( 1979 ) . pang , nucl . phys . * a128 * , 497 ( 1965 ) ; j. a. evans , g. g. dussel , e. e. macqueda and r. p. j. perazzo , nucl . phys . * a367 * , 77 ( 1981 ) ; g. g. dussel , e. e. macqueda , r. p. j. perazzo and j. a. evans , nucl . a450 * , 164 ( 1986 ) . b. errea , ph.d . thesis , consejo superior de investigaciones cientficas , unpublished ( 2009 ) . s. lerma h. , b. errea , j. dukelsky and w. satula , phys . 99 * , 032501 ( 2007 ) . n. sandulescu , b. errea , j. dukelsky , phys . rev . * c80 * , 044335 ( 2009 ) . a. poves and g. martinez - pinedo , phys . b430 * 203 , ( 1998 ) . j. p. elliott , proc . london ser . a * 245 * , 128 , 562 ( 1956 ) .	a shell - model study of proton - neutron pairing in @xmath0 shell nuclei using a parametrized hamiltonian that includes deformation and spin - orbit effects as well as isoscalar and isovector pairing is reported . by working in a shell - model framework we are able to assess the role of the various modes of proton - neutron pairing in the presence of nuclear deformation without violating symmetries . results are presented for @xmath1ti , @xmath2ti , @xmath3ti , @xmath3v and @xmath4cr to assess how proton - neutron pair correlations emerge under different scenarios . we also study how the presence of a one - body spin - obit interaction affects the contribution of the various pairing modes . * pacs numbers : * 21.60.cs , 21.60.fw , 02.30.ik +
the ic 65 group of four late type galaxies ( ic 65 , ugc 608 , ugc 622 , and an `` edge - on '' galaxy - pgc 138291 ) has been studied by van moorsel ( 1983 ) in the 21 cm hi line with the wsrt . he noted : `` each of these four galaxies show asymmetric features in their hi distribution , especially ugc 622 . ... the presence of hi asymmetries and the small dispersion in radial velocities makes it likely that all detected four galaxies are members of one group '' at the distance of @xmath6 mpc ( @xmath7 km / s / mpc is assumed ) . because of its unfavourable location close to the zone of avoidance ( @xmath8 ) this group was not registered in earlier catalogs of nearby groups , and has been registered as the lyon group of galaxies ( lgg ) 16 ( garcia 1993 ) . it fulfills selection criteria of the hickson compact groups . van moorsel ( 1983 ) detected one more hi - rich lsb anonymous ( a ) galaxy located @xmath9 to the ne of ic 65 ( a 0101 + 4744 in the fig . 1 ) , having systemic velocity , which is close to the group mean . this encourages us to look for further dwarf companions in the area of this particular group . we carried out a systematic search for new lsb dwarf member candidates of this particular group by utilizing the digitized poss ii ( dss 2 ) blue and red frames , which were previously linearized and photometrically calibrated by us . the searched area was @xmath10 , i.e. 0.67 @xmath11 0.67 mpc at the distance of the group . the detection of objects was performed by means of running the sextractor software ( bertin & arnouts 1996 ) in double - image mode on the blue and red dss 2 frames . galaxies were selected in the sextracted catalog by several steps using the stellarity index , diameters and surface brightness ( sb ) characteristics . in a crowded environment of the ic 65 group the automatic selection was insufficient . we needed to take the morphologies into account . in effect , the final list of 105 galaxies was selected , which is believed to be complete for objects with diameters @xmath12 at the 25.5 @xmath13 isophote . 0.1 cm ( 160,70 ) in order to disentangle the possible new dwarf members of the group from distant field galaxies , we applied two criteria : 1 ) the binggeli s ( 1994 ) empirical relation between the central sb and absolute magnitude , common both for de and dirr galaxies ( fig . 2a ) , and 2 ) the empirical light concentration parameter , as proposed by trentham et al . ( 2001 ) , versus sextracted colour index ( fig . 2b ) . as a result , we have selected four lsb irregular galaxies , which show the lowest light concentration ( i.e. the largest scale lengths ) , the bluest color ( @xmath14 ) , and fit the empirical binggeli s relation for dwarf galaxies . there are 7 more galaxies with lsb morphologies , but all of them with @xmath15 . at fainter magnitudes and/or smaller diameters we actually lose the ability to distinguish possible group members from the background galaxies on sb grounds . + table 1 lists the basic data for certain group members and for new selected anonymous ( a ) probable dwarf members . all magnitudes are corrected for galactic absorption . llllcrllll galaxy & type & @xmath16 & @xmath17 & @xmath18 & @xmath19 & @xmath20 & @xmath21 & grd@xmath22 & @xmath23 + ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8) & ( 9 ) & ( 10 ) + + ugc 608 & sabdm & 2.@xmath240 0.@xmath249 & 128@xmath25 & 2755 & 14.34 & -18.26 & 0.74 & 0 & 0.3 : + pgc 138291 & s(dm ) & 1.3 : 0.3 : & 165 & 2598 & 16.5 : & -16.4 : & 1.1 : & 0 & 0.2 : + a0100 + 4756 & dirr & 0.6 : 0.4 : & 30 : & & 18.5 : & -14.4 : & 0.75 & 0 & 0.2 : + a0100 + 4734 & dirr & 0.8 : 0.3 : & 108 : & & 16.86 & -16.1 : & 0.74 & + & 0.25 + ugc 622 & scd : & 1.1 0.7 & 159 & 2714 & 13.76 & -19.21 & 1.27 & + : & 0.61 + ic 65 & sab(s)bc & 4.4 1.2 & 153 & 2614 & 12.69 & -20.24 & 1.07 & 0 & 0.56 + a0101 + 4744 & dirr & 0.7 : 0.2 : & 100 & 2760 & 17.56 & -15.4 & 0.78 & + & 0.1 : + a0101 + 4752 & dirr & 0.4 : 0.3 : & 20 : & & 18.0 : & -14.9 : & 0.66 & + & 0.2 : + @xmath22 @xmath21 radial color gradient ( grd ) : positive gradient ( + ) and zero gradient ( 0 ) broad - band @xmath26 and @xmath27 frames were taken during two observing periods in 1995 and 1999 with the calar alto 1.23 m telescope and 1024 @xmath11 1024 pixel ccd camera , yielding a field of view of 8.7@xmath28 with a scale of 0.51@xmath29/px . the exposure times were 100 - 600 sec with seeing conditions typically between 1.5@xmath30 ( fwhm ) . for magnitude calibration and color correction , standard stars were observed simultaneously in the star cluster ngc 7790 ( christian et al . deep images ( 1800 sec in @xmath31 and 1200 sec in @xmath32 ) , which were taken in 1995 under nonphotometric conditions , were calibrated by means of comparing total fluxes of a sample of nonsaturated stellar images in the photometric and nonphotometric frames . after standard correction procedures the ccd frames were smoothed by means of the adaptive filtering technique described in lorenz et al . ( 1993 ) and run within the midas environment . next , the sky background was fitted by a tilted plane , using the least - squares method . after background subtraction the galaxies were interactively cleaned from disturbing objects using the interactive polygon editor . the surface brightness ( sb ) distribution was analyzed by two methods . first , we calculated the equivalent light profile by slicing the smoothed galaxy image at predefined intensity levels and counting the pixel intensities in the areas between successive isophotes . the differential counts reduced to a unit area give the azimuthally averaged intensity profile of the galaxy as a function of the equivalent radius . this procedure permits to enhance signal - to - noise ratio for the faint outskirts of the galaxies and to proceed the sb profile until @xmath33 . a set of isophotal , effective and asymptotic photometric parameters was determined on the basis of the equivalent light profiles and light growth curves . the second approach should receive some information about the two - dimensional structure . for this purpose , we used the ellipse fitting algorithm of bender & mllenhoff ( 1986 ) as available in midas . we obtained a set of radial profiles - surface brightness ( sb ) , axis ratio ( b / a ) , position angle ( pa ) and decentering ( @xmath34 and @xmath35 ) - in each particular passband . the color - index profile was calculated by means of combining these particular sb profiles . the sb profiles depend on the morphological type and bear some information about the possible environmental influences on the dynamical evolution of galaxies . preliminary inspection of the derived sb profiles reveals that those of the lsb dwarf galaxies show minor deviations from the pure exponential mostly in their inner parts , because of the occurrence of star - forming knots , if any . the outer ( generally rather noisy ) profile might represent the older , underlying disk of the dwarf galaxy and therefore provides physically meaningful parameters when fitted by an exponential density law . + the luminous group members show evidence of multiple components in their sb profiles . however , the conventional bulge / disk decomposition looks straightforward only for the parent galaxy ic 65 and the late type lsb galaxy ugc 608 , both of which show evidence of a faint bulge and/or a bar component . another bright member galaxy ugc 622 shows a typical type ii profile ( freeman 1970 ) , and the `` edge - on '' galaxy is evidently a pure late type disk galaxy . the obvious first step is to fit the sb profiles with the exponential intensity law @xmath36 . the scaling parameters ( @xmath37 ) have been obtained from the sb profiles using the `` marking the disk '' method . the disk fitting range was defined for each galaxy individually in outer part of the profile . in the case of ic 65 and ugc 608 , both of which show reliable excess light above the exponential disk model , this extra light was modeled with a sersic ( 1968 ) power law @xmath38 , which has been found proper when describing the light distribution of spiral bulges ( e.g. andredakis et al . the main photometric results are presented in the form of a ) a set of 2d images ( broad - band optical and nir image , color - index image , hi contour image ( when available ) ; b ) 1d radial profiles ( sb and color - index profiles , etc . ) ; c ) tables with conventional photometric and exponential model parameters . the 2d images show important morphological aspects such as the bulge contribution , the presence of bars and the shape of spiral arms . we apply contours of the laplacian - filtered image in order to reveal specific morphological features in luminous parts of the galaxies . for the bright galaxies we extracted individual @xmath39 and @xmath40 frames from the 2mass database , and showed the @xmath2 composite image , which characterizes the distribution of old stellar populations in these galaxies . we also showed the color - index image of each bright galaxy and of those dwarf galaxies for which reliable colors could be obtained above the noise level . it is especially useful to visualize the stellar population and dust distribution . the parameters describing the axis - ratio ( @xmath41 ) and the orientation ( @xmath42 ) of the galaxy isophotes were calculated as an average obtained from fitting ellipses typically to the range of isophotes 24 - 25.5 @xmath13 . in order to quantify the irregularities in optical morphology of the outer disk of regular galaxies , we calculated the _ decentering degree _ as the displacement of the center of the most external isophote with respect to the luminosity center , normalized to the last measured radius ( marquez & moles 1999 ) . here we present an abridged version of the obtained results for the parent galaxy ic 65 ( fig . 3 ) as well as for all four new probable dwarf members of the group ( figs . 4 - 7 ) . more results for all studied galaxies are given in vennik & hopp ( 2004 ) . to our present knowledge , groups of galaxies are a proper place to study interactions since the galaxy encounters in groups occur at lower velocities , compared to the clusters of galaxies , and may lead to efficient merging events . dominant luminous galaxies in the group can potentially reveal the history and evolutionary status of the respective system . old groups usually contain one or several luminous ellipticals , which may have formed in merging episodes and could have structures reminiscent of past interaction events . on the other hand , pure spiral groups should be dynamically younger aggregates . their members could reveal ongoing interaction through morphological distortions and in their possibly triggered star - formation activity ( tanvuia et al . the dynamically youngest groups are good places to look for galaxies evolving at @xmath43 0 . the analysis ( of available photometric and kinematic data ) for the ic 65 group of galaxies leads to the following results : @xmath44 this particular compact group of four luminous late type spiral galaxies is well isolated both in redshifts and in projection to the sky . we could identify four lsb dwarf companion candidates on deep dss 2 frames according to their sb , color and morphology . + @xmath44 available hi imaging data show that all bright members and at least one new dwarf companion are rich in hi with hi - mass to blue luminosity ratios in the range of 0.4 1.2 . their hi halos are typically twice as extended as the optical holmberg radii of individual galaxies . the outer hi isophotes ( with column densities in the range of 2.3 8.6 @xmath45 atoms @xmath46 ) generally appear disturbed , especially in ic 65 and in ugc 622 ( see fig . 1 ) . + @xmath44 optical morphology of bright galaxies generally appears to be regular , with minor disturbances in outer isophotes : e.g. in ic 65 the decentering degree is @xmath47 8.2 % and in ugc 622 @xmath47 6.3% , which are significantly larger than the average decentering of 2.4 @xmath48 2.7 % , found by marquez & moles ( 1999 ) for isolated spirals . + @xmath44 all bright group members ( except pgc 138291 , which we could not study in such detail ) consist of many blue star - forming knots and plumes , especially ugc 608 . + @xmath44 ugc 622 is a scd galaxy without a classical bulge , but containing a tiny bright nucleus . it shows type ii sb profile with two exponential sections of different scale lengths . + @xmath44 new probable dwarf companions are all of irregular appearance with blue knots , especially a 0101 + 4744 and a 0100 + 4734 , for which a chain of blue sf patches give a cometary appearance . + @xmath44 significant positive color gradients were found in lsb dwarf galaxies of short scale lengths , and they are related to starbursts . late type dwarfs generally show a stellar population gradient , i.e. younger and bluer stars are more centrally concentrated and redder stars dominate the underlying older disk component . morphologically disruptive interactions are expected to be more common in low - velocity - dispersion ( @xmath49 ) less evolved groups such as the local group and the m 81 group ( zabludoff 2001 ) . the ic 65 group of galaxies obviously belongs to the class of less evolved groups : with its @xmath49 = 70 km / s , observed drift in optical isophotes and obvious tidal features in hi outer isophotes , as well as with the evidence of enhanced sf activity in fragile dwarf companion candidates let us conclude that the ic 65 group of galaxies is probably dynamically young and possibly at the stage of its collapse . andredakis , y.c . , peletier , r.f . , balcells , m. , 1995 , mnras , 275 , 874 + bender r. , mllenhoff c. , 1987 , a&a 177 , 71 + bertin , e. , arnouts , s. , 1996 , a&as , 117,393 + binggeli b. , 1994 , in : g.hensler , c.theis , j.s.gallagher ( eds ) , `` the panchromatic view of galaxies '' , editions frontier , p. 173 + christian c.a . , adams m. , barnes j.v . , butcher h. , mould j.r . , siegel m. , 1985 , pasp 97 , 363 + ferguson , h.c . , binggeli , b. , 1994 , a&arv . , 6 , 67 + freeman , k.c , 1970 , apj , 160 , 811 + garcia , a.m. , 1993 , a&as , 100,47 + lorenz h. , richter g.m . , capaccioli m. , longo g. , 1993 , a&a 277 , 321 + marquez , i. , moles , m. , 1999 , a&a , 344 , 421 + sersic , j.l . , 1968 , atlas de galaxias australes . cordoba + tanvuia , l. , kelm , b. , focardi , p. , rampazzo , r. , zeilinger , w.w . , 2003 , aj , 126 , 1245 + trentham , n. , tully , r.b . , verheijen , m.a.w . , 2001 , mnras , 325 , 385 + van moorsel , g.a . , 1983 , a&as , 54 , 1 + vennik , j. , hopp , u. , 2004 , in preparation + zabludoff , a. , 2001 , asp conf . proc . 240 , 547 +	in the framework of a larger program for studying the photometric properties of nearby groups of galaxies we have carried out surface photometry of four bright known members and a number of new possible lsb companions of the ic 65 ( lgg 16 ) group of galaxies . the purpose of the present study is twofold : ( a ) a search for new lsb dwarf members and measuring of their global photometric characteristics , and ( b ) a search for possible effects of mutual interactions on the morphology and photometric characteristics of luminous and faint group members . this study is based on our @xmath0 ccd observations performed with the 1.23 m and 2.2 m telescopes at calar alto , which were supplemented by the deep @xmath1 frames obtained from the second palomar digital sky survey ( dss 2 ) , and the nir @xmath2 frames from the 2mass database . in addition , we have used the hi imaging data from literature and the results of new pilot hi observations of selected dwarf - galaxy candidates made by w. huchtmeier with the effelsberg 100 m radiotelescope . in total , 105 galaxies have been detected with sextractor software on the linearized dss 2 @xmath3 blue and red frames centered on lgg 16 , and classified by their surface brightness , color and morphology . we have discovered four new probable lsb dwarf members of the group , two of which have been marginally detected in hi . the isophotal analysis in optical wavelengths revealed the presence of minor disturbances in the outer isophotes of some bright regular member galaxies . the hi emission appears generally more disturbed , compared to the emission in the optical passbands . the new dwarf companion candidates ( with @xmath4 ) generally contain blue star - forming regions with typical ( cousins ) colours of @xmath5 . the drift of optical isophotes together with the warping of hi isophotes outside the optical image of regular galaxies , as well as evidence for the enhanced star - forming activity in fragile dwarf galaxies indicate that this dense group of at least 8 late type galaxies is gravitationally bound and mutually interacting . 0.1 cm ( 150,200 )
in order to determine the positions and motions of astronomical objects on the sky , astrometry uses light signals ( photons ) which are emitted by the celestial objects . these light rays propagate from the celestial light source through the gravitational field of the solar system and do finally arrive at the observer . therefore , the precise determination of the trajectories of light signals through the warped space - time of solar system is a fundamental assignment of a task in relativistic astrometry . according to the theory of general relativity @xcite light rays propagate along null - geodesics governed by the geodesic equation , @xmath0 where ( [ geodetic_equation1 ] ) represents the geodesic equation and the constraint ( [ isotropic_condition1 ] ) must be imposed for null - geodesics which states that the tangent four - vector along light rays is isotropic . in ( [ geodetic_equation1 ] ) and ( [ isotropic_condition1 ] ) the four - coordinates of a light - signal @xmath1 depend on affine parameter @xmath2 , and the christoffel symbols in ( [ geodetic_equation1 ] ) are related to the metric @xmath3 of curved space - time , @xmath4 with metric signature @xmath5 . the geodesic equation ( [ geodetic_equation1 ] ) and the isotropic condition ( [ isotropic_condition1 ] ) are valid in any reference system . with the aid of the zeroth component of ( [ geodetic_equation1 ] ) , the geodesic equation and the isotropic condition can be expressed in terms of coordinate time @xmath6 rather than the affine parameter @xmath2 as follows @xcite , @xmath7 while the zeroth component in ( [ geodetic_equation2 ] ) vanishes identically . the equations in ( [ geodetic_equation2 ] ) and ( [ isotropic_condition2 ] ) are more appropriate in order to integrate the geodesic equation and also in view of the fact that real astrometric measurements do by all means imply the use of concrete reference systems . in line with the resolutions of international astronomical union ( iau ) @xcite , the barycentric celestial reference system ( bcrs ) is adopted , which is the standard global chart in modern - day astrometry . the origin of the spatial axes of the bcrs is located at the barycenter of the solar system , the harmonic coordinates of the bcrs are denoted by @xmath8 where @xmath6 is the bcrs coordinate time and @xmath9 are the three - dimensional coordinates referred to the spatial axes of the bcrs , and obey the harmonic gauge condition ( de donder gauge ) : @xmath10 where @xmath11 is the determinant of metric tensor . for a unique solution of the geodesic equation ( [ geodetic_equation2 ] ) mixed initial - boundary conditions must be imposed @xcite : @xmath12 where the dot in ( [ initial_boundary_condition_2 ] ) denotes total derivative with respect to coordinate time . the first condition ( [ initial_boundary_condition_1 ] ) defines the spatial coordinates of the photon at the moment @xmath13 of emission of light . the second condition ( [ initial_boundary_condition_2 ] ) defines the unit - direction of the light ray at past null infinity , that means the unit - tangent vector along the light path in the infinite past hence at infinite spatial distance from the origin of the global coordinate system . then , the exact solution of ( [ geodetic_equation2 ] ) for the trajectory of the light ray , propagating from the light source through the solar system towards the observer , can formally be written as follows , @xmath14 where the term @xmath15 denotes gravitational corrections to the unperturbed light ray . in case of weak gravitational fields it is useful to decompose the metric tensor as follows , @xmath16 where @xmath17 is the metric of minkowskian space and for any components of the metric perturbations @xmath18 . because the gravitational fields are weak in the solar system , the orbital motions of the solar system bodies are slow ( virial theorem ) , @xmath19 and @xmath20 ( notations are given in the appendix [ appendix_notation ] ) , hence an expansion of the metric in terms of inverse powers of the speed of light can be applied , called post - newtonian expansion or weak - field slow - motion approximation @xcite , which for the covariant and contravariant components reads @xmath21 where @xmath22 with @xmath23 ; e.g. eqs . ( 4.17 ) - ( 4.19 ) in @xcite . notice that the post - newtonian expansion in ( [ post_newtonian_metric_b ] ) describes the metric in the near - zone of the solar system defined by @xmath24 where @xmath25 is a characteristic wavelength of gravitational radiation emitted by the solar system . it should be mentioned that , according to the famous theorem in @xcite , the post - newtonian expansion of the metric tensor is , in fact , non - analytic because it contains logarithmic terms . however , in the near - zone the post - newtonian expansion in inverse powers of the speed of light is valid up to 4pn approximation , that means logarithmic terms in metric coefficients emerge at the order of @xmath26 @xcite . the post - newtonian expansion of the metric in ( [ post_newtonian_metric_b ] ) inherits a corresponding post - newtonian expansion of the light trajectory ( [ introduction_5 ] ) , which up to terms of the order @xmath27 reads @xmath28 where the label @xmath29 , @xmath30 , and @xmath31 refer to terms of the order @xmath32 , @xmath33 , and @xmath34 , respectively . the expressions for @xmath35 and @xmath36 for a light trajectory in the field of @xmath37 arbitrarily moving bodies of finite size have recently been determined in @xcite . in these investigations each individual body @xmath38 is allowed to move along its own arbitrary worldline @xmath39 and the global metric of the solar system has been described in terms of the full set of time - dependent intrinsic mass - multipoles @xmath40 and full set of time - dependent intrinsic spin - multipoles @xmath41 , allowing for arbitrary shape , inner structure and rotational motion of the massive bodies of the solar system . about the magnitude of these terms in time - delay and light deflection we refer to table ii and table iii in @xcite . however , rapidly growing accuracy in astrometric measurements demands to account for post - post - newtonian terms @xmath42 as well . in particular , it is well - known that present - day precision in astrometry has reached a level of a few micro - arcseconds ( @xmath43 ) in angular observations of stars @xcite and a level of a few nano - seconds ( @xmath44 ) in measurements of time delay @xcite . such extremely high - precision astrometry necessitates to account for 2pn effects in the theory of light propagation @xcite . on the other side , results about the post - post - newtonian terms @xmath42 in ( [ introduction_6 ] ) are fairly rare . so far , 2pn effects in light propagation have mainly been determined for the case of mass monopoles at rest @xcite , that means where the position of the mass monopole remains constant : @xmath45 . in this respect , an important progress in calculating post - post - newtonian effects on light propagation in the monopole field has been achieved in @xcite where an explicit 2pn solution for light trajectories in the schwarzschild field as function of coordinate time has been found and later been confirmed within several progressing investigations @xcite . also alternative approaches for the calculation of directions of light rays and their propagation time in 2pn approximation have been developed , which avoid the peculiarities of solving the null geodesic equations , based on the eikonal concept @xcite , on the synge s world function @xcite or on the time transfer function formalism @xcite . an ambitious goal in astrometric measurements in near future is to aim at sub - micro - arcsecond ( sub- ) level in angular determination and sub - nano - second ( sub - ns ) level in time delay measurements . for instance , several space - based astrometry missions are under discussion which have been proposed to the european space agency ( esa ) which aim at precisions on sub - nano - arcsecond ( sub - nas ) level in angular determination of celestial objects @xcite . such extremely high - precision astrometry needs to account for the impact of the motion of massive bodies on the light propagation in 2pn approximation . the problem , however , of light propagation in the field of moving monopoles in 2pn approximation has not been considered yet , aside from the investigation in @xcite which was not intended for the problem of light propagation in the solar system . for this reason , we will consider the problem of light propagation through the gravitational field of one pointlike body in slow but otherwise arbitrary motion in the 2pn approximation . the article is organized as follows : in section [ section2 ] the geodesic equation in 2pn approximation is presented , in section [ section3 ] the metric of one massive pointlike body in arbitrary motion in 2pn approximation is given , in section [ section4 ] and section [ section5 ] the first and second integration of geodesic equation is represented . the observable effects of total light deflection and time delay are given in section [ observables ] . a summary and outlook can be found in section [ section6 ] . the notation in use is given in the appendix [ appendix_notation ] . the solar system is composed of @xmath37 massive bodies of finite size which move according their mutual gravitational interaction . in our investigation we will consider one of these massive bodies and approximate the body as pointlike object with newtonian rest mass @xmath46 . by inserting the metric ( [ post_newtonian_metric_b ] ) into ( [ geodetic_equation2 ] ) we obtain the geodesic equation in 2pn approximation , which in terms of global coordinate time reads @xcite @xmath47 where we have taken into account that in general @xmath48 @xcite . the last term in ( [ geodesic_equation3 ] ) , i.e. the term @xmath49 , is a peculiarity in the sense that this term is seemingly of the order @xmath27 , but by inspection of ( [ metric_4 ] ) one realizes that the first integration ( [ integral_1 ] ) of this term results into @xmath50 which is of the order @xmath34 , and , therefore , can not be neglected . furthermore , the following relations have been used , @xmath51 which result from @xmath52 . the metric perturbations in ( [ post_newtonian_metric_b ] ) are functions of the field - points @xmath53 , while in the geodesic equation ( [ geodesic_equation3 ] ) the metric perturbations are of relevance at the coordinates of the photon @xmath54 . consequently , the derivatives in ( [ geodesic_equation3 ] ) are taken along the light ray : @xmath55 the geodesic equation in 2pn approximation in ( [ geodesic_equation3 ] ) can be solved by iteration and allows to determine the coordinate velocity ( first integration ) and the light trajectory ( second integration ) up to terms of the order @xmath27 : @xmath56 as mentioned above , the 1pn and 1.5pn terms in ( [ light_trajectory_2pn_a ] ) and ( [ light_trajectory_2pn_b ] ) have recently been determined in @xcite and @xcite , respectively , for the case of @xmath37 bodies in slow but otherwise arbitrary motion and the bodies may have arbitrary shape and inner structure and can be in arbitrary rotational motion . the aim of this investigation is to determine the 2pn terms for the case of one pointlike body in arbitrary slow - motion . ( [ second_integration ] ) through the gravitational field of one pointlike massive body @xmath57 moving along an arbitrary worldline in slow motion @xmath58 . at the same instant of coordinate time the body s position is @xmath39 ( gray sphere ) . however , since gravitational action travels with the finite speed of light , the light ray at @xmath54 is influenced by the gravitational field generated by the body at its retarded position @xmath59 ( black sphere ) . the spatial vector @xmath60 is defined by eq . ( [ retarded_time_2 ] ) and points from the massive body @xmath57 at its retarded position toward the exact photon s position at instant @xmath6 . ] we shall assume that the one - body system is isolated ( fock - sommerfeld boundary conditions ) , that means flatness of the metric at spatial infinity and the constraint of no - incoming gravitational radiation is imposed at minkowskian past null infinity @xmath61 , which in terms of trace - reversed metric perturbation @xmath62 read as follows @xcite , @xmath63 where @xmath64 . in addition , @xmath65 should be bounded in this limit @xcite , that means any component of the metric tensor obeys the constraint @xmath66 where @xmath67 is some positive number related to the total rest mass of the gravitational system . according to ( [ post_newtonian_metric_b ] ) and eqs . ( [ g00 ] ) - ( [ metric_w3 ] ) , the metric perturbations for the gravitational fields of one pointlike body in slow motion read : @xmath68 @xmath69 where @xmath70 and @xmath71 while its absolute value @xmath72 , and we introduce the unit - vector @xmath73 . the constraints ( [ asymptotic_1 ] ) - ( [ asymptotic_3 ] ) restrict the time - dependence of the acceleration so that it vanishes at past null infinity : @xmath74 . the bcrs metric coefficients @xmath75 in the mass - monopole approximation for @xmath37 slowly moving bodies were given by eqs . ( 8) and ( 51 ) - ( 55 ) in @xcite . the same metric coefficients were also given by eqs . ( 39.63a ) - ( 39.63c ) in @xcite . in the limit of one monopole in slow - motion they agree with our metric coefficients in eqs . ( [ metric_1 ] ) , ( [ metric_2 ] ) , ( [ metric_3 ] ) and ( [ metric_5 ] ) . we also notice that in the limit of one body @xmath57 at rest , the metric in ( [ metric_1 ] ) - ( [ metric_6 ] ) agrees with the metric in eqs . ( 25 ) in @xcite if the body is assumed to be located at the origin of the global reference system . for further details consult @xcite and the appendix [ appendix_metric ] . the three - vector @xmath76 in ( [ vector_a ] ) is an arbitrary spatial field - point . but according to ( [ geodesic_equation_3 ] ) , as soon as the partial derivatives in the geodesic equation ( [ geodesic_equation3 ] ) are performed , the field - point @xmath76 has to be identified with the exact spatial position of the light signal @xmath54 , that means after all partial derivatives are performed we have @xmath77 of course , one has strictly to distinguish between ( [ vector_a ] ) and ( [ vector_b ] ) , but nevertheless the same notation @xmath78 for these expressions is in use and will certainly not cause any confusion . the first integration of geodesic equation yields the coordinate velocity of the photon , @xmath79 where @xmath80 is given by ( [ geodesic_equation3 ] ) and the boundary condition ( [ initial_boundary_condition_2 ] ) must be imposed . the geodesic equation ( [ geodesic_equation3 ] ) is solved by iteration , that means in first iteration the integration is performed along the unperturbed light ray and in the second iteration the integration proceeds along the light ray in 1pn approximation . owing to the fact that the metric , thence the geodesic equation , depends on the arbitrary worldline of the body , a solution of geodesic equation is obtained by means of integration by parts with respect to coordinate time . one may show that , after a finite set of partial integrations , the remaining non - integrated terms of such an approach are terms beyond 2pn approximation . the solution for the coordinate velocity of the photon is , first of all , given in terms of the spatial position of the massive body at coordinate time , @xmath39 . since gravitational action propagates with the finite speed of light , it is meaningful to reexpress this solution in terms of retarded time of the massive body s position @xmath59 . the retarded time is defined by an implicit relation , @xmath81 where @xmath82 with @xmath83 and @xmath54 being the exact photon trajectory . further details are given in the appendix [ appendix_retarded_time ] . accordingly , we may define an impact vector of the incident light ray associated with the body s position at retarded instant of time , given by @xmath84 and weak gravitational field means @xmath85 where @xmath86 . for grazing light rays the impact vector at the retarded position equals the radius of the massive body , while in general it will be larger : @xmath87 . it should also be remarked that the solution for the coordinate velocity of the photon takes the most simple form in terms of @xmath59 . in this way one obtains the following 2pn solution for the photon s coordinate velocity in the field of one arbitrarily moving pointlike body : @xmath88 where the arguments of the vectorial functions are given in appendix [ appendix_1pn_solution ] . one may demonstrate that in case of body at rest ( [ first_integration ] ) agrees with @xcite , and the terms up to order @xmath34 agree with @xcite . in view of the complexity of the 2pn solution , all those terms in ( [ vectorial_function_a2 ] ) have been combined in some small parameter @xmath89 which can be estimated by @xmath90 which amounts to be less than @xmath91 for grazing light rays at jupiter and even less for all the other solar system bodies . we emphasize that the solution in ( [ first_integration ] ) does not depend on the acceleration of the massive body , as long as it is given in terms of the retarded position of the massive body . this fact is related to the case of an arbitrarily moving and radiating electron in classical electrodynamics where the radiation fields do not depend on the acceleration of the electron if the worldline of the electron is expressed in terms of its retarded position @xcite . finally , we notice that the vectorial function @xmath92 in eq . ( [ vectorial_function_a1 ] ) agrees with eq . ( 46 ) in @xcite and the vectorial function @xmath93 in eq . ( [ vectorial_function_a3 ] ) agrees with eq . ( 48 ) in @xcite ( in general theory of relativity @xmath94 ) . the second integration of geodesic equation yields the light trajectory of the light signal , @xmath95 where @xmath96 is given by ( [ first_integration ] ) and the boundary condition ( [ initial_boundary_condition_1 ] ) must be imposed . note that ( [ first_integration ] ) is given in terms of retarded time , but in order to proceed with the integration in ( [ integral_2 ] ) all terms in ( [ first_integration ] ) must have to be reexpressed in terms of coordinate time by means of relations ( [ retarded_time_b ] ) - ( [ retarded_time_d ] ) . like in case of the first integration in ( [ integral_1 ] ) , the second integration in ( [ integral_2 ] ) is performed by iteration . furthermore , since the worldline of the body remains arbitrary , @xmath39 , the integration is performed by means of integration by parts . in this way , one obtains the light trajectory in terms of the spatial position of the massive body at coordinate time , @xmath39 , but can be rewritten in terms of retarded time of the position of massive body , @xmath59 , which is also from the physical point of view more appropriate because gravitational action travels with the finite speed of light . altogether , we obtain the following 2pn solution for the photon s trajectory in the field of one arbitrarily moving pointlike body , for an illustration see fig . [ diagram ] : @xmath97 where the arguments of the vectorial functions are given in appendix [ appendix_1pn_solution ] . one may demonstrate that in case of body at rest ( [ second_integration ] ) agrees with @xcite , and the terms up to order @xmath34 agree with @xcite . in view of the complexity of the 2pn solution , all those terms in ( [ vectorial_function_b2 ] ) have been combined in some small parameter @xmath98 which can be estimated by @xmath99 which yields @xmath100 for jupiter and even less for all the other solar system bodies . we emphasize that the solution in ( [ second_integration ] ) does not depend on the acceleration of the massive body , as long as the solution is given in terms of the retarded position of the massive body . this important fact resembles the case of an arbitrarily moving electron , where its radiation field does not depend on the acceleration as long as the worldline of the electron is given in terms of its retarded position @xcite . we notice that the vectorial function @xmath101 in eq . ( [ vectorial_function_b1 ] ) agrees with eq . ( 50 ) in @xcite and the vectorial function @xmath102 in eq . ( [ vectorial_function_b3 ] ) agrees with eq . ( 51 ) in @xcite ( in general theory of relativity @xmath94 ) . in this section we briefly consider the observable effects of total light deflection and time delay which are of upmost relevance for astrometry and belong to the classical tests of relativity . the total light deflection of a light signal propagating through the gravitational field of one arbitrarily moving body is defined by the angle between the coordinate light velocity at @xmath103 . from ( [ first_integration ] ) we first of all obtain up to terms of the order @xmath27 : @xmath104 in the limit of one body at rest and located at the origin of reference system , the expression in ( [ total_light_deflection_nu ] ) agrees with eq . ( 64 ) in @xcite . the impact vector @xmath105 and the impact vector @xmath106 are related to each other subject to ( [ solution_1pn_a ] ) . then , from ( [ total_light_deflection_nu ] ) one obtains for the total light deflection up to terms of the order @xmath27 : @xmath107 , \nonumber\\ \label{light_deflection}\end{aligned}\ ] ] where @xmath108 and @xmath109 and terms @xmath110 have been omitted in ( [ light_deflection ] ) in view of the estimate in ( [ estimate_epsilon1 ] ) . furthermore , in ( [ light_deflection ] ) the impact vector at @xmath111 and @xmath112 of the unperturbed light ray has been used : @xmath113 which in the case of a motionless body at the origin of coordinate system coincides with the impact vector defined by eq . ( 55 ) in @xcite . the expression in ( [ light_deflection ] ) depends on the direction of the light ray @xmath114 , on the coordinates of the light source @xmath115 and on the mass , position and velocity of the massive body @xmath116 and it generalizes the corresponding 2pn expression for a body at rest @xcite , cf . ( 3.2.44 ) in @xcite or eq . ( 65 ) in @xcite . the occurrence of the third term in the brackets in ( [ light_deflection ] ) is caused by the fact that the total light deflection , which is a coordinate - independent observable , is expressed in terms of coordinate - dependent quantities . this assertion can be shown by introducing a coordinate independent impact vector similar to the one given by eq . ( 57 ) in @xcite . but , as emphasized above , the use of concrete reference systems is inevitable in real astrometric data reduction . a light signal which propagates through the curved space of a massive body takes a longer time to travel from one space - time point to another space - time point compared to the flat minkowskian space . let s assume the light source and the observer to be located at @xmath117 and @xmath118 , respectively , and to be at rest with respect to the global reference system , and we may define a spatial distance @xmath119 . then , from ( [ second_integration ] ) one obtains the following expression for the time delay up to terms of the order @xmath27 : @xmath120 where @xmath121 ; the terms @xmath122 were neglected in view of the estimate in ( [ estimate_epsilon2 ] ) . the expression in ( [ shapiro ] ) generalizes the corresponding 2pn expression for one monopole at rest @xcite and it generalizes the expression in eqs . ( 146 ) - ( 148 ) in @xcite which is valid for arbitrarily moving monopoles but in 1.5pn approximation . present - day astrometry has reached a level of a few micro - arcseconds in angular determination of celestial objects and prospective astrometry aims at sub - micro - arcsecond or even nano - arcsecond level of precision . associated therewith is the precise determination of light trajectories through the warped space - time of the solar system as one central issue in relativistic astrometry . an exact solution for the light ray is , however , not possible because of the involved structure of the metric of the solar system and one has , therefore , to resort on approximation schemes . the gravitational fields of the solar system are weak , @xmath19 , and the velocities of the bodies are slow , @xmath20 , so that an expansion of the metric tensor of the solar system in inverse powers of the speed of light becomes meaningful as given by eq . ( [ post_newtonian_metric_b ] ) , @xmath123 up to terms of the order @xmath124 . this so - called post - newtonian expansion ( weak - field slow - motion approximation ) implicitly assumes that all retardations are small , which is well - justified inside the near - zone of the solar system . a corresponding expansion of the light trajectory is given by eq . ( [ introduction_6 ] ) and reads @xmath125 up to terms of the order @xmath124 . one of the most intricate problems in the relativistic theory of light propagation concerns the impact of the motion of the massive bodies on light trajectory . in recent investigations @xcite the 1pn and 1.5pn terms , @xmath35 and @xmath36 , have been determined for the case of @xmath37 arbitrarily moving bodies having full mass - multipole and spin - multipole structure . the rapid advance in astrometric measurements enforces one to account for post - post - newtonian effects @xmath42 in the theory of light propagation as well . the 2pn terms in ( [ summary_2 ] ) are only known for the case of one monopole at rest , first been determined in @xcite and later been confirmed within several ongoing investigations @xcite . but little is known about these terms in ( [ summary_2 ] ) for the case of moving bodies . so far , the only investigation in 2pn approximation regarding light trajectory in the field of moving bodies has been performed in @xcite which was , however , not intended for light propagation inside the solar system . in our investigation , the problem of light propagation in the field of one arbitrarily moving pointlike monopole has been considered . especially , an analytical solution in post - post - newtonian approximation for coordinate velocity @xmath96 and trajectory @xmath54 of the light ray is presented . according to the recommendations of iau @xcite the metric is given in terms of harmonic coordinates . because of the fact that the worldline @xmath39 of the massive body is arbitrarily , an integration of the geodesic equation in ( [ geodesic_equation3 ] ) is only possible by means of integration by parts . the first integration ( [ integral_1 ] ) and the second integration ( [ integral_2 ] ) has been performed in terms of coordinate time . in this respect one has to keep in mind that the post - newtonian expansion of the metric ( [ summary_1 ] ) and of the light ray ( [ summary_2 ] ) inherits that all retardations are small , but they are not negligible . instead , the fact remains that gravitational action travels with the speed of light also inside the near - zone of the solar system . the phrase _ smallness of retardation effects _ in the solar system means that a series expansion of the retarded time is meaningful , as given by eqs . ( [ retarded_time_a2 ] ) - ( [ retarded_time_d ] ) . by means of these relations the solution , first of all given in terms of the instantaneous position of the body @xmath39 , can be expressed in terms of the retarded position of the body @xmath59 , where the first integration and the second integration of geodesic equation adopt the most simple form , as given by eqs . ( [ first_integration ] ) and ( [ second_integration ] ) , respectively . the expressions for the observables of total light deflection and of shapiro time delay are given by eqs . ( [ light_deflection ] ) and ( [ shapiro ] ) . the case of @xmath37 arbitrarily moving pointlike bodies is rather involved and needs special consideration . but in view of the fact that the impact of two - body effects on light deflection is less than @xmath126 in the solar system @xcite , one might assert that the 2pn light trajectory in the field of @xmath37 pointlike monopoles in arbitrary slow - motion can be obtained from our solution just by a summation over @xmath37 individual bodies , at least for an envisaged accuracy on nas - level . these aspects will have to be scrutinized within more detailed prospective analyses . this work was supported by the deutsche forschungsgemeinschaft ( dfg ) . throughout the article the following notation is in use . * @xmath127 is the newtonian constant of gravitation . * @xmath128 is the vacuum speed of light . * @xmath46 denotes the rest mass of body @xmath57 . * @xmath70 is the schwarzschild radius . * @xmath129 denotes the equatorial radius of body @xmath57 . * @xmath130 denotes the orbital velocity of massive body @xmath57 . * @xmath131 . + * @xmath132 . * lower case latin indices take values 1,2,3 . * lower case greek indices take values 0,1,2,3 . * the three - dimensional coordinate quantities ( three - vectors ) referred to the spatial axes of the reference system are in boldface : @xmath133 . * the contravariant components of three - vectors @xmath134 . * the contravariant components of four - vectors @xmath135 . * the absolute value of a three - vector . @xmath136 . * the scalar product of two three - vectors . @xmath137 with kronecker delta @xmath138 . * the vector product of two three - vectors . reads @xmath139 with levi - civita symbol @xmath140 . gravitational action travels with the finite speed of light which implies that the gravitational field at some field point @xmath76 is generated by the pointlike body at its position @xmath59 at the retarded instant of time defined by @xmath141 where @xmath142 . in the near - zone of the solar system @xcite one may assume that all retardations are small , hence a series expansion of ( [ retarded_time_a1 ] ) becomes meaningful , @xmath143 which will later be used for the series expansion of the metric tensor . using ( [ retarded_time_a2 ] ) and the series expansion of the retarded position of the body which up to terms of the order @xmath33 reads @xmath144 we find the following relations : @xmath145 where @xmath146 and @xmath147 is the velocity and acceleration of the body , respectively . these relations agree with eqs . ( 47 ) - ( 49 ) in @xcite up to the term proportional to the acceleration @xmath148 . these relations have been obtain for @xmath149 , but we notice that the relations in ( [ retarded_time_a2 ] ) and ( [ retarded_time_b ] ) - ( [ retarded_time_d ] ) remain its validity for @xmath150 , because they root on the expansion in ( [ retarded_position ] ) . the series - expansions in ( [ retarded_time_b ] ) - ( [ retarded_time_d ] ) are useful as long as the retardations are small which is well - justified in the near - zone of the solar system . it especially constraints the accelerations , @xmath151 for any moment of time . for our intention we need the 2pn metric in harmonic gauge ( [ harmonic_gauge ] ) for the case of one arbitrarily but slowly moving pointlike monopole . the 2pn metric contains terms proportional to @xmath127 and terms proportional to @xmath152 , which are considered in what follows . the terms proportional to @xmath127 can easily be obtained from the metric for one arbitrarily moving pointlike monopole in post - minkowskian approximation , which has been given by eq . ( 10 ) in @xcite , by eq . ( 11 ) in @xcite , and also by eq . ( 43 ) in @xcite : @xmath153 where @xmath154 is the lorentz factor and @xmath155 denotes monopole . the vector pointing from the retarded position @xmath59 of the body @xmath57 towards the field - point @xmath76 reads : @xmath156 let us note that in ( [ arbitrarily_moving_body_4 ] ) we present the covariant components of the metric perturbations , while in @xcite the contravariant components have been used . accordingly , the covariant components of the four - velocity of the body are @xmath157 , and @xmath158 is the three - velocity of the body in the global system . let us also draw the attention to the fact , that the metric tensor in ( [ arbitrarily_moving_body_4 ] ) does not depend on the acceleration but only on the velocity of body @xmath57 because it is given in terms of retarded time , see also comment below eq . ( [ estimate_epsilon1 ] ) . the metric in ( [ arbitrarily_moving_body_4 ] ) is valid for an arbitrarily moving body which could even be in ultra - relativistic motion . we are interested in the case of a slowly - moving body , and a corresponding series expansion of ( [ arbitrarily_moving_body_4 ] ) in terms of the small parameter @xmath20 yields up to order @xmath27 : @xmath159 the retarded time - argument in ( [ comparison_5 ] ) - ( [ comparison_15 ] ) has to be replaced by the global coordinate time using the relations in ( [ retarded_time_b ] ) - ( [ retarded_time_d ] ) . before going further , one has to realize that the acceleration of some body @xmath57 is proportional to @xmath127 according to the equations of motion for @xmath37 pointlike bodies , @xmath160 where the terms of order @xmath32 are given by the einstein - infeld - hoffmann equations @xcite . here , however , we can not use the equations of motion ( [ equations_of_motion ] ) because we consider the metric of one body @xmath57 in arbitrary motion and the physical origin of the motion of the body is not relevant for the moment being . especially , we do not have some kind of equations of motion like in an @xmath37-body system ( just imagine accelerating rockets tied to that body ) . therefore , we are enforced to keep the acceleration terms explicitly in eqs . ( [ retarded_time_b ] ) - ( [ retarded_time_d ] ) . of course , if one would go back and consider an @xmath37-body system under the influence of their mutual gravitational interaction , then one could make use of the equations of motion ( [ equations_of_motion ] ) and then such an acceleration term would appear as term of the order @xmath152 in the metric tensor . according to these considerations , to order @xmath127 we obtain : @xmath161 while @xmath162 and @xmath163 we recognize that in ( [ appendix_metric_4 ] ) - ( [ appendix_metric_6 ] ) there are terms proportional to the acceleration of the body . the metric in eqs . ( [ appendix_metric_1 ] ) - ( [ appendix_metric_6 ] ) agrees with the metric given by eqs . ( 7.2a ) - ( 7.2c ) in @xcite for all terms proportional to @xmath127 and up to the order @xmath27 . but we notice that in eqs . ( 7.2a ) - ( 7.2c ) in @xcite there are no acceleration terms , because they have been rewritten by means of the equations of motion of an @xmath37-body system ( [ equations_of_motion ] ) , cf . ( 3.11 ) in @xcite and the text above that equation . another point to mention concerns the expression in ( [ appendix_metric_5 ] ) . for an @xmath37-body system it is a strict law that there are no terms to power @xmath164 in @xmath165 @xcite , because in an @xmath37-body system , instead of ( [ appendix_metric_5 ] ) , we would have a summation over all bodies , @xmath166 where @xmath167 is the total newtonian momentum of the @xmath37-body system , which is strictly conserved to order @xmath32 , that means @xmath168 @xcite . therefore , in an @xmath37-body system there is in fact no term to power @xmath164 in @xmath165 @xcite . but in our case of one single body which moves along an arbitrary worldline without to resort on the equations of motion ( [ equations_of_motion ] ) , there is no conservation of total newtonian momentum , hence we have to keep that term in ( [ appendix_metric_5 ] ) . nevertheless , there is an important difference regarding the acceleration terms : in 2pn approximation the acceleration term in ( [ appendix_metric_5 ] ) would disappear for an @xmath37-body system , while the acceleration term in ( [ appendix_metric_4 ] ) and ( [ appendix_metric_6 ] ) could be rewritten in a form proportional to @xmath152 , but they remain to be of the order @xmath34 , hence they would not disappear for an @xmath37-body system in 2pn approximation . the metric of a system of two pointlike bodies under the influence of their mutual gravitational interaction has been determined in 2.5pn approximation in @xcite , that means @xmath169 , @xmath170 and @xmath171 up to terms of the order @xmath26 , @xmath172 , and @xmath173 , respectively . recently , the metric of @xmath37 pointlike bodies has been determined in @xcite in 2pn approximation for the light rays , that is : @xmath3 up to terms of the order @xmath27 . in order to find all terms proportional to @xmath152 , we may issue the results from ref . @xcite , but have to take the limit @xmath174 for all bodies except body @xmath57 . in this way we obtain from eqs . ( 47 ) - ( 49 ) ( with @xmath175 ) in @xcite : @xmath176 the metric perturbations @xmath177 are given by @xmath178 with @xmath179 given by ( [ appendix_metric_4 ] ) - ( [ appendix_metric_6 ] ) and @xmath180 given by ( [ appendix_metric_g2_1 ] ) and ( [ appendix_metric_g2_2 ] ) . let us draw the attention to the fact , that if we insert the equations of motion ( [ equations_of_motion ] ) into the last term in ( [ appendix_metric_4 ] ) , then we would get the next - to - last term in eq . ( 47 ) in @xcite . similarly , if we insert the equations of motion ( [ equations_of_motion ] ) into the last term in ( [ appendix_metric_6 ] ) , then we would get the next - to - last term in the second line of eq . ( 49 ) in @xcite . but we emphasize again , that we are not allowed to do such kind of replacements ( which in @xcite were called order - reduced form of the metric , cf . text above eq . ( 3.11 ) ibid ) , because we do not consider an @xmath37-body system but a system of one body which moves arbitrarily along its worldline without taking resort on the equations of motion in ( [ equations_of_motion ] ) . collecting all metric coefficients in eqs . ( [ appendix_metric_1 ] ) - ( [ appendix_metric_6 ] ) and in eqs . ( [ appendix_metric_g2_1 ] ) - ( [ metric_g_gg ] ) , the post - post newtonian metric for light rays can also expressed in terms of so - called potentials ( @xmath181 , @xmath182 , @xmath183 ) in the following form , cf . ( a1 ) - ( a3 ) in @xcite or eq . ( 2 ) in @xcite : @xmath184 where the potentials read @xmath185 recalling @xmath186 . the newtonian and the first post - newtonian solution for the light ray appears as argument in the vectorial functions of the 2pn solution in ( [ first_integration ] ) and ( [ second_integration ] ) . in this appendix we will present these expressions . in newtonian approximation we have @xmath187 furthermore , the light trajectory in the field of one arbitrarily moving body in the first post - newtonian approximation can be obtained from @xcite by means of relations ( [ retarded_time_b ] ) - ( [ retarded_time_c ] ) and reads : @xmath188 m. soffel , s.a . klioner , g. petit , et al . , _ the iau 2000 resolutions for astrometry , celestial mechanics and metrology in the relativistic framework : explanatory supplement _ , astron . j. * 126 * ( 2003 ) 2687 . richter , r.a . matzner , _ second - order contributions to gravitational deflection of light in the parametrized post - newtonian formalism . ii . photon orbits and deflections in three dimensions _ , d * 26 * ( 1982 ) 2549 . a. hees , s. bertone , c. le poncin - lafitte , _ relativistic formulation of coordinate light time , doppler and astrometric observables up to the second post - minkowskian order _ , phys . d * 89 * ( 2014 ) 064045 . formulation for observed and computed values of deep space network data types for navigation _ , john wiley , also available as deep space communications and navigation series , jpl publication 00 - 7 , 2000 . s. zschocke , m.h . soffel , _ gravitational field of one uniformly moving extended body and @xmath37 arbitrarily moving pointlike bodies in post - minkowskian approximation _ , class . quantum grav . * 31 * ( 2014 ) 175001 . kopeikin , b. mashhoon , _ gravitomagnetic effects in the propagation of electromagnetic waves in variable gravitational fields of arbitrary - moving and spinning bodies _ , phys . rev . d * 65 * ( 2002 ) 064025 .	an analytical solution for the light trajectory in the near - zone of the gravitational field of one pointlike body in arbitrary slow - motion in the post - post - newtonian approximation is presented in harmonic gauge . expressions for total light deflection and time delay are given . the presented solution is a further step toward high - precision astrometry aiming at nano - arcsecond level of accuracy .
the peculiar velocity of the solar system through the universe has been determined relative to the frame of reference provided by the cosmic microwave background radiation ( cmbr ) , to be @xmath4 km / s in the direction @xmath5 ( lineweaver et al . 1996 ; hinshaw et al . a study using the number counts of radio sources ( blake and wall 2002 ) had found the velocity to be consistent with that from the cmbr . here we determine the peculiar motion from the anisotropy in sky brightness due to radio sources . this provides an independent check on the interpretation of cmbr dipole anisotropy being due to motion of the solar system . also cmbr provides information about the isotropy of the universe for redshift @xmath6 , but the radio source population refers to a much later epoch @xmath7 . thus it also provides an independent check on the cosmological principle where isotropy of the universe is assumed for all epochs . we have used the nvss catalogue ( nrao vla sky survey , condon et al . 1998 ) for our investigations . this survey covers whole sky north of declination @xmath8 , a total of 82% of the celestial sphere , at 1.4 ghz . there are about 1.8 million sources in the catalogue with a flux - density limit @xmath9 mjy . figure 1 shows a plot of the bright sources ( @xmath10 mjy ) from the nvss catalogue in galactic co - ordinates . there is a strip of excess sources near the galactic equator . the large gap corresponds to the southern declination limit of the survey . an observer moving with a velocity @xmath11 , will find sources in the forward direction brighter by a factor @xmath12 , due to doppler boosting , where @xmath13 is the doppler factor , @xmath14 is the velocity of light and @xmath15 ( @xmath16 ) is the spectral index defined by @xmath17 . here we have used the non - relativistic formula for the doppler factor as cmbr observations indicate that @xmath18 . the increased flux density due to doppler boosting in the forward direction will cause a telescope of a given sensitivity limit to detect comparatively a larger number of sources . the contribution to the observed sky brightness at a given flux - density level @xmath19 comes from sources having the rest - frame flux - density @xmath20 . with the integral source counts of extragalactic radio source population following a power law @xmath21 ( @xmath22 , ellis and baldwin 1984 ) , the number of the sources observed therefore will be higher by @xmath23 , a factor independent of @xmath19 ( as long as @xmath24 and @xmath15 can be deemed to be independent of @xmath19 level ) . the observed sky brightness will thus change by an overall factor @xmath23 due to doppler boosting . also due to the aberration of light , the apparent position of a source at angle @xmath25 will be shifted in the forward direction by an angle @xmath26 , as a result there will be a higher number density per steradian in the forward direction as compared to that in the backward direction . this excess in number density @xmath27 . thus as a combined effect of doppler boosting and the aberration , the observed sky brightness ( an integrated emission from discrete sources per unit solid angle ) will vary as @xmath28 which ( for @xmath29 ) can be written as @xmath30 , a dipole anisotropy over the sky with amplitude @xmath31(v / c)$ ] ( ellis and baldwin 1984 ) . to find the velocity of the solar system , we consider all sources to lie on the surface of a sphere of unit radius , and let @xmath32 be the position vector of @xmath33 source with respect to the centre of the sphere . an observer stationary at the centre of the sphere will find the sky brightness to be uniformly distributed in all directions ( due to the assumed isotropy of the universe ) and therefore should get @xmath34 . on the other hand for a moving observer , the forward shift in position due to aberration and the doppler boosting of flux density in the forward direction , implies that the vectorial sum @xmath35 will yield a net vector in the direction of motion , thereby fixing the direction of the dipole . if @xmath36 is the polar angle of the @xmath33 source with respect to the dipole direction , then the magnitude of the vectorial sum can be written as @xmath37 . writing @xmath38 and converting the summation into an integration over the sphere , we get @xmath39\frac{v}{c}.\ ] ] the formula is equally valid for samples with finite upper and lower flux - density limits . here @xmath40 for a sky fully covered by the sample , and @xmath41 when there are finite gaps in the sky coverage and may need to be determined numerically for individual cases . as the nvss catalogue has a gap of sources for dec @xmath42 , in that case our assumption of @xmath34 for a stationary observer does not hold good . however if we drop all sources with dec @xmath43 as well , then with equal and opposite gaps on opposite sides @xmath34 is valid for a stationary observer . further we also excluded all sources from our sample which lie in the galactic plane ( @xmath44 ) as the excess of galactic sources towards the galactic centre ( see fig . 1 ) is likely to contaminate the determination of velocity . of course exclusion of such strips , which affect the forward and backward measurements identically , to a first order do not have systematic effects on the results ( ellis and baldwin 1984 ) . before proceeding with the actual source sample we used the monte carlo technique to create an artificial radio sky with about two million sources ( number density of sources similar to that in the nvss catalogue ) distributed at random positions in the sky . while the sky positions for each simulation were allotted randomly for each source , for the flux - density distribution we took the observed nvss sample , the latter justified because the source counts remain unchanged when integrated over the whole sky . the aberration merely shifts the apparent positions in the sky without adding or removing any sources and the effects of doppler boosting on the differential source counts at any flux - density level are equal and opposite in the forward and backward hemispheres , therefore the total source counts summed over the whole sky remain the same . on this we superimposed doppler boosting and aberration effects of our assumed motion , choosing a different velocity vector for each simulation ( in a number of cases we also used the velocity vector from cmbr measurements ( lineweaver et al . 1996 ; hinshaw et al . 2009 ) for monte carlo simulations ) . the resultant artificial sky was then used to recover back the velocity vector under conditions similar to our nvss case ( e.g. , with @xmath45 gaps in the sky ) , and thence obtained velocity vector was compared with the value actually used in that particular simulation . this not only verified our procedure but also allowed us to make an estimate of errors in the dipole co - ordinates as a large number of simulations ( @xmath46 ) were run starting with different random sky positions and for a different velocity vector each time . from these simulations we also estimated @xmath47 ( see eq . ( 1 ) ) , for the effect of the gaps in our samples . our results are presented in table 1 , which is almost self - explanatory . as a relatively small number of strong sources at high flux - density levels could introduce large statistical fluctuations in the sky brightness , we have restricted our sample to below 1000 mjy level . at the lower end we have restricted it to 15 mjy levels as the completeness of the sample at weaker flux - density levels could be doubtful ( blake and wall 2002 ) . the error in @xmath48 due to statistical fluctuations comprises two components . the variance in the flux - density distribution about the average value @xmath49 among sources contributes an error @xmath50 , while the statistical fluctuations in number density in the two hemispheres contributes @xmath51 . as these two are statistically independent , the net error in @xmath52 then is @xmath53 . the effects of error in the dipole direction on @xmath54 can be estimated this way . a shift @xmath55 in the dipole direction causes an exchange of spherical wedges of solid angle @xmath56 near the equator ( i.e. for @xmath57 ) between the forward and backward hemi - spheres . however the dipole anisotropy being minimum there ( @xmath58 ) the effect on speed estimate is minimal . it will result in a fractional change in @xmath59 of about @xmath60 . in all cases it will be a systematic effect , resulting in the speed being slightly underestimated ( @xmath61 for a typical error of 12 - 15 deg in the dipole direction ) . from table 1 the direction the velocity vector ( with our best estimate ra@xmath62 , dec@xmath63 or in galactic co - ordinates @xmath64 ) is quite in agreement with those determined from the cmbr ( ra@xmath65 , dec@xmath66 or @xmath5 with errors less than a degree ) ( lineweaver et al . 1996 ; hinshaw et al . however the estimates of @xmath11 ( @xmath67 km / s ) appear much higher than the cmbr value ( @xmath2 km / s ) by a factor @xmath1 at a statistically significant ( @xmath3 ) level . to guard against the possibility that some systematic effects like local clustering ( mainly the virgo super - cluster ) might have affected the dipole magnitude , we restricted our region of the sky brightness to that outside the super - galactic plane by rejecting sources with low super - galactic latitude , @xmath68 . we determined the dipole progressively excluding sources in the latitude steps of 5 degrees , and from a comparison of all these cases ( @xmath69 ) no systematic changes were seen in the computed dipole magnitude ( table 2 ) . thus it does not seem that the observed @xmath11 , a factor of @xmath1 larger than the cmbr , has resulted from a local clustering . an earlier attempt using the number counts of radio sources ( blake and wall 2002 ) had found a peculiar velocity seemingly consistent with that from the cmbr observations , but from the sky brightness anisotropy we got the velocity @xmath1 times the cmbr value . to ascertain that the difference somehow is not between the dipoles arising from the sky brightness and the number counts , we have determined the velocity from the number counts as well , using a technique slightly different from that of blake and wall ( 2002 ) . first the direction of the dipole was determined from @xmath70 , the three vector components ( @xmath71 ) being essentially the same as those of the dipole position determined from the three @xmath72 spherical harmonic coefficients ( blake and wall 2002 ) . then the dipole magnitude was calculated from the fractional difference @xmath73(v / c)=2k { \cal d}/3 $ ] , similar to that for @xmath52 in the case of sky brightness . the results are summarized in table 3 . comparing with table 1 we notice that the observed anisotropies in both the sky brightness and the number counts yield similar velocities with magnitudes @xmath1 times the cmbr value in both cases . however while in number counts the weaker sources , because of their much larger numbers ( @xmath74 ) , dominate the dipole determination , in the case of sky brightness , the contribution of each source being proportional to its flux density , the dipole determination depends equally on the stronger sources , @xmath75 ( for @xmath76 ) . a comparison of our results for source counts ( table 3 ) with those of blake and wall ( 2002 ) shows that the direction estimates of the dipole match exceedingly well , with an almost one to one correspondence for various bins . in fact , at a first look , even the magnitudes of dipoles , with average value @xmath77 in our case as compared to @xmath78 in blake and wall ( 2002 ) , seem to match very well . however , there is an essential difference . the dipole magnitude defined by blake and wall ( 2002 ) as @xmath79 , is a factor of 2 larger than @xmath48 defined in our case . thus while the dipole expected from the cmbr value in blake and wall ( 2002 ) is @xmath80 , in our case it is only half of that ( @xmath81 . the tabulated values of blake and wall ( 2002 ) seem about a factor of 2 larger than the cmbr prediction in all bins ( @xmath78 vs. @xmath80 ) though only at @xmath82 level . but in our case the observed dipole is @xmath1 times the cmbr prediction ( @xmath83 vs. @xmath84 ) at @xmath3 level . thus effectively we are finding dipole magnitude to be double of that by blake and wall ( 2002 ) , which is quite surprising since the basic data used ( nvss ) is the same even if the techniques differ . in a consistent case our tabulated @xmath48 for all bins should have been @xmath80 ( i.e. , half of the tabulated values of blake and wall 2002 ) , which definitely is not the case . the major difference in the two techniques is in the masking out of certain areas in sky which may otherwise contribute excess local sources to cause an under or overestimate of the dipole magnitude . however , it will be very surprising if this were making the difference of 2 for almost all flux bins . after all the local cluster sources should have different source counts than the truly far - off sources and should be affecting different flux bins quite differently ; the difference should have been especially discernible if the overall source counts are going to get affected by a factor of @xmath1 or so . it will be all the more intriguing if it had to happen without causing any shifts in the direction of the dipole which in both cases is found to be the same as that of the cmbr dipole . to us it appears that more likely the actual dipole determination in both cases ( blake and wall 2002 and ours ) is essentially the same ( with the different techniques and differential masking procedure making @xmath85 difference ) and that the difference of 2 creeps in while relating the dipole magnitude to the velocity as the two formulae differ by a factor of 2 . we can only state that because of the discrepancy we are finding with the already known results ( lineweaver et al . 1996 ; blake and wall 2002 ; hinshaw et al . 2009 ) we had to be extra careful in the magnitude scaling as well as in checking our code thoroughly , through monte - carlo simulations and otherwise . one way to conclusively eliminate the possibility that our large dipole value could be a consequence of some unknown or ignored local clustering in certain regions of the sky is to determine the projection of the inferred velocity for different polar angles with respect to the dipole direction and to verify if the observed velocity components are following the @xmath86 relation . this is because any local clustering would affect the magnitude of the components being determined from different regions of the sky very differently . we determined the velocity components for three different polar angles for the last flux - density bin ( @xmath87 mjy ) which had the largest number of sources . starting from the dipole direction we first divided the sky into six equal - area zones . then computing fractional difference in source counts between symmetrically placed pairs of sky zones , we determined the peculiar velocity components for three different polar angles . figure 2 shows a plot of the the three components , which seem to fit very well with the expected @xmath88 variation of @xmath89 km / s , the estimated speed for that bin in table 3 . also plotted is the projection of the cmbr value ( 369 km / s ) for a comparison . suppose a relatively local ( but perhaps on a scale much larger than that of virgo supercluster but at @xmath90 ) over - density generating the gravitational field responsible for the large observed motion , yields on an over - density also in the distribution of radio sources , and it gives a significant contribution to the excess sky brightness and to the excess counts in the direction of motion . the observed dipole amplitude will then be the sum of the contribution due to the actual motion with the added contribution of the source over - density masquerading as high solar motion in our analysis , in that case the derived velocity could be an upper limit . but fig . 2 would be still difficult to explain there . if some excess in the sources due to local clustering in certain regions of the sky were indeed instrumental in causing the tabulated large ( @xmath89 km / s ) value , then the three velocity component values could not have been influenced by the same factor almost identically by the local clustering as these were determined from different slices of the sky with absolutely no overlaps . this is a clinching evidence that some local clustering is not the cause of the inferred high velocity values . the nvss survey could be affected by a number of systematics , caused by changes in observing conditions when sweeping large areas on the sky , that may produce excess power on large angular scales . for example , blake and wall ( 2002 ) have noted that dim sources in nvss show some declination dependence which largely disappears above 15 mjy , but nevertheless one can not be totally sure that this and other possible systematics ( e.g. calibration changes in time during the observing months ) do not introduce a spurious modulation of the source density on large scales that may partially mimic a dipole . again , except in a very contrived situation we could not have obtained consistent results for the projected velocity values . the expected cmbr values are way below the observed ones in fig . 2 , but one must ensure that there are no scaling problems either . actually the plotted values are arrived at directly from the observed asymmetry in number counts in a straightforward calculation . for example , for the first plotted point ( @xmath91 ) the observed number counts in the respective sky zones give a fractional number @xmath92 , which multiplied with @xmath93 gives 1089 ( km / s ) as the projected velocity value , what is plotted in fig . similar are the calculations for the other plotted points , and it does not seem that there could be anything amiss in the scaling . the evidence seems irrefutable that the velocity inferred from the radio source distribution is indeed much larger than that from the cmbr . the fact that the directions of the dipole from the radio source data and the cmbr measurements are matching well , implies that the cause of the dipoles is common and the motion of the solar system seems to be the only reasonable explanation for that . but such a statistically significant difference in the estimates of the magnitude of the velocity vector is puzzling . assuming that the cmbr dipole estimates do not suffer from any residual errors during subtraction of galactic and other contributions , one can not escape the conclusion that there is a genuine discrepancy in the two dipoles and that the reference frame defined by the radio source population at @xmath7 does not coincide with that defined by the cmbr originating at @xmath6 . here we may add that there is some evidence that the motion of the local group of galaxies may be different when measured with respect to different reference frames ( lauer and postman 1994 ; giovanelli et al . 1998 ) at @xmath94 and at @xmath95 . however , while an anisotropy at @xmath96 scale might still be called `` local '' , an anisotropy at @xmath97 is nevertheless `` global '' as it encompasses a substantial fraction of the universe . on the other hand if there are differential perturbations to the hubble flow on the scale of the distribution of the radio source population vis - - vis that of the cmbr , the implication will be serious as any such anomaly would imply anisotropy on a universal scale . this would violate the cosmological principle where the isotropy of the universe is assumed for all epochs , and on which the whole modern cosmology rests upon . 20 blake , c. , & wall , j. 2002 , , 416 , 150 condon , j. j. , cotton , w. d. , greisen , e. w. , yin , q. f. , perley , r. a. , taylor , g. b. , & broderick , j. j. 1998 , , 115 , 1693 ellis , g. f. r. , & baldwin , j. e. 1984 , , 206 , 377 giovanelli , r. , haynes , m. p. , freudling , w. , da costa , l. n. , salzer , j. j. , wegner , g. 1998 , , 505 , l91 hinshaw , g. et al . , 2009 , , 180 , 225 lauer , t. r. , & postman , m. 1994 , , 425 , 418 lineweaver , c. h. , tenorio , l. , smoot , g. f. , keegstra , p. , banday , a. j. , & lubin , p. 1996 , , 470 , 38	according to the cosmological principle , the universe should appear isotropic , without any preferred directions , to an observer whom we may consider to be fixed in the co - moving co - ordinate system of the expanding universe . such an observer is stationary with respect to the average distribution of the matter in the universe and the sky brightness at any frequency should appear uniform in all directions to such an observer . however a peculiar motion of such an observer , due to a combined effect of doppler boosting and aberration , will introduce a dipole anisotropy in the observed sky brightness ; in reverse an observed dipole anisotropy in the sky brightness could be used to infer the peculiar velocity of the observer with respect to the average universe . we determine the peculiar velocity of the solar system relative to the frame of distant radio sources , by studying the anisotropy in the sky brightness from discrete radio sources , i.e. , an integrated emission from discrete sources per unit solid angle . our results give a direction of the velocity vector in agreement with the cosmic microwave background radiation ( cmbr ) value , but the magnitude ( @xmath0 km / s ) is @xmath1 times the cmbr value ( @xmath2 km / s ) at a statistically significant ( @xmath3 ) level . a genuine difference between the two dipoles would imply anisotropic universe , with the anisotropy changing with the epoch . this would violate the cosmological principle where the isotropy of the universe is assumed for all epochs , and on which the whole modern cosmology is based upon .
table i : simulation parameters , and fitting parameters . 1 : comparison between the analytical expression of the soft - core center - of - mass gaussian potential , eq . ( [ pot ] ) , and computer simulations data , vs. the c.o.m . interpolymer distance normalized by the polymer end - to - end distance , @xmath72 . 2 : center - of - mass mean - square displacement as a function of time . best fit of the molecular dynamics simulation data ( filled circle ) with the rouse equation ( dashed line ) , and with the intermolecular diffusion equation , eq . ( [ rai ] ) , for polymer melts ( full line ) . the short - dashed lines indicate the longest rouse relaxation time , @xmath73 . fig . 3 : monomer and center - of - mass mean - square displacements as a function of time . best fit of the molecular dynamics simulation data ( filled circle for c.o.m . and open squares for monomer dynamics ) with the rouse equation ( dashed line ) , and with the diffusion equations for semiflexible polymer melts ( full line ) . the short - dashed lines indicate the longest rouse relaxation time , @xmath73 .	we present a generalized langevin equation for the dynamics of interacting semiflexible polymer chains , undergoing slow cooperative dynamics . the calculated gaussian intermolecular center - of - mass and monomer potentials , wich enter the gle , are in quantitative agreement with computer simulation data . the experimentally observed , short - time subdiffusive regime of the polymer mean - square displacements , emerges here from the competition between the intramolecular and the intermolecular mean - force potentials . 2 despite its intrinsic complexity , the dynamics of low molecular weight unentangled polymer fluids has been considered for a long time to be a well understood problem in polymer physics . it is commonly accepted that when the degree of polymerization , @xmath0 , does not exceed the entanglement value , @xmath1 , a molecule is free to diffuse in the liquid , and follows fickian dynamics . in this case , the rouse approach successfully describes several key features of the polymer dynamics . these include the scaling with @xmath0 of the bulk viscosity and the single - chain diffusion coefficient @xmath2 , with @xmath3 the boltzmann constant , @xmath4 the temperature , and @xmath5 the monomer friction coefficient@xcite . upon closer examination , however , the dynamics of unentangled polymer fluids still presents important unresolved questions . in the rouse theory , local intermolecular interactions are completely neglected ; the single chain dynamics is driven by intramolecular entropic restoring forces and segmental friction , while the dynamics of the surrounding chains provide a heat bath . in this description the polymer center of mass is free to diffuse following brownian dynamics , and the mean - square displacement is expected to scale linearly in time at all time scales . in reality , a single polymer in a fluid spans a volume @xmath6 , with @xmath7 the polymer radius of gyration , and @xmath8 the statistical segment length . inside this volume are contained an average of @xmath9 chains , that interact with each - other through the potential of mean force@xcite . the range of the potential is the same as that of the correlation hole@xcite in the pair correlation function @xmath10 , a distance of order @xmath11 in polymer fluids@xcite . such interactions on length scales shorter than or equal to @xmath11 leads to the rouse equation s failure to adequately describe short - time polymer dynamics . the inconsistency of the rouse equation with computer simulation data of short - time unentangled@xcite and entangled polymer dynamics in the melt state , and in concentrated solutions@xcite , has been known for many years . in the short - time regime , @xmath12 , the single - chain center - of - mass ( c.o.m . ) mean - square - displacement , @xmath13 , of unentangled polymers shows anomalous diffusion . for times @xmath14 @xmath13 crosses over from the short - time ballistic dynamics to long - time fickian diffusion ( @xmath15 ) through an intermediate regime ( @xmath16)@xcite . only at long times ( @xmath17 ) does the system obey rouse dynamics and fickian diffusion is recovered . entangled polymer fluids show the same behavior in the short - time regime where rouse dynamics is expected to hold@xcite . the short - time anomalous dynamics only appears in melts and in solutions above the polymer overlap concentrations@xcite , further supporting the hypothesis that their short - time anomalous dynamics is a consequence of intermolecular interactions . in a recent paper , we derived a generalized langevin equation for the cooperative dynamics ( cdgle ) of interacting , completely flexible polymer fluids@xcite , where explicitly intramolecular and intermolecular forces are included . the cdgle is here implemented to treat the heterogeneous dynamics of unentangled semiflexible polymer melts with finite - size chains . the theory is found to be in excellent agreement with computer simulation data performed by g.grest and coworkers@xcite . the gle for a group of @xmath18 molecules undergoing slow cooperative dynamics is derived from the first - principle liouville equation by projecting the dynamics of the entire fluid onto the phase - space of the slow variables , i.e. the coordinates of @xmath19 molecules statistically comprised inside the volume defined by the range of the mean - force potential , @xmath20 , with @xmath21 the monomer density . an effective segment @xmath22 in molecule @xmath23 is driven by the intramolecular potential , @xmath24 with @xmath25 , the intermolecular time - dependent potential of mean - force , @xmath26 $ ] , and the random interactions with the surrounding fluid , where the projected random force is given by @xmath27 . the segment position @xmath28 follows the equation of motion@xcite : @xmath29 + { \bf f}_a^{q(i)}(t ) \ , \label{cdgle}\end{aligned}\ ] ] with @xmath30 . the effective monomer friction coefficient , @xmath31 , is given by the linear combination of the bare friction , @xmath32 , conventionally assumed to be the rouse friction , and the intramolecular ( @xmath23 ) and intermolecular ( @xmath33 ) memory functions containing time and space correlation of the random forces : @xmath34 the coupling of the random forces acting on different molecules is discarded in the traditional single - chain approaches@xcite , and it is negligible for systems of uncorrelated molecules , for which the pair distribution function @xmath35 . with the purpose of describing the dynamics of finite - size semiflexible polymers , eq . ( [ cdgle ] ) is reformulated in a matrix formalism , and semiflexibility is introduced following the approach developed by bixon and zwanzig at the simple rouse level@xcite . the new equation of motion applies to the overdamped dynamics of a fluid of interacting molecules of any kind , including semiflexible polymers of finite length . it recovers the traditional single - molecule equation@xcite in the limit @xmath36 , and/or with the fluid described as a continuum , @xmath37 . in the case @xmath38 and @xmath39 , it recovers the diffusive equation for binary interacting particles @xcite . thus our equation correctly reproduces the dynamics of systems characterized by dominant intramolecular or intermolecular interactions . since eq . ( [ cdgle ] ) is obtained by projecting the fluid dynamics onto the extended set of intra- and intermolecular slow variables , the memory function contributions are minimized and can be neglected in first approximation@xcite . in addition , if only the intramolecular polymer coordinates are chosen as slow variables , and the memory function is discarded , this procedure recovers the traditional rouse equation . the memory function contribution to the intramolecular equation becomes relevant for melts of high molecular weight polymers , where it describes the cross - over from unentangled to entangled polymer dynamics@xcite . the gle is decoupled in center - of - mass diffusion and intramolecular dynamics by a transformation of variables from beads to bond monomer coordinates@xcite . the solution of the c.o.m . diffusive equation requires the derivation of an analytical expression for the c.o.m . intermolecular potential . we start from the four - point distribution function which relates a pair of monomers belonging to different polymers , and their c.o.m . positions . the distribution is factorized in the product of two - point distribution functions and integrated over the monomer coordinates @xmath40 . in the special case of polymer molecules we adopt the thread - prism expression of the monomer pair distribution function@xcite , and the gaussian intramolecular distribution@xcite , @xmath41^{3/2 } exp[- 3 r_a^2/(2 r_g^2)]$ ] . the effective potential acting between the centers - of - mass of a pair of molecules inside a region of slow cooperative dynamics is approximated by @xmath42 e^{-3 r^2/(4 r_g^2 ) } \ . \nonumber\end{aligned}\ ] ] @xmath43 is gaussian and finite for all distances , with a range on the order of the radius of gyration . at full polymer - polymer overlap , @xmath44 decreases with increasing reduced fluid density , @xmath45 , reflecting the transition towards an incompressible system . @xmath44 also decreases with increasing polymer molecular weight due to the increasing interpenetrability of the polymer chains . figure 1 shows a comparison between eq . ( [ pot ] ) and the potential extracted from the trajectories of molecular dynamics computer simulations for melts of unentangled polyethylene chains @xcite . the simulations are performed at constant volume and constant temperature , using the experimental densities at atmospheric pressure . we analyze trajectories from unentangled polyethylene samples with increasing degree of polymerization , as reported in table i ( the entanglement degree of polymerization @xmath46 ) . the only fitting parameter in the quantitative comparison , is the potential at complete polymer overlap , @xmath47 , which is compared in table i to @xmath44 . all the samples are in good agreement with eq . ( [ pot ] ) . the value of @xmath47 agrees quantitatively with the simulations for the high @xmath0 samples . in the low @xmath0 samples the presence of the fine structure at short distance , due to the local monomer packing , does nt allow us to make a quantitative comparison . the data show a decrease of @xmath44 with increasing molecular weigth or density , in a greement with eq . ( [ pot ] ) , and with a recent study on dilute and semidilute polymer solutions@xcite . eq . ( [ pot ] ) represents an improvement on the mean - field formula derived by flory and krigbaum , which predicts the opposite trend@xcite . however in solution both simulations and renormalization group calculations shows that the @xmath48 scaling regime is finite . in our study of polymer melt dynamics , the small range of @xmath0 available does not allow us to investigate this feature . from eq . ( [ pot ] ) we derive the intermolecular force between the c.o.m . of two polymers . in eq . ( [ cdgle ] ) this is moltiplied by the number of interacting polymers @xmath49 . the time evolution of the intermolecular center - of - mass potential follows the time - dependence of the interpolymer distance . with the purpose of obtaining an analytical solution of eq . ( [ cdgle ] ) , we approximate @xmath50 with its statistical average over the polymer configuration space , @xmath51 \nonumber \\ & & e^{-3 < r^2(t)>/(4 r_g^2 ) } r(t ) \label{gt } \ .\end{aligned}\ ] ] when eq . ( [ gt ] ) is introduced in eq . ( [ cdgle ] ) , the latter reduces to a set of coupled rouse equations with explicit intermolecular contributions @xmath52 + { \bf f}^{i}(t ) \ , \label{rai}\end{aligned}\ ] ] which obey the fluctuation - dissipation condition @xmath53 , where @xmath54 . @xmath55 is the coordinate of the c.o.m . of the dynamical aggregate . the advantage of adopting the approximate form of the intermolecular force is that the gle becomes solvable using the standard techniques of transformation to normal mode coordinates . the set of coupled equations , eq . ( [ rai ] ) , reduces to two gles in the relative and cooperative variables , from which the relative many - chain intermolecular distance @xmath56 , where @xmath13 is the single chain mean - square displacement . the many - chain cooperative diffusion coefficient , @xmath57 , approaches zero in melts of high molecular weight polymer chains , at low temperature , and/or at high liquid density . since the intermolecular force depends at each instant on the value of the intermolecular distance @xmath58 , we solve eq . ( [ rai ] ) self - consistently for small time increments ( @xmath59 ) . by this procedure we take into account the change in the intermolecular interaction with the dynamical evolution of the system . using eq . ( [ rai ] ) , we calculate the c.o.m . mean - square displacement , @xmath60 . since the appearance of anomalous c.o.m . dynamics is related to the amplitude of @xmath61 ( when @xmath62 single chain uncorrelated dynamics is recovered ) the anomalies are confined in the time scale @xmath63 , which is the time necessary for the system to travel outside the range of the mean - force potential ( @xmath64 ) . we test the validity of our approach by comparing the c.o.m . mean - square displacement calculated from the solution of eq . ( [ rai ] ) , with the data from computer simulations for the samples previously analyzed . the values of @xmath21 , @xmath4 , @xmath0 , @xmath8 and @xmath65 are the input to the equation together with the effective value of the potential at initial time and zero intermolecular distance , @xmath47 , calculated from the fitting of the potential . here the statistical segment is much larger than the chemical bond because of the intrinsic stiffness of the polyethylene molecule . figures 2 and 3 show that the diffusive equation with the soft - core intermolecular potential reproduces quantitatively the anomalous subdiffusive c.o.m . dynamics for all of the samples investigated , and in the complete range of time scales . the slope of the subdiffusive c.o.m . regime is reported in table i , and it is found to decrease with increasing degree of polymerization . the degree of dynamical anomaly increases with increasing @xmath0 and decreasing @xmath4 , as the system moves towards its glass transition . for the range of @xmath0 investigated in this study a scaling regime has yet to be reached . to describe the polymer flexibility we adopt a freely rotating chain model for segments of length @xmath66 with an effective stiffness parameter @xmath67 ( @xmath68 , @xmath69 ) . the value of @xmath70 is obtained from the polymer mean - square end - to - end distance calculated from the simulation data , and reproduces the experimental characteristic ratio for a finite polyethylene chain . the introduction of the explicit description of the local stiffness does not affect the dynamics at the c.o.m . level . this is because both the flexible model with renormalized segments and the the explicit semiflexible description are devised to reproduce the global chain properties . in figures 2 and 3 the c.o.m . diffusion is identical if we use a semiflexible or a renormalized flexible model . the monomer dynamics depends strongly on the local chemical structure , and chain stiffness . however , at the monomer level , the intermolecular contributions are negligible . our mode analysis shows that only the lowest index , global modes are affected by the intermolecular potential in agreement with previous studies@xcite . in analogy with the static picture , we argue that the insensitivity of the local dynamics to the intermolecular forces is due to the presence of similar and compensating monomer - monomer intramolecular and intermolecular excluded - volume interactions@xcite . in the regime of @xmath63 , the data show a dynamics slightly faster than the rouse prediction of @xmath71 . this effect is predicted by our diffusion equation with included local polymer semiflexibility . figure 3 shows that the monomer mean - square displacement calculated with the semiflexible approach is in better agreement with the simulations than the rouse function . the inclusion , or exclusion , of the intermolecular contribution , does not modify the monomer diffusion . in conclusion we show that the center - of - mass intermolecular potential in polymer melts can be described at a good level of approximation by an effective gaussian potential . the introduction of the local intermolecular forces in the gle strongly improves the description of the global dynamical properties of polymer melts in the short - time domain . the intermolecular forces , however , do not affect the dynamics on the local monomer scale , which is instead strongly dependent on the local polymer semiflexibility . we are grateful to g.s.grest for sharing the computer simulation trajectories . acknowledgment is made to the donors of the petroleum research fund , administrated by the acs , for partial support of this research . we also acknowledge the support of the national science foundation under the grant dmr-9971687 . m.doi , s.f.edwards , _ the theory of polymer dynamics _ ( oxford university press , oxford , 1986 ) . j.p.hansen , i.r.mcdonald , _ theory of simple liquids _ ( academic press , london , 1991 ) . p .- g . de gennes , _ scaling concepts in polymer physics _ ( cornell university press , ithaca , 1979 ) . k.s.schweizer , j.g.curro , adv . chem . phys . * 97 * , 1 ( 1997 ) . w.paul , g.d.smith , d.y.yoon , b.farago , s.rathgeber , a.zirkel , l.willner , d.richter , phys . rev . lett . * 80 * , 2346 ( 1998 ) . a.kopf , b.duenweg , w.paul , j. chem . phys . * 107 * , 6945 ( 1997 ) . k.kremer , g.s.grest , j. chem . phys . * 92 * , 5057 ( 1990 ) . a.kolinski , j.skolnick , r.yaris , j. chem . phys . * 86 * , 1567 ( 1987 ) . m.mondello , g.s.grest , j. chem . phys . * 106 * , 9327 ( 1997 ) . k.s.kostov , k.f.freed , e.b.webb iii , m.mondello , g.s.grest , j.chem.phys . * 108 * , 9155 ( 1998 ) . m.guenza , j. chem . phys . * 110 * , 7574 ( 1999 ) . m.bixon , r.zwanzig , j.chem.phys . * 60 * , 1896 ( 1978 ) . f.j.vesely , h.a.posch , mol.phys . * 64 * , 97 ( 1988 ) . k.s.schweizer , j.chem.phys . * 91 * , 5802 ( 1989 ) and references therein . m.guenza , a.perico , macromol . * 25 * , 5942 ( 1992 ) . p.g.bolhuis , a.a.louis , j.p.hansen , j. chem . phys . * 114 * , 4296 ( 2001 ) ; a.a.louis , p.g.bolhuis , j.p.hansen , e.j.meijer , phys.rev.lett . * 85 * , 2522 ( 2000 ) . d.richter , l.willner , a.zirkel , b.farago , l.j.fetters , j.s.huang , phys . rev . lett . * 71 * , 4158 ( 1993 ) . [ cols="^,^,^,^,^,^,^",options="header " , ]
processes involving particle production in the presence of classical electromagnetic fields @xmath1 can be considered in the context of semi - classical approximation , in which the source particle is described by a classical current with a prescribed trajectory and the emitted particles are considered as fully quantized fields . for the case of accelerated protons , the inertially forbidden decays via the strong interaction channel like , for instance , @xmath2 and @xmath3 , and via the weak channel , as @xmath4 , have been studied in detail in the semi - classical approximation . for the case of motion under magnetic fields , the validity of the semi - classical approximation is warranted if , @xmath5 where @xmath6 is the magnitude of the magnetic field measured in the instantaneous reference frame of the proton , @xmath7 is the usual lorentz factor , and @xmath8 g is a critical value for the magnetic field , denoting the region where a full quantized analysis is mandatory . provided that condition ( [ chifactor ] ) is obeyed , the semi - classical calculations are accurate and the associated emitted power depends only on the proper acceleration of the source @xcite . for the case of a magnetic field @xmath9 , for instance , the proper acceleration of a ( relativistic ) proton is @xmath10 and , consequently , eq . ( [ chifactor ] ) can be rewritten as @xmath11 the parameter @xmath12 can be defined in a frame independent way as @xmath13 , where @xmath14 stands for the proton momentum . for the case of inertially forbidden particle production , as those ones explicitly listed above , one naturally expects a strong suppression for quasi - inertiall @xmath15 protons . we review here the role played by the proper acceleration @xmath0 of the source in the massive particle production , with emphasis on the ranges favoring the massive particle emission channels . we also discuss the possibility of observing signals of curvature pion radiation from accelerated protons in the magnetosphere of strongly magnetized pulsars . unless otherwise stated , natural units where @xmath16 and the spacetime signature @xmath17 are adopted through this work . we consider here the following kind of decay process @xmath18 where @xmath19 and @xmath20 are the source particles , described by a classical current corresponding to the states of a two level system following a prescribed classical trajectory with proper acceleration @xmath0 , and @xmath21 are the emitted particles , described by quantized fields . ( for further details , see @xcite , for instance . ) we assume that the respective masses @xmath22 of the emitted particles @xmath21 obeys @xmath23 , where @xmath24 and @xmath25 stands for , respectively , the masses of the source particles @xmath26 and @xmath27 . if @xmath28 , the process is known to be forbidden for inertial trajectories ( @xmath29 ) . it is intuitive to expect that , if the source is supposed to follow a prescribed trajectory , the momentum @xmath30 of the emitted particle @xmath31 , measured in the rest frame of the source , must be constrained to @xmath32@xcite . moreover , the mean energy @xmath33 of the emitted particles ( also measured in the proton s reference frame ) is of the order of the source proper acceleration@xcite , _ i.e. _ , @xmath34 then , from the condition @xmath35 and @xmath36 without loss of generality , let us assume a uniformly accelerated trajectory along the @xmath37 direction with worldline given by @xmath38 following the same procedure employed in the refs . @xcite , one obtains the differential decay rate as function of the energy of the emitted pion @xmath39 where @xmath40 is the modified bessel function of order @xmath41 , @xmath42 mev is the @xmath43 mass , and @xmath44 is the effective coupling constant . the total emitted power is @xmath45 where @xmath46 stands for the meijer g - function@xcite . the corresponding normalized energy distribution , @xmath47 is plotted in fig.([energy ] ) , field for different values of the proper acceleration of the proton . full line : a = 150 mev , dashed line : a = 500 mev.,scaledwidth=45.0% ] from which it is clear that mean energy of the emitted particles are @xmath48 . from ( [ threshold ] ) and fig.([energy ] ) , one can argue that , if the acceleration is such that @xmath49 the corresponding process ( [ general ] ) becomes energetically favored . in fact , more precise estimates reveal that the threshold is @xmath50 , where @xmath51 . hence , the neutral pion production ( [ scalarmeson1 ] ) is expected do be favored for @xmath52 mev . for the physical relevant case of protons in circular motion , one has also exactly the same threshold ( [ scalarmeson ] ) . for such case , the formulas for the neutral pion production ( [ scalarmeson1 ] ) can be obtained from the previously one calculated in @xcite for scalar emissions @xmath53 we notice that , in order to obtain that the total power emission ( [ powerscalar ] ) from eq . ( 4.7 ) of @xcite , one must perform the following modifications on the particle states : @xmath54 ) , @xmath55 is related to the strong coupling constant @xmath56 by @xmath57 the large difference between the strong and the electromagnetic coupling constants will imply a large difference also in the respective emitted powers . the comparison between ( [ powerscalar ] ) and the larmor formula for the usual synchrotron radiation shows that @xmath58 , where @xmath59 is the electromagnetic coupling constant . hence , if the proper acceleration of a proton reach the value @xmath60 , the channel ( [ scalarmeson1 ] ) is expect to dominate over the usual synchrotron radiation emission . it must be emphasized that the formalism presented here can not be applied when the charge currents corresponding to initial and final states are different . neutral and charged particles follow distinct trajectories in the presence of an electromagnetic field . the related channel @xmath61 for example , requires a different semi - classical formalism . the formalism and the respective calculations can be found in @xcite ( see also @xcite ) . pulsar magnetospheres offer many interesting possibilities to test these results involving high energy accelerated particles@xcite . let us consider , for this purpose , the simplest polar cap models of pulsars @xcite . for such models , a proton interacts with the dipolar magnetic field of the pulsar and is accelerated by an electric field parallel to the magnetic field lines . the transverse to the magnetic field component @xmath62 of the proton momentum vanish rapidly due to the usual synchrotron emission . the protons are accelerated along the magnetic field lines by the electric potential drop of the pulsar , producing photons by curvature radiation . a particle following a circular trajectory of radius @xmath63 has proper centripetal acceleration given by @xmath64 . the curvature radius @xmath63 of the magnetic field line of a pulsar is related to the radius @xmath65 of the star and to the light cone radius , @xmath66 through @xmath67 , where @xmath68 is the star period . then , one has @xmath69 the @xmath70 factor accounts for the efficiency of the polar cap acceleration mechanism @xcite @xmath71 . also , @xmath72 is the maximum available potential drop near the surface of the star @xcite . taking for a typical young pulsar @xmath73 m , one has from ( [ acceleration ] ) and ( [ scalarmeson ] ) , the following condition on the star surface magnetic field @xmath74 for the occurrence of pion emission with non negligible intensity . it is easily verified that even for @xmath75 , the condition ( [ final ] ) is far above of the known pulsars of the atnf catalogue@xcite , see fig . 2 . in fact , a systematic search in the atnf pulsar catalogue for strongly magnetized stars that could accelerate relativistic protons up to the curvature pion production threshold reveals that the best candidate is the 16 ms pulsar j0537 - 6910 , but the corresponding characteristic parameter @xmath76 is too small to give origin to observable signals @xcite . for the 1700 objects in the atnf pulsar catalogue@xcite with known period @xmath77s and surface magnetic field @xmath78 g . the shadowed region corresponds to the condition ( 3.3 ) . all the pulsars in the atnf catalogue are far below the pion production threshold.,title="fig:",scaledwidth=45.0%][bp ] we have briefly reviewed the role played by the source proper acceleration in the semi - classical approximation for the process of massive particle emission . unfortunately , we have from eq.([final ] ) and fig . 2 that the necessary threshold for pion production by protons in circular motion is far above the typical acceleration experimented by protons in pulsar magnetospheres . nevertheless , it would be also interesting to investigate , in this context , other radiative processes like photopion production and the inverse compton scattering . the authors are grateful to a. t. dias , c. o. escobar , g. e. a. matsas , r. opher , s. razzaque , and v. pleitez for enlightening and helpful discussions . herpay t and patks a 2008 j. phys . g 35 025201 fregolente d and saa a 2008 phys . d77 103010 berezinsky v , dolgov a and kachelrie m 1995 phys . lett . b 351 261 herpay t , razzaque s , patks a and mszros p 2008 j. cosm . 25 tokuhisa a and kajino t 1999 astrophys . j. 525 l117 ruderman m a and sutherland p g 1975 astrophys . j. 196 51 harding a k , usov v v and muslimov a g 2005 astrophys . j. 622 531 vietri m 2008 foundations of high - energy astrophysics theoretical astrophysics ( chicago , il : chicago univ . press ) arons j 2003 astrophys . j. 589 871 zhang b , dai z g , mszros p , waxman e and harding a k 2003 astrophys . j. 595 346 harding a k and muslimov a g 2001 astrophys . j. 556 987 manchester r n , hobbs g b , teoh a and hobbs m 2005 astron . j. 129 1993 [ arxiv : astro - ph/0412641 ] .	non - electromagnetic emissions from high energy particles in extreme environments has been studied in the literature by using several variations of the semi - classical formalism . the detailed mechanisms behind such emissions are of great astrophysical interest since they can alter appreciably the associated energy loss rates . here , we review the role played by the source proper acceleration @xmath0 in the particle production process . the acceleration @xmath0 determines the typical scale characterizing the particle production and , moreover , if the massive particle production is inertially forbidden , it will be strongly suppressed for @xmath0 below a certain threshold . in particular , we show that , for the case of accelerated protons in typical pulsar magnetospheres , the corresponding accelerations @xmath0 are far below the pion production threshold .
the physical mechanisms behind the heating of the solar corona and the acceleration of the fast solar wind are two major unresolved issues in solar physics . thermal heating alone is not sufficient to bring models into agreement with observations of the lower corona and at earth @xcite . the acceleration of the solar wind occurs predominantly within a few solar radii of the surface @xcite . additionally , solar and heliospheric observatory ( soho ) observations have shown that ions are heated below 4 @xmath0 @xcite . @xcite suggested that because the locations for the heating of the corona and the acceleration of the solar wind are the same , it is possible that the same mechanism could contribute to both . one source of heating is magnetic reconnection associated with flares and nanoflares at the solar surface . using extreme ultraviolet observations , @xcite have suggested the heating of active regions is impulsive , and therefore could be associated with nanoflares . in the high speed solar wind , reconnection events current sheets and filaments have been suggested as a heating mechanism @xcite . in the lower corona , the region of interest in this work , damping of alfvn waves is known to produce nonthermal acceleration and to bring models into agreement with observations both near the sun and at large distances @xcite . the possible damping mechanisms for alfvn waves in the photosphere to lower corona are numerous and include nonlinear damping @xcite , turbulent cascade @xcite , phase mixing @xcite , landau damping @xcite , neutral collisional @xcite , ion - cyclotron damping @xcite , and surface alfvn wave damping @xcite ( and references herein ) . global magnetohydrodynamics ( mhd ) models that do not include waves employ different methods such as empirical heating functions or varied polytropic index distribution to drive and heat the solar wind and are able to match well with ulysses , yohkoh , helios and advanced composition explorer data @xcite . to include alfvn wave damping in a realistic model of the solar environment , physical damping mechanisms must be specified . some mhd models use alfvn waves without specifying a damping mechanism . global @xcite and local @xcite models include additional equations for the waves and prescribe an empirical damping length . these studies are benchmarked with ulysses and helios observations . alternatively , one can include the wave energy and momentum without damping @xcite , and match soho extreme ultraviolet and yohkoh soft x - ray observations . of the numerous possible mechanisms for damping alfvn waves , those which we expect to be important for low frequency waves are nonlinear turbulent damping , phase mixing , surface alfvn waves . in this paper , we study the contribution of surface alfvn wave damping by utilizing a 3d global mhd simulation of a thermally - driven solar wind @xcite . in this model , the lower corona is the inner boundary ( chromosphere and transition region are not resolved . ) waves are not explicitly included in the model , nor can low frequency waves be resolved due to spatial and temporal resolution limits ( to be discussed further in section 2.2 ) . we consider waves with frequencies lower than those that are completely damped in the chromosphere and on the order of those dominating the heliosphere : @xmath1 hz ( periods 3 seconds to 3 days ) . alfvn waves have been detected in the lower layers of the solar atmosphere using both ground based observations @xcite and the hinode spacecraft @xcite with periods 2 - 4 minutes ( @xmath4 hz ) . ground based observations indicate the presence of alfvn waves in the corona with periods of five minutes ( @xmath5 hz ) @xcite . at 1 au , the dominant wave power is in waves with periods of 1 - 3 hours ( @xmath6 to @xmath7 hz ) @xcite . in this study we are interested in a damping mechanism that acts in the lower corona @xcite . in the chromosphere , alfvn waves with frequencies above 0.6 hz are damped by ion - neutral collisional damping and frequencies below @xmath8 hz were unaffected @xcite . @xcite found that waves below @xmath8 hz were not damped by any mechanism in the chromosphere . @xcite also found that nonlinear damping occurred over the extended corona . @xcite found waves @xmath9 hz were reflected by a gradient in the background alfvn profile , and dissipated not in the lower layers of the atmosphere , but over distance of a few solar radii . in the corona , the ion - cyclotron frequency is @xmath10 hz , so cyclotron resonance damping is not relevant for low frequency waves . phase mixing of outgoing and reflected incoming alfvn waves has also been studied @xcite . using a combination of three damping mechanisms ( nonlinear damping , surface alfvn wave damping , and phase mixing ) , @xcite were able to match observations of mass loss rates and terminal velocities for cool , giant stars . they applied their model to the sun and were able to obtain coronal heating and match wind velocity and alfvn wave power density observations in a 1d simulation @xcite . in this paper , we extend their work on surface alfvn wave damping in a 3d simulation of the solar corona . surface alfvn waves form on a magnetic interface - a finite thickness boundary separating two regions of plasma with a strong inhomogeneity in magnetic field and/or density . the alfvn wave in each region can interact , damp and transfer energy into the resonant layer separating the two plasmas ( resonant absorption ) . @xcite first utilized surface alfvn waves and resonant absorption as a mechanism to heat coronal loops . the transfer of mhd wave energy by resonant absorption was also studied in @xcite and @xcite . an alternative dissipation mechanism for surface alfvn waves is nonlinear wave steepening @xcite . these and other efforts , e.g. @xcite have resulted in damping lengths which depend on the frequency of the waves , the nature of the magnetic interface , and the local plasma parameters ( density , magnetic field and velocity ) . utilizing these relations , the profile of the damping length in the wind has been estimated @xcite . all previous studies made assumptions about the wind . for example , @xcite calculated nonlinear viscous damping of surface alfvn waves in polar coronal holes . they assumed two values of the superradial expansion of the magnetic field lines , profiles for density ( based on observations ) , and a single frequency ( 0.01 hz ) . they obtained one profile , and concluded that the nonlinear damping of the surface alfvn waves in region of strong magnetic field expansion should contribute significantly to the heating in the solar wind . the surface alfvn wave damping length depends on the profile of the background alfvn speed . as was shown in a survey of alfvn speed profiles from several mhd models @xcite , the alfvn profiles for @xcite and @xcite are almost identical below 10 @xmath0 , and the profiles were different from mhd models using empirical heating functions . the profile for @xcite is similar to these two , but differs very low in the corona . @xcite concluded that the inclusion of alfvn waves with empirical damping brought mhd models better in alignment with local heating studies that had the best agreement with observations . in the present study , we quantify the surface alfvn wave damping length for use in an mhd wave - driven model . we expect to find surface alfvn waves in the border between the fast and slow solar wind for two reasons : a ) the gradient in density and b ) the superradial expansion of the open magnetic field required to fill the space over closed streamers . we will focus on the superradial expansion of the field lines and compare our 3d mhd model with the observational study of @xcite . our calculations show that waves with periods less than 1 hour ( frequency greater than 2.8 @xmath3 hz ) are damped in the region between 1 - 4 @xmath0 , the region of interest for both solar wind acceleration and coronal heating @xcite . no other major source of damping has been suggested for these waves in this region . we demonstrate the importance of the 3d geometry in our results . we show that the contribution of the damping of surface alfvn waves to the wind is on the order of magnitude as ( or in some cases larger than ) turbulence . it is important to note that this study of waves in a solar wind solution was not self consistent - we did not consider any back effects on the waves from the plasma . the paper is organized as follows : in section 2 we describe the theory and numerical simulation background . in section 3 we calculate the location of coronal hole boundaries , and the damping length profile along those lines . in section 4 we estimate the energy surface alfvn waves will deposit below 4 @xmath0 . we also calculate the heating invoked in a semi - empirical thermodynamical model , compare with the wave flux and discuss the results . finally , conclusions can be found in section 5 . the inhomogeneity in density and/or magnetic field that gives rise to the surface alfvn waves can be described analytically as either a discontinuity or finite layer @xcite , such as a flux tube . for the case of a rapidly expanding flux tube of width @xmath11 , ( where @xmath11 is much smaller than the radius of the flux tube ) , surface alfvn waves form on the inner and outer surfaces . these waves can interact , damp and deposit energy into the surrounding plasma with a damping rate @xcite , @xmath12 where @xmath13 is the average wave number , @xmath14 is the frequency and @xmath15 and @xmath16 are the alfvn wave frequency on either side of the flux tube ( 1 representing inside and 2 outside . ) we assume the width of the flux tube to be much smaller than the radius . this allows us to take @xmath17 , as in @xcite . if the frequency on the outside is much larger than the frequency inside ( i.e. , a strong inhomogeneity ) , then the damping rate is @xmath18 the surface alfvn wave damping length can be written as the alfvn speed @xmath19 divided by the damping rate , @xmath20 the initial damping length @xmath21 can be written as @xmath22 which , by taking @xmath17 , can be simplified to @xmath23 utilizing the relation that @xmath24 ( where @xmath25 is the superradial expansion factor of the field line ) , and fixing the frequency of the waves to be constant with height , the damping length in the inertial frame is @xmath26 where @xmath27 is alfvn mach number , @xmath28 is the solar wind speed , @xmath29 is the magnetic field strength , @xmath30 is the mass density . the subscript 0 indicates the variable is to be evaluated at the reference height . in the present study the model does not treat the chromosphere or the transition region ; the lower corona is the inner boundary . the cell size at the inner boundary is @xmath31 , and so we chose @xmath32=1.04@xmath0 as our reference height to assure that calculations would not include the solar surface inner boundary cells . we quantify the expansion of open field lines by @xmath25 , given by . @xmath33 where @xmath34 is the cross sectional area of the flux tube at distance @xmath35 . a value of 2 for @xmath25 indicates pure radial expansion . the lines which border closed field lines must open faster than radial to fill the space above the closed loops . in studies where @xmath25 is not a function of @xmath35 , typical values chosen were 2 - 6 for open field regions from 1 - 10 @xmath0 @xcite . here we determine @xmath25 explicitly as a function of height from 1.04 - 10 @xmath0 . a similar parameter for the expansion of a field line is the superradial diverging factor or superradial enhancement factor @xmath36 , as in @xmath37 we will use a 3d mhd model as a laboratory to estimate the contribution of the surface waves to the wind from 1.04 - 10 @xmath0 . we obtain steady state solar wind solutions by using the solar corona component of the space weather modeling framework ( swmf ) , developed by the university of michigan @xcite . this 3d global magnetohydrodynamics ( mhd ) model incorporates michelson doppler imager ( mdi ) magnetograms to generate an initial magnetic field configuration with the potential field source surface model ( see @xcite for details ) . an initial density is assumed on the solar surface ( @xmath38 @xmath39 ) , and the mhd equations are evolved in local time steps ( 12,000 iterations ) and time accurate calculations ( for ten minutes ) to achieve steady state solutions for solar minima conditions in a 24 @xmath40 24 @xmath40 24 @xmath0 domain . the steady state was generated with carrington rotation ( cr ) 1912 . solar wind solutions from swmf were validated in @xcite from cr1916 - 1929 by comparing with advanced composition explorer and wind satellite data ( near 1 au ) . cr1912 was chosen to allow for comparisons to an observational study of the expansion of open field lines @xcite . waves occur naturally as a perturbation to the mhd equations , and so their presence may be expected when solving the mhd equations in space and time . however , in global simulations waves have to be included explicitly @xcite due to time and spatial limitations . the time step of this simulation ( 0.2 seconds ) is less than the smallest period considered in this analysis ( 3 seconds ) . additionally , the grid resolution is not enough to spatially resolve the waves . @xcite ( herein referred to as do99 ) characterized the large - scale solar magnetic topology during solar minima conditions ( august 1996 ) using data from the solar and heliospheric observatory s ( soho ) ultraviolet coronagraph spectrometer instrument . they analyzed the latitudinal dependence of two line emission intensities and found the values were constant within the large polar coronal holes but suddenly increased at the border of the holes and equatorial streamers . do99 used this increase in intensity to identify the colatitude of coronal hole boundary ( chb ) , i.e. the border between open and closed magnetic field lines . in the present study , we identify the chb locations as the open field line with the largest colatitude ( the angle measured from the pole to the equator ) in the steady state described in the previous section . we compare our calculation of the field expansion with the observed values from do99 . figure 1 shows the 3d coronal hole boundary ( chb ) field lines obtained from the model . the different colors refer to : red as northeast ( ne ) ; blue as southeast ( se ) ; green as southwest ( sw ) ; and purple as northwest ( nw ) . the solar surface is shown colored by the radial component of the magnetic field . on 1996 august 17 , there was an active region near the center of the disk , and mostly quiet sun at the intersection of the plane of the sky with the photosphere . the first two rows of table [ tbl1 ] provide chb colatitudes ( angle measured from the pole to the chb footpoint ) found in do99 and this study . we find that 3 of the 4 simulated chb are higher in latitude ( i.e. , smaller colatitude ) compared to do99 . as we discuss below , these differences could be due to projection effects . the superradial expansion factor @xmath25 and superradial enhancement factor @xmath36 were calculated as a function of height for each chb field line . in figure 2 we show @xmath36 for the chb lines in this study . additionally , we have included the @xmath36 profiles which correspond to the minimum and maximum asymptotic values of @xmath36 for chbs determined in do99 . rows 3 and 4 in table [ tbl1 ] provide the asymptotic superradial enhancement factor value for each line . we find that both @xmath25 and @xmath36 cover a larger range of values compared to dob99 . only the sw line from our simulation falls in the range from do99 and only from 3 - 6 @xmath0 . in general , we find that 3 of the 4 boundary lines ( all except sw ) have lower values in our simulation compared to do99 . the best agreement for the location of the chb between the studies is the sw line . the southern hemisphere from the simulation is more similar to the results deduced from observations in the plane of the sky in do99 than the northern hemisphere . overall , we find that @xcite had larger values of @xmath36 , and a small range of values , for the same field lines . as we show below , this result is due to 3d vs. 2d projection effects . in order to quantify the significance of 2d projection effects in the context of the expansion of field lines , we calculated @xmath36 for four field lines shown in figure 3a . the projection of the field lines on the plane of the sky is shown in figure 3b . contours of solar wind speed ( in @xmath41 are shown in figures 3a and b , and the solar surface is shown as white sphere . figure 3c provides the superradial enhancement factor @xmath36 , calculated according to equation [ eq : f ] for each field line ( labeled a , b , c and d as in figure 3b ) . the calculation of the 3d line is shown as solid lines , and the 2d projection line calculation is shown as dashed lines . figure 3c demonstrates that a 2d projection of a 3d field line on the plane of the sky can overestimate the divergence of the line . the 3d lines from a - d all approach values of f 4 while the 2d estimates vary between 4 and 13 . this projection effect explains why the values for @xmath36 from our simulation are smaller than those determined in the observational study of do99 . it is crucial , therefore , to do studies of surface alfvn wave damping in a 3d simulation in order to capture the true divergence of the field lines . in figure 4 , we plot @xmath42 ( see eq . [ eq : l ] ) which was calculated using parameters @xmath30 , @xmath29 and @xmath43 from the steady state solution . figure 4a presents @xmath42 for the coronal hole boundary field lines in figure 1 with frequency @xmath44 hz , normalized to the initial damping length @xmath45 of the sw line ( chosen because of the agreement with do99 ) . this normalization allows for comparison of the profile features from different source regions as a function of height . in figure 4b we feature only the sw line and present @xmath42 for several frequencies , from @xmath46 hz . it can be seen in figure 4b that frequencies above @xmath47 ( short dashed line ) will be appreciably damped within a few solar radii of the surface . figure 4a shows distinctly different profiles from the southern and northern chb lines . we examined the source region of each footpoint and found that the se and sw lines originated near small active regions in which the radial component of the magnetic field @xmath48 g. both northern hemisphere lines originated from quiet sun regions ( @xmath49 g ) . for swmf and other mhd models , @xcite showed that the alfvn profile will contain a maximum , or hump , if the source region is quiet sun . the profile from an active region in global models begins at a maximum value , and drops to less than a few hundred @xmath50 within one solar radius from the surface . the profile of @xmath42 is controlled by the alfvn speed profile . the normalized profiles in fig . 4a show that the position corresponding to @xmath51 is closest to the sun for active regions . the profiles from quiet sun source regions have a plateau , pushing @xmath51 further from the sun . the implication of this result can be seen in the equation relating the alfvn wave energy density , @xmath52 if the damping length is @xmath53 or less , then the waves will be damped close to the sun . therefore , the presence of the hump means the energy of the surface alfvn wave can travel further into the corona before substantial damping occurs . this means that the quiet sun region will damp more surface waves at further distances , so it is more efficient in carrying the wave momentum out into the corona . active regions will damp closest to the sun . in the previous section we considered how surface alfvn waves at the solar surface with the frequency range @xmath54 to @xmath55 hz @xcite would be damped in our background solar wind environment . we found that waves with frequency above @xmath56 hz were appreciably damped below 4@xmath0 . we now consider the contribution of their wave flux to the energy of the wind . there will be a contribution to the momentum of the wind as well , but in this analysis we ignore this contribution . we assume that the wave damping will contribute solely to heating the wind ; therefore we derive here an upper limit on their contribution to the heating of the plasma . the spectra of the surface alfvn waves @xcite is : @xmath57 where @xmath58 , @xmath59 is the mean frequency in the observed range and the power index corresponding to the low frequency waves we are considering is @xmath60=0.6 @xcite . we assume that this flux of surface alfvn waves is propagating along open field lines during solar minima . this flux will decrease with distance as @xmath61 where @xmath42 is the damping length , for each frequency . with the damping lengths calculated in the previous section , we calculate how much flux is lost between @xmath62 ( our self - imposed reference height ) and @xmath63 . the height of @xmath63 was chosen because the contribution to solar wind acceleration and coronal heating must be deposited within a few solar radii of the sun @xcite . the contribution to the solar wind at each frequency is @xmath64 where the limits are @xmath65 and @xmath66 , and the definition of @xmath67 from equation [ eq : l ] has been included ( @xmath68 , @xmath25 , @xmath69 , are all functions of r. ) we replace @xmath45 in eq . [ eq : fluxlost2 ] with eq . [ eq : l0 ] . the total flux lost is found using @xmath70d\omega}\ ] ] where the limits are @xmath71 hz and @xmath72 hz . next we compare the potential contribution of the wave flux to the heating in the model ( eq . [ eq : heat2 ] ) . it should be stressed that we are not doing a self - consistent calculation : we are estimating wave flux from a model that does not include waves , and we are not considering any feedback of the waves on the plasma . the simulation analyzed in this paper is that of a thermally driven solar wind . the polytropic index @xmath73 in the model varies in space by utilizing the wang - sheeley - arge model and the conservation of energy along a solar wind field line using the bernoulli equation . this serves to artificially heat the wind @xcite in a manner mimicking turbulence @xcite . figure 5 shows the distribution of @xmath73 in the plane of the sky on 1996 august 17 . in this section we quantify the heating due to this variable gamma , and do a comparison with the energy deposited by damped surface alfvn waves , as in eq . [ eq : fluxlost4 ] . the first law of thermodynamics attributes changes in the internal energy @xmath74 of a gas to work done on or by the gas @xmath75 , and heat added to or removed from the gas @xmath76 . in the case of an ideal gas , the change in internal energy can be written as @xmath77 , where @xmath78 is the specific heat at constant volume . the work is expressed as @xmath79 , where @xmath80 is the pressure and @xmath81 is the volume . the first law can therefore be written as @xmath82 by introducing the ideal gas equation of state and assuming that the ratios of specific heats are constants , one can derive a polytropic equation , @xmath83 where @xmath60 is referred to as the polytropic index . the notation stems from @xcite to clarify that this index can ( but need not ) be the ratio of specific heats , and that we are not necessarily considering an adiabatic process . the symbol @xmath84 is typically used for the ratio of specific heats , and in the case of an adiabatic expansion ( no heating enters or leaves the system ) , @xmath85 . an isothermal wind expansion would be characterized by @xmath86 . observations of the solar wind have indicated that @xmath60=1.46 - 1.58 in the heliosphere @xcite . a value closer to unity is adopted in some global mhd models in the region near the sun in order to generate fast solar wind and match temperature observations in the heliosphere @xcite . all previous discussion had the underlying assumption that @xmath60 was constant with height . if that condition is not met , then the polytropic index is referred to as an effective ( or local ) polytropic index and written as @xmath73 @xcite . the polytropic equation ( eq . [ eq : poly ] ) is modified to @xmath87 such that the relationship between density and pressure is not simple . the variation of @xmath73 with height has been utilized to heat the solar wind used in this paper . we will characterize the additional heating provided by the prescribed distribution of @xmath73 in our solar wind simulation , and argue that surface alfvn waves damped near the sun could replace this artificial heating and move the model towards a more physical treatment of the solar environment . for a solar wind with a constant ratio of specific heats @xmath84 , the conservation of energy can be written as @xcite : @xmath88 = \rho { \bf g}\cdot { \bf u } + q\ ] ] where p is the thermal pressure , q is the additional heating function , and the energy density is @xmath89 recent global mhd studies adopt an exponential function for the form of q with several free parameters in order to benchmark the model with observations during solar minima conditions @xcite . substituting eq . [ eq : enden ] into eq . [ eq : energy_conserve ] , and setting time derivatives to zero for a steady solar wind , we find : @xmath90 = \rho { \bf g}\cdot { \bf u } + q.\ ] ] as discussed at the beginning of this section , there is no heating function q in the simulation used in this paper , and the ratio of specific heats @xmath84 is replaced by the effective gamma @xmath73 , @xmath91 = \rho { \bf g}\cdot { \bf u}.\ ] ] although @xmath73 has both latitudinal and azimuthal dependence below 4@xmath0 , we consider only at the radial variation , and so we replace @xmath92 by @xmath93 . we assume that we have exactly the same solar wind solution in the two cases we are considering : that of a variable gamma and of an additional volumetric heating function with @xmath94 . in order to quantify the amount of heating in the model with variable gamma , we subtract equation [ eq : energy_conserve3 ] from [ eq : energy_conserve2 ] : @xmath95.\ ] ] equation [ eq : heat1 ] can be written as : @xmath96.\ ] ] knowing how @xmath73 , @xmath80 and @xmath97 vary along any radial line , and setting @xmath94 we can integrate equation [ eq : heat2 ] between @xmath98=1.04@xmath0 and @xmath99=4@xmath0 to find the heat input along any field line : @xmath100 this equation gives the heat deposited into the system between the two heights . we compare @xmath76 with the flux of damped surface alfvn waves ( equation [ eq : fluxlost4 ] . ) table [ tbl1 ] provides @xmath101 and @xmath76 for the chb field lines . the expansion @xmath36 of the nw line from the simulation has the best match to the observations ( 5.55 compared to 6.0 ) . the surface alfvn wave flux along this line , and along the ne lin , is larger than the heating q by an order of magnitude . the geometrical properties of the sw line also match well with observations , however the wave flux for it ( and also the se line ) account for 20 @xmath102 of the required heating . the distinction between the southern and northern lines is the source region : they come from a stronger magnetic field region . near an active region , the second term in eq . [ eq : heat2 ] ( which includes the pressure and radial velocity ) is larger than a quiet sun region . therefore , we expect q to be larger than the surface alfvn wave flux along lines from active regions . as we assumed all of the wave flux goes to heating , this procedure gives an upper limit on the contribution of the damping of surface alfvn waves along an open magnetic field line to the heating along that line . a random sampling of 7 open field lines in the northern hemisphere with footpoints in the plane of the sky ( see figure 3 ) yielded @xmath101 that were on the order , or an order of magnitude larger than the @xmath76 . this work is the first study to look at surface alfvn waves in a 3d environment without assuming a priori a geometry of the field lines or magnetic and density profiles and strengths . we showed the calculation of the expansion of field lines must be done in a 3d environment . our calculations show that waves with periods less than 1 hour ( frequency greater than 2.8 @xmath3 hz ) are damped in the region between 1 - 4 @xmath0 , the region of interest for both solar wind acceleration and coronal heating @xcite . we showed that the quiet sun region will damp surface waves at further distances , so it is more efficient in carrying the wave momentum out into the corona . surface waves formed on flux tubes with footpoints in an active region will damp closer to the sun . the required heating from an active region was found to be larger than the damping of surface alfvn wave flux , therefore another mechanism ( such as turbulence ) may be the dominant heating in these regions , with surface alfvn waves contributing approximately 20@xmath102 of the heating . we estimated damping of surface alfvn waves in the border between open and closed field lines at heights 1.04 - 4 @xmath0 due to the superradial expansion of the field lines . as some of the wave flux would go to the momentum of the wind , we provide an upper limit on the contribution of surface alfvn waves to the heating of the solar wind . in the region between open and closed field lines , within a few solar radii of the surface , no other major source of damping has been suggested for the low frequency waves we consider here . our results demonstrate that it is not necessary to have turbulence in order to heat the solar wind - and that it is imperative to include the physics of surface alfvn wave damping in solar wind models in order to more physically model the heating . surface alfvn waves could also be present in the solar wind , in the flux tube structures said to fill interplanetary space @xcite . another environment which could support these waves are corotating interaction regions ( cirs ) , due to the inhomogeneity in density present in these structures @xcite . both of these topics will be addressed in future works . it is important to note that this study of waves in a solar wind solution was not self consistent - we did not consider any back effects on the waves from the plasma . we simply tried to estimate if the waves could produce the heating required to create the solar wind solution from the model . in the future we will pursue other damping mechanisms with the goal of incorporating the key mechanisms of wave damping in self - consistently in global mhd models to improve the lower corona . [ eq : f ] ) for the lines in fig . [ fig1 ] ( colors correspond ) from 1.04 - 10 solar radii ( rs ) . we also show the profiles which correspond to the minimum ( dashed line ) and maximum ( dash - dot line ) values of @xmath36 determined in the observational study by @xcite . we find that @xmath36 covers a larger range of values compared to the observational study . [ fig2 ] ] ) are shown on the plane , and the solar surface is shown as a white sphere . the expansion factor calculated from the 3d field line and 2d projections for each line is shown in ( c ) . the 2d projection can overestimate the expansion factor as compared to the 3d calculation . [ fig3],title="fig : " ] ) are shown on the plane , and the solar surface is shown as a white sphere . the expansion factor calculated from the 3d field line and 2d projections for each line is shown in ( c ) . the 2d projection can overestimate the expansion factor as compared to the 3d calculation . [ fig3],title="fig : " ] ) are shown on the plane , and the solar surface is shown as a white sphere . the expansion factor calculated from the 3d field line and 2d projections for each line is shown in ( c ) . the 2d projection can overestimate the expansion factor as compared to the 3d calculation . [ fig3],title="fig : " ] ) profiles for the coronal hole boundaries in fig . 1 ( colors correspond ) , normalized to the @xmath21 of the sw line ( green ) from 1.04 - 10 solar radii ( rs ) . note the plateau in the profiles for lines whose source region on the sun is quiet sun , differing from field lines with footpoints near active regions , whose profiles drop quickly . b ) profiles for the sw coronal hole boundary field line corresponding to different frequencies . from the bottom up to top profile : @xmath103 hz ( dash - dot - dot line ) ; @xmath104 hz ( dash - dot line ) ; @xmath44 hz ( dot line ) ; @xmath105 hz ( long dash line ) ; @xmath47 hz ( short dash line ) ; @xmath106 hz ( thin solid line ) ; @xmath107 hz ( thick solid line ) . waves with frequencies above @xmath47 are appreciably damped below 4rs . [ fig4],title="fig : " ] ) profiles for the coronal hole boundaries in fig . 1 ( colors correspond ) , normalized to the @xmath21 of the sw line ( green ) from 1.04 - 10 solar radii ( rs ) . note the plateau in the profiles for lines whose source region on the sun is quiet sun , differing from field lines with footpoints near active regions , whose profiles drop quickly . b ) profiles for the sw coronal hole boundary field line corresponding to different frequencies . from the bottom up to top profile : @xmath103 hz ( dash - dot - dot line ) ; @xmath104 hz ( dash - dot line ) ; @xmath44 hz ( dot line ) ; @xmath105 hz ( long dash line ) ; @xmath47 hz ( short dash line ) ; @xmath106 hz ( thin solid line ) ; @xmath107 hz ( thick solid line ) . waves with frequencies above @xmath47 are appreciably damped below 4rs . [ fig4],title="fig : " ] , the effective polytropic index , in the plane of the sky on 1996 august 17 . @xmath73 is specified on the solar surface ( shown as white sphere ) using the bernoulli integral , and is interpolated to 1.1 at 2.5 rs ( inner black circle ) . above 2.5rs , @xmath73 varies linearly until 12.5rs ( outer black circle ) , above which it has the value 1.5 . the white lines show the boundaries of grid refinement . [ fig5 ] ] belcher , j. w. & davis , l. , jr . 1971 , , 76 , 16 , 3534 borovsky , j. e. 2008 , , 113 , a08110 cirtain , j. w. et al . 2007 , science , 318 , 1580 cohen , o , et al . 2007 , , 654 , l163 cohen , o. , sokolv , i. v. , roussev , i. i. & gombosi , t. i. 2008 , , 113 , a03104 cranmer , s. r & van ballegooijen , a. a. 2005 , , 156 , 265 cranmer , s. r. , van ballegooijen , a. a. , & edgar , r. j. 2007 , , 171 , 520 de pontieu , b. , martens , p. c. h. & hudson , h. s. 2001 , , 558 , 859 de pontieu , b. et al . 2007 , science , 318 , 1574 dobrzycka , d. , cranmer , s. r. , panasyuk , l. s. & kohl , j. l. 1999 , , 104 , a5 , 9791 esser , r. et al . 1999 , , 510 , l63 evans , r. m. , opher , m. , manchester , w. b. & gombosi , t. i. 2008 , , 687 , 1355 grall , r. r. , coles , w. a. , klinglesmith , m. t. , breen , a. r. , williams , p. j. s. , markkanen , j. & esser , r. 1996 , nature , 379,429 groth , c. p. t. , de zeeuw , d. l. , gombosi , t. i. & powell , k. g. 2000 , , 105 , a11 , 25053 hartmann , l. & macgregor , k. b. 1980 , , 242 , 260 hasegawa , a. & uberoi , c. 1982 , the alfvn wave , u. s. department of energy technical information center , oak ridge , tn heyvaerts , j. & priest , e. r. 1983 , , 117 , 220 hollweb , j. v. 1971 , , 27 , 20 hollweg , j. v. 1986 , , 91 , a4 , 4111 hollweg , j. v. 1987 , , 312 , 880 ionson , j. a. 1978 , , 226 , 650 . isenberg , p. a. , lee , m. a. & hollweg , j. v. 2001 , , 106 , a4 , 5649 jatenco - pereira , v. & opher , r. 1989 , , 209 , 327 jatenco - pereira , v. , opher , r. & yamamoto , l. c. 1994 , , 432 , 409 jess , d. b , mathioudakis , m. , erdlyi , r. , crockett , p. j. , keenan , f. p. & christian , d. j. 2009 , science , 323 , 1582 kohl , j. l. et al . 1998 , , 501 , l127 leake , j. e. , arber , t. d. & khodachenko , m. l. 2005 , , 442 , 1091 lee , m. a. & roberts , b. 1986 , , 301 , 430 lionello , r. , linker , j. a. & mikic , z. 2009 , , 690 , 902 manchester , w. b. et al . 2004 , , 109 , a02107 matthaeus , w. h. , zank , g. p. , oughton , s. , mullan , d. j. & dmitruk , p. 1999 , , 523 , l94 matthaeus , w. h. , dmitruk , p. , oughton , s. & mullan , d. 2003 , solar wind ten : proceedings of the tenth international solar wind conference , 10.1063/1.1618627 mikic , z. , linker , j. a. , schnack , d. d. , lionello , r. & tarditi , a. 1999 , physics of plasmas , 6 , 5 moore , r. l. , musielak , z. e. , suees , s. t & an , c. -h . 1991 , , 378 , 347 narain , u. & sharma , r. k. 1998 , solar physics , 181 , 287 ofman , l. & davila , j. m. 1997 , , 103 , a10 , 23677 okamoto , t. j. et al . 2007 , science , 318 , 1577 parker , e. n. 1963 , interplanetary dynamical processes , wiley - interscience , new york parker , e. n. 1965 , , 4 , 666 parker , e. n. 1991 , , 376 , 355 patsourakos , s. & klimchuk j. a. 2009 , , 696 , 760 roussev , i. i. et al . 2003 , , 595 , l57 ruderman , m . s. 1992 , , 399 , 724 suzuki , t. k. & inutsuka , s. 2005 , , 632 , l49 suzuki , t. k. & inutsuka , s. 2006 , , 111 , a06101 tomczyk , s. et al . 2007 , science , 317 , 1192 tomczyk , s. & mcintosh , s. w. 2009 , , in press toth , g. et al . 2005 , , 110 , a12226 totten , t. l. , freeman , j. w. & arya , s. 1995 , , 100 , a1 , 13 tsubouchi , k. 2009 , , 114 , a02101 tu , c. y. , marsch , e. & thieme , k. m. 1989 , 17 , 283 usmanov , a. v. & goldstein , m. l. 2003 , , 108 , a9 , 1354 usmanov , a. v. & goldstein , m. l. 2006 , , 111 , a07101 verdini , a. & velli , m. 2007 , , 662 , 669 wentzel , d. g. 1979 , , 227 , 319 wentzel , d. g. 1989 , , 336 , 1073 crrrrrrrrrrr & ne & nw & se & sw + @xmath108 @xcite & 29.0@xmath109 & 31.0@xmath109 & 23.3@xmath109 & 28.0@xmath109 + @xmath108 this study & 37.2@xmath109 & 22.3@xmath109 & 19.2@xmath109 & 25.2@xmath109 + @xmath110 @xcite & 6.56 & 6.00 & 7.30 & 6.5 + @xmath110 this study & 2.56 & 5.55 & 4.71 & 7.62 + @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 + @xmath101 @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath119 +	here we investigate the contribution of surface alfvn wave damping to the heating of the solar wind in minima conditions . these waves are present in regions of strong inhomogeneities in density or magnetic field ( e. g. , the border between open and closed magnetic field lines ) . using a 3-dimensional magnetohydrodynamics ( mhd ) model , we calculate the surface alfvn wave damping contribution between 1 - 4 @xmath0 ( solar radii ) , the region of interest for both acceleration and coronal heating . we consider waves with frequencies lower than those that are damped in the chromosphere and on the order of those dominating the heliosphere : @xmath1 hz . in the region between open and closed field lines , within a few @xmath0 of the surface , no other major source of damping has been suggested for the low frequency waves we consider here . this work is the first to study surface alfvn waves in a 3d environment without assuming a priori a geometry of field lines or magnetic and density profiles . we demonstrate that projection effects from the plane of the sky to 3d are significant in the calculation of field line expansion . we determine that waves with frequencies @xmath22.8 @xmath3 hz are damped between 1 - 4 @xmath0 . in quiet sun regions , surface alfvn waves are damped at further distances compared to active regions , thus carrying additional wave energy into the corona . we compare the surface alfvn wave contribution to the heating by a variable polytropic index and find that it an order of magnitude larger than needed for quiet sun regions . for active regions the contribution to the heating is tweny percent . as it has been argued that a variable gamma acts as turbulence , our results indicate that surface alfvn wave damping is comparable to turbulence in the lower corona . this damping mechanism should be included self consistently as an energy driver for the wind in global mhd models .
the interaction of a charged particle with a gravitational wave ( gw ) has been studied by using a hamiltonian formalism ( @xcite-@xcite ) . this approach is similar to the one used for the study of the interaction of ions with electrostatic waves in magnetized plasmas ( @xcite-@xcite ) . the existence of chaos in phase space has been shown as a consequence of the overlap of resonances . recently , the acceleration of the charged particles due to the chaotic diffusion has been associated with bursts @xcite . however , such a diffusion is obtained for relatively high energy particles and becomes efficiently present only for large amplitude values of the gw ( @xmath0 ) . furthermore , the existence of a parametric resonance in the general problem of interaction between particles and waves is discussed in the classical book of landau and lifshitz @xcite . for the particular problem kleidis _ et al . _ @xcite showed that a parametric resonance arises for a set of frequencies of the gw of the form @xmath1 and an estimation of its width is given . also , the parametric resonance has been associated with the existence of chaotic motion . in the present paper we study the possible fast acceleration of low energy particles for small amplitudes of the gw by taking into account two special properties of the system . the first is the parametric resonance which , as we shall show , arises only for @xmath2 . the second property is the integrability of the system , and subsequently the absence of chaos , when the gw propagates in a parallel direction with that of the magnetic field . in section 2 we review the hamiltonian formalism of the model which describes the interaction . in section 3 , we study systematically the rise of a parametric resonance and in section 4 we prove the integrability of the system in the case of parallel propagation . it is shown in section 5 that when the above integrable case is in parametric resonance fast acceleration of particles occurs . the astrophysical implications are discussed in section 6 . the dynamics of a charged particle in a curved space - time and in the presence of a uniform magnetic field can be studied by considering the hamiltonian ( see @xcite ) @xmath3 where @xmath4 and @xmath5 denote the generalized coordinates and momenta of the particle , respectively . the metric @xmath6 defines the curved space - time and @xmath7 are the components of the vector potential @xmath8 of the magnetic field . it can be considered that the curvature of space - time is due to a gravitational wave ( gw ) . by considering a linearly polarized gw propagating in a direction @xmath9 of angle @xmath10 with respect to the axis @xmath11 , a possible form of the tensor @xmath6 is the following @xcite @xmath12 where @xmath13 is the minkovski space - time and @xmath14 is the amplitude of the gravitational wave . by using the metric tensor derived in @xcite and expanding its components up to @xmath15 we obtain the nonzero components of the tensor @xmath16 in the following form @xmath17 where @xmath18 the constant parameter @xmath19 is associated with the frequency @xmath20 of the gw . in the case of a uniform magnetic field @xmath21 , which corresponds to @xmath22 , we may choose a system of units where the larmor frequency of the particles is @xmath23 . in this case @xmath19 is the ratio of the gw frequency to the larmor frequency i.e. @xmath24 . by substituting in ( [ hoo ] ) the above specific expressions and performing a canonical transformation to new canonical variables ( @xmath25 ) ( for details see @xcite ) we get @xmath26 the variables @xmath27 are associated with the gyro motion of the particles in the magnetic field , @xmath28 , where @xmath29 is the particle s energy and @xmath30 its velocity . the quantity @xmath31 is a constant of motion due to the symmetries of the original system . for the hamiltonian ( [ ham1 ] ) we can easily obtain the periodic solution @xmath32 the solution ( [ po ] ) corresponds to rest particles since it provides @xmath33 ( the energy of the rest mass ) . because of the restriction @xmath34 we should set @xmath35 in order our system to possess the above special solution . thus , the hamiltonian is now written in the form @xmath36 and the solution ( [ po ] ) is periodic of period @xmath37 . by considering the variations @xmath38 corresponding to the variables @xmath39 and @xmath40 , respectively , the variational equations for the periodic solution ( [ po ] ) are @xmath41 furthermore , let @xmath42 be the monodromy matrix . considering the unperturbed system ( i.e. @xmath43 ) we obtain from the monodromy matrix @xmath42 the stability index @xmath44 the condition @xmath45 , ensures the linear stability of the periodic orbits for all values @xmath19 except for @xmath46 . these values are candidates for the generation of a parametric resonance . in the perturbed system we examine the stability by following the methodology described by yakubovich and starzhinskii @xcite , p.357 . the variational equations ( [ veqs0 ] ) are of the form @xmath47 and the fundamental @xmath48 matrix of solutions is @xmath49 where @xmath50 is a @xmath51-periodic matrix , @xmath52 is the matrix of coefficients of the linear system ( [ veqs0 ] ) and @xmath53 are constant matrices calculated from the variational equations ( [ veqs1 ] ) on the periodic orbit . after some calculations we find @xmath54 . the nonzero components of the matrix @xmath55 are @xmath56 where @xmath57 and @xmath58 . the stability of the periodic solution is given by the eigenvalues of the matrix @xmath59 defined in ( [ veqs1 ] ) , which are @xmath60 conclusively for every value of @xmath14,@xmath10 and @xmath61 the eigenvalues @xmath62 are purely imaginary and the periodic orbit that corresponds to the rest particles is stable . therefore , for arbitrarily small amplitude values @xmath14 , instability ( parametric resonance ) may occur only at @xmath2 . at this value the above method fails to classify the periodic orbit and we proceed to a numerical investigation . namely , we obtain the monodromy matrix by integrating numerically the variational equations beside the canonical equations of the hamiltonian ( [ ham2 ] ) for various values of the amplitude @xmath14 of the gravitational wave and of the propagation angle @xmath10 . for some typical values of the propagation angle @xmath10.,width=302 ] ) for @xmath63 , @xmath64 and ( a ) @xmath65 , ( b ) @xmath2 , ( c ) @xmath66.,width=491 ] as it is shown in figure 1 , the periodic orbit becomes unstable in a v - shaped region of parameters centered at @xmath2 . we observe that the borders of the parametric resonance can be approximated by the fitting formula @xmath67 and the instability region broadens as @xmath68 , i.e. towards the parallel propagation . in figure 2 we present the poincare sections of ( [ ham2 ] ) on the plane @xmath69 for @xmath70 . the periodic orbit correspond to the fixed point of the sections . we see that at the parametric resonance at @xmath2 a pitchfork bifurcation of the periodic orbit ( [ po ] ) occurs . two stable periodic orbits bifurcate , which correspond to @xmath71 , while the periodic orbit ( [ po ] ) becomes unstable and the associated stable and unstable manifolds form a separatrix curve of figure - eight . the poincare section of system ( [ ham2 ] ) can be defined also in the plane of variables @xmath72 for @xmath73 . then the topology of the section in fig.2b is represented in the section @xmath72 as it is shown in fig.3a . the unstable periodic orbit ( [ po ] ) , indicated by p@xmath74 , appears at @xmath75 and its separatrix curve is located in the enegry domain @xmath76 . for any initial conditions the particle s energy @xmath77 shows regular bounded oscillation of librational type around @xmath78 or rotational type above the separatrix @xmath79 . the stable periodic orbit , which is generated by the pitchfork bifurcation and indicated as p@xmath80 , corresponds to energy value @xmath81 . as it is illustrated in fig.3b , the value @xmath81 depends on both parameters @xmath14 and @xmath10 . it holds always @xmath82 and also we can obtain by fitting that @xmath83 for @xmath84 . consequently , as @xmath68 , the energy oscillations increase rapidly their amplitude and at @xmath85 all particles are ejected to infinite energy independently of the amplitude of the gw . this special case is studied in the following section . plane for @xmath2 , @xmath63 and @xmath64 . ( b ) the energy @xmath81 of the bifurcated stable periodic solution as a function of the propagation angle @xmath10 . as @xmath68 we obtain that @xmath83.,scaledwidth=100.0% ] we consider @xmath85 , i.e. the gw propagates in parallel direction with the magnetic field . then the hamiltonian ( [ ham2 ] ) takes the form @xmath86 the corresponding canonical equations are : @xmath87 from equation ( [ h0a ] ) we obtain directly that @xmath88 and subsequently equations ( [ h0c ] ) and ( [ h0d ] ) consist a linear non - autonomous system . providing a solution @xmath89 the system turns to be integrable for any values of the parameters @xmath14 and @xmath19 . the poincare section in fig.4a , indicates the regular oscillations of the particle s energy for all initial conditions . such a situation changes at the parametric resonance for @xmath2 and the dynamics is depicted by the poincare section in fig.4b . we obtain that the stable and the unstable manifolds which originate from the unstable periodic orbit ( [ po ] ) are located at @xmath90 and @xmath91 ( @xmath92 ) , respectively , and extend from @xmath33 up to infinity . the dynamics becomes purely hyperbolic . ) for ( a ) @xmath93 , @xmath94 and ( b ) @xmath2 , @xmath95 . the up and down directed arrows indicate the unstable and stable manifolds , respectively.,width=510 ] the linear system of eqs . ( [ h0c ] ) and ( [ h0d ] ) is not directly solvable . by differentiating and introducing the variable transformation @xmath96 and @xmath97 , we derive the separable system of second order equations @xmath98 if we introduce the new time variable @xmath99 and omit the terms @xmath100 the above equations take the form @xmath101 where @xmath102 stands for either @xmath103 or @xmath104 , the minus sign refers to @xmath104 and the plus sign to @xmath103 and prime denotes differentiation with respect to @xmath105 . equation ( [ meq ] ) is a mathieu type equation with solutions of the form @xmath106 , where @xmath107 is a periodic function @xcite . in the parameter space @xmath108 stable and unstable regions are defined where @xmath109=0 $ ] and @xmath109\neq 0 $ ] respectively . for the actual system ( [ ham20 ] ) the unstable region is the v - shaped region shown in fig . 1 for @xmath85 . in fig.5a we present an example of the oscillations of the variable @xmath110 which correspond to the trajectory along the unstable manifold for @xmath2 . such a variation is associated with an increase of the particle s energy @xmath111 . from ( [ ham20 ] ) it follows that @xmath112 the increase of @xmath77 is exponential ( see fig . 4b ) and the numerical study suggest that for @xmath113 , the associated exponent coincides with the amplitude of the gw , i.e. @xmath114 of the gyro motion of a particle in the parametric resonance ( b ) the exponential increase of energy @xmath115 for the evolution along the unstable manifold . the corresponding parameters are @xmath2 , @xmath63 and @xmath116.,width=453 ] in previous studies , mentioned in the introduction , the acceleration of the particles is a result of the chaotic diffusion . however , wide chaotic regions are formed only for high energy levels and in the presence of very strong gravitational waves ( @xmath117 ) @xcite . especially , for small values of the propagation angle @xmath118 chaotic regions appear in phase space only as very thin bounded zones around islands formed according to the poncare - birkhoff theorem . consequently , acceleration of low energy particles under the perturbation caused by weak gws can occur only through the parametric resonance ( @xmath2 ) at the integrable case ( @xmath85 ) . in a more realistic situation , the direction and the magnitude of the magnetic field met by the particles is not constant . according to the results obtained in the previous sections , the charged particles will accelerate when the magnetic field becomes almost parallel to the direction of the gravitational wave and have a magnitude such that @xmath119 . a simple simulation that illustrates this situation is described below . we consider that the propagation angle between the gw , with constant @xmath2 , and the direction of the magnetic field varies in time linearly as @xmath120 the time variable @xmath121 is the coordinate time @xmath122 in the original form ( [ hoo ] ) of the system . the slope parameter @xmath123 controls the time span spent by the particle in an interval ( @xmath124 ) and , particularly , the critical time interval @xmath125 where @xmath126 . a typical example of the evolution of a particle under the above condition is shown in fig.6a . we observe that the particle , which starts with low energy @xmath127 ( particularly @xmath128 ) and @xmath129 , obeys small oscillations for approximately @xmath130 . when the system approaches the integrable hyperbolic dynamics at @xmath85 , the particle accelerates and a significant jump in particle s energy occurs . leaving the hyperbolic domain ( @xmath131 ) the particle is captured in a regular phase space domain with corresponding energy @xmath132 significantly greater than @xmath133 ( particularly , @xmath134 ) . the total increase in energy , @xmath135 , depends on the critical time interval @xmath125 and the @xmath10-width , @xmath136 , where the hyperbolic structure is effective . for the particular simple model it holds that @xmath137 . certainly , the most significant parameter for the acceleration is the amplitude @xmath14 of the gw which is associated with the exponential rate of the acceleration along the unstable manifolds in phase space according to the estimation ( [ gammaexp ] ) . in fig.6b we present the relation between @xmath14 and @xmath135 derived by the numerical simulation for some particular values of the parameter @xmath123 . we observe that , if the critical time @xmath125 is sufficiently large ( or @xmath123 is small ) , the energy jump is notable even for very small values of @xmath14 . the same qualitative results are obtained when we keep constant the propagation angle @xmath85 and change the parameter @xmath19 passing through the value @xmath2 . we have shown that gravitational waves with frequency @xmath20 and amplitude @xmath138 can accelerate charged particles to very high energies when two conditions are met ( a ) the gravitational wave propagates along the external magnetic field ( @xmath139 ) and ( b ) the ratio of the gravitational wave frequency to gyro frequency is about 2 ( @xmath140 . the model used has a number of important simplifications : ( 1 ) we assume that the magnetic field is constant ( 2 ) the energy absorption by the particles is small compared to the energy curried by the gravitational wave and ( 3 ) we ignore the collective plasma effects associated with the particle acceleration . retaining the above simplifications we may proceed to estimate the energy gained by the particles in a realistic astrophysical system . we assume that a volume of length @xmath141 is close to a source of gravitational waves . the volume is filled with plasma with density @xmath142 and the magnetic field is in a fully turbulent state . the gravitational wave is crossing the volume in few seconds and every time it travels along the ambient field ( @xmath143 ) with almost zero strength its energy increases . the zero strength magnetic field , which is associated to null surfaces , is an important requirement since the gravitational wave has low frequency ( of few khz ) and acceleration is possible only for @xmath119 ) . the energy increase is fast and can be substantial ( @xmath144 ) if the amplitude of the gravitational wave is @xmath145 . therefore the particles diffuse in energy space not by small random steps , as it was the case in the stochastic interaction studied earlier ( @xcite- @xcite ) , but through large random energy jumps . it is beyond the scope of this article to present a detailed analysis of the above process but we can present a rough estimate of the energy transferred from the gravitational wave to the plasma . let us assume that a particle travel a distance @xmath146 ( its trajectory is not a straight line ) before exiting the interaction volume . let us also assume that only in small parts of its trajectory ( @xmath147 a few kilometers ) the particle is able to meet the conditions needed for its acceleration . for a small portion of protons , e.g @xmath148 ) , which participate in the interaction , these regions can be a fraction of @xmath149 and the energy gain will be @xmath150 ( or @xmath151 ) , where @xmath152 is the mass of proton . the number of particles participating in this interaction can be as many as @xmath153 the total energy lost by the gravitational wave is , approximately , @xmath154 ergs , which constitute a small fraction of its total energy @xmath155 ergs ( see @xcite ) . according to the model , the acceleration conditions ( @xmath139 , @xmath119 ) can be met sporadically along the trajectory of a charged particle in the vicinity of a gravitational wave source when the magnetized plasma is in a turbulent state . a few particles can then reach very high energies in a few seconds absorbing only a very small fraction of the gravitational wave . it remains to be seen , by performing further simulations , if the process described above for particle acceleration can be significant in a realistic astrophysical environment . in this case the electromagnetic emission from the accelerated particles can be a precursor of the gravitational wave for easier detection .	the acceleration of charged particles in the presence of a magnetic field and gravitational waves is under consideration . it is shown that the weak gravitational waves can cause the acceleration of low energy particles under appropriate conditions . such conditions may be satisfied close to the source of the gravitational waves if the magnetized plasma is in a turbulent state . university of thessaloniki , dept of physics , 54124 thessaloniki , greece , email : [email protected]
recent experimental advances have permitted to probe electron transport processes at the atomic scale.@xcite junctions can be formed that support current flow through atom - sized constrictions or even single molecules . atomic vibrations become detectable and very dependable on the environmental temperature . according to the distance between electrodes , the conductance can vary several orders of magnitude when the applied voltages are small , typically below the ev scale . this behavior is due to the exponential dependence of current with distance when the conductance is due to an electron tunneling process . however , at short electrode distances , the current levels off and saturates : the contact regime is reached . the conductance is maximum in this case and a high - conductance regime is attained . the physics in these two regimes can be very different . the low - conductance regime has been thoroughly studied with the scanning tunneling microscope ( stm ) . the initial inelastic effects were realized by showing the increase in conductance on an acetylene molecule when the bias voltage matched the c h stretch mode.@xcite the proof that the mode was indeed excited was the isotopical effect that the changes of conductance showed when replacing c@xmath0h@xmath0 by c@xmath0d@xmath0 . this finding paved the way to vibrational spectroscopy with sub - ngstrm spatial resolution , permitting the identification of chemical components of matter on the atomic scale.@xcite the first experimental evidence of mode excitation in the high - conductance regime was achieved in monatomic gold wires.@xcite the conductance of the wires showed clear reductions at thresholds that were proven to originate in the backscattering of electrons from some selected vibrations of the wires.@xcite similarly , experiments with the break junction geometry have also revealed signatures in the conductance related to several vibrational modes of a single h@xmath0 molecule trapped between the electrodes.@xcite the emerging picture is that in the tunneling or low - conductance regime , the excitation of vibrations leads to increases in conductance at the corresponding voltage thresholds , while in the contact or high - conductance regime , the effect of vibrations is to reduce the conductance . theoretical studies in the weak electron - vibration coupling regime have shown that the lowest order expansion@xcite is capable of correlating this behavior with a single parameter : the eigenchannel transmission probability @xmath1.@xcite in the simplified case of a single electronic level connected with two electrodes under symmetrical conditions , the inelastic effects ( of a vibrationally mediated on - site modulation ) go from increases in the conductance for @xmath2 to decreases for @xmath3 . in this way , the behavior of the inelastic conductance would define the crossover from tunneling to contact . there is experimental evidence showing that this picture is indeed more complex . the excitation of the o o stretch mode of the chemisorbed o@xmath0 molecule on ag(110 ) @xcite leads to a decrease of the tunneling current ( instead of an increase ) in opposition with most cases in the low - conductance regime.@xcite the aim of the present work is to analyze the continuous evolution from tunneling to contact in a model system constituted by a junction of gold atoms , which provides an almost perfect realization of a single transmission channel system . the definition of when a given atomic system correspond to one of the two cases analyzed above is already problematic , hence we address this issue by investigating the behavior of different properties of the junction with the interatomic distance . initially , we are interested in studying the crossover from tunneling to contact by evaluating the total energy , the strain , and the modification of vibrational modes as the electrode distance decreases . this allows us to find a range of distances where the junction behaves as either two independent systems or a strongly coupled one . the second part of this work evaluates the effect of the interatomic distance in electron transmission ; this allows us to study the elastic conductance within landauer s formalism . the correlation of the transmission against the interatomic relaxation permits a clear identification of both regimes as well as the transition region . finally , the inelastic properties of the conductance are studied in the different regimes . the inelastic signals are interpreted in a simplified model that captures the calculated behavior and illustrates the fundamental concepts . the continuous transition from tunneling to contact is experimentally challenging , since most metallic point contacts ( including au ) usually exhibits a sudden jump in the conductance when the surfaces are brought into contact.@xcite on the other hand , experiments with a low - temperature stm on cu(111 ) and ag(111 ) surfaces have shown that both sharp jumps as well as smooth variations can be obtained in the crossover from tunneling to contact : when the tip is approached over a clean surface one observes a jump in conductance ( related to the transfer of the tip - atom to the surface ) , whereas over an isolated metallic adatom the evolution is smooth and reversible.@xcite to our knowledge there is no measurement of the evolution of the inelastic signals in the formation of a metallic point contact , likely owing the relative weak effect ( conductance changes expected to be less than 1% ) and difficulties with noise . despite this we envision that our idealized model system is not unrealistic , and can provide a useful framework for investigating the complicated interplay between chemical bonding , electron conduction , atomic vibrations , etc . our first - principles treatment further addresses all of these issues in a unified way to provide quantitative predictions . representation of two au(100 ) surfaces sandwiching two pyramids pointing towards each other . the characteristic electrode separation @xmath4 is measured between the second - topmost surface layers , since the surface layer itself is relaxed and hence deviates on the decimals from the bulk values . the interatomic distance between the apex atoms is denoted @xmath5 . ] the present work can be divided by the different methods that we have used . in order to study the structural properties of the atomic junction the standard dft siesta@xcite method is used . the elastic conductance is evaluated from the transmission function of the atomic junction calculated with transiesta,@xcite and the inelastic contribution is performed using the method presented in ref . the system representing the atomic junction is depicted in fig . [ fig : tunnel2contact ] . we consider a periodic supercell with a @xmath6 representation of two au(100 ) surfaces sandwiching two pyramids pointing towards each other . the characteristic electrode separation @xmath4 will be measured between the second - topmost surface layers , since the surface layer itself is relaxed and hence deviates on the decimals from the bulk values . the corresponding calculations with the siesta method are carried out using a single zeta plus polarization ( szp ) basis with a confining energy of 0.01 ry ( corresponding to the 5@xmath5 and 6@xmath7 states of a free au atom ) , the generalized gradient approximation ( gga ) for the exchange - correlation functional , a cutoff energy of 200 ry for the real - space grid integrations , and the @xmath8-point approximation for the sampling of the three - dimensional brillouin zone . the interaction between the valence electrons and the ionic cores are described by a standard norm - conserving troullier - martins pseudo - potential generated from a relativistic atomic calculation . the calculations of the vibrations are performed by diagonalization of the dynamical matrix extracted from finite differences ( with corrections for the egg - box effect , i.e. , the movement of basis orbitals following the displaced atoms with respect to the real space integration grid).@xcite as the active atoms we consider initially for pedagogical purposes just the two apex atoms and compare afterwards the results when the vibrational region is enlarged . the transport calculation naturally considers infinite electrodes by including the dft self - energy calculated for infinite atomistic leads in the conduction equations.@xcite since we are here interested in the low - bias regime ( of the order of the vibrational frequencies ) it suffices to calculate the electronic structure in equilibrium in order to describe the elastic transport properties . while the transmission generally involves a sampling over @xmath9-points we here approximate it with its @xmath8-point value ; this has previously been shown to be a reasonable approximation for supercells of similar dimensions in the case of atomic gold wires.@xcite finally , the inelastic transport calculations are performed using the nonequilibrium green s function ( negf ) formalism combined with the electrode couplings @xmath10 extracted from the transiesta calculations and the electron - vibration couplings @xmath11 ( corresponding to modes @xmath12 with energies @xmath13 ) from the finite - difference method.@xcite according to the lowest order expansion ( loe)@xcite the inelastic current reads @xmath14 \nonumber \\ & & + \sum_\lambda { \cal i}^\mathrm{sym}_\lambda(v , t , \langle n_\lambda\rangle ) \ , \mathrm{tr}\big [ \mathbf{g}^\dag \mathbf{\gamma}_l \mathbf{g } \big\ { \mathbf{m}^\lambda \mathbf{a}_r \mathbf{m}^\lambda + \frac{i}{2 } ( \mathbf{\gamma}_r \mathbf{g}^\dag \mathbf{m}^\lambda \mathbf{a } \mathbf{m}^\lambda - \mathrm{h.c . } ) \big\ } \big ] \nonumber \\ & & + \sum_\lambda{\cal i}^\mathrm{asym}_\lambda(v , t)\ , \mathrm{tr}\big [ \mathbf{g}^\dag \mathbf{\gamma}_l \mathbf{g } \big\ { \mathbf{\gamma}_r \mathbf{g}^\dag \mathbf{m}^\lambda ( \mathbf{a}_r-\mathbf{a}_l ) \mathbf{m}^\lambda + \mathrm{h.c . } \big\ } \big ] , \label{eq : current1 } \\ { \cal i } ^\mathrm{sym}_\lambda & = & \frac{e } { \pi \hbar } \left ( { 2 e v } \langle n_\lambda \rangle + \frac{\hbar \omega_\lambda-{e v}}{e^{\beta(\hbar \omega_\lambda - e v)}-1}- \frac{{\hbar \omega_\lambda}+{e v}}{e^{\beta(\hbar \omega_\lambda+e v)}-1}\right ) , \label{eq : currentnormal } \\ { \cal i}^{\mathrm{asym}}_\lambda&= & \frac{e}{\hbar}\int_{-\infty}^{\infty } \frac{d\varepsilon}{2\pi } \left[n_\textrm{f}(\varepsilon ) -n_\textrm{f}(\varepsilon - e v)\right ] \ , { \cal h}_{\varepsilon ' } \{{n_\textrm{f}(\varepsilon'+\hbar \omega_\lambda ) -n_\textrm{f}(\varepsilon'-\hbar \omega_\lambda)}\}(\varepsilon ) , \label{eq : currenthilbert}\end{aligned}\ ] ] where g@xmath15h is the conductance quantum , @xmath16 the external bias voltage , @xmath17 the occupation of mode @xmath12 , @xmath18 the fermi function , @xmath19 the hilbert transform , and @xmath20 the inverse temperature . the retarded green s function @xmath21 , the spectral function @xmath22 , as well as the electrode couplings @xmath10 are all evaluated at the fermi energy in the loe scheme . for convenience we have also defined the quantities @xmath23 such that @xmath24 . the sums in eq . ( [ eq : loecurrent ] ) runs over all modes @xmath12 in the vibrational region . for a symmetric system ( such as the present one for the atomic junction ) it can be shown that the asymmetric terms in the current expression vanishes . furthermore , at low temperatures ( @xmath25 ) and in the externally damped limit ( @xmath26 ) the inelastic conductance change from each mode @xmath12 ( beyond the threshold voltage @xmath27 ) is given by @xmath28.\end{aligned}\ ] ] from this expression we note that @xmath29 can be either positive or negative , depending on sign of the trace . where the force experience a significant increase . ( a ) , ( b ) , and ( c ) are three representative electrode separations of the three regimes considered in this article . ] as the electrode separation is decreased , we relax in each step the apex atoms , the base atoms of the pyramids , and the first - layer atoms until residual forces are smaller than 0.02 ev / . this allows us to obtain the evolution of the ( kohn - sham ) total energy @xmath30 of the system as a function of the electrode distance , see fig . [ fig : etotvslength ] . we find that the energy is reduced ( of the order 1 ev ) by the attractive interaction between the apex atoms , due to the formation of a covalent bond at short distances , fig . [ fig : etotvslength](a ) . the slope of the energy presents a rapid change for distances shorter than @xmath31 . this is more clearly seen in the lower part of fig . [ fig : etotvslength ] where the strain or force on the unit cell is represented . this force is evaluated as the numerical derivative of the total energy with respect to electrode separation . here , the onset of chemical interactions is clearly seen around @xmath31 , fig . [ fig : etotvslength](b ) , where the force experiences a significant increase reaching a maximum at @xmath32 . the evolution of the interaction between the apex atoms with distance is also revealed in the study of the vibrational modes . this is presented in fig . [ fig : freqvslength ] , where the blue connected data points correspond to the 6 modes where only the apex atoms vibrate , and the red crosses to the 30 modes where also the pyramid bases vibrate . these modes follow different behavior with the electrode separation . in the following we analyze the simplest case of just the two apex atoms . generally , two longitudinal stretch modes ( represented with connected circles in fig . [ fig : freqvslength ] ) line up the highest in energy . for an electrode distance larger than @xmath33 these correspond to the isolated ( i.e. , decoupled and hence degenerate ) stretch modes of each apex atom , fig . [ fig : freqvslength](c ) . as the electrodes are approached , the attractive apex - apex interaction leads to a slight displacement of the apex atoms away from the base of the pyramids . the consequence is a small weakening of the apex - atom coupling to the base which results in decreasing frequencies , i.e. , to softening of the modes . another consequence of the increasing interaction is the splitting of the degenerate modes into a symmetric ( out of phase ) and an antisymmetric ( in phase ) mode . we will refer to these as the alternating bond length ( abl ) mode@xcite and the center of mass ( cm ) mode , respectively . when the electrode separation reaches the region between @xmath34 and @xmath31 the frequencies drop significantly , fig . [ fig : freqvslength](b ) . this points again at the chemical interaction crossover that we presented in the previous paragraph : now the interaction between the apex atoms becomes comparable with the interaction with the electrodes and hence weakens the stretch modes initially set by the interaction between the apex atom with the base of the pyramid . as the apex - apex interaction grows larger , the modes start to increase in frequency and further show an significant split , fig . [ fig : freqvslength](a ) . the behavior of the two stretch modes of fig . [ fig : freqvslength ] is easily understood with a simple one - dimensional elastic model of two masses , each coupled to infinite - mass system with a spring constant @xmath35 , and interconnected by another spring constant @xmath36 . the frequencies of the two stretch modes are then @xmath37 ( in phase ) and @xmath38 ( out of phase ) , where @xmath39 is the mass of each atom . note that in the tunneling regime the apex - apex interaction is attractive , cf . [ fig : etotvslength ] , which would correspond to a negative value of @xmath36 . when the bond has been formed , @xmath36 can be represented as classical ( positive ) spring constant . this model essentially captures the evolution of the stretch modes . in particular , the sign change of @xmath36 at the chemical instability explains the mode crossing between @xmath34 and @xmath31 , fig . [ fig : freqvslength](b ) , and why the cm mode has a higher frequency than the abl mode in the tunneling case , and vice versa in the contact case . the analysis of the modes with electrode distance thus permits us to recover the same range of distances with the chemical crossover as in the preceding section concerning the total energy and strain . this identification is also possible from the more realistic calculation that includes the vibration of the base atoms . ( blue disks ) and apex - apex distance @xmath5 ( red crosses ) versus electrode separation @xmath4 . in the tunneling regime the transmission decays exponentially with separation as indicated with the dashed line ( corresponding to a tunneling parameter @xmath40 ) . the point at ( a ) corresponds well with the contact region of transmission one and closest apex separation , ( b ) is near half transmission and the instability in apex separation , ( c ) is finally the tunneling regime , where the apex atoms are independent . ] in this study we are interested in the low - bias regime . hence the elastic conductance is determined via landauer s formula by the transmission at the fermi energy @xmath41 . as expected for the gold contact , we find that the total transmission is essentially due to a single eigenchannel ( for the geometries considered here the contribution from the secondary channel is at least three orders of magnitude smaller ) . figure [ fig : transvsapexdist ] plots the transmission @xmath1 and the apex - apex distance @xmath5 as a function of electrode separation @xmath4 . in the tunneling regime the transmission is characterized by an exponential decay with separation . it is instructive to compare this with the transmission probability @xmath42 for a rectangular one - dimensional barrier , where @xmath43 is a characteristic tunneling length , @xmath44 the apparent barrier height , and @xmath45 the barrier width ( valid for @xmath46 ) . as shown in fig . [ fig : transvsapexdist ] , an exponential fit to the calculated tunneling data leads to a tunneling parameter @xmath40 which would correspond to an apparent barrier height of the order @xmath47 ev . compared with measurements of the work function on perfectly flat au surfaces ( 5.315.47 ev)@xcite this value is certainly high . on the other hand @xmath44 is not very well determined from an exponential fit to data spanning only one decade . another contribution to a relatively large barrier height could be geometric effects from the pyramidal shapes . the deviation from the exponential tunneling behavior ( visible around @xmath31 ) is a clear indication of the crossover to contact . the contact regime is characterized by a constant transmission equal to unity since an atomic gold junction has effectively only one conduction channel . the value @xmath48 to define the crossover between contact and tunneling is somewhat arbitrary , however a detailed comparison with the previous section justifies this definition . indeed , fig . [ fig : transvsapexdist ] also shows the behavior of the apex - atom distance @xmath5 with electrode separation , permitting us to make contact with the chemical crossover defined previously . between @xmath34 and 16.0 we find that the apex - apex distance has changed by almost 0.7 . this shows that at these electrode distances , there is an instability that drive the formation of a covalent bond between apex atoms . this agrees with the conclusion from both total energy , strain and frequency calculations that the crossover takes place between 15.8 and 16.0 . from fig . [ fig : transvsapexdist ] we identify a transmission 1/2 associated with @xmath49 ( @xmath50 ) hence permitting us to identify the crossover from tunneling to contact with the chemical crossover . the behavior of the inelastic contributions to conductance is very different in the two studied regimes . in the tunneling regime the opening of the inelastic channel enhances the conductance of the system , while the creation of a vibrational excitation in a high conductance regime is a source of backscattering that decreases the conductance . figure [ fig : d2idv2 ] shows the calculated change in conductance ( second derivative of the current with respect to bias voltage @xmath51 ) for the contact , crossover and tunneling regions . these three typical cases labeled ( a ) , ( b ) , and ( c ) , respectively are indicated in the previous figs . [ fig : etotvslength][fig : transvsapexdist ] for easy reference . we investigate how the inelastic conductance change depends on how many atoms in the junction that are considered active : in fig . [ fig : d2idv2 ] the thick blue line is the spectrum corresponding to only the two apex atoms vibrating , the dashed red curve to the 10 pyramid atoms vibrating , and the dotted black curve to the pyramids and first - layer atoms vibrating ( 42 atoms ) . in this way we follow the convergence of the calculations as the vibrational region is gradually enlarged . the essential data from these calculations are summarized in tab . [ tab : d2idv2 ] . from the simplest case when only the two apex atoms are vibrating , we arrive at the conclusion that only the two longitudinal stretch modes contribute to the change in conductance , leading to the qualitative known result of increase of the conductance in tunneling regime and decrease in contact . the crossover case fig . [ fig : d2idv2](b ) presents a combination of an increase in conductance from the abl mode and a decrease from the cm mode . this behavior is a signature of the different processes of conduction . in the tunneling case , the tunneling process is determined by the more slowly - decaying components of the electron wave function of the surface . because of the exponential tunneling probability dependence on distance a mode that modulates the tunneling gap is expected to contribute positively to the conductance.@xcite indeed this is the case for the abl mode . furthermore , the cm mode that correspond to a fixed apex - apex distance can not contribute positively , neither the transverse modes because none of them _ decrease _ the apex - apex distance from the equilibrium position during a vibration period . instead , the cm mode is found to contribute negatively to the conductance , cf . tab . [ tab : d2idv2 ] . a simplified model presented in the next section will explain this observation . in the contact case , the electronic structure responsible for the conduction process is largely concentrated upon the apex atom , hence the transport is being modified by the motion of basically only these atoms . indeed both the abl and cm modes lead to drops in the conductance as is evident from fig . [ fig : d2idv2](a ) and tab . [ tab : d2idv2 ] . the transverse modes give essentially no signal ; this is similar to the findings for atomic gold wires where the transverse modes can not couple because of symmetry.@xcite figure [ fig : d2idv2 ] shows how the inelastic spectrum is modified if we increase the vibrational region by allowing more atoms to vibrate . in the tunneling and contact cases we see that the single main peak splits up into a number of peaks , indicating that the apex vibrations are actually coupled with the vibrations in the bulk . for the contact case the broadening of the signals is expected to be directly influenced by the phonon density of states of the bulk . as was first shown by yanson,@xcite the spectroscopy of microcontacts at low temperatures a technique nowadays referred to as point contact spectroscopy reveals a signal in @xmath51 which is a direct measurement of the eliashberg function @xmath52 , i.e. , roughly speaking the product of the squared electron - phonon coupling matrix element @xmath53 and the phonon density of states @xmath54 , averaged over the fermi sphere.@xcite in the case of microcontacts the measured signal is predominantly due to the transverse modes . this is in contrast to our case of the atomic point contact , where we only find signals from the longitudinal modes . however , from fig . [ fig : d2idv2](a ) we notice a signal broadening by increasing the vibrational region , pointing towards the vibrational coupling to the bulk modes . in the crossover region between tunneling and contact , fig . [ fig : d2idv2](b ) shows a dramatic change depending on the size of the vibrational region . different modes give positive or negative contributions in the conductance , but in such a way that they lead to an overall absence of ( or relatively small ) variation in the conductance , cf . tab . [ tab : d2idv2 ] . comparing the total change in conductance induced by all modes ( for the tunneling , crossover , and contact situations ) , we find that the calculations with different vibrational regions give almost the same results . as found in tab . [ tab : d2idv2 ] , we thus conclude that to a first approximation we can describe @xmath55 , i.e. , the overall conductance change can be estimated with the minimal vibrational region ( the two apex atoms ) . this simple approach does however not accurately describe details of the @xmath51 spectrum . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] the effect of the tunneling to contact crossover has important implications in the inelastic conductance since in the first case the inelastic effects trend to increase and in the second case to diminish the electron conduction . from the results of the previous section , we have seen that this crossover roughly takes place at the same range of distances as in the elastic conductance case . by looking at the transmission in the elastic conductance case , we conclude that when the transmission is @xmath48 we should also be near the crossover between tunneling to contact in the inelastic case one . this finding is similar to the crossover found for the single - state impurity model analyzed in ref . however , in the present case , the system is not obviously modeled with a single - state impurity . instead we can easily reproduce the same kind of analysis for a slightly more sophisticated model , where two impurities are connected to reservoirs and interacts via a hopping term between them . under symmetric conditions this system is described by @xmath56,\quad \mathbf{\gamma}_l = \left[\begin{array}{cc } \gamma & 0 \\ 0 & 0 \end{array}\right],\quad \mathbf{\gamma}_r = \left[\begin{array}{cc } 0 & 0 \\ 0 & \gamma \end{array}\right],\end{aligned}\ ] ] where the hamiltonian @xmath57 includes on - site energies @xmath58 and a hopping matrix element @xmath59 . the level broadening functions @xmath60 describes the coupling of the sites to the contacts @xmath61 with strength @xmath62 ( which in the wide band approximation are considered energy independent ) . the corresponding retarded green s function is @xmath63^{-1}\nonumber\\ & = & \frac{2}{(2\delta\varepsilon + i \gamma)^2 - 4 t^2}\left [ \begin{array}{cc } 2\delta\varepsilon+i \gamma & 2 t \\ 2 t & 2\delta\varepsilon+ i\gamma \end{array } \right],\qquad\end{aligned}\ ] ] where in our case @xmath64 holds since the level positions would be close to the fermi energy @xmath41 ( the on - resonance case ) . the transmission becomes @xmath65 where perfect transmission @xmath66 is obtained for @xmath67 . inspired by our electron - phonon coupling matrices obtained from the full dft calculations , we assign the following forms to the longitudinal abl and cm mode couplings @xmath68,\quad \mathbf{m}_\textrm{cm } = \left [ \begin{array}{cc } m_3 & 0 \\ 0 & -m_3 \end{array } \right].\end{aligned}\ ] ] the abl mode is symmetric and generally described by two coupling strengths : @xmath69 represents an on - site modification via a change in the electrode coupling , whereas @xmath70 is a modulation of the hopping between the apexes . correspondingly , the cm mode which is asymmetric bears an asymmetric on - site modulation @xmath71 and no hopping modulation ( fixed apex - apex distance ) . with these expressions we can simply evaluate eq . ( [ eq : conductance ] ) to find the following inelastic conductance changes @xmath72 we first discuss the conclusions to be drawn about the abl mode . notice that @xmath73 is only weakly dependent on the the on - site coupling element @xmath69 and vanishes on resonance ( @xmath74 ) . in the tunneling limit ( @xmath75 ) we find that @xmath76 i.e. , the abl mode gives a _ positive _ contribution to the conductance proportional to the square of coupling strength @xmath70 . in the contact limit ( @xmath77 ) we find @xmath78 i.e. , the abl mode gives here a _ negative _ contribution . the exact crossover between an increase and a decrease is determined by solving @xmath79 , which indeed yields @xmath48 as is the case for the single - site case.@xcite next , we see from eq . ( [ eq : cmcontrib ] ) that the conductance change @xmath80 from the cm mode is always negative ( i.e. , the cm mode backscatters even in the tunneling regime ) . these results thus permit to rationalize the crossover from tunneling to contact for the inelastic conductance as found numerically in sec . [ sec : inelasticconductance]as taking place around a transmission of @xmath48 . the evolution of the inelastic signals from the tunneling to contact regimes has been studied through dft calculations . we have compared our results with the crossover between bonding and rupture of the atomic junction by studying the geometric and electronic structures of the junction , together with the strength of the atomic bonds and the associated vibrational frequencies . this permitted us to find a typical transition distance between electrodes where a small change leads to a large readjustment of the apex - apex atom distance , as well as a change of the strength of interactions as revealed by the total energy , the strain and the frequencies of the system s modes . the conductance has been calculated by first evaluating the transmission of electrons through the system , and second by calculating the conductance change due to the excitation of vibrations . as found in previous studies@xcite the change in conductance due to inelastic effects permits to characterize the crossover from tunneling to contact . the most notorious effect being a decrease of conductance in the contact regime to an increase in the tunneling one when the bias voltage exceeds the vibrational thresholds . our dft - based calculations show that the effect of vibrational modes in the @xmath81 spectra is rather complex , in particular when modes localized in the contact region are permitted to extend into the electrodes . as an example , we find that certain modes can give rise to decreases in conductance when in the tunneling regime , opposite to the above mentioned result . whereas details in the inelastic spectrum depends sensitively on the size of the vibrational region , we find that the magnitude of the overall change in conductance can actually be reasonably described with just the minimal case where only the apex atoms vibrate . this means that while the modes are rather delocalized the region of inelastic scattering is localized around the apex atoms . by comparing our results with a simplified model , we conclude that in this single eigenchannel problem the tunneling to contact crossover takes place exactly at @xmath48 , in agreement with the findings for the elastic conduction process and the chemical crossover . hence , we can trace back the origin of the conduction process , both in the presence and absence of vibrational excitation , to the same type of underlying electron structure that determine the electrode s chemical interaction and the electron conductance.@xcite the authors acknowledge many valuable discussions with a .- this work , as a part of the european science foundation eurocores programme sasmec , was partially supported by funds from the snf and the ec 6th framework programme . computational resources were provided by the danish center for scientific computing ( dcsc ) .	the evolution of electron conductance in the presence of inelastic effects is studied as an atomic gold contact is formed evolving from a low - conductance regime ( tunneling ) to a high - conductance regime ( contact ) . in order to characterize each regime , we perform density functional theory ( dft ) calculations to study the geometric and electronic structures , together with the strength of the atomic bonds and the associated vibrational frequencies . the conductance is calculated by first evaluating the transmission of electrons through the system , and secondly by calculating the conductance change due to the excitation of vibrations . as found in previous studies [ paulsson _ et al . _ , phys . rev . b. * 72 * , 201101(r ) ( 2005 ) ] the change in conductance due to inelastic effects permits to characterize the crossover from tunneling to contact . the most notorious effect being the crossover from an increase in conductance in the tunneling regime to a decrease in conductance in the contact regime when the bias voltage matches a vibrational threshold . our dft - based calculations actually show that the effect of vibrational modes in electron conductance is rather complex , in particular when modes localized in the contact region are permitted to extend into the electrodes . as an example , we find that certain modes can give rise to decreases in conductance when in the tunneling regime , opposite to the above mentioned result . whereas details in the inelastic spectrum depend on the size of the vibrational region , we show that the overall change in conductance is quantitatively well approximated by the simplest calculation where only the apex atoms are allowed to vibrate . our study is completed by the application of a simplified model where the relevant parameters are obtained from the above dft - based calculations .
studying the mass distribution of massive clusters at high redshifts is important as their existence puts strong constraints on possible cosmological models ( e.g. eke et al . 1996 ; bahcall & fan 1998 ; donahue et al . 1998 ) . furthermore , these systems may be young and thus much can be learned about the formation of massive clusters of galaxies . massive structures in the universe induce a systematic distortion of the images of faint background sources , providing a powerful diagnostic to study their mass distributions ( e.g. mellier 1999 ) . the weak lensing signal allows us to map the projected mass surface density of the lens ( e.g. kaiser & squires 1993 ) , and determine its mass . although the weak lensing mass estimate does not require assumptions about the dynamical state or geometry , it does require a good estimate for the redshift distribution of the faint sources . since tyson , wenk , & valdes ( 1990 ) succesfully measured the systematic distortion of faint galaxies by a foreground cluster , many massive clusters of galaxies have been studied ( e.g. bonnet , mellier , & fort 1994 ; fahlman et al . 1994 ; squires et al . these studies concentrated on massive clusters with redshifts @xmath8 . smail et al . ( 1994 ) tried to measure the weak lensing signal for the optically selected , high redshift cluster cl 1603 + 43 @xmath9 . its observed velocity dispersion is @xmath10 km / s ( postman , lubin , & oke 1998 ) . smail et al . ( 1994 ) used source galaxies down to a limiting magnitude of @xmath11 , but failed to detect a signal . their results are consistent with a lower mass for the cluster . this is supported by the cluster s x - ray luminosity of @xmath12 ergs / s ( castander et al . 1994 ) , which is an order of magnitude lower than the x - ray luminosity of the brightest , high redshift , x - ray selected clusters the x - ray emission of clusters of galaxies provides an efficient means to select massive clusters of galaxies . since the work of smail et al . ( 1994 ) , several high redshift , x - ray luminous clusters have been discovered from einstein ( gioia & luppino 1994 ) and rosat observations ( henry et al . 1997 ; ebeling et al . 1999 ; rosati et al . 1998 ) . luppino & kaiser ( 1997 ; lk97 hereafter ) successfully detected the weak lensing signal from the x - ray selected , high redshift cluster ms 1054 - 03 using deep ground based imaging . clowe et al . ( 1998 ) reported on the detection of weak lensing signals for the x - ray selected , high redshift clusters ms 1137 + 66 and rxj1717 + 67 , from deep keck imaging . however , these studies suffered from the lack of knowledge of the redshift distribution of the faint sources . lensing studies of high redshift clusters are difficult because the lensing signal is low and most of the signal comes from small , faint galaxies . this is the regime where hst observations have great advantage over ground based imaging , because the background galaxies are much better resolved . as a result , higher number densities of sources can be reached , the images are not crowded , and more importantly the correction for the circularization by the psf is much smaller . especially for high redshift clusters , where the redshift of the lens approaches that of the sources , the lensing signal depends strongly on the redshift distribution of the sources . in this paper we present our weak lensing analysis of the cluster ms 1054 - 03 using a mosaic of deep wfpc2 images . it is the second cluster we studied using a mosaic of deep hst observations . in hoekstra ( 1998 , hfks98 hereafter ) we presented the analysis of cl 1358 + 62 , a rich cluster of galaxies at a redshift of @xmath13 . ms 1054 - 03 @xmath14 is the most distant cluster in the einstein extended medium sensitivity survey ( gioia et al . 1990 ; gioia & luppino 1994 ) . it is extremely rich and has a restframe x - ray luminosity , @xmath15=0.3 and @xmath16 . this gives a scale of @xmath17 kpc at the distance of ms 1054 . ] @xmath18 ergs / s ( donahue et al . 1998 ) , which makes it one of the brightest x - ray clusters known . from their asca observations donahue et al ( 1998 ) find a high x - ray temperature of @xmath19 kev . our space based observations allow us to study the cluster mass distribution in much more detail than lk97 . photometric redshift distributions inferred from the hubble deep field north ( fernndez - soto , lanzetta , & yahil 1999 ) and south ( chen et al . 1998 ) give a good approximation of the redshift distribution of the sources used in this weak lensing analysis . as a result it is possible to obtain well calibrated mass estimates for high redshift clusters of galaxies from multi - colour imaging , even if the availability of spectroscopic redshift data for the sources is limited . the observations and data reduction are outlined in section 2 . in section 3 we briefly discuss the method we use to analyse the galaxies . section 4 deals with the light distribution of the cluster . in section 5 we present the measurements of the lensing induced distortions of the faint galaxies and the reconstruction of the projected mass surface density . section 6 deals with the redshift distribution of the faint galaxies . our mass determination is presented in section 7 . in this section we also discuss several effects that complicate accurate mass determinations . the results for the mass - to - light ratio are given in section 8 . in section 9 we examine in more detail the substructure of the cluster . the lensing signal due to the cluster galaxies themselves is studied in section 10 . in this analysis we use wfpc2 images taken with the hubble space telescope ( go proposal 7372 , pi franx ) . figure [ layout ] shows the layout of the mosaic constructed from the 6 pointings of the telescope . the integration time per pointing was 6400s in both the @xmath20 and @xmath21 filter . for the weak lensing analysis we omitted the data of the planetary camera because of the brighter isophotal limit . the total area covered by the observations is approximately 26 arcmin@xmath22 . for each pointing the exposures of a given passband were split into two sets . the images in each set were offset from each other by integer pixels ( @xmath23 pixels ) . both sets were reduced and combined separately , resulting in two images ( for details see van dokkum et al . 1999 ; van dokkum 1999 ) . these two images were designed to be offset by 5.5 pixels , to allow the construction of an interlaced image ( e.g. fruchter & hook 1998 ; van dokkum 1999 ) , which has a sampling that is a factor @xmath24 better than the original wfpc2 images . except for the images in the first pointing in f814w , all offsets were within 0.1 pixel of the requested value . we found that we had to omit the shape measurements in the @xmath21 band for pointing 1 , because of the deviating offset for this exposure . however , we could use the @xmath20 image , and as the noise in the shear estimate is dominated by the intrinsic ellipticities of the galaxies , the loss in signal - to - noise is minimal . our analysis technique is based on that developed by kaiser , squires , & broadhurst ( 1995 ; ksb95 hereafter ) and lk97 , with a number of modifications which are described in hfks98 . the first step in the analysis is the detection of the galaxy images , for which we used the hierarchical peak finding algorithm from ksb95 . we selected objects with a significance @xmath25 larger than @xmath26 over the local sky . the detected objects were analysed , yielding estimates for the sizes , magnitudes and shapes of the objects . the detection and analysis were done per wfpc2 chip and filter . as a new feature , our software computes an estimate for the error on the polarization from the data . this error estimate is discussed in appendix a. the resulting catalogs were inspected visually in order to remove spurious detections , like diffraction spikes , hii regions in resolved galaxies , etc . 4402 objects were detected in the @xmath20 images , whereas the @xmath21 images yielded 4290 stars and galaxies . the combined catalog consisted of 5848 stars and galaxies , of which 2844 were detected in both filters . for all detected objects we estimated the apparent magnitude using an aperture with a diameter of @xmath27 pixels ( where @xmath28 is the gaussian scalelength of the object as given by the peak finding algorithm ) . we converted the measured counts to @xmath20 and @xmath21 magnitudes , zero - pointed to vega , using the zeropoints given in the hst data handbook ( stsci , baltimore ) . we added 0.05 magnitude to the zero - points to account for the charge transfer efficiency ( cte ) effect . figure [ objects ] shows the apparent magnitude versus the half light radius of the object . moderately bright stars are easily identified as they are located in the vertical locus . bright stars saturate and hence their measured sizes increase . using figure [ objects ] we selected a sample of moderately bright stars , which were used to examine the psf . to obtain a catalog of galaxies we removed the stars from the catalog . at faint levels the discrimination between stars and galaxies is not possible , but by extrapolating the number counts of moderately bright stars to faint magnitudes , we find that the stars contribute less than 2% to the total counts . we attempted to determine colours for all detected objects ( including galaxies detected in only one filter ) , using the same aperture in both filters . the software failed to determine colours for some of the galaxies . most of these were extremely faint , and detected only in the @xmath21 band . for the weak lensing analysis we only used galaxies for which colours were determined . due to this selection we lost 30% of the galaxies fainter than @xmath29 ( 533 galaxies ) . the shape parameters of the galaxies were corrected for the psf ( cf . section 3.1 ) and the camera distortion we used the estimated errors on the shape measurements ( cf . appendix a ) to combine the results from the @xmath20 and @xmath21 images in an optimal way , resulting in a final catalog of 4971 galaxies . the better sampling of the interlaced images improved the object detection and analysis significantly . hfks98 showed that due to the poor sampling of wfpc2 images only the shapes of galaxies with sizes @xmath30 could be measured reliably . the images used here allow us to use all detected galaxies . hfks98 found a number density of 79 galaxies arcmin@xmath31 from 3600s exposures in both @xmath20 and @xmath21 . the observations presented here yield a number density of 191 galaxies arcmin@xmath31 . this higher number density is due to both the longer integration time , and the improved sampling of the interlaced images . an analysis of the effect of small pointing errors in the interlaced images shows that these are not important for our analysis ( cf . appendix b ) . figure [ counts ] shows the observed number counts of galaxies in our observations versus apparent magnitude . for comparison we also show the counts from the hdf - north ( williams et al . this indicates that our catalog of galaxies is complete down to @xmath32 and @xmath33 . several observational effects systematically distort the images of the galaxies . to obtain an accurate and unbiased estimate of the weak lensing signal these effects have to be corrected for . to do so , we follow the scheme outlined in ksb95 , lk97 , and hfks98 . hfks98 showed that the psf of wfpc2 is highly anisotropic at the edges of the chips . we selected a sample of moderately bright stars from the observations of ms 1054 . figure [ psf814 ] shows the observed polarization of these stars in the f814w band . the anisotropy pattern is similar to the pattern presented in hfks98 . because the psf changes slightly with time , we fitted the shape parameters of our new observations to a scaled model of the psf polarization measured from the globular cluster m4 and added a first order polynomial : @xmath34 this procedure gave excellent fits to the observed psf anisotropy in our observations of ms 1054 . the psf also circularizes the images of the small , faint galaxies , thus lowering the amplitude of the lensing signal . compared to ground based observations , the small faint galaxies are much better resolved in hst observations , resulting in much smaller corrections for the circularization by the psf . this allows us to measure a well calibrated lensing signal . to obtain estimates for the real shapes of the faint galaxies , we determined the pre - seeing shear polarizability @xmath35 ( lk97 ; hfks98 ) . the measurements of @xmath35 for individual galaxies are rather noisy , and therefore we binned the data as a function of size of the object . the results are presented in figure [ corfac ] , which shows the required correction factor to obtain the true ellipticity of a galaxy from the observed polarization . the upper panel shows the cumulative number of objects as a function of object size . the objects for which the peak finder assigned a size smaller than the psf were analysed with a weight function matched to the psf . these objects end up in the bin corresponding to the smallest objects . the correction for objects with sizes comparable to the size of the psf is large . the psf in ground based images is much larger , and therefore large correction factors are required for objects that are still well resolved in hst images . from a @xmath36 deep ground based image we found that @xmath37 for galaxies of @xmath38 , whereas galaxies of similar brightness in our hst images have @xmath39 . as a test of our procedure we compared the distortion measured from the @xmath20 images and the @xmath21 images ( which included the correction for the circularization ) using only objects for which shape parameters could be measured in both images . we compared the average tangential distortions and found that the results for the two filters agree well to one another ( we found @xmath40 from the @xmath20 data and @xmath41 for the @xmath21 images ) . recently it has been found that the charge transfer efficiency problem for wfpc2 has worsened . this can systematically change the shapes of the faint galaxies , as charge is smeared along the ccd columns . this could result in a detectable change in the shapes of the galaxy images . we compared the shapes of the objects in the regions where the pointings overlap ( cf . figure [ layout ] ) . unfortunately the number of objects was low , and the contribution of shot noise to the shape estimates substantial . we did not detect a systematic change in the shapes , but we can not put strong constraints on the size of the effect . we also simulated the effect , assuming that the change in shape increased linearly with row number for each chip , with no change at low row numbers and a shear of 10% at high row numbers . we found that this resulted in a neglegible change in our mass estimate . figure [ colmag ] shows a plot of the colour of the galaxies versus their @xmath21 magnitude . the cluster is visible as the concentration of galaxies with @xmath42 . the galaxies on this colour - magnitude relation are generally early type galaxies , which are among the reddest objects in the cluster . we selected 324 galaxies brighter than @xmath43 , lying close to the colour - magnitude relation , to study the cluster light distribution . this catalog was compared to the spectroscopic catalog ( van dokkum 1999 ) . the contamination by field galaxies is small , and we removed galaxies with redshifts incompatible to that of the cluster . many of these belong to a small group at @xmath44 ( van dokkum 1999 ) . we have use sextractor ( bertin & arnouts 1996 ) to determine total magnitudes for these galaxies . figure [ lumdis ] shows grey scale plots of the luminosity and number density weighted light distributions . the images have been smoothed with a gaussian with a fwhm of 20 arcseconds . the galaxy number density and the luminosity weighted distributions look quite similar . the cluster light distribution is elongated east - west , and three clumps can be identified . figure [ lumdis ] also shows gray scale images of the smoothed number density when we split the sample of all detected galaxies into a bright ( @xmath45 ) and a faint sample ( @xmath46 ) . the cluster is clearly visible in the bright sample , but does not show up in the number counts of the faint galaxies ( which have been corrected for the area covered by large bright galaxies ) . figure [ lumdis]d shows several significant peaks , indicative of some clustering at high redshift . we estimate the luminosity in the redshifted @xmath47 band following van dokkum ( 1999 ) . the direct transformation from the hst filters to the passband corrected @xmath47 band is given by @xmath48 where @xmath49 denotes the passband corrected @xmath47 band magnitude . the luminosity is given by @xmath50 where @xmath51 is the solar absolute @xmath47 magnitude , @xmath52 is the distance modulus , and @xmath53 is the extinction correction in the @xmath21 filter towards ms 1054 - 03 . the redshift of @xmath54 for ms 1054 gives a distance modulus of @xmath55 . taking galactic extinctions from burstein & heiles ( 1982 ) we use a value of 0.03 for @xmath53 . at faint magnitudes , the colour - magnitude relation is less clearly defined . to estimate the total light of the cluster , an estimate of the amount of light contributed by galaxies fainter than @xmath43 is required . to do so , we fitted a luminosity function to the sample of bright , colour selected cluster galaxies . we found that a schechter function with @xmath56 and @xmath57 @xmath58 could fit the observed counts of the sample of bright , colour selected cluster galaxies . for nearby clusters one typically finds @xmath59 ( eg . binggeli , sandage , & tammann 1988 ) , indicating that the galaxies in ms 1054 - 03 are approximately twice as bright as nearby cluster galaxies . this is in good agreement with the results of van dokkum et al . ( 1998 ) who studied the fundamental plane in ms 1054 - 03 . the luminosity function indicates that we miss approximately 5% of the light when not including the faint galaxies . ms 1054 - 03 is a high redshift cluster and one might expect a relatively high fraction of blue cluster galaxies , which are too blue to be included in our colour selected sample of cluster galaxies . comparison of our sample to a sample of bright , spectroscopically confirmed members ( van dokkum 1999 ) indicates that we miss 16% of the total light in the @xmath47 band . the cumulative light profile as a function of radius from the centre for the sample of bright galaxies is presented in figure [ lumprof ] . the profile is multiplied by a factor @xmath60 to account for the light from the faint cluster galaxies and the bluest galaxies . the solid line corresponds to the observed profile , whereas the dashed line shows a isothermal profile for comparison . we estimate the total luminosity within an aperture of 1 @xmath2 mpc to be @xmath61 . due to the intrinsic ellipticities of the galaxies , their individual shape estimates are very noisy measurements of the lensing induced distortion . we therefore need to average the measurements of a large number of galaxies to obtain a useful estimate for the distortion @xmath62 . to estimate the average distortion , we weight each object with the inverse square of the uncertainty in the distortion , which includes the contribution of the intrinsic ellipticies of the galaxies and the shot noise in the shape measurement ( cf . appendix a ) : @xmath63 selecting galaxies with @xmath64 and @xmath65 ( the source sample from table [ table1 ] ) , and smoothing using a gaussian with a fwhm of 20 arcsec , we obtain the distortion field presented in figure [ shearfield ] . the sticks indicate the direction of the distortion , and the length is proportional to the amplitude of the distortion . it shows that on average the faint galaxies tend to align tangentially to the ( elongated ) cluster mass distribution , as expected from gravitational lensing . the average tangential distortion @xmath66 in radial bins is a useful measure of the lensing signal . it is defined as @xmath67 , where @xmath68 is the azimuthal angle with respect to the assumed centre of the mass distribution . we assume that the position of the brightest cluster galaxy defines the centre of the cluster . at large radii from the cluster centre , where we do not have data on complete circles , we account for possbile azimuthal variation of @xmath66 by fitting @xmath69 to the observed distortions . this procedure gives the correct value for an ellipsoidal mass distribution . although the cluster appears elongated , no large azimuthal variations in the tangential distortion were found . the average tangential distortion as a function of radius is presented in figure [ gtall]a . in the absence of lensing , the points should scatter around zero . the figure shows a systematic tangential alignment of the sources . the dependence of the signal on radius is quite complicated , indicative of a complex mass distribution . the results of the two - dimensional mass reconstruction ( cf . section 5.1 ) show that the sharp decrease in the tangential distortion around 40 is due to substructure in the cluster mass distribution . as a test , figure [ gtall]b shows the results when the phase of the distortion is increased by @xmath70 ( i.e. rotating the sources by @xmath71 ) . in this case the signal should vanish if it is due gravitational lensing , as is observed . we divide our sample of galaxies in several subsamples , based on their apparent magnitude and colours . table [ table1 ] lists some details of the subsamples . it shows that even our bright sample in fact consists of rather faint galaxies . [ cols="<,^,^,^,^,^ " , ] combining measurement of the mass with the estimate of the total light in the passband corrected @xmath47 band within an aperture of 1 @xmath2 mpc radius ( cf . section 4 ) yields @xmath72 . this result has to be corrected for luminosity evolution in order to compare it to values found for lower redshift clusters . studies of the fundamental plane of clusters of galaxies at various redshifts have quantified the luminosity evolution ( e.g. van dokkum & franx 1996 ; kelson et al . 1997 ; van dokkum et al . 1998 ) . the fundamental plane of ms 1054 was studied by van dokkum et al . they found that the mass - to - light ratio of early type galaxies in ms 1054 - 03 is @xmath73 lower than the present day value . under the assumption that the total luminosity of the cluster has changed by the same amount , the corrected mass - to - light ratio of ms 1054 is @xmath74 . the resulting mass - to - light ratio depends on the cosmological parameters . in table [ tableml ] the observed mass - to - light ratios in the passband corrected @xmath47 band are listed ( column ( 4 ) ) . column ( 5 ) lists the mass - to - light ratio corrected for luminosity evolution as determined from the evolution of the fundamental plane for ms 1054 - 03 for the various cosmologies considered here ( van dokkum et al . 1998 ) . hfks98 measured a mass - to - light ratio of @xmath75 for cl 1358 + 62 , which changes to @xmath76 after correction for luminosity evolution . carlberg , yee , & ellingson ( 1997 ) studied a sample of 16 rich clusters of galaxies , and they find a value of @xmath77 for the average mass - to - light ratio . using @xmath78 as found for nearby early type galaxies ( jrgensen , franx , & kjaergaard 1995 ) this transforms to @xmath79 in the @xmath47 band . this comparison indicates that the mass - to - light ratio of ms 1054 is similar to what is found from other studies , suggesting that the range in cluster mass - to - light ratios is rather small . lk97 argued that ms 1054 - 03 might be a relatively dark cluster , but our analysis does not confirm their results . we also studied the radial dependence of the mass - to - light ratio using equation [ eqzetac ] and an estimate for the average surface density in the control annulus . this allows us to estimate the average mass - to - light ratio @xmath80 as a function of radius . to avoid confusion with the clumps in the mass distribution , we present the results for radii larger than 50 arcseconds in figure [ moverlbar ] ( solid points ) . the open circles indicate the mass - to - light ratios measured for the three clumps in the mass distribution , using the results listed in table [ tableclump ] . the results indicate that the light is more concentrated towards the centre of the cluster than the mass . the data enable us to study the observed substructure in the cluster mass distribution in detail . figure [ masscen]a shows the mass reconstruction of the central 3.3 by 2.7 arcminute of the hst mosaic . the origin coincides with the assumed cluster centre . figure [ masscen]b is the overlay of the reconstruction and the @xmath21 image . to allow a direct comparison of the mass reconstruction with the light distribution , we also present the overlay of the smoothed light distribution in figure [ masscen]c . this shows that the projected mass distribution compares well with the light distribution . next we estimate the masses of the three clumps from the observed distortions ( see table [ tableclump ] ) . we use two measures : * a simultaneous fit of three isothermal spheres to the full distortion field ; * the @xmath81-statistic ( equation [ eqzetac ] ) . a simultaneous fit of three sis models to the observed distortion shows that the masses of the three clumps are similar , with corresponding velocity dispersions of @xmath82 km / s . comparison with the estimate of the total cluster mass indicates that most of the mass is in the three clumps . for the @xmath81 statistic we need to estimate the average dimensionless surface density in a control annulus around each clump ( we use @xmath83 and @xmath84 as at these radii the effect of the substructure is expected to be small ) . to do so , we assume that the surface density profile at large radii is isothermal with a velocity dispersion of @xmath85 km / s as found above . the results are listed in table [ tableclump ] . the results from the @xmath81-statistic correspond to the total projected mass in the aperture , which also includes mass from the other clumps . therefore we consider the results of the fitting procedure ( column ( 1 ) ) as the best estimates for the mass . however , to estimate the average mass - to - light ratio of the clumps within a radius of 250 @xmath2 kpc we do use the @xmath81 statistic , as both mass and light are measured in the same aperture . the mass - to - light ratios are around @xmath86 , somewhat lower than the average mass - to - light ratio within an 1 @xmath2 mpc radius aperture around the cluster centre . new x - ray telescopes like chandra or xmm will have the sensitivity to study the x - ray properties of high redshift clusters in detail . ms 1054 - 03 will be an interesting target as we know how the mass is distributed . thus high resolution measurements of the x - ray emission and temperature gradients can provide important information about the thermodynamics of cluster formation . the redshift catalog of van dokkum ( 1999 ) enables us to study the velocity structure of the cluster . the galaxies more that 100 west of the centre show a velocity difference of about 800 km / s with respect to the average velocity . however , their contribution to the total velocity dispersion is small . the three clumps are located closer to the centre , and can not be separated in velocity . therefore the radial motion of the clumps appears to be small compared to the velocity dispersion of the cluster . both outer clumps in the mass distribution are in projection 500@xmath2 kpc away from the central clump . if we assume that all motion is perpendicular to the line of sight , and using a relatively slow approach ( 500 km / s ) we estimate the timescale for relaxation to be on the order of 1 gyr . therefore the cluster should be relaxed at a redshift of @xmath87 . if the physical separations of the clumps are larger , the relaxation time increases correspondingly . to study the mass distribution of ms 1054 - 03 , we concentrated on the lensing signal on relatively large scales . however , the cluster galaxies themselves introduce a small signal as well . measuring this galaxy - galaxy lensing signal can be used to study the halo properties of field galaxies ( e.g. brainerd , blandford , & smail 1996 ; schneider & rix 1997 , hudson et al . 1998 ) . a robust method to examine the cluster galaxies is to measure the average tangential distortion around an ensemble of cluster members . the results are not very sensitive to the precise cluster mass distribution , provided one confines the analysis to small radii . due to the clustering of the lenses , the signal at large radii is dominated by the smooth cluster signal . when measuring the signal out to large radii , a careful removal of the cluster signal is required . we subtracted the contribution of the smooth cluster mass distribution from the observed distortion using the two - dimensional mass reconstruction , and computed the shear @xmath88 taking the @xmath89 correction into account . we used a sample of galaxies brighter than @xmath90 @xmath91 , with colours close to the cluster colour - magnitude relation to study the masses of the cluster galaxies . to minimize the effect of the large scale cluster mass distribution , we excluded galaxies near the centres of the three clumps , resulting in a sample of 241 galaxies . in total , 4300 cluster galaxy - background galaxy pairs were used to measure the distortion . massive galaxies produce a larger distortion than less massive galaxies and therefore we normalized the amplitudes of the signals from the various galaxies correspondingly . to do so , we assumed that the shear @xmath92 , i.e. we assumed a faber - jackson scaling relation . the results are presented in figure [ galgal ] . a fit of a singular isothermal model to the data , yielded a detection of the lensing signal at the 99.8% confidence limit . we found for the einstein radius of an @xmath5 galaxy ( @xmath93 for ms 1054 - 03 ) a value of @xmath94 arcseconds , which corresponds to a velocity dispersion of @xmath95 km / s . the mass - to - light ratio within a radius of @xmath96 kpc is @xmath97 . correcting this result for luminosity evolution yields @xmath98 . if we neglect the effect of the smooth cluster contribution , we find a slightly higher mass for the galaxies : @xmath99 km / s . this indicates that the mass estimate determined using the direct averaging method is indeed robust . natarayan & kneib ( 1998 ) studied the halo properties of the cluster galaxies in ac 114 . they used a maximum - likelihood analysis to constrain the mass and the extent of the galaxy halos . constraining the extent of the halos of the cluster galaxies in ms 1054 is a delicate task which has not been attempted here . it requires a careful analysis of the smooth cluster mass distribution in a maximum likelihood analysis . we have measured the weak gravitational lensing of faint , distant galaxies by ms 1054 - 03 , a rich cluster of galaxies at @xmath0 . the data consist of a two - colour mosaic of 6 interlaced images taken with the wfpc2 camera on the hubble space telescope . the expected lensing signal is low for high redshift clusters , and most of the signal comes from small , faint galaxies . compared to ground based observations , these galaxies are much better resolved in hst observations . high number densities of sources are reached , and the correction for the circularization by the psf is much smaller . this enables us to measure an accurate and well calibrated estimate of the lensing induced distortion of the faint background galaxies . several effects complicate an accurate analysis when the distortions or the surface density are non - neglegible ( e.g. in the cluster cores ) . we do not measure the shear , but the distortion @xmath62 . neglecting the effect of a non - zero surface density results in an overestimate of the lensing signal . furthermore , the method used ( ksb95 ; lk97 ) underestimates the distortion when it is larger than @xmath100 . although we correct for these effects , it is best to restrict the analysis to the weak lensing regime , where the convergence @xmath101 is low , and the distortions are small . converting the lensing signal into a mass estimate requires knowledge of the redshift distribution of the faint sources . we have explored the use of photometric redshift distributions inferred from the northern ( fernndez - soto et al . 1999 ) and southern ( chen et al . 1998 ) hubble deep fields . comparison of the results for the two fields suggest field to field variations which lead to a 10% systematic uncertainty in the mass estimate for ms 1054 - 03 . our results for the mass and mass - to - light ratio of ms 1054 - 03 agree well with other estimators , suggesting that these photometric redshift distributions are a fair approximation of the true distribution . this gives us confidence that we are able to measure a well calibrated weak lensing mass for this high redshift cluster . using the @xmath81 statistic ( clowe et al . 1998 ) , we find a mass @xmath102 within an aperture of 1 @xmath2 mpc radius . under the assumption of an isothermal mass distribution , this corresponds to a velocity dispersion of @xmath3 km / s . a fit of an isothermal sphere model to the observed tangential distortion at large radii , yields a velocity dispersion of @xmath103 km / s . however , this result depends on the assumed radial surface density profile . the uncertainties in the mass estimates only reflect the contribution from the intrinsic ellipticities of the sources . the weak lensing mass estimate is in good agreement with the x - ray estimate from donahue et al . ( 1998 ) and the observed velocity dispersion ( tran et al . 1999 ; van dokkum 1999 ) . the observed average mass - to - light ratio in the passband corrected @xmath47 band is @xmath104 ( measured within a 1 @xmath2 mpc radius aperture ) . we use the results from van dokkum et al . ( 1998 ) to correct our estimate for luminosity evolution . the resulting mass - to - light ratio of @xmath105 can be compared to results obtained for nearby clusters , and is in excellent agreement with the findings of carlberg et al . ( 1997 ) . our high resolution mass reconstruction shows three distinct peaks in the mass distribution , which are in good agreement with the irregular light distribution . the results of lk97 lacked the resolution to detect this substructure . in projection the clumps are approximately 500 @xmath2 kpc apart , and we estimate the timescale for virialization to be at on the order of @xmath106 1 gyr . thus the cluster should be virialized at a redshift @xmath107 . we found that the clumps are of roughly equal mass , with corresponding velocity dispersions of @xmath108 km / s . to study the masses of the cluster members , we have measured the average tangential shear around a sample of 241 bright cluster galaxies . assuming the faber - jackson scaling relation we find a velocity dispersion of @xmath7 km / s for an @xmath5 galaxy ( @xmath109 for ms 1054 - 03 ) . the mass - to - light ratio within a radius of @xmath96 kpc is @xmath97 . correcting this result for luminosity evolution yields @xmath98 . the number of known massive high redshift clusters of galaxies is increasing rapidly . the results presented here demonstrate that these systems can be studied with sufficient signal - to - noise to be of interest for studies of cluster formation . the improved sampling and throughput of the advanced camera for surveys , that will be installed on the hst in the near future , will be very useful for weak lensing analyses of high redshift clusters of galaxies , especially when the results are combined with forthcoming x - ray observations . the combination of detailed weak lensing analyses and x - ray observations by chandra and xmm will provide important information on the thermodynamics of cluster formation . the authors thank pieter van dokkum for making the data available in reduced form , which has been invaluable for this analysis . bahcall , n.a . , & fan , x. 1998 , apj , 504 , 1 bertin , e. , & arnouts , s. 1996 , a&as , 117 , 393 binggeli , b. , sandage , a. , & tammann , g.a . 1988 , ara&a , 26 , 509 brainerd , t.g . , blandford , r.d . , & smail , i. 1996 , apj , 466 , 623 broadhurst , t. j. , taylor , & a. n. , peacock , j.a . 1995 , apj , 438 , 49 burstein , d. , & heiles , c. 1982 , aj , 87 , 1165 castander , f. , ellis , r. , frenk , c. , dressler , a. , & gunn , j. 1994 , apj , 424 , l79 carlberg , r.g . , yee , h.c . , & ellingson 1997 , apj , 478 , 462 chen , h .- w . , fernndez - soto , a. , lanzetta , k.m . , pascarelle , s.m . , puetter , r.c . , yahata , n. , & yahil , a. , preprint , astro - ph/9812339 _ http://www.ess.sunysb.edu/astro/hdfs/home.html _ clowe , d. , luppino , g.a . , kaiser , n. , henry , j.p . , & gioia , i.m . 1998 , apj , 497 , l61 donahue , m. , voit , g.m . , gioia , i. , luppino , g.a . , hughes , j.p . , & stocke , j.t . 1998 , apj , 502 , 550 ebeling , h. , jones , l.r . , perlman , r. , scharf , c. , horner , d. , wegner , g. , malkan , m. , & mullis , c.r . 1999 , submitted to apj eke , v.r . , cole , s. , & frenk , c.s . 1996 , mnras , 282 , 263 gunn , j. , hoessel , j. , & oke , j.b . 1986 , apj , 306 , 30 fahlman , g. , kaiser , n. , squires , g. , & woods , d. 1994 , apj , 431 , l71 fernndez - soto , a. , lanzetta , k.m . , & yahil , a. 1999 , apj , 513 , 34 fruchter , a.s . , & hook , r.n . 1998 , preprint , astro - ph/9808087 gioia , i.m . , maccacaro , t. , schild , r.e . , wolter , a. , stocke , j.t . , morris , s.l . , henry , j.p . 1990 , apjs , 72 , 567 gioia , i.m . , & luppino , g.a . 1994 , apjs , 94 , 583 gioia , i.m . , henry , j.p . , mullis , c.r . , ebeling , h. , & wolter , a. 1998 , aj , accepted gorenstein , m.v . , shapiro , i.i . , & falco , e.e 1988 , apj , 327 , 693 holtzman , j.a . , hester , j.j . , casertano , s. , trauger , j.t . , watson , a.m. , et al . 1995 , pasp , 107,156 henry , j.p . , gioia , i.m . , mullis , c.r . , clowe , d.i . , luppino , g.a . , boehringer , h. , briel , u.g . , voges , w. , & huchra , j.p . 1997 , aj , 114,1293 hoekstra , h. , franx , m. , kuijken , k. , & squires , g. 1998 , apj , 504 , 636 ( hfks98 ) jrgensen , i. , franx , m. , & kjaergaard , p. 1995 , mnras , 273 , 1097 . kaiser , n. , & squires , g. 1993 , apj , 404 , 441 kaiser , n. , squires , g. , & broadhurst , t. 1995 , apj , 449 , 460 kelson , d.d . , van dokkum , p.g . , franx , m. , illingworth , g.d . , & fabricant , d. 1997 , apj , 478 , l13 kuijken , k. 1999 , a&a , in press ( astro - ph/9904418 ) luppino , g.a . , kaiser , n. 1997 , apj , 475 , 20 ( lk97 ) mellier , y. 1999 , preprint , astro - ph/9812172 neumann , d. , arnoud , m. , joy , m. , & patel , s.k . 1999 , in preparation postman , m. , lubin , l.m . , & oke , j.b . 1998 , aj , 116 , 560 rhodes , j. , refregier , a. , & groth , e. 1999 , preprint , astro - ph/9905090 rosati , p. , della ceca , r. , norman , c. , & giacconi , r. 1998 , apj , 492 , l21 schneider , p. , & rix , h .- w . 1997 , apj , 474 , l25 seitz , c. , & schneider , p. 1997 , a&a , 318 , 687 smail , i. , ellis , r. , fitchett , m. , & edge , a. 1994 , mnras , 270 , 245 squires , g. & kaiser , n. 1996 , apj , 473 , 65 tyson , j.a . , wenk , r.a . , & valdes , f. 1990 , apj , 349 , l1 tran , k .- v . h. , kelson , d.d . , van dokkum , p.g . , franx , m. , illingworth , g.d . , & magee , d. 1999 , preprint , astro - ph/9902349 van dokkum , p.g . 1999 , phd thesis , groningen university van dokkum , p.g . , & franx , m. 1996 , mnras , 281 , 985 van dokkum , p.g . , franx , m. , kelson , d.d . , & illingworth , g. 1998 apj , apj , 504 , l17 van dokkum , p.g . , franx , m. , fabricant , d. , kelson , d.d . , & illingworth , g. d 1998 , apj , preprint , astro - ph/9906152 williams , r.e . , 1996 , aj , 112 , 1335 willick , j.a . 1999 , preprint , astro - ph/9904367 the weak lensing signal is obtained by averaging the shape measurements of a number of faint galaxies . the noise in the shape estimate differs from object to object , and therefore it is important to weight the measurements properly . for bright galaxies the error on the shear measurement is dominated by the intrinsic ellipticity distribution of the galaxies . at fainter magnitudes , shot - noise increases the uncertainty in the measurements . as these galaxies require large seeing corrections , the measurement error on the distortion can be quite high . this is illustrated in figure [ scatter ] . figure [ scatter]a shows the observed scatter in the polarization as a function of apparent magnitude in the @xmath21 band . as fainter objects suffer more from the circularization by the psf one would expect the scatter to decrease towards fainter magnitudes . however , the increasing contribution by shot noise ( indicated by the solid line ) cancels out the effect of the circularization . figure [ scatter]b shows the observed scatter in the ellipticities ( shear ) of the objects . the scatter at faint magnitudes increases rapidly due to the combination of shot noise and large psf corrections . the solid line indicates the expected scatter due to the combination of shot noise and psf correction . we estimate the weighted mean of the distortion using : @xmath110 where the weight @xmath111 is the inverse of the variance of the distortion : @xmath112 where @xmath113 is the scatter due to the intrinsic ellipticity of the galaxies , @xmath114 is the pre - seeing shear polarizability and @xmath115 is the error estimate for the polarization , which is derived below . this procedure , however , gives only the optimal estimate for the average ellipticity , but does not account for the fact that the nearby bright galaxies are lensed less efficiently than faint high redshift galaxies . to get the best lensing estimate it is therefore also necessary to weight with @xmath116 as well this is demonstrated in figure [ weight]a , where we show the weight function as a function of apparent magnitude ( solid line ) . the dashed line shows the weight function hen we include the strength of the lensing signal . this effectively decreases the weight of bright ( nearby ) galaxies and increases the relative weight of the faint ( distant ) galaxies . it is also interesting to see which magnitude range contributes most to the measurement of the distortion . in figure [ weight]b we plot the weight function for each magnitude bin times the number of galaxies in that bin . this gives the total contribution of each bin to the measurement of the distortion . although the bright galaxies have the highest weight function ( cf . figure [ weight]a ) , they are not so numerous to contribute significantly to the lensing signal . at the faint end both incompleteness and increasing noise causes the contribution to decrease rapidly . figure [ weight]b also shows that including the strength of the lensing signal @xmath117 changes the profile only slightly . it shows that galaxies with @xmath21 magnitudes between 24 and 26.5 contribute most to the lensing signal . here we derive an expression to estimate the error on the polarization measurement , using the observations . the observed quadrupole moments are given by : @xmath118 which are combined to form the two - component polarization . @xmath119 due to noise , the true image @xmath120 is changed to the observed image @xmath121 : @xmath122 where we take @xmath123 . the first term in equation a5 gives the true image , the second term is the noise in the background , and the third contribution comes from the shot noise due to the image . we therefore observe quadrupole moments given by : @xmath124 as @xmath125 , this shows that @xmath126 . thus noise does not introduce a bias in the average measurement of the quadrupole moments . however , it does introduce a neglegible bias in the polarization . we now can estimate the variance in the observed quadrupole moments using equation a6 and noting that @xmath127 , @xmath128 , and @xmath129 , where @xmath130 is the dirac delta function . this yields : @xmath131 we now want to estimate the errors on the measurements of the polarizations @xmath132 and @xmath133 and express them in terms of the errors on the quadrupole moments . this finally gives : @xmath134\end{aligned}\ ] ] and @xmath135.\end{aligned}\ ] ] from simulations and real images we find that the errors on both @xmath132 and @xmath133 are similar . this estimate for the error on the polarization can be calculated from the observations and allows us to properly weight the signals of the galaxies using equation a2 . the images of small , faint galaxies in wfpc2 observations are poorly sampled . hfks98 showed that as a result the smallest objects can not be corrected reliably for the psf effects . however , several techniques can be used to improve the sampling of wfpc2 observations by dithering the exposures ( fruchter & hook 1998 ) . when the shifts are half pixel in both @xmath136 and @xmath137 ( modulo integer pixel shifts ) one can construct an interlaced image ( fruchter & hook 1998 ) which has a @xmath24 better sampling . interlaced images were constructed from the observations of ms 1054 - 03 . doing so , we assumed that the offsets were exactly half pixel . due to pointing errors and the camera distortion the real offsets are slightly different . in this section we examine the effect of imperfect offsets on the shape measurements . to do so , we examine the change in the shape of a model psf . although interlacing is not a convolution , it affects large objects ( as these are well sampled even before interlacing ) less than small objects . by examining the effects on the psf , we study a worst case scenario . the effects should be less for most of the galaxies used in the weak lensing analysis . we use a ten times oversampled model psf calculated using tiny tim . this psf is shifted and rebinned to wfpc2 sampling and convolved with the pixel scattering function , in order to realistically mimic the wfpc2 psf . we start with a reference image in which the star is centered exactly on a pixel . this image is shifted @xmath138 and an interlaced image is created assuming the shift was exactly ( 0.5 , 0.5 ) pixels . the resulting image is analysed and the measured ellipticities are compared to the correct value . the change in ellipticity as a function of the applied offset is presented in figure [ interlace ] ( left panel ) . when the reference star is placed at the intersection of four pixels the results are different , as is indicated in figure [ interlace ] ( right panel ) . these two extreme cases should cover the full range of possible configurations . in figure [ interlace ] we also plot the observed offsets for our data . the open circles represent the offsets of the @xmath21 images , and the solid triangles show the offsets for the @xmath20 images . only one @xmath21 pointing has such a deviating offset that the shape measurements can not be used for our weak lensing analysis . in all other images the offsets are such that no significant bias is introduced by the interlacing . even if the offset is perfect in the centre of the image , telescope distortions will result in imperfect offsets at the edges . we simulated the amplitude of the effect using the coefficients for the telescope distortions from holtzman et al . ( 1995 ) . figure [ interlace2 ] shows the introduced ellipticity for our worst offset ( 5.71 , 5.89 ) . the bias in the observed ellipticities is clear in this case . these data ( the @xmath21 image of the first pointing ) were not used in the weak lensing analysis . the second worse offset ( 5.53 , 5.60 ) is also shown in this figure . the ellipticity introduced in this case is small . only at the edges of the chips the camera distortion introduces a small effect . as the effect is even smaller for galaxies , we feel certain that we can use the interlaced images for our weak lensing analysis .	we have measured the weak gravitational lensing of faint , distant background galaxies by ms 1054 - 03 , a rich and x - ray luminous cluster of galaxies at a redshift of @xmath0 , using a two - colour mosaic of deep wfpc2 images . the small corrections for the size of the psf and the high number density of background galaxies obtained in these observations result in an accurate and well calibrated measurement of the lensing induced distortion . the strength of the lensing signal depends on the redshift distribution of the background galaxies . we used photometric redshift distributions from the northern and southern hubble deep fields to relate the lensing signal to the mass . the predicted variations of the signal as a function of apparent source magnitude and colour agrees well with the observed lensing signal . the uncertainty in the redshift distribution results in a 10% systematic uncertainty in the mass measurement . we determine a mass of @xmath1 within an aperture of radius 1 @xmath2 mpc . under the assumption of an isothermal mass distribution , the corresponding velocity dispersion is @xmath3 . for the mass - to - light ratio we find @xmath4 after correcting for pass - band and luminosity evolution . the errors in the mass and mass - to - light ratio include the contribution from the random intrinsic ellipticities of the source galaxies , but not the ( systematic ) error due to the uncertainty in the redshift distribution . however , the estimates for the mass and mass - to - light ratio of ms 1054 - 03 agree well with other estimators , suggesting that the mass calibration works well . the reconstruction of the projected mass surface density shows a complex mass distribution , consistent with the light distribution . the results indicate that ms 1054 - 03 is a young system . the timescale for relaxation is estimated to be at least 1 gyr . we have also studied the masses of the cluster galaxies , by averaging the tangential shear around the cluster galaxies . using the faber - jackson scaling relation , we find the velocity dispersion of an @xmath5 galaxy ( @xmath6 for ms 1054 - 03 ) is @xmath7 km / s .
let @xmath7 be a directed graph ( digraph ) , with @xmath5 edges and @xmath4 vertices . @xmath8 is _ strongly connected _ if there is a directed path from each vertex to every other vertex . the _ strongly connected components _ of @xmath8 are its maximal strongly connected subgraphs . two vertices @xmath9 are _ strongly connected _ if they belong to the same strongly connected component of @xmath8 . a vertex ( resp . , an edge ) of @xmath8 is a _ strong articulation point _ ( resp . , a _ strong bridge _ ) if its removal increases the number of strongly connected components . a digraph @xmath8 is @xmath2-vertex - connected if it has at least three vertices and no strong articulation points ; @xmath8 is @xmath2-edge - connected if it has no strong bridges . the @xmath2-vertex- ( resp . , @xmath2-edge- ) connected components of @xmath8 are its maximal @xmath2-vertex- ( resp . , @xmath2-edge- ) connected subgraphs . , with strong articulation points and strong bridges shown in red ( better viewed in color ) . ( b ) the @xmath2-vertex - connected components of @xmath8 . ( c ) the @xmath2-vertex - connected blocks of @xmath8 . ( d ) the @xmath2-edge - connected components of @xmath8 . ( e ) the @xmath2-edge - connected blocks of @xmath8 . ] c0.0c0.10c0.25c0.15c0.20c0.15c0.20 + & ( a ) @xmath8 & ( b ) @xmath10 & ( c ) @xmath11 & ( d ) @xmath12 & ( e ) @xmath13 & differently from undirected graphs , in digraphs @xmath2-vertex and @xmath2-edge connectivity have a much richer and more complicated structure . to see an example of this , let @xmath0 and @xmath1 be two distinct vertices and consider the following natural @xmath2-vertex and @xmath2-edge connectivity relations , defined in @xcite . vertices @xmath0 and @xmath1 are said to be _ @xmath2-vertex - connected _ ( resp . , _ @xmath2-edge - connected _ ) , and we denote this relation by @xmath14 ( resp . , @xmath15 ) , if there are two internally vertex - disjoint ( resp . , two edge - disjoint ) directed paths from @xmath0 to @xmath1 and two internally vertex - disjoint ( resp . , two edge - disjoint ) directed paths from @xmath1 to @xmath0 ( note that a path from @xmath0 to @xmath1 and a path from @xmath1 to @xmath0 need not be edge- or vertex - disjoint ) . @xmath2-vertex - connected block _ ( resp . , _ @xmath2-edge - connected block _ ) of a digraph @xmath7 is defined as a maximal subset @xmath16 such that @xmath17 ( resp . , @xmath18 ) for all @xmath19 . in undirected graphs , the @xmath2-vertex- ( resp . , @xmath2-edge- ) connected blocks are identical to the @xmath2-vertex- ( resp . , @xmath2-edge- ) connected components . as shown in figure [ fig : example ] , this is not the case for digraphs . put in other words , differently from the undirected case , in digraphs @xmath2-vertex- ( resp . , @xmath2-edge- ) connected components do not encompass the notion of pairwise @xmath2-vertex ( resp . , @xmath2-edge ) connectivity among its vertices . we note that pairwise @xmath2-connectivity is relevant in several applications , where one is often interested in local properties , e.g. , checking whether two vertices are @xmath2-connected , rather than in global properties . it is thus not surprising that @xmath2-connectivity problems on directed graphs appear to be more difficult than on undirected graphs . for undirected graphs it has been known for over 40 years how to compute all bridges , articulation points , @xmath2-edge- and @xmath2-vertex - connected components in linear time , by simply using depth first search @xcite . in the case of digraphs , however , the very same problems have been much more challenging . indeed , it has been shown only few years ago that all strong bridges and strong articulation points of a digraph can be computed in linear time @xcite . furthermore , the best current bound for computing the @xmath2-edge- and the @xmath2-vertex - connected components in digraphs is not even linear , but it is @xmath20 , and it was achieved only very recently by henzinger et al . @xcite , improving previous @xmath21 time bounds @xcite . finally , it was shown also very recently how to compute the @xmath2-edge - connected blocks of digraphs in linear time @xcite . in this paper , we complete the picture on @xmath2-connectivity for digraphs by presenting the first algorithm for computing the @xmath2-vertex - connected blocks in @xmath3 time . our bound is asymptotically optimal and it improves sharply over a previous @xmath21 time bound by jaberi @xcite . as a side result , our algorithm constructs an @xmath6-space data structure that reports in constant time if two vertices are @xmath2-vertex - connected . additionally , when two query vertices @xmath0 and @xmath1 are not @xmath2-vertex - connected , our data structure can produce , in constant time , a `` witness '' by exhibiting a vertex ( i.e. , a strong articulation point ) or an edge ( i.e. , a strong bridge ) that separates them . we are also able to compute in linear time a sparse certificate for @xmath2-vertex connectivity , i.e. , a subgraph of the input graph that has @xmath6 edges and maintains the same @xmath2-vertex connectivity properties . our algorithm follows the high - level approach of @xcite for computing the @xmath2-edge - connected blocks . however , the algorithm for computing the @xmath2-vertex - connected blocks is much more involved and requires several novel ideas and non - trivial techniques to achieve the claimed bounds . in particular , the main technical difficulties that need to be tackled when following the approach of @xcite are the following : * first , the algorithm in @xcite maintains a partition of the vertices into approximate blocks , and refines this partition as the algorithm progresses . unlike @xmath2-edge - connected blocks , however , @xmath2-vertex - connected blocks do not partition the vertices of a digraph , and therefore it is harder to maintain approximate blocks throughout the algorithm s execution . to cope with this problem , we show that these blocks can be maintained using a more complicated forest representation , and we define a set of suitable operations on this representation in order to refine and split blocks . we believe that our forest representation of the @xmath2-vertex - connected blocks of a digraph can be of independent interest . * second , in @xcite we used a properly defined _ canonical decomposition _ of the input digraph @xmath8 , in order to obtain smaller _ auxiliary _ digraphs ( not necessarily subgraphs of @xmath8 ) that maintain the original @xmath2-edge - connected blocks of @xmath8 . a key property of this decomposition was the fact that any vertex in an auxiliary graph @xmath22 is reachable from a vertex outside @xmath22 only through a single strong bridge . in the computation of the @xmath2-vertex - connected blocks , we have to decompose the graph according to strong articulation points , and so the above crucial property is completely lost . to overcome this problematic issue , we need to design and to implement efficiently a different and more sophisticated decomposition . * third , differently from @xmath2-edge connectivity , @xmath2-vertex connectivity in digraphs is plagued with several degenerate special cases , which are not only more tedious but also more cumbersome to deal with . for instance , the algorithm in @xcite exploits implicitly the property that two vertices @xmath0 and @xmath1 are @xmath2-edge - connected if and only if the removal of any edge leaves @xmath0 and @xmath1 in the same strongly connected component . unfortunately , this property no longer holds for @xmath2-vertex connectivity , as for instance two mutually adjacent vertices are always left in the same strongly connected component by the removal of any other vertex , but they are not necessarily @xmath2-vertex - connected . to handle this more complicated situation , we introduce the notion of _ vertex - resilient blocks _ and prove some useful properties about the vertex - resilient and @xmath2-vertex - connected blocks of a digraph . another difference with @xcite is that now we are able to provide a witness for two vertices not being @xmath2-vertex - connected . this approach can be applied to provide a witness for two vertices not being @xmath2-edge - connected , thus extending the result in @xcite . as in @xcite , some basic components of our algorithms are flow graphs and dominator trees , that we review in section [ sec : dominators ] . in section [ sec : blocks ] we prove some useful properties of the vertex - resilient and @xmath2-vertex - connected blocks that allow us to represent them by a forest . our linear - time algorithms for computing the vertex - resilient blocks and the @xmath2-vertex - connected blocks are described in sections [ section : vertex - resilient - blocks ] and [ section:2vc - blocks ] . we describe the computation of the sparse certificate in section [ section : sparse - certificate ] . a _ flow graph _ is a digraph such that every vertex is reachable from a distinguished start vertex . let @xmath7 be the input digraph , which we assume to be strongly connected . ( if not , we simply treat each strongly connected component separately . ) for any vertex @xmath23 , we denote by @xmath24 the corresponding flow graph with start vertex @xmath25 ; all vertices in @xmath26 are reachable from @xmath25 since @xmath8 is strongly connected . the _ dominator relation _ in @xmath27 is defined as follows : a vertex @xmath28 is a _ dominator _ of a vertex @xmath1 ( @xmath28 _ dominates _ @xmath1 ) if every path from @xmath25 to @xmath1 contains @xmath28 ; @xmath28 is a _ proper dominator _ of @xmath1 if @xmath28 dominates @xmath1 and @xmath29 . the dominator relation is reflexive and transitive . its transitive reduction is a rooted tree , the _ dominator tree _ @xmath30 : @xmath28 dominates @xmath1 if and only if @xmath28 is an ancestor of @xmath1 in @xmath30 . if @xmath31 , @xmath32 , the parent of @xmath1 in @xmath30 , is the _ immediate dominator _ of @xmath1 : it is the unique proper dominator of @xmath1 that is dominated by all proper dominators of @xmath1 . an edge @xmath33 is a _ bridge _ in @xmath27 if all paths from @xmath25 to @xmath1 include @xmath33 . lengauer and tarjan @xcite presented an algorithm for computing dominators in @xmath34 time for a flow graph with @xmath4 vertices and @xmath5 edges , where @xmath35 is a functional inverse of ackermann s function @xcite . subsequently , several linear - time algorithms were discovered @xcite . italiano et al . @xcite showed that the strong articulation points of @xmath8 can be computed from the dominator trees of @xmath27 and @xmath36 , where @xmath25 is an arbitrary start vertex and @xmath37 is the digraph that results from @xmath8 after reversing edge directions ; similarly , the strong bridges of @xmath8 correspond to the bridges of @xmath27 and @xmath36 . let @xmath38 be a rooted tree whose vertex set is @xmath26 . tree @xmath38 has the _ parent property _ if for all @xmath39 , @xmath0 is a descendant of the parent of @xmath1 in @xmath38 . tree @xmath38 has the _ sibling property _ if @xmath0 does not dominate @xmath1 for all siblings @xmath0 and @xmath1 . the parent and sibling properties are necessary and sufficient for a tree to be the dominator tree @xcite . let @xmath0 and @xmath1 be two distinct vertices in a digraph . by menger s theorem @xcite , @xmath15 if and only if the removal of any edge leaves @xmath0 and @xmath1 in the same strongly connected component , i.e. , two vertices are @xmath2-edge - connected if and only if they are resilient to the deletion of a single edge . the situation for @xmath2-vertex connectivity is more complicated . indeed , menger s theorem implies that @xmath14 only if the removal of any vertex different from @xmath0 and @xmath1 leaves them in the same strongly connected component , while the converse holds only when @xmath0 and @xmath1 are not adjacent . for instance , two mutually adjacent vertices are left in the same strongly connected component by the removal of any other vertex , although they are not necessarily @xmath2-vertex - connected . to handle this situation , we use the following notation , which was also considered in @xcite . vertices @xmath0 and @xmath1 are said to be _ vertex - resilient _ , denoted by @xmath40 if the removal of any vertex different from @xmath0 and @xmath1 leaves @xmath0 and @xmath1 in the same strongly connected component . we define a _ vertex - resilient block _ of a digraph @xmath7 as a maximal subset @xmath16 such that @xmath41 for all @xmath19 . see figure [ fig : vrb - example ] . note that , as a ( degenerate ) special case , a vertex - resilient block might consist of a singleton vertex only : we denote this as a _ trivial vertex - resilient block_. in the following , we will consider only non - trivial vertex - resilient blocks . since there is no danger of ambiguity , we will call them simply vertex - resilient blocks . we remark that two vertices @xmath0 and @xmath1 that are vertex - resilient are not necessarily @xmath2-vertex - connected : this is indeed the case for vertices @xmath42 and @xmath43 in the digraph of figure [ fig : example](a ) . if , however , @xmath0 and @xmath1 are not adjacent then @xmath14 if and only if @xmath40 . .,scaledwidth=22.0% ] we next provide some basic properties of the vertex - resilient blocks and the @xmath2-vertex - connected blocks . in particular , we show that any digraph has at most @xmath44 vertex - resilient ( resp . , @xmath2-vertex - connected ) blocks and , moreover , that there is a forest representation of these blocks that enables us to test vertex - resilience ( resp . , @xmath2-vertex - connectivity ) between any two vertices in constant time . this structure is reminiscent of the representation used in @xcite for the biconnected components of an undirected graph . [ lemma : vr ] let @xmath28 , @xmath0 , @xmath45 , and @xmath46 be distinct vertices such that @xmath47 , @xmath48 , @xmath49 and @xmath50 . then also @xmath51 and @xmath52 . assume , for contradiction , that @xmath45 and @xmath46 are not vertex - resilient . then there is a strong articulation point @xmath1 such that every path from @xmath46 to @xmath45 contains @xmath1 , or every path from @xmath45 to @xmath46 contains @xmath1 ( or both ) . without loss of generality , suppose that @xmath1 is contained in every path from @xmath46 to @xmath45 . since @xmath28 and @xmath0 are distinct , we can assume that @xmath53 . ( if @xmath54 then we swap the role of @xmath28 and @xmath0 . ) then , @xmath55 implies that there is a path @xmath56 from @xmath46 to @xmath28 that avoids @xmath1 , and similarly , @xmath57 implies that there is a path @xmath58 from @xmath28 to @xmath45 that avoids @xmath1 . so , @xmath56 followed by @xmath58 gives a path from @xmath46 to @xmath45 that does not contain @xmath1 , a contradiction . hence @xmath51 . the fact that @xmath52 follows by repeating the same argument for @xmath28 and @xmath0 . [ corollary : vrb ] let @xmath59 and @xmath60 be two distinct vertex - resilient blocks of a digraph @xmath7 . then @xmath61 . follows immediately from lemma [ lemma : vr ] . we denote by @xmath62 the vertex - resilient blocks that contain @xmath28 . define the _ block graph _ @xmath63 of @xmath8 as follows . the vertex set @xmath64 consists of the vertices in @xmath26 and also contains one _ block node _ for each vertex - resilient block of @xmath8 . the edge set @xmath65 consists of the edges @xmath66 where @xmath67 . thus , @xmath43 is an undirected bipartite graph . next we show that it is also acyclic . [ lemma : vb - graph - path ] let @xmath28 and @xmath0 be any vertices that are connected by a path @xmath56 in @xmath43 . then , for any vertex @xmath68 not on @xmath56 , @xmath28 and @xmath0 are strongly connected in digraph @xmath69 . it suffices to show that @xmath8 contains a path @xmath58 from @xmath28 to @xmath0 that avoids @xmath1 . the same argument shows that @xmath8 contains a path from @xmath0 to @xmath28 that avoids @xmath1 . let @xmath70 . then @xmath71 , for @xmath72 , so there is a path @xmath73 in @xmath8 from @xmath74 to @xmath75 that avoids @xmath1 . then the catenation of paths @xmath76 gives a path in @xmath8 from @xmath28 to @xmath0 that avoids @xmath1 . [ lemma : vb - graph ] graph @xmath43 is acyclic . suppose , for contradiction , that @xmath43 contains a cycle @xmath77 . we show that all vertices @xmath78 belong to the same vertex - resilient block @xmath59 . let @xmath79 be two vertices on a minimal cycle @xmath77 of @xmath43 that are adjacent to a block node @xmath59 . ( such @xmath28 , @xmath0 , and @xmath59 exist since @xmath43 is bipartite . ) then , @xmath28 and @xmath0 can not be the only vertices in @xmath26 that are on @xmath77 , since otherwise they would be adjacent to another block @xmath60 on @xmath77 , violating corollary [ corollary : vrb ] . therefore , @xmath77 contains a vertex @xmath80 . clearly , @xmath81 , otherwise the edge @xmath82 would exist contradicting the minimality of @xmath77 . hence , there is a vertex @xmath83 such that all paths from @xmath84 to @xmath1 contain a common strong articulation point or all paths from @xmath1 to @xmath84 contain a common strong articulation point . suppose , without loss of generality , that a vertex @xmath45 is contained in every path from @xmath84 to @xmath1 . let @xmath56 be the path that results from @xmath77 by removing @xmath59 . let @xmath85 and @xmath86 be the subpaths of @xmath56 from @xmath28 to @xmath1 and from @xmath0 to @xmath1 , respectively . then @xmath87 or @xmath88 ( or both ) . suppose @xmath87 ; if not then swap the role of @xmath28 and @xmath0 . then , by lemma [ lemma : vb - graph - path ] there is a path @xmath58 in @xmath8 from @xmath28 to @xmath1 that avoids @xmath45 . also , since @xmath89 , there is a path @xmath90 in @xmath8 from @xmath84 to @xmath28 that avoids @xmath45 . then the catenation of @xmath90 and @xmath58 gives a path in @xmath8 from @xmath84 to @xmath1 that avoids @xmath45 , a contradiction . [ lemma : blocks - number ] the number of vertex - resilient blocks in a digraph @xmath8 is at most @xmath44 . we prove the lemma by showing that forest @xmath43 contains at most @xmath44 block nodes . since @xmath43 is a forest we can root each tree @xmath38 of @xmath43 at some arbitrary vertex @xmath91 . every level of @xmath38 contains either only vertices of @xmath26 or only block nodes , because @xmath43 is bipartite . moreover , every block node is adjacent to at least two vertices of @xmath26 , due to the fact that each ( non - trivial ) vertex - resilient block in @xmath8 contains at least @xmath2 vertices . hence , every leaf of @xmath38 is a vertex in @xmath26 . now consider a partition of @xmath38 into vertex disjoint paths @xmath92 , such that each @xmath73 leads from some vertex or block node to a leaf descendant . the number of block nodes in each @xmath73 is at most equal to @xmath93 . also , in the path @xmath73 starting at @xmath91 the number of block nodes in @xmath73 is less than @xmath93 . we conclude that there at most @xmath44 block nodes in @xmath43 . [ lemma : blocks - size ] the total number of vertices in all vertex - resilient blocks is at most @xmath94 . by lemmas [ lemma : vb - graph ] and [ lemma : blocks - number ] , the block graph @xmath43 is a forest with at most @xmath95 vertices . each occurrence of a vertex @xmath0 in a block @xmath59 corresponds to an edge @xmath96 of @xmath43 . therefore , the total number of vertices in all vertex - resilient blocks equals the number of edges in @xmath43 , and the lemma follows . and its vertex - resilient block forest @xmath43 . the strong articulation points and the strong bridges of @xmath8 are shown in red . ( better viewed in color.)[fig : vrb - forest],title="fig:",scaledwidth=43.0% ] and its vertex - resilient block forest @xmath43 . the strong articulation points and the strong bridges of @xmath8 are shown in red . ( better viewed in color.)[fig : vrb - forest],title="fig:",scaledwidth=55.0% ] [ lemma : vb - graph - path-2 ] let @xmath28 and @xmath0 be any vertices that are not vertex - resilient but are connected by a path @xmath56 in @xmath43 . then , for any vertex @xmath97 on @xmath56 , @xmath28 and @xmath0 are not strongly connected in digraph @xmath69 . we prove the lemma by contradiction . let @xmath56 be a path that connects @xmath28 and @xmath0 in @xmath43 . by lemma [ lemma : vb - graph ] , this path is unique for @xmath28 and @xmath0 . first suppose that @xmath56 contains only one other vertex @xmath98 , so @xmath99 . then @xmath100 and @xmath101 . now suppose that @xmath28 and @xmath0 are strongly connected in @xmath69 . this fact , together with lemma [ lemma : vb - graph - path ] , imply that @xmath28 and @xmath0 are strongly connected in @xmath102 for all @xmath103 . but this contradicts the assumption that @xmath28 and @xmath0 are not vertex - resilient . now suppose that path @xmath56 contains more than one vertex in @xmath104 . let @xmath105 , where @xmath106 . by the argument above , @xmath107 and @xmath108 are not strongly connected in @xmath109 for all @xmath110 . suppose that @xmath28 and @xmath0 are strongly connected in @xmath111 for a fixed @xmath110 . by lemma [ lemma : vb - graph - path ] , @xmath28 and @xmath107 , and @xmath108 and @xmath0 , are strongly connected in @xmath111 . but then , @xmath107 and @xmath108 are also strongly connected in @xmath111 , a contradiction . we consider @xmath43 as a forest of rooted trees by choosing an arbitrary vertex as the root of each tree . then @xmath100 if and only if @xmath28 and @xmath1 are siblings or one the grandparent of the other . see figure [ fig : vrb - forest ] . we can perform both tests in constant time simply by storing the parent of each vertex in @xmath43 . thus , we can test in constant time if two vertices are vertex - resilient . note that we can not always apply lemma [ lemma : vb - graph - path-2 ] to find a strong articulation point that separates two vertices @xmath28 and @xmath1 that are not vertex - resilient . indeed , two vertices that are strongly connected but not vertex - resilient may not even be connected by a path in the forest @xmath43 ( see , e.g. , vertices @xmath112 and @xmath113 in figure [ fig : vrb - forest ] ) . so if we wish to return a witness that @xmath28 and @xmath1 are not vertex - resilient , we can not rely on @xmath43 . we deal with this problem in section [ sec : queries ] . now we turn to @xmath2-vertex - connected blocks and provide some properties that enable us to compute them via the vertex - resilient blocks . [ lemma:2vc - resilient - sb ] let @xmath0 and @xmath1 be two distinct vertices of @xmath8 such that @xmath40 . then , @xmath0 and @xmath1 are not @xmath2-vertex connected if and only if at least one of the edges @xmath114 and @xmath115 is a strong bridge in @xmath8 . menger s theorem @xcite implies that if @xmath0 and @xmath1 are not adjacent then @xmath14 if and only if @xmath40 . if , on the other hand , @xmath40 but @xmath0 and @xmath1 are not @xmath2-vertex - connected , then at least one of the edges @xmath114 and @xmath115 exists in @xmath8 and is a strong bridge . the following corollary , which relates @xmath2-vertex - connected , @xmath2-edge - connected and vertex - resilient blocks , is an immediate consequence of lemma [ lemma:2vc - resilient - sb ] . [ cor:2vc - resilient ] for any two distinct vertices @xmath0 and @xmath1 , @xmath14 if and only if @xmath40 and @xmath15 . by corollary [ cor:2vc - resilient ] we have that the @xmath2-vertex - connected blocks are refinements of the vertex - resilient blocks , formed by the intersections of the vertex - resilient blocks and the @xmath2-edge - connected blocks of the digraph @xmath8 . since the @xmath2-edge - connected blocks are a partition of the vertices of @xmath8 , these intersections partition each vertex - resilient block . from this property we conclude that lemmas [ lemma : vr ] , [ lemma : vb - graph ] , and [ lemma : blocks - number ] and corollary [ corollary : vrb ] also hold for the @xmath2-vertex - connected blocks . in this section we present new algorithms for computing the vertex - resilient blocks of a digraph @xmath8 . we can assume that @xmath8 is strongly connected , so @xmath116 . if not , then we process each strongly connected component separately ; if @xmath41 then @xmath28 and @xmath0 are in the same strongly connected component @xmath117 of @xmath8 , and moreover , any vertex on a path from @xmath28 to @xmath0 or from @xmath0 to @xmath28 also belongs in @xmath117 . we begin with a simple algorithm that removes a single strong articulation point at a time . in order to get a more efficient solution , we need to consider simultaneously how different strong articulation points divide the vertices into blocks , which we do with the help of dominator trees . we achieve linear running time by combining the simple algorithm with the dominator - tree - based division , and by applying suitable operations on the block forest structure . algorithm , illustrated in figure [ fig : simplevrb ] , is an immediate application of the characterization of the vertex - resilient blocks in terms of strong articulation points . let @xmath28 and @xmath0 be two distinct vertices . we say that a strong articulation point @xmath45 _ separates @xmath28 from @xmath0 _ if all paths from @xmath28 to @xmath0 contain @xmath45 . in this case @xmath28 and @xmath0 belong to different strongly connected components of @xmath118 . this observation implies that we can compute the vertex - resilient blocks by computing the strongly connected components of @xmath118 for every strong articulation point @xmath45 . to do this efficiently we define an operation that refines the currently computed blocks . let @xmath119 be a set of blocks , let @xmath120 be a partition of a set @xmath121 , and let @xmath45 be a vertex not in @xmath122 . for each block @xmath123 , substitute @xmath59 by the sets @xmath124 of size at least two , for all @xmath125 . in section [ section:2vc - blocks ] , where we will compute the @xmath2-vertex - connected blocks from the vertex - resilient blocks and the @xmath2-edge - connected blocks , we will use the notation @xmath126 as a shorthand for @xmath127 with @xmath128 . [ lemma : refine ] let @xmath129 be the total number of elements in all sets of @xmath119 ( @xmath130 ) , and let @xmath131 be the number of elements in @xmath122 . then , the operation @xmath127 can be executed in @xmath132 time . a simple way to achieve the claimed bound is to number the sets of the partition @xmath120 , each with a distinct integer i d in the interval @xmath133 $ ] . consider a block @xmath59 . each element @xmath134 is assigned a label that is equal to the i d of the set @xmath125 that contains @xmath0 if @xmath135 , and zero otherwise . then , the computation of the sets @xmath124 for all @xmath125 can be done in @xmath136 time with bucket sorting . [ lemma : simplevrb ] algorithm runs in @xmath137 time , where @xmath138 is the number of strong articulation points of @xmath8 . this is @xmath139 in the worst case . the strong articulation points of @xmath8 can be computed in linear time by @xcite . in each iteration of step 3 , we can compute the strongly connected components of @xmath118 in linear time @xcite . as we discover the @xmath140-th strongly connected component , we assign label @xmath140 ( @xmath141 ) to the vertices in @xmath142 . by lemma [ lemma : blocks - number ] , the number of vertex - resilient blocks of @xmath8 is at most @xmath44 . therefore , since the total number of blocks ( trivial and non - trivial ) can not decrease during any iteration , @xmath119 contains at most @xmath44 blocks in each execution of step 3 . by induction on the number of iterations , it follows that the algorithm maintains the invariant that any two distinct blocks in @xmath119 have at most one element in common , and that the corresponding block graph is a forest . therefore , by lemma [ lemma : blocks - number ] , the total number of elements in all blocks is at most @xmath94 . so , by lemma [ lemma : refine ] , each iteration of step 3 takes @xmath6 time . this yields the desired @xmath137 running time , where @xmath143 is the number of strong articulation points of @xmath8 . since a digraph may have up to @xmath4 strong articulation points , this is @xmath139 in the worst case . we will show how to obtain a faster algorithm by applying the framework developed in @xcite for the computation of the @xmath2-edge - connected blocks , namely by using dominator trees and auxiliary graphs . as already mentioned , auxiliary graphs need to be defined in a substantially different way , which complicates several technical details . as a warm up , first consider the computation of @xmath144 , i.e. , the vertex - resilient blocks that contain a specific vertex @xmath0 . consider the flow graph @xmath145 with start vertex @xmath0 and its reverse digraph @xmath146 , obtained after reversing edge directions . let @xmath1 be a vertex other than @xmath0 . clearly , @xmath0 and @xmath1 are vertex - resilient if and only if @xmath0 is the only proper dominator of @xmath1 in both @xmath145 and @xmath146 , i.e. , @xmath147 and @xmath148 . now let @xmath28 be a sibling of @xmath1 in both @xmath149 and @xmath150 . the fact that @xmath148 and @xmath151 implies that for any vertex @xmath152 there is path from @xmath1 to @xmath28 through @xmath0 that avoids @xmath45 . so @xmath1 and @xmath28 are in a common vertex - resilient block that contains @xmath0 if and only if they lie in the same strongly connected component of @xmath153 . this observation implies the following linear - time algorithm to compute the vertex - resilient blocks that contain @xmath0 . compute the dominator trees @xmath149 and @xmath150 of @xmath145 and @xmath146 respectively . let @xmath154 ( resp . , @xmath155 ) be the set of children of @xmath0 in @xmath145 ( resp . , @xmath146 ) . set @xmath156 and initialize the set of blocks @xmath157 . compute the strongly connected blocks @xmath158 of @xmath153 . let @xmath120 be the set that contains the nonempty restrictions of the @xmath142 sets to @xmath122 , i.e. , @xmath120 contains the nonempty sets @xmath159 . finally , execute @xmath160 . note that all the vertex - resilient blocks can be computed in @xmath21 time by applying the above algorithm to all vertices @xmath0 . to avoid the repeated applications of this algorithm we develop a new concept of _ auxiliary graphs _ for @xmath2-vertex connectivity . before doing that , we state two properties regarding information that a dominator tree can provide about vertex - resilient blocks and paths . [ lemma : vertex - resilient - necessary ] let @xmath7 be a strongly connected graph , and let @xmath23 be an arbitrary start vertex . any two vertices @xmath45 and @xmath46 are vertex - resilient only if they are siblings in @xmath30 or one is the immediate dominator of the other in @xmath27 . immediate . [ lemma : blocks - paths ] let @xmath91 be a vertex , and let @xmath0 be any vertex that is not a descendant of @xmath91 in @xmath30 . then there is a path from @xmath0 to @xmath91 that does not contain any proper descendants of @xmath91 in @xmath30 . moreover , all simple paths from @xmath0 to any descendant of @xmath91 in @xmath30 contain @xmath91 . let @xmath56 be any path from @xmath0 to @xmath91 . ( such a path exists since graph @xmath8 is strongly connected . ) let @xmath28 be the first vertex on @xmath56 such that @xmath28 is a descendant of @xmath91 . then either @xmath161 or @xmath28 is a proper descendant of @xmath91 . in the first case the lemma holds . suppose @xmath28 is a proper descendant of @xmath91 . since @xmath0 is not a descendant of @xmath91 in @xmath30 , there is a path @xmath58 from @xmath25 to @xmath0 in @xmath8 that does not contain @xmath91 . then @xmath58 followed by the part of @xmath56 from @xmath0 to @xmath28 is a path from @xmath25 to @xmath28 that avoids @xmath91 , a contradiction . as in @xcite , _ auxiliary graphs _ are a key concept in our algorithm that provides a decomposition of the input digraph @xmath8 into smaller digraphs ( not necessarily subgraphs of @xmath8 ) that maintain the original vertex - resilient blocks . in @xcite we used a _ canonical decomposition _ of the input digraph , in order to obtain auxiliary graphs that maintain the @xmath2-edge - connected blocks . a key property of this decomposition was the fact that any vertex in an auxiliary graph @xmath22 is reachable from a vertex outside @xmath22 only though a single strong bridge . in the computation of the vertex - resilient blocks , however , we have to decompose the input digraph according to strong articulation points , and thus the above property is completely lost . to overcome this critical issue , we apply a different and more involved decomposition . let @xmath25 be an arbitrarily chosen start vertex in @xmath8 . recall that we denote by @xmath27 the flow graph with start vertex @xmath25 , by @xmath36 the flow graph obtained from @xmath27 after reversing edge directions , by @xmath30 and @xmath162 the dominator trees of @xmath27 and @xmath36 respectively , and by @xmath154 and @xmath155 the set of children of @xmath0 in @xmath30 and @xmath162 respectively . for each vertex @xmath91 , let @xmath163 denote the level @xmath164 descendants of @xmath91 , i.e. , @xmath165 , @xmath166 , etc . for each vertex @xmath167 that is not a leaf in @xmath30 we build the _ auxiliary graph @xmath168 of @xmath91 _ as follows . the vertex set of @xmath22 is @xmath169 and it is partitioned into a set of _ ordinary _ vertices @xmath170 and a set of _ auxiliary _ vertices @xmath171 . the auxiliary graph @xmath22 results from @xmath8 by contracting the vertices in @xmath172 as follows . all vertices that are not descendants of @xmath91 in @xmath30 are contracted into @xmath91 . for each vertex @xmath173 , we contract all descendants of @xmath1 in @xmath30 into @xmath1 . see figure [ fig : auxiliary ] . we use the same definition for the auxiliary graph @xmath174 of @xmath25 , with the only difference that we let @xmath25 be an ordinary vertex . also note that when we form @xmath174 from @xmath8 , no vertex is contracted into @xmath25 . in order to bound the size of all auxiliary graphs , we eliminate parallel edges during those contractions . [ lemma : auxiliary - graphs - size ] the auxiliary graphs @xmath22 have at most @xmath175 vertices and @xmath176 edges in total . a vertex of @xmath8 may appear in at most four auxiliary graphs . therefore , the total number of edges in all auxiliary graphs excluding type-(b ) shortcut edges @xmath177 with @xmath178 is at most @xmath179 . a type-(b ) shortcut edge @xmath177 with @xmath178 of @xmath22 corresponds to a unique vertex in @xmath180 , so there are at most @xmath4 such edges . cc , the dominator tree @xmath30 of flow graph @xmath27 , the auxiliary graph @xmath181 and the dominator tree @xmath182 of the flow graph @xmath183 . ( the edges of the dominator tree @xmath182 are shown directed from child to parent . ) the auxiliary vertices of @xmath42 are shown gray . [ fig : auxiliary],title="fig:",scaledwidth=50.0% ] & , the dominator tree @xmath30 of flow graph @xmath27 , the auxiliary graph @xmath181 and the dominator tree @xmath182 of the flow graph @xmath183 . ( the edges of the dominator tree @xmath182 are shown directed from child to parent . ) the auxiliary vertices of @xmath42 are shown gray . [ fig : auxiliary],title="fig:",scaledwidth=50.0% ] + @xmath8 & @xmath30 + + , the dominator tree @xmath30 of flow graph @xmath27 , the auxiliary graph @xmath181 and the dominator tree @xmath182 of the flow graph @xmath183 . ( the edges of the dominator tree @xmath182 are shown directed from child to parent . ) the auxiliary vertices of @xmath42 are shown gray . [ fig : auxiliary],title="fig:",scaledwidth=50.0% ] & , the dominator tree @xmath30 of flow graph @xmath27 , the auxiliary graph @xmath181 and the dominator tree @xmath182 of the flow graph @xmath183 . ( the edges of the dominator tree @xmath182 are shown directed from child to parent . ) the auxiliary vertices of @xmath42 are shown gray . [ fig : auxiliary],title="fig:",scaledwidth=23.0% ] + @xmath181 & @xmath182 [ lemma : blocks - auxiliary - paths ] let @xmath0 and @xmath1 be two vertices in @xmath184 . any path @xmath56 from @xmath0 to @xmath1 in @xmath8 has a corresponding path @xmath185 from @xmath0 to @xmath1 in @xmath22 , and vice versa . moreover , if @xmath0 and @xmath1 are both ordinary vertices in @xmath22 , then @xmath185 contains a strong articulation point if and only if @xmath56 does . the correspondence between paths in @xmath8 and paths in @xmath22 follows from the definition of the auxiliary graph . next we prove the second part of the lemma . let @xmath185 be the path in @xmath22 that corresponds to a path @xmath56 from @xmath0 to @xmath1 in @xmath8 , where both @xmath0 and @xmath1 are ordinary vertices in @xmath22 . by the construction of the auxiliary graph , we have that if @xmath185 contains a strong articulation point then so does @xmath56 . for the contraposition , suppose @xmath56 contains a strong articulation point @xmath45 . consider the following cases : * @xmath186 . then , by the construction of the auxiliary graph , we have @xmath187 . * @xmath45 is a descendant of a vertex @xmath188 . vertex @xmath84 is a strong articulation point since it is either @xmath45 or a proper descendant of @xmath45 . then , by lemma [ lemma : blocks - paths ] , the part of @xmath56 from @xmath0 to @xmath45 contains @xmath84 . so , @xmath185 also contains @xmath84 by the construction of the auxiliary graph . * @xmath45 is not a descendant of @xmath91 . in this case , we have @xmath189 . since @xmath0 and @xmath1 are ordinary vertices of @xmath22 , @xmath190 is not empty and therefore @xmath91 is a strong articulation point . by lemma [ lemma : blocks - paths ] , the part of @xmath56 from @xmath45 to @xmath1 contains @xmath91 . so , @xmath185 also contains @xmath91 by the construction of the auxiliary graph . hence , in every case @xmath185 contains a strong articulation point and the lemma follows . [ corollary : auxialiry - graphs ] each auxiliary graph @xmath22 is strongly connected . follows from the construction of @xmath22 , lemma [ lemma : blocks - auxiliary - paths ] , and the fact that @xmath8 is strongly connected . the next lemma shows that auxiliary graphs maintain the vertex - resilient relation of the original digraph . [ lemma : vertex - resilient - auxiliary ] let @xmath0 and @xmath1 be any two distinct vertices of @xmath8 . then @xmath0 and @xmath1 are vertex - resilient in @xmath8 if and only if they are both ordinary vertices in an auxiliary graph @xmath22 and they are vertex - resilient in @xmath22 . suppose first that @xmath0 or @xmath1 is @xmath25 . without loss of generality assume @xmath191 . then by lemma [ lemma : vertex - resilient - necessary ] we have that @xmath192 , so @xmath0 and @xmath1 are both ordinary vertices of @xmath174 . now consider that @xmath193 . from lemma [ lemma : vertex - resilient - necessary ] we have that @xmath0 and @xmath1 belong in a set @xmath194 so they are both ordinary vertices of @xmath22 . clearly if all paths from @xmath0 to @xmath1 in @xmath22 contain a common vertex ( strong articulation point ) , then so do all paths from @xmath0 to @xmath1 in @xmath8 by lemma [ lemma : blocks - auxiliary - paths ] . now we prove the converse . suppose all paths from @xmath0 to @xmath1 in @xmath8 contain a common vertex @xmath28 . if @xmath195 then also all paths from @xmath0 to @xmath1 in @xmath22 contain @xmath28 by the proof of lemma [ lemma : blocks - auxiliary - paths ] . so suppose @xmath178 . then @xmath0 is not an ancestor of @xmath1 in @xmath30 , since otherwise there would be a path from @xmath0 to @xmath1 that avoids @xmath28 . first consider that @xmath28 is a ( proper ) descendant of @xmath91 in @xmath30 . since @xmath0 is not an ancestor of @xmath1 in @xmath30 , there is a vertex @xmath196 that is an ancestor of @xmath28 . by lemma [ lemma : blocks - paths ] , all paths from @xmath0 to @xmath28 in @xmath8 , and thus all paths from @xmath0 to @xmath1 , contain @xmath45 . by lemma [ lemma : blocks - auxiliary - paths ] this is also true for all paths from @xmath0 to @xmath1 in @xmath22 . finally , if @xmath28 is not a descendant of @xmath91 , lemma [ lemma : blocks - paths ] implies that all paths from @xmath28 to @xmath1 in @xmath8 contain vertex @xmath91 . hence , all paths from @xmath0 to @xmath1 in @xmath8 contain @xmath91 , and so do all paths from @xmath0 to @xmath1 in @xmath22 by lemma [ lemma : blocks - auxiliary - paths ] . now we specify how to compute all the auxiliary graphs @xmath197 in @xmath3 time . observe that the edge set @xmath198 contains all edges in @xmath7 induced by the vertices in @xmath184 ( i.e. , edges @xmath199 such that @xmath195 and @xmath200 ) . we also add in @xmath198 the following types of _ shortcut _ edges that correspond to paths in @xmath8 . ( a ) if @xmath8 contains an edge @xmath177 such that @xmath201 is a descendant of @xmath91 in @xmath30 and @xmath200 then we add the shortcut edge @xmath202 where @xmath84 the is an ancestor of @xmath28 in @xmath30 such that @xmath188 . ( b ) if @xmath8 contains an edge @xmath177 such that @xmath28 but not @xmath0 is a descendant of @xmath91 in @xmath30 then we add the shortcut edge @xmath203 where @xmath84 the nearest ancestor of @xmath28 in @xmath30 such that @xmath204 ( @xmath205 if @xmath195 ) . we note that we do not keep multiple ( parallel ) shortcut edges . see figure [ fig : auxiliary ] . we use the same definition for the auxiliary graph @xmath174 of @xmath25 , with the only difference that we let @xmath25 be an ordinary vertex . we also note that @xmath174 does not contain type-(b ) shortcut edges . to construct the auxiliary graphs @xmath168 we need to specify how to compute the shortcut edges of type ( a ) and ( b ) . to do this efficiently we need to test ancestor - descendant relations in @xmath30 . there are several simple @xmath206-time tests of this relation @xcite . the most convenient one for us is to number the vertices of @xmath30 from @xmath207 to @xmath4 in preorder , and to compute the number of descendants of each vertex . then , vertex @xmath0 is a descendant of @xmath91 if and only if @xmath208 , where , for any vertex @xmath45 , @xmath209 and @xmath210 are , respectively , the preorder number and the number of descendants of @xmath45 in @xmath30 . suppose @xmath177 is an edge of type ( a ) . we need to find the ancestor @xmath84 of @xmath28 in @xmath30 such that @xmath188 . we process all such arcs of @xmath22 as follows . we create a list @xmath211 that contains the edges @xmath177 of type ( a ) , and sort @xmath211 in increasing preorder of @xmath28 . we create a second list @xmath212 that contains the vertices in @xmath180 , and sort @xmath212 in increasing preorder . then , the shortcut edge of @xmath177 is @xmath202 , where @xmath84 is the last vertex in the sorted list @xmath212 such that @xmath213 . thus the shortcut edges of type ( a ) can be computed in linear time by bucket sorting and merging . now we consider the edges of type ( b ) . for each vertex @xmath173 we need to test if there is an edge @xmath177 in @xmath8 such that @xmath28 is a proper descendant of @xmath1 and @xmath0 is not a descendant of @xmath91 in @xmath30 . in this case , we add in @xmath22 the edge @xmath214 . to do this test efficiently , we assign to each edge @xmath177 a tag @xmath215 which we set equal to the preorder number of the nearest common ancestor of @xmath28 and @xmath0 in @xmath30 . we can do this easily by using the parent property and the @xmath206-time test of the ancestor - descendant relation as follows : @xmath216 if @xmath28 is an ancestor of @xmath0 in @xmath30 , @xmath217 if @xmath0 is an ancestor of @xmath28 in @xmath30 , and @xmath218 otherwise . at each vertex @xmath31 in @xmath30 we store a label @xmath219 which is the minimum tag of among the edges @xmath115 . using these labels we compute for each @xmath31 in @xmath30 the values @xmath220 . these computations can be done in @xmath221 time by processing the tree @xmath30 in a bottom - up order . now consider the auxiliary graph @xmath22 . we process the vertices in @xmath180 . for each such vertex @xmath1 we add the shortcut edge @xmath214 if @xmath222 . [ lemma : auxiliary - graphs - construction ] we can compute all auxiliary graphs @xmath22 in @xmath221 time . our linear - time algorithm is illustrated in figure [ fig : fastvrb ] . it uses two levels of auxiliary graphs and applies one iteration of algorithm for each auxiliary graph of the second level . the algorithm uses different dominator trees , and applies lemma [ lemma : vertex - resilient - necessary ] in order to identify the vertex - resilient blocks . since different dominator trees may define different blocks ( which by lemma [ lemma : vertex - resilient - necessary ] are supersets of the vertex - resilient blocks ) , we will use an operation that we call @xmath223 to combine the different blocks . we begin by computing the dominator tree @xmath30 for an arbitrary start vertex @xmath25 . for any vertex @xmath0 , we let @xmath224 denote the set containing @xmath0 and the children of @xmath0 in @xmath30 , i.e. , @xmath225 . lemma [ lemma : vertex - resilient - necessary ] gives an initial division of the vertices into blocks that are supersets of the vertex - resilient blocks . specifically , the vertex - resilient blocks that contain @xmath0 are subsets of @xmath224 or @xmath226 ( for @xmath227 ) . during the course of the algorithm , each vertex @xmath0 becomes associated with a set of blocks @xmath228 that contain @xmath0 , which are subsets of @xmath224 and @xmath226 if @xmath227 . the blocks are refined by applying the @xmath229 operation of section [ section : vr - simple ] and operation @xmath223 that we define next , and at the end of the algorithm each set of blocks @xmath228 will be equal to @xmath144 . let @xmath59 be a block and @xmath38 be a tree with vertex set @xmath230 . for any vertex @xmath231 , let @xmath232 be the set containing @xmath0 and the children of @xmath0 in @xmath38 . return the set that consists of the blocks @xmath233 of size at least two , for all @xmath231 . [ lemma : split ] let @xmath129 be the number of vertices in @xmath234 . then , the operation @xmath235 can be executed in @xmath236 time . we number the vertices of @xmath38 in preorder . let @xmath237 be the preorder number of @xmath231 . let @xmath238 be the parent of @xmath239 in @xmath38 , where @xmath91 is the root of @xmath38 . we associate each vertex @xmath239 in @xmath59 with two labels @xmath240 and @xmath241 , and create two corresponding pairs @xmath242 and @xmath243 . also , if @xmath244 , we associate @xmath91 with one label @xmath245 , and create a corresponding pair @xmath246 . each block created by the @xmath223 operation consists of a set of at least two vertices @xmath134 that are associated with a specific label . we can find these blocks by sorting the pairs @xmath247 by label , which can be done in @xmath236 time with bucket sort . at a high level , the algorithm begins with a `` coarse '' block tree , induced by the @xmath224 sets of @xmath30 , which is then refined by the blocks defined from the dominator trees of the auxiliary graphs . the final vertex - resilient block forest is then computed by considering the strongly connected components of the second level auxiliary graphs , after removing their designated start vertex . the algorithms needs to keep track of the blocks that contain a specific vertex , and , conversely , of the vertices that are contained in a specific block . to facilitate this search we explicitly store the adjacency lists of the current block forest @xmath43 . recall that @xmath43 is bipartite , so the adjacency list of a vertex @xmath0 stores the blocks that contain @xmath0 , and the adjacency list of a block node @xmath59 stores the vertices in @xmath59 . initially @xmath43 contains one block for each set @xmath224 , for all vertices @xmath0 that are not leaves in @xmath30 . these blocks are later refined by executing the @xmath223 and @xmath229 operations , which maintain the invariant that @xmath43 is a forest , and that any two distinct blocks have at most two vertices in common . when we execute a @xmath223 or a @xmath229 operation we can update the adjacency lists of @xmath43 , while maintaining the bounds given in lemmas [ lemma : refine ] and [ lemma : split ] . also , since during the execution of the algorithm the number of blocks can only increase , @xmath43 contains at most @xmath44 blocks at any given time . this fact implies that lemma [ lemma : blocks - number ] holds , so the total number of vertices and edges in @xmath43 is @xmath6 . [ lemma : fastvrb - correctness ] algorithm is correct . let @xmath28 and @xmath0 be any vertices . if @xmath28 and @xmath0 are vertex - resilient in @xmath8 , then by lemma [ lemma : vertex - resilient - auxiliary ] they are vertex - resilient in both auxiliary graphs of @xmath8 and @xmath22 that contain them as ordinary vertices . this implies that the algorithm will correctly include them in the same block in step 1 and will not separate them in steps 3.3 and 3.5 . so suppose that @xmath28 and @xmath0 are not vertex - resilient . then , without loss of generality , we can assume that all paths from @xmath28 to @xmath0 contain a common strong articulation point . thus , @xmath248 . we argue that all the blocks that contain @xmath28 and all the blocks that contain @xmath0 will be separated in some step of the algorithm . first we observe that @xmath28 and @xmath0 can appear together in at most one of the blocks constructed in step 1 . also , @xmath28 and @xmath0 can remain in at most one block after each @xmath223 operation ( @xmath28 and @xmath0 can have at most one identical label @xmath249 ) . so suppose that @xmath28 and @xmath0 are still contained in one common block just before the execution of step 3.5 . we will show that @xmath28 and @xmath0 will be separated after the @xmath229 operation executed in step 3.5.3 . since @xmath28 and @xmath0 were not separated by a @xmath223 operation , they are either siblings or one is the parent of the other in @xmath250 . also , since @xmath248 we have the following cases . \(a ) @xmath251 . then @xmath28 and @xmath0 are both ordinary vertices of the auxiliary graph @xmath181 with @xmath252 . lemma [ lemma : vertex - resilient - auxiliary ] implies that @xmath22 contains a strong articulation point @xmath45 that separates @xmath28 from @xmath0 . we argue that @xmath45 is a proper ancestor of @xmath28 in @xmath250 . if not , then @xmath253 contains a path @xmath254 from @xmath28 to @xmath91 that avoids @xmath45 . since @xmath255 , @xmath42 contains a path @xmath58 from @xmath91 to @xmath0 that avoids @xmath45 . thus @xmath256 is a path in @xmath42 from @xmath28 to @xmath0 that avoids @xmath45 , a contradiction . now we claim that @xmath257 is also a strong articulation point that separates @xmath28 from @xmath0 . suppose the claim is false . then @xmath258 , so @xmath45 is a proper ancestor of @xmath259 in @xmath250 . let @xmath56 be a path from @xmath28 to @xmath0 that avoids @xmath259 . then @xmath45 is on @xmath56 since @xmath45 separates @xmath28 from @xmath0 . let @xmath260 be the part of @xmath56 from @xmath28 to @xmath45 . also , since @xmath45 is a proper ancestor of @xmath259 in @xmath250 , @xmath253 has a path @xmath261 from @xmath91 to @xmath45 that avoids @xmath259 . then @xmath256 is a path in @xmath42 from @xmath28 to @xmath91 that avoids @xmath259 , a contradiction . the claim implies that @xmath28 and @xmath0 are located in different strongly connected components of @xmath262 , so they are contained in different blocks computed in step 3.5.3 . \(b ) @xmath263 . then @xmath28 and @xmath0 are both ordinary vertices of the auxiliary graph @xmath181 . lemma [ lemma : vertex - resilient - auxiliary ] implies that @xmath22 contains a strong articulation point @xmath45 that separates @xmath28 from @xmath0 . by the same arguments as in case ( a ) , it follows that @xmath257 is a strong articulation point that separates @xmath28 from @xmath0 . so again @xmath28 and @xmath0 will be located in different blocks after step 3.5.3 . [ lemma : fastvrb - time ] algorithm runs in @xmath221 time . we account for the total time spent on each step that algorithm executes . step 1 takes @xmath221 time by @xcite , and step 2 takes @xmath221 time by lemma [ lemma : auxiliary - graphs - construction ] . from lemma [ lemma : auxiliary - graphs - size ] we have that the total number of vertices and the total number of edges in all auxiliary graphs @xmath42 of @xmath8 are @xmath6 and @xmath221 respectively . then , again by lemma [ lemma : auxiliary - graphs - size ] , the total size ( number of vertices and edges ) of all auxiliary graphs @xmath264 for all @xmath42 , computed in step 3.4 , is still @xmath221 and they are also computed in @xmath221 total time by lemma [ lemma : auxiliary - graphs - construction ] . now consider the @xmath223 operations . all these operations that occur during step 3.3 for a specific auxiliary graph @xmath22 operate on the same tree @xmath38 , which can be preprocessed once , as in lemma [ lemma : split ] , for all @xmath223 operations . therefore , the total preprocessing time for all @xmath223 operations is @xmath6 . excluding the preprocessing time for @xmath38 , a @xmath265 operation takes time proportional to the number of vertices in @xmath59 . therefore all @xmath223 operations take @xmath6 time in total by lemmas [ lemma : blocks - size ] and [ lemma : split ] . in step 3.5.1 we examine the adjacency lists of the ordinary vertices @xmath266 and find the corresponding blocks that contain at least such two ordinary vertices . then we examine the adjacency lists of each such block . so , the adjacency lists of each vertex @xmath0 and each block that contains @xmath0 can be examined at most three times . hence , step 3.5.1 takes @xmath6 time in total . finally , steps 3.5.2 and 3.5.3 take @xmath221 time in total by @xcite and lemmas [ lemma : blocks - size ] and [ lemma : refine ] . algorithm computes the vertex - resilient blocks of the input digraph @xmath8 and stores them in the block forest @xmath43 of section [ sec : blocks ] , which makes it straightforward to test in constant time if two query vertices @xmath0 and @xmath1 are vertex - resilient . here we show that if @xmath0 and @xmath1 are not vertex - resilient , then we can report a witness of this fact , that is , a strong articulation point @xmath45 such that @xmath0 and @xmath1 are not in the same strongly connected component of @xmath102 . using this witness , it is straightforward to verify in @xmath221 time that @xmath0 and @xmath1 are not vertex - resilient ; it suffices to check that @xmath0 is not reachable from @xmath1 in @xmath102 or vice versa . to obtain this witness , we would like to apply lemma [ lemma : vb - graph - path-2 ] , but this requires @xmath0 and @xmath1 to be in the same tree of the block forest . fortunately , we can find the witness fast by applying lemmas [ lemma : vertex - resilient - necessary ] and [ lemma : blocks - paths ] , which use information computed during the execution of . we do that as follows . first consider the simpler case where @xmath191 . if lemma [ lemma : vertex - resilient - necessary ] does not hold for @xmath25 and @xmath1 in @xmath30 then @xmath267 is a strong articulation point that separates @xmath25 from @xmath1 . otherwise , @xmath268 , and @xmath25 and @xmath1 are both ordinary vertices in the auxiliary graph @xmath269 . then @xmath25 and @xmath1 can not satisfy lemma [ lemma : vertex - resilient - necessary ] in @xmath270 , so @xmath271 is a strong articulation point that separates @xmath1 from @xmath25 . now consider the case where @xmath272 . suppose first that @xmath0 and @xmath1 do not satisfy lemma [ lemma : vertex - resilient - necessary ] in @xmath30 . then @xmath32 is not an ancestor of @xmath0 or @xmath273 is not an ancestor of @xmath1 ( or both ) . assume , without loss of generality , that @xmath32 is not an ancestor of @xmath0 . by lemma [ lemma : blocks - paths ] , all paths from @xmath0 to @xmath1 pass through @xmath32 , so @xmath32 is a strong articulation point that separates @xmath0 from @xmath1 . on the other hand , if lemma [ lemma : vertex - resilient - necessary ] holds for @xmath0 and @xmath1 in @xmath30 , then @xmath0 and @xmath1 are both ordinary vertices in an auxiliary graph @xmath181 , where @xmath274 if @xmath275 , @xmath276 if @xmath277 , and @xmath278 otherwise . by lemma [ lemma : vertex - resilient - auxiliary ] , @xmath0 and @xmath1 are not vertex - resilient in @xmath42 . if they violate lemma [ lemma : vertex - resilient - necessary ] for @xmath182 then we can find a strong articulation point that separates them as above . finally , assume that lemma [ lemma : vertex - resilient - necessary ] holds for @xmath0 and @xmath1 in @xmath182 . now @xmath0 and @xmath1 are both ordinary vertices in an auxiliary graph @xmath279 . from the proof of lemma [ lemma : fastvrb - correctness ] we have that @xmath280 or @xmath281 and that @xmath259 is a strong articulation point that separates @xmath0 and @xmath1 . all the above tests can be performed in constant time . it suffices to store the dominator tree @xmath30 of @xmath27 , and the dominator trees @xmath182 of all auxiliary graphs @xmath282 . the space required for these data structures is @xmath6 by lemma [ lemma : auxiliary - graphs - size ] . [ theorem : vrb ] let @xmath8 be a digraph with @xmath4 vertices and @xmath5 edges . we can compute the vertex - resilient blocks of @xmath8 in @xmath3 time and store them in a data structure of @xmath6 space . given this data structure , we can test in @xmath206 time if any two vertices are vertex - resilient . moreover , if the two vertices are not vertex - resilient , then we can report in @xmath206 time a strong articulation point that separates them . we can compute the @xmath2-vertex - connected blocks of the input digraph @xmath7 by applying corollary [ cor:2vc - resilient ] as follows . given the vertex - resilient blocks @xmath119 and the @xmath2-edge - connected blocks @xmath120 of @xmath8 , we simply execute @xmath126 . this takes @xmath6 time by lemma [ lemma : refine ] . also , since the @xmath2-vertex - connected blocks have a block forest representation , we can test if two given vertices are @xmath2-vertex - connected in @xmath206 time as described in section [ sec : blocks ] . if we only wish to answer queries of whether two vertices @xmath0 and @xmath1 are @xmath2-vertex - connected , without computing explicitly the @xmath2-vertex and the @xmath2-edge - connected blocks , then we can use a simpler alternative , as suggested by lemma [ lemma:2vc - resilient - sb ] . first , we test if @xmath0 and @xmath1 are vertex - resilient in @xmath206-time as in section [ sec : queries ] , and if they are not , then we can report a strong articulation point that separates them . if , on the other hand , @xmath0 and @xmath1 are vertex - resilient then we need to check if @xmath8 contains @xmath114 or @xmath115 as a strong bridge . we can do this easily using the same information as in section [ sec : queries ] , namely the dominator tree @xmath30 of @xmath27 , and the dominator trees @xmath182 of all auxiliary graphs @xmath282 . for instance , if @xmath114 is a strong bridge in @xmath8 , then it will appear as an edge in one of the dominator trees . therefore , it suffices to mark the edges of dominator trees that are strong bridges , and then check if @xmath0 is the parent of @xmath1 or @xmath1 is the parent of @xmath0 in @xmath30 or in @xmath250 , where @xmath181 is the auxiliary graph of @xmath8 such that @xmath274 if @xmath275 , @xmath276 if @xmath277 , and @xmath278 otherwise . [ theorem:2vcb ] let @xmath8 be a digraph with @xmath4 vertices and @xmath5 edges . we can compute the @xmath2-vertex - connected blocks of @xmath8 in @xmath3 time and store them in a data structure of @xmath6 space . given this data structure , we can test in @xmath206 time if any two vertices are @xmath2-vertex - connected . moreover , if the two vertices are not @xmath2-vertex - connected , then we can report in @xmath206 time a strong articulation point or a strong bridge that separates them . here we show how to extend algorithm so that it also computes in linear time a sparse certificate for the vertex - resilient and the @xmath2-vertex - connected relations . that is , we compute a subgraph @xmath283 of the input graph @xmath8 that has @xmath6 edges and maintains the same vertex - resilient and @xmath2-vertex - connected blocks as the input graph . we can achieve this by applying the same approach we used in @xcite for computing a sparse certificate for the @xmath2-edge - connected blocks . as in section [ section : vertex - resilient - blocks ] we can assume without loss of generality that @xmath8 is strongly connected , in which case subgraph @xmath283 will also be strongly connected . the certificate uses the concept of _ independent spanning trees _ @xcite . a spanning tree @xmath38 of a flow graph @xmath27 is a tree with root @xmath25 that contains a path from @xmath25 to @xmath0 for all vertices @xmath0 . two spanning trees @xmath59 and @xmath284 rooted at @xmath25 are _ independent _ if for all @xmath0 , the paths from @xmath25 to @xmath0 in @xmath59 and @xmath284 share only the dominators of @xmath0 . every flow graph @xmath27 has two such spanning trees , computable in linear time @xcite . moreover , the computed spanning trees are _ maximally edge - disjoint _ , meaning that the only edges they have in common are the bridges of @xmath27 . during the execution of algorithm , we maintain a list ( multiset ) @xmath285 of the edges to be added in @xmath283 . the same edge may be inserted into @xmath285 multiple times , but the total number of insertions will be @xmath6 . then we can use radix sort to remove duplicate edges in @xmath6 time . we initialize @xmath285 to be empty . during step 1 of algorithm we compute two independent spanning trees , @xmath286 and @xmath287 of @xmath27 and insert their edges into @xmath285 . next , in step 3.1 we compute two independent spanning trees @xmath288 and @xmath289 for each auxiliary graph @xmath183 . for each edge @xmath177 of these spanning trees , we insert a corresponding edge into @xmath285 as follows . if both @xmath28 and @xmath0 are ordinary vertices in @xmath183 , we insert @xmath177 into @xmath285 since it is an original edge of @xmath8 . otherwise , @xmath28 or @xmath0 is an auxiliary vertex and we insert into @xmath285 a corresponding original edge of @xmath8 . such an original edge can be easily found during the construction of the auxiliary graphs . finally , in step 3.5 , we compute two spanning trees for every connected component @xmath142 of each auxiliary graph @xmath262 as follows . let @xmath290 be the subgraph of @xmath291 that is induced by the vertices in @xmath142 . we choose an arbitrary vertex @xmath292 and compute a spanning tree of @xmath293 and a spanning tree of @xmath294 . we insert in @xmath285 the original edges that correspond to the edges of these spanning trees . the sparse certificate @xmath283 has the same vertex - resilient blocks and @xmath2-vertex - connected blocks as the input digraph @xmath8 . we first argue that the execution of algorithm on @xmath283 and produces the same vertex - resilient blocks as the execution of algorithm on @xmath8 . the correctness of algorithm implies that it produces the same result regardless of the choice of start vertex @xmath25 . so we assume that both executions choose the same start vertex @xmath25 . we will refer to the execution of algorithm with input @xmath8 ( resp . @xmath283 ) as @xmath295 ( resp . @xmath296 ) . first we note that @xmath283 is strongly connected since it contains a spanning tree of @xmath27 and a spanning tree for the reverse of each auxiliary graph @xmath22 . moreover , the fact that @xmath283 contains two independent spanning trees of @xmath8 implies that @xmath8 and @xmath283 have the same dominator tree with respect to the start vertex @xmath25 that are computed in step 1 . hence , the auxiliary graphs computed in step 2 of algorithm have the same sets of ordinary and auxiliary vertices in both executions @xmath295 and @xmath296 . hence , step 3.1 computes the same dominator trees @xmath297 and @xmath250 in both executions , and therefore steps 3.2 and 3.3 compute the same blocks . the same argument as in steps 1 and 2 implies that both executions @xmath295 and @xmath296 compute in step 3.4 auxiliary graphs @xmath264 with the same sets of ordinary and auxiliary vertices . finally , by construction , the strongly connected components of each auxiliary graph @xmath262 are the same in both executions of @xmath295 and @xmath296 . we conclude that @xmath295 and @xmath296 compute the same vertex - resilient blocks as claimed . next , observe that since the independent spanning trees computed in steps 1 and 3.1 of the extended version of are maximally edge - disjoint , @xmath283 maintains the same strong bridges as @xmath8 . then , by corollary [ cor:2vc - resilient ] , @xmath283 also has the same @xmath2-vertex - connected blocks as @xmath8 . we presented the first linear - time algorithms for computing the vertex - resilient and the @xmath2-vertex - connected relations among the vertices of a digraph . we showed how to represent these relations with a data structure of @xmath6 size , so that it is straightforward to check in constant time if any two vertices are vertex - resilient or @xmath2-vertex - connected . moreover , if the answer to such a query is negative , then we can provide a witness of this fact in constant time , i.e. , a vertex ( strong articulation point ) or an edge ( strong bridge ) of @xmath8 that separates the two query vertices . an experimental study of the algorithms described in this paper is presented in @xcite , where it is shown that they perform very well in practice on very large graphs ( with millions of vertices and edges ) . we leave as an open question if the @xmath2-edge - connected or the @xmath2-vertex - connected components of a digraph can be computed faster than @xmath20 . w. di luigi , l. georgiadis , g. f. italiano , l. laura , and n. parotsidis . 2-connectivity in directed graphs : an experimental study . in _ proc . 17th wks . on algorithm engineering and experiments _ , pages 173187 , 2015 .	given a directed graph , two vertices @xmath0 and @xmath1 are @xmath2-_vertex - connected _ if there are two internally vertex - disjoint paths from @xmath0 to @xmath1 and two internally vertex - disjoint paths from @xmath1 to @xmath0 . in this paper , we show how to compute this relation in @xmath3 time , where @xmath4 is the number of vertices and @xmath5 is the number of edges of the graph . as a side result , we show how to build in linear time an @xmath6-space data structure , which can answer in constant time queries on whether any two vertices are @xmath2-vertex - connected . additionally , when two query vertices @xmath0 and @xmath1 are not @xmath2-vertex - connected , our data structure can produce in constant time a `` witness '' of this property , by exhibiting a vertex or an edge that is contained in all paths from @xmath0 to @xmath1 or in all paths from @xmath1 to @xmath0 . we are also able to compute in linear time a sparse certificate for @xmath2-vertex connectivity , i.e. , a subgraph of the input graph that has @xmath6 edges and maintains the same @xmath2-vertex connectivity properties as the input graph .
we address the issue of front propagation in a reaction - transport equation of kinetic type , @xmath1 here , the density @xmath2 describes a population of individuals in a continuum setting , and @xmath3 is the macroscopic density . the subset @xmath4 is the set of all possible velocities . individuals move following a velocity - jump process : they run with speed @xmath5 , and change velocity at rate 1 . they instantaneously choose a new velocity following the probability distribution @xmath6 . unless otherwise stated , we assume in this paper that @xmath7 is symmetric and @xmath8 satisfies the following properties : @xmath9 , and @xmath10 in addition , individuals are able to reproduce , with rate @xmath11 . new individuals start with a random velocity chosen with the same probability distribution @xmath6 . we could have chosen a different distribution without changing the main results , but we do not for the sake of clarity of the presentation . finally , we include a quadratic saturation term , which accounts for local competition between individuals , regardless of their speed . the main motivation for this work comes from the study of pulse waves in bacterial colonies of _ escherichia coli _ @xcite . kinetic models have been proposed to describe the run - and - tumble motion of individual bacteria at the mesoscopic scale @xcite . several works have been dedicated to derive macroscopic equations from those kinetic models in the diffusion limit @xcite . recently it has been shown that for some set of experiments , the diffusion approximation is not valid , so one has to stick to the kinetic equation at the mesoscopic scale to carefully compare with data @xcite . there is one major difference between this motivation and model . pulse waves in bacterial colonies of _ e. coli _ are mainly driven by chemotaxis which create macroscopic fluxes . growth of the population can be merely ignored in such models . in model however , growth and dispersion are the main reasons for front propagation , and there is no macroscopic flux due to the velocity - jump process since the distribution @xmath8 satisfies @xmath12 . for the sake of applications , we also refer to the growth and branching of the plant pathogen _ phytophthora _ by mean of a reaction - transport equation similar to @xcite . there is a strong link between and the classical fisher - kpp equation @xcite . in case of a suitable balance between scattering and growth ( more scattering than growth ) , we can perform the parabolic rescaling @xmath13 in , @xmath14 the diffusion limit yields @xmath15 , where @xmath16 is solution to the fisher - kpp equation ( see @xcite for example ) , @xmath17 we recall that for nonincreasing initial data decaying sufficiently fast at @xmath18 , the solution of ( [ kpp ] ) behaves asymptotically as a travelling front moving at the minimal speed @xmath19 @xcite . in addition , this front is stable in some weighted @xmath0 space @xcite . therefore it is natural to address the same questions for . we give below the definition of a travelling wave for equation . [ def : deftw ] we say that a function @xmath2 is a smooth travelling front solution of speed @xmath20 of equation if it can be written @xmath21 , where the profile @xmath22 satisfies @xmath23 in fact the profile @xmath24 is a solution to the stationary equation in a moving frame , @xmath25 the existence of travelling waves in reaction - transport equations has been adressed by schwetlick @xcite for a similar class of equations . first , the set @xmath7 is bounded and @xmath8 is the uniform distribution over @xmath7 . second , the nonlinearity can be chosen more generally ( either monostable as here , or bistable ) , but it depends only on the macroscopic density @xmath26 ( * ? ? ? ( 4 ) ) . for the monostable case , using a quite general method he was able to prove existence of travelling waves of speed @xmath27 for any @xmath28 , a result very similar to the fisher - kpp equation . we emphasize , that although the equations differ between schwetlick s work and ours , they coincide as far as the linearization in the regime of low density @xmath29 is concerned . on the contrary to schwetlick , we do not consider a general nonlinearity and we restrict to the logistic case , but we consider general velocity kernels @xmath30 . more recently , the rescaled equation ( [ eq : kinkpp2 ] ) has been investigated by cuesta , hittmeir and schmeiser @xcite in the parabolic regime @xmath31 . using a micro - macro decomposition , they construct possibly oscillatory travelling waves of speed @xmath32 for @xmath33 small enough ( depending on @xmath34 ) . in addition , when the set of admissible speeds @xmath7 is bounded , and @xmath35 they prove that the travelling wave constructed in this way is indeed nonnegative . lastly , when @xmath8 is the measure @xmath36 for some @xmath37 , equation ( [ eq : kinkpp ] ) is analogous to the reaction - telegraph equation for the macroscopic density @xmath26 ( up to a slight change in the nonlinearity however ) . this equation has been the subject of a large number of studies in the applied mathematics community @xcite . recently , the authors prove the existence of a minimal speed @xmath38 such that travelling waves exist for all speed @xmath39 @xcite . moreover these waves are stable in some @xmath0 weighted space , with a weight which differs from the classical exponential weight arising for the fisher - kpp equation . as the reaction - telegraph equation involves both parabolic and hyperbolic contributions , the smoothness of the wave depends on the balance between these contributions . in fact there is a transition between a parabolic ( smooth waves ) and a hyperbolic regime ( discontinuous waves ) , see remark [ rem : two velocities ] below . the authors also prove the existence of supersonic waves , having speed @xmath40 ( see remark [ rem : supersonic ] ) . the aim of the present paper is to investigate the existence and stability of travelling waves for equation ( [ eq : kinkpp ] ) for arbitrary kernels @xmath8 satisfying ( [ eq : hypm ] ) . for the existence part , we shall use the method of sub- and supersolutions , which do not rely on a perturbation argument . the stability part relies on the derivation of a suitable weight from which we can build a lyapunov functional for the linearized version of . the crucial assumption for the existence of travelling waves is the boundedness of @xmath7 . we prove in fact that in the case @xmath41 there exists no ( positive ) travelling wave . we then investigate the spreading rate for some particular choices of @xmath8 ( gaussian distribution , cauchy s distribution ) . unfortunately we are only able to give partial answer to this last question . in the last stage of writting of this paper , we realized that this issue was already addressed by mndez , campos and gmez - portillo for a slightly different equation admitting the same linearization near the front edge @xcite . our results are in agreement with their predictions . [ existence - tw ] assume that the set @xmath7 is compact , and that @xmath42 satisfies . let @xmath43 . there exists a speed @xmath44 such that there exists a travelling wave @xmath24 solution of of speed @xmath27 for all @xmath45 . the travelling wave is nonincreasing with respect to the space variable : @xmath46 . moreover , if @xmath47 then there exists no travelling wave of speed @xmath48 . the minimal speed @xmath38 is given through the following implicit dispersion relation : for each @xmath49 there is a unique @xmath50 such that @xmath51 then we have the formula @xmath52 [ rem : two velocities ] in the special case of two possible velocities only @xcite , corresponding to @xmath53 , two regimes have to be distinguished , namely @xmath54 and @xmath55 . in the case @xmath56 the travelling wave with minimal speed vanishes on a half - line . there , the speed of the wave is not characterized by the linearized problem for @xmath57 . note that this case is not contained in the statement of theorem [ existence - tw ] since it is assumed that @xmath42 . this makes a clear difference between the case of integrable @xmath8 and the case of a measure with atoms . [ rem : supersonic ] we expect that travelling waves exist for any @xmath39 , although this seems to contradict the finite speed of propagation when @xmath58 . in fact _ supersonic _ waves corresponding to @xmath58 should be driven by growth mainly , as it is the case in a simplified model with only two speeds @xcite . a simple argument to support the existence of such waves consists in eliminating the transport part , and seeking waves driven by growth only , @xmath59 . integrating with respect to @xmath60 yields a logistic equation for @xmath61 , @xmath62 , which as a solution connecting @xmath63 and @xmath64 for any positive @xmath27 . however these waves are purely artificial and we do not address this issue further . we now define @xmath65 and investigate the dependence of the minimal speed with respect to the velocity kernel @xmath66 . in the following proposition , we give some general bounds on the minimal speed . [ prop : bound on c * ] assume that @xmath7 is symmetric and that @xmath67 for all @xmath5 . the minimal speed satisfies the following properties , 1 . _ [ scaling ] _ for @xmath68 , define @xmath69 , then @xmath70 2 . _ [ rearrangement ] _ denote @xmath71 the schwarz decreasing rearrangement of the function @xmath8 ( see @xcite for a definition of this notion ) and @xmath72 the schwarz increasing rearrangement of the density distribution @xmath8 , then @xmath73 3 . _ [ comparison ] _ if @xmath54 then @xmath74 on the other hand , if @xmath55 then @xmath75 4 . _ [ diffusion limit ] _ in the diffusion limit @xmath13 we recover the kpp speed of the wave , @xmath76 in the case of a bounded set of velocities , we prove that for suitable initial data @xmath77 , the front spreads asymptotically with speed @xmath38 , in a weak sense . [ prop : spreadingbounded ] assume that @xmath7 is bounded and that @xmath78 . let @xmath79 such that @xmath80 for all @xmath81 . let @xmath82 be the solution of the cauchy problem . then 1 . if there exists @xmath83 such that @xmath84 for all @xmath85 and @xmath5 , then for all @xmath86 , @xmath87 2 . if there exists @xmath88 and @xmath89 such that @xmath90 for all @xmath91 and @xmath5 , then for all @xmath92 , @xmath93 where @xmath38 is the minimal speed of existence of travelling waves given by theorem [ existence - tw ] we also establish linear and nonlinear stability in suitable weighted @xmath0 spaces . the keypoint is to derive a correct weight which enables to build a lyapunov functional for the linear problem . we construct a semi - explicit weight @xmath94 , but we believe it is not the optimal one in some sense ( see remark [ rk : phi not optimal ] ) . let @xmath24 be a travelling wave of speed @xmath27 , and let @xmath95 the perturbation of @xmath24 in the moving frame . neglecting the nonlinear contributions , we are led to investigate the linear equation @xmath96 [ linstability ] there exists a weight @xmath97 such that the travelling front of speed @xmath98 is linearly stable in the weighted space @xmath99 . more precisely , the following lyapunov identity holds true for any solution @xmath100 of the linear equation , @xmath101 the weight @xmath97 is explicitly given in definition [ eq : def weight phi ] . using a comparison argument , in the spirit of @xcite , together with the explicit formula of the dissipation for the linearized system , we prove a nonlinear stability result . [ nonlinstability ] let @xmath102 $ ] . let @xmath103 be a travelling wave with speed @xmath98 . let @xmath82 be a solution of . suppose that the initial data satisfies @xmath104 then the following lyapunov identity holds true for the perturbation @xmath105 , @xmath106 \left\vert u \right\vert^2 e^{- 2 \phi ( z , v ) } d z dv \leq 0\,,\ ] ] where @xmath97 denotes the same weight as in proposition [ linstability ] . we expect that nonlinear stability holds true for any @xmath107 $ ] . however this would require to redefine the weight @xmath97 , since we believe it is not the optimal one . boundedness of @xmath7 is a crucial hypothesis in order to build the travelling waves . we believe that it is a necessary and sufficient condition . we make a first step to support this conjecture by investigating the case @xmath41 . we first prove infinite speed of spreading of the front under the natural assumption @xmath108 . as a corollary there can not exist travelling wave in the sense of definition [ def : deftw ] . note that there exist travelling waves with less restrictive conditions than definition [ def : deftw ] , at least in the diffusive regime @xcite . these fronts are expected to oscillate as @xmath109 . we expect that such oscillating fronts do exist far from the diffusive regime . in the case where @xmath41 and @xmath8 is gaussian , we plotted the dispersion relation in the complex plane @xmath110 , for an arbitrary given @xmath111 . we observed that it selects two complex conjugate roots , supporting the fact that oscillating fronts should exist ( results not shown ) . [ prop : spreadingunbounded ] assume that @xmath112 for all @xmath113 . let @xmath79 such that @xmath80 for all @xmath81 and there exists @xmath88 and @xmath114 such that @xmath90 for all @xmath91 and @xmath5 . let @xmath82 be the solution of the cauchy problem . then for all @xmath111 , @xmath115 we can immediately deduce from this result the non - existence of travelling waves when @xmath116 , by taking such a travelling wave as an initial datum @xmath117 in order to reach a contradiction . assume that @xmath118 for all @xmath113 . then equation does not admit any travelling wave solution . next we investigate specific choices for the distribution @xmath8 , both numerically and theoretically . in the case of a gaussian distribution , we expect a spreading rate following the power law @xmath119 ( see also @xcite ) . to support this guess , we prove in fact that spreading occurs _ at most _ with this rate . for this purpose we build a supersolution which is spreading with this rate . this issue has been addressed by mndez , campos and gmez - portillo @xcite in a physical paper for a slightly different equation , where the nonlinearity lies in the diffusion kernel instead of the growth rate . they conjectured that , as for the kpp equation , the front speed is determined through the linearization of the equation near the unstable steady state @xmath64 . we believe that the linearization should give the power law of the propagation , but it is not clear to us that it will give the exact location of the transition . however , it turns out that the linearized equations in @xcite and in the present paper are the same . then , performing some fourier - laplace transform of the solution , mendez et al derived heuristically the power law of the propagation . in particular , for a gaussian kernel , they found out that the spreading rate is @xmath119 . [ thm : unbounded ] let @xmath120 for all @xmath113 . let @xmath79 such that @xmath80 for all @xmath81 . assume that there exists @xmath121 such that @xmath122 let @xmath82 be the solution of the cauchy problem . then for all @xmath123 , one has @xmath124 in the case of a cauchy distribution @xmath125 , we obtain faster spreading rate under similar assumptions ( see proposition [ supersolunbounded - cauchy ] ) , namely for all @xmath123 , one has @xmath126 the phenomenon of accelerating fronts have raised a lot of attention in the literature of reaction - diffusion equations . this phenomenon occurs for the fisher - kpp equation when the initial datum decays more slowly than any exponential @xcite ; for a variant of the fisher - kpp equation where the diffusion operator is replaced by a nonlocal dispersal operator with fat tails @xcite , or by a nonlocal fractional diffusion operator @xcite . recently , accelerating fronts have been conjectured to occur in a reaction - diffusion - mutation model which extends the fisher - kpp equation to a population with heterogeneous diffusion coefficient @xcite . there is some subtlety hidden behind this phenomenon of infinite speed of spreading . in fact the diffusion limit of the scattering equation ( namely @xmath127 ) towards the heat equation makes no difference between bounded or unbounded velocity sets ( see @xcite and the references therein ) . however very low densities behave quite differently , which can be measured in the setting of large deviations or wkb limit . this can be observed even in the case of a bounded velocity set . in @xcite the large deviation ( wkb ) limit of the scattering equation is performed . it differs largely from the classical eikonal equation obtained from the heat equation . the case of unbounded velocities is even more complicated @xcite . to conclude , let us emphasize that low densities are the one that drive the front here ( pulled front ) . so the diffusion limit is irrelevant in the case of unbounded velocities , since very low density of particles having very large speed makes a big difference . we first recall some useful results concerning the cauchy problem associated with : well - posedness and a strong maximum principle . these statements extend some results given in @xcite . they do not rely on the boundedness of @xmath7 . [ global existence : theorem 4 in @xcite ] [ cauchypbm ] let @xmath117 a measurable function such that @xmath128 for all @xmath81 . then the cauchy problem has a unique solution @xmath129 in the sense of distributions , satisfying @xmath130 the next result refines the comparison principle of @xcite in order to extend it to sub and supersolutions in the sense of distributions and to state a strong maximum principle . its proof is given in appendix . [ prop - mp ] assume that @xmath131 are respectively a super- and a subsolution of , _ i.e. _ @xmath132 in the sense of distributions . assume in addition that @xmath133 satisfies @xmath134 for all @xmath135 . then @xmath136 for all @xmath137 . assume in addition that @xmath7 is an interval , and that @xmath78 . if there exists @xmath138 such that @xmath139 , then one has @xmath140 for all @xmath141 such that @xmath142 . if @xmath116 , then this statement reads as in the parabolic framework : if @xmath143 and @xmath144 at @xmath145 , then @xmath146 for all @xmath147 . in the case @xmath148 $ ] we have to take into account finite speed of propagation , obviously . we assume throughout this section that @xmath149 is compact . we construct the travelling waves for @xmath150 . the proof is divided into several steps . it is based on a sub and supersolutions method . the aim of this first step is to solve the linearized equation of at @xmath151 , in the regime of low density @xmath57 . such an achievement gives information about the speed and the space decreasing rate of a travelling wave solution of the nonlinear problem , as for the fisher - kpp equation . the linear problem for writes @xmath152 we seek a solution having exponential decay at @xmath153 . more specifically we separate the variables in our ansatz : @xmath154 , with @xmath155 . the next proposition gathers the results concerning the linear problem . [ propdisp ] there exists a minimal speed @xmath38 such that for all @xmath156 , there exists @xmath157 such that @xmath158 is a nonnegative solution of . the profile @xmath159 is explicitely given by @xmath160 the admissible @xmath161 are solutions of the following dispersion relation , @xmath162 moreover , among all possible @xmath163 for a given @xmath27 , the minimal one @xmath164 is well defined and isolated . here appears the crucial assumption on the boundedness of @xmath7 . if this condition is not fulfilled , it is never possible to ensure that the profile @xmath159 is nonnegative since the denominator is linear with respect to @xmath60 . * * plugging the ansatz @xmath165 into yields @xmath166 the profile is given by @xmath167 the dispersion relation reads @xmath168 , or equivalently . moreover , we require the profile @xmath169 to be nonnegative , which gives the condition @xmath170 for all @xmath5 , which implies @xmath171 . from now on , we focus on the existence of solutions @xmath161 of , with @xmath172 and @xmath173 . let us denote @xmath174 so that we look for solutions of @xmath175 . * # step 2 . * technically speaking , for all @xmath172 , the function @xmath176 is analytic over @xmath177 indeed , as @xmath178 is integrable for all @xmath179 , it is clear that @xmath180 is the analytic development of @xmath181 for @xmath182 . next we observe that @xmath183 is decreasing for all @xmath184 , and that @xmath185 is strictly convex . moreover , the function @xmath181 satisfies the following properties : @xmath186 { } 0 \ , .\end{aligned}\ ] ] the last property relies on the lebesgue s dominated convergence theorem since @xmath187 . * # step 3 . * assume first that @xmath188 . then fatou s lemma gives @xmath189 as a consequence , @xmath190 is well defined and finite for all @xmath172 . it follows from the earlier properties that @xmath191 and @xmath192 . moreover , the regularity and monotonicity properties of @xmath181 guarantee that @xmath193 is continuous and decreasing . hence , there exists @xmath38 such that @xmath194 and there exists @xmath195 such that @xmath196 . next , for all @xmath197 , as @xmath183 is decreasing , one has @xmath198 for all @xmath86 . thus , as @xmath199 , there exists @xmath163 such that @xmath200 for all @xmath86 . second , consider a general @xmath42 possibly vanishing at @xmath201 . to recover the first step , we define for @xmath202 a new maxwellian by @xmath203 over @xmath7 ( and @xmath64 outside of @xmath7 ) , where @xmath204 is the measure of @xmath7 . then @xmath205 since @xmath206 , and thus the earlier step yields that there exists a sequence @xmath207 of minimal speeds associated with @xmath208 . we also associate @xmath209 with @xmath210 through ( [ eq - defi ] ) . we define @xmath211 and we now show that it is the minimal speed . * take @xmath92 . then for all @xmath212 and for some arbitrarily large @xmath179 so that @xmath213 , one has @xmath214 because @xmath215 as @xmath216 , we get @xmath217 thus @xmath200 has no solution for @xmath218 if @xmath92 . * assume that @xmath219 . then one has @xmath220 when @xmath179 is large enough and thus for all @xmath179 sufficiently large , there exists @xmath221 such that @xmath222 . up to extraction , one may assume that @xmath223 converges to some @xmath224 $ ] . fatou s lemma yields @xmath225 . hence , there exists a solution @xmath226 $ ] of @xmath200 and obviously @xmath227 since @xmath199 . * lastly , if @xmath228 , we know that for all @xmath229 , there exists @xmath230 $ ] such that @xmath231 . assuming that @xmath232 $ ] as @xmath233 , we get @xmath234 . [ minlambda ] for all @xmath235 , the quantity @xmath236 is well - defined . moreover , for all @xmath237 , if @xmath238 is small enough , then @xmath239 . we know from the definition of @xmath38 that for all @xmath150 , the set @xmath240 is not empty . thus , we can take a minimizing sequence @xmath241 which converges towards the infimum of @xmath242 . as this sequence is bounded , one can assume , up to extraction , that @xmath243 . then lebesgue s dominated convergence theorem gives @xmath244 . hence @xmath245 . next , we have already noticed in the proof of proposition [ propdisp ] that @xmath246 for all @xmath86 . as @xmath247 , the definition of @xmath164 yields @xmath248 . the conclusion follows from the strict convexity of the function @xmath185 . in this step we construct sub and supersolutions for . we fix @xmath250 and we denote @xmath251 without ambiguity . [ lem - super ] let @xmath252 then @xmath253 is a supersolution of , that is , it satisfies in the sense of distributions : @xmath254 [ lem - sub ] there exist @xmath255 and @xmath238 such that if @xmath256 then @xmath257 is a subsolution of , that is satisfies in the sense of distributions : @xmath258 first , @xmath259 and @xmath260 both clearly satisfy since @xmath261 . next , as @xmath253 is continuous , it immediately follows from the jump formula that , as a minimum of two supersolutions , it is a supersolution of ( [ eq - super ] ) in the sense of distributions . the same arguments as in the proof of lemma [ lem - super ] yield that it is enough to prove that ( [ eq - subper ] ) is satisfied by @xmath257 over the open set @xmath262 . as @xmath86 , proposition [ propdisp ] gives @xmath263 small enough such that @xmath264 and @xmath265 . we compute the linear part : @xmath266 to prove the lemma , we now have to choose a relevant @xmath267 such that @xmath268 holds for all @xmath269 . as @xmath270 for all @xmath271 , one has @xmath272 and thus it is enough to choose @xmath267 such that @xmath273 on the other hand for all @xmath271 such that @xmath274 , we have @xmath275 , meaning that @xmath276 plugging this estimate into , it is enough to choose @xmath267 such that @xmath277 this concludes the proof since such a @xmath267 obviously exists . let @xmath279 , where @xmath38 denotes the minimal speed of proposition [ propdisp ] . in order to prove the existence of travelling waves , we will prove that the solution of the following evolution equation , corresponding to equation in the moving frame at speed @xmath27 , converges to a travelling wave as @xmath280 : @xmath281 the well - posedness of equation immediately follows from proposition [ cauchypbm ] . let now derive some properties of the function @xmath82 from proposition [ prop - mp ] . [ lem - sandwich ] for all @xmath282 , one has @xmath283 as @xmath257 is a subsolution of ( [ eq - approx ] ) and @xmath253 is a supersolution of ( [ eq - approx ] ) , with @xmath284 for all @xmath285 , this result is an immediate corollary of proposition [ prop - mp ] . [ lem - monotonicity ] for all @xmath286 , the function @xmath287 is nonincreasing . take @xmath288 and define @xmath289 . then as @xmath253 is nonincreasing in @xmath290 , one has @xmath291 for all @xmath292 . proposition [ prop - mp ] yields that @xmath293 for all @xmath135 . [ lem - timedec ] for all @xmath294 , the function @xmath295 is nonincreasing . take @xmath296 and define @xmath297 . then lemma [ lem - sandwich ] yields that @xmath298 for all @xmath292 . hence , proposition [ prop - mp ] gives @xmath299 for all @xmath137 . [ lem - lipsch ] the family @xmath300 is uniformly continuous with respect to @xmath81 . moreover , for any @xmath301 , the continuity constants does not depend on @xmath302 . we begin with the space regularity . let @xmath303 . the function @xmath304 is such that @xmath305 is lipschitz - continuous with respect to @xmath290 . therefore there exists a constant @xmath306 such that for all @xmath81 , we have @xmath307 as @xmath308 , it is easily checked that @xmath309 is a supersolution of . hence proposition [ prop - mp ] yields that @xmath310 hence the function @xmath311 is lipschitz continuous with respect to @xmath290 . since the function @xmath311 is bounded from above , @xmath312 is also lipschitz continuous with respect to @xmath290 . the lipschitz constant is uniform with respect to @xmath313 and @xmath314 . we now come to the velocity regularity . for the sake of clarity we first consider the case where @xmath8 is @xmath315 on @xmath7 . the function @xmath316 is @xmath315 too . we introduce @xmath317 . it satisfies the following equation @xmath318 multiplying the equation by @xmath319 we obtain @xmath320 the l.h.s . is linear with respect to @xmath321 and satisfies the maximum principle . the r.h.s . is uniformly bounded since @xmath322 and @xmath82 is uniformly lipschitz with respect to @xmath290 . obviously the constant @xmath323 is a supersolution . we deduce that @xmath324 is uniformly bounded over @xmath325 . in the case where @xmath8 is only continuous over the compact set @xmath7 , thus uniformly continuous , we shall use the method of translations again . however we have to be careful since @xmath7 is bounded . let @xmath326 . we introduce @xmath327 . the function @xmath328 satisfies the following equation @xmath329 let @xmath330 . there exists @xmath331 such that for @xmath332 we have @xmath333 and @xmath334 . moreover we have obviously @xmath335 . we conclude using the same argument as in the @xmath315 case . the modulus of uniform continuity is uniform with respect to @xmath313 and @xmath314 . we are now in position to prove the existence of travelling waves of speed @xmath27 , except for the minimal speed @xmath38 . gathering lemmas [ lem - sandwich ] , [ lem - monotonicity ] and [ lem - timedec ] , we know that @xmath336 is well - defined for all @xmath292 , that @xmath337 is nonincreasing in @xmath290 for all @xmath60 and that @xmath338 . let now prove that @xmath24 defines a travelling wave solution of . as @xmath82 satisfies , converges pointwise and is bounded by the locally integrable function @xmath253 , it follows from lebesgue s dominated convergence theorem that @xmath24 satisfies in the sense of distributions . moreover , lemma [ lem - lipsch ] ensures that @xmath24 is continuous with respect to @xmath339 . we next check the limits towards infinity . let @xmath340 . thanks to @xmath341 , the lebesgue dominated convergence theorem gives @xmath342 . on the other hand , we get @xmath343 in the sense of distributions . integrating over the compact set @xmath7 , we deduce that @xmath344 , _ i.e. _ that @xmath345 or @xmath63 . as @xmath24 is nonincreasing and @xmath338 , one necessarily has @xmath346 and @xmath347 . finally , ( [ eq - limit ] ) gives @xmath348 and @xmath349 for all @xmath5 . this gives the appropriate limits . consider a decreasing sequence @xmath350 converging towards @xmath38 . we already know that for all @xmath179 , equation admits a travelling wave solution @xmath351 , with @xmath352 and @xmath353 , and @xmath354 is nonincreasing . up to translation , we can assume that @xmath355 . moreover , lemma [ lem - lipsch ] ensures that the functions @xmath356 are uniformly continuous with respect to @xmath81 since the continuity stated in lemma [ lem - lipsch ] is uniform with respect to @xmath302 for any @xmath301 . thanks to the ascoli theorem and a diagonal extraction process , we can assume that the sequence @xmath356 converges locally uniformly in @xmath81 to a function @xmath24 . clearly @xmath24 satisfies in the sense of distributions . moreover , as @xmath24 is nonincreasing with respect to @xmath290 , one could recover the appropriate limits at infinity with the same arguments as in the proof of the existence of travelling waves with speeds @xmath86 . [ lem : posharnack ] assume that @xmath357 . for all @xmath358 there exists @xmath359 and a nonnegative , arbitrarily small , compactly supported function @xmath360 which is a subsolution of @xmath361 for the sake of clarity we emphasize the dependence of the function @xmath181 upon the growth rate @xmath11 : @xmath362 we denote by @xmath363 the smallest speed such that there exists a solution @xmath49 of @xmath364 ( see proposition [ propdisp ] ) . let @xmath365 . by continuity we can choose @xmath366 so small that @xmath367 . we claim that there exists @xmath368 such that @xmath369 , with @xmath370 and @xmath371 . indeed we know from the proof of proposition [ propdisp ] [ step 3 ] that @xmath372 under the assumption @xmath373 . using a continuity argument we also have the strict inequality @xmath374 , uniformly with respect to @xmath366 . the complex function @xmath375 is analytic in a neighborhood of @xmath376 . hence , the rouch theorem yields that there exists @xmath377 such that the equation @xmath378 has a solution @xmath379 with @xmath380 arbitrarily close to @xmath376 . we denote by @xmath381 the corresponding velocity profile , @xmath382 by continuity we can choose @xmath383 and @xmath380 such that @xmath384 holds for all @xmath5 . by the very definition of @xmath385 , we have @xmath386 . we denote @xmath387 . recall that we have the strict inequality @xmath374 , uniformly with respect to @xmath366 . using a continuity argument we can impose that @xmath388 . now define the real function @xmath389 by @xmath390,\ ] ] one has @xmath391 and @xmath392 for all @xmath5 . thus , there exists an interval @xmath393\subset { \mathbb{r}}$ ] and a bounded domain @xmath394\times v$ ] such that : @xmath395 on the other hand , as @xmath388 , there exists a constant @xmath396 such that @xmath397 hence , one can choose @xmath398 small enough such that @xmath399 for all @xmath400 we deduce from @xmath401 the following identities , @xmath402 taking real part on both sides , we get for @xmath403 , @xmath404 hence @xmath405 is a subsolution of for all @xmath400 on @xmath406 . we deduce that the truncated function @xmath407 is a subsolution of over @xmath408 . assume that @xmath409 is a travelling wave solution of of speed @xmath410 . according to lemma [ lem : posharnack ] , there exists @xmath411 and a nonnegative compactly supported subsolution @xmath412 of with speed @xmath383 . as @xmath24 is positive and continuous , we can decrease @xmath412 so as to obtain @xmath413 . let @xmath414 and @xmath415 . these two functions are respectively a solution and a subsolution of . as @xmath416 for all @xmath81 , proposition [ prop - mp ] implies @xmath417 taking @xmath418 and letting @xmath280 , we get @xmath419 this is a contradiction . let @xmath86 . consider first the initial datum @xmath420 and let @xmath421 the solution of the cauchy problem . denote by @xmath24 a travelling wave of minimal speed @xmath38 . there exists @xmath422 such that @xmath423 for all @xmath81 . it is straightforward to check that @xmath424 is a supersolution of . hence , the comparison principle of proposition [ prop - mp ] implies that @xmath425 for all @xmath141 . in particular we have , @xmath426 as @xmath427 for all @xmath5 and @xmath86 , we get @xmath428 for all @xmath5 . on the other hand , as @xmath429 is nonincreasing with respect to @xmath430 it follows from the comparison principle that @xmath431 is nonincreasing ( see lemma [ lem - monotonicity ] ) . thus @xmath432 for all @xmath433 and the conclusion follows . for a general initial datum @xmath117 satisfying the hypotheses of proposition [ prop : spreadingbounded ] , one has @xmath434 for all @xmath81 and thus @xmath435 for all @xmath141 , from which the conclusion follows . let @xmath92 . the same arguments as in the first step yield that we can assume that @xmath436 let @xmath412 a compactly supported subsolution of associated with a speed @xmath437 . since @xmath412 can be chosen arbitrarily small , up to translation of @xmath412 , we can always assume that @xmath438 . let @xmath133 the solution of the cauchy problem associated with the initial datum @xmath439 . the comparison principle yields @xmath440 for all @xmath441 . let @xmath442 . this function satisfies @xmath443 clearly @xmath412 is a ( stationary ) subsolution of this equation . the comparison principle proposition [ prop - mp ] yields that @xmath444 is nondecreasing for all @xmath81 ( see also lemma [ lem - timedec ] for a similar proof ) . let @xmath445 . this function is clearly a weak solution of @xmath446 moreover , we have @xmath447 and @xmath448 . [ lem : sliding ] we have @xmath449 . * * first we prove that @xmath450 is positive everywhere . take @xmath451 such that @xmath452 . as @xmath453 satisfies , proposition [ prop - mp ] yields @xmath454 for all @xmath141 such that @xmath455 . as @xmath456 , for all @xmath294 one can take @xmath147 large enough so that @xmath457 . therefore @xmath458 . we thus conclude that @xmath450 is positive over @xmath408 . * # step 2 . * next we prove that @xmath459 . let @xmath460 . define @xmath461 , and @xmath462 as @xmath463 is compactly supported and @xmath450 is positive over @xmath408 and continuous , we have @xmath464 when @xmath465 is small enough . therefore @xmath466 . we argue by contradiction . assume that @xmath467 . the definition of @xmath468 yields that @xmath469 and that @xmath470 . as @xmath463 is compactly supported , this infimum is indeed reached : there exists @xmath471 such that @xmath472 . assume that @xmath473 and take @xmath474 such that @xmath475 . we introduce @xmath476 and @xmath477 . as @xmath478 , proposition [ prop - mp ] gives @xmath479 for all @xmath141 such that @xmath480 , that is : @xmath481 as @xmath482 , for all @xmath430 , one can take @xmath147 large enough so that @xmath483 , leading to @xmath484 for all @xmath81 , a contradiction since equality holds at @xmath485 . hence , @xmath486 , which is also a contradiction since @xmath450 is positive while @xmath463 is compactly supported . we conclude that @xmath487 , namely @xmath488 . evaluating this inequality at @xmath489 , one gets @xmath490 for all @xmath491 . as @xmath492 under the assumption @xmath78 , we have proved in fact that @xmath493 * # step 3 . * as @xmath494 , we can define @xmath495 we know from the previous step that this quantity is positive . if @xmath496 , then the same types of arguments as in step 2 lead to a contradiction . hence @xmath497 , meaning that @xmath498 . as @xmath499 , we conclude that @xmath500 . as a consequence of lemma [ lem : sliding ] we obtain @xmath501 this implies in particular that @xmath502 for all @xmath339 by a sandwiching argument . moreover , as @xmath117 is nonincreasing with respect to @xmath290 , @xmath503 is nonincreasing and thus @xmath504 for all @xmath505 , from which the conclusion follows since @xmath506 . recall that the dispersion relation giving the speed @xmath27 as a function of the exponential decay @xmath163 is @xmath507 , where @xmath508 is defined in . let introduce @xmath509 the function associated with the dilated velocity profile @xmath510 . the function @xmath509 clearly satisfies the scaling relation @xmath511 , therefore we get @xmath512 from the very definition of @xmath38 . we use the symmetry of the kernel @xmath513 to write @xmath514 let define @xmath515 it is an increasing function over @xmath516 , thus @xmath517 . the hardy - littlewood inequality ( * ? ? ? 3 ) yields @xmath518 the dispersion relation is nonincreasing with respect to @xmath27 . it follows immediately that @xmath519 we use the symmetry of the kernel @xmath513 . for @xmath49 the dispersion relation writes @xmath520 since the function @xmath521 is convex on its domain of definition , jensen s inequality yields @xmath522 we recognize the dispersion relation associated with the two - speed model @xcite . we deduce @xmath523 this second - order polynomial has a negative value at @xmath524 , therefore @xmath525 is necessarily greater than the vanishing value , @xmath526 minimizing the right - hand - side with respect to @xmath163 , we deduce that @xmath527 implies : @xmath528 on the other hand we clearly obtain from , @xmath529 by comparison of the relations , as in the proof of proposition [ prop : bound on c](b ) , we deduce that the speed corresponding to a given maxwellian @xmath30 supported on @xmath530 is smaller than the speed corresponding to @xmath531 . this peculiar case is analysed in @xcite . the minimal speed in this case is @xmath532 the dispersion relation for the rescaled equation reads @xmath533 the previous result guarantees that @xmath534 is bounded from above for @xmath535 , @xmath536 for a given @xmath49 , we perform a taylor expansion of up to second order , @xmath537 uniformly for @xmath538 $ ] , since @xmath7 is bounded . therefore , solving the relation dispersion for the minimal speed boils down to solving @xmath539 we deduce @xmath540 therefore the minimal speed verifies @xmath541 . in this subsection , we focus on the linearized problem around some travelling wave moving at speed @xmath542 . we recall that we consider a solution @xmath100 of the equation associated with the linearization around a travelling wave : @xmath543 where the notation @xmath544 always stands in the sequel for a function of the @xmath545 variable . we shall prove stability of the wave in a suitable @xmath0 framework , inspired by @xcite . we search for an ansatz @xmath546 . the function @xmath547 satisfies : @xmath548 from , we shall derive the dissipation inequality stated in proposition . we test against @xmath547 to obtain the kinetic energy estimate : @xmath549 we aim at choosing a weight @xmath97 such that the dissipation is coercive in @xmath0 norm . let define the symmetric kernel @xmath550 as follows @xmath551 we seek a function @xmath97 such that @xmath552 for a suitable positive function @xmath267 , in the sense of kernel operators . for this purpose we focus on the eigenvalues of the kernel operator @xmath553 . [ eq : def weight phi ] we introduce @xmath554 the smallest solution of the following dispersion relation @xmath555 and we define @xmath556 through the differential equation @xmath557 finally we define @xmath558 recall that @xmath559 , so that the weight @xmath97 is well - defined as soon as @xmath560 is well - defined . a small argumentation is required to prove that @xmath561 is well - defined too . for a given @xmath27 and @xmath562 , define @xmath563 the function @xmath564 is continuous , and satisfies the following properties , @xmath565\,,\\ & g(\lambda ) = \int_{v } \frac { ( 1+r ) m(v ) - r f ( z , v ) } { 1 + \lambda ( c - v ) } dv = 1 - \int_{v } \frac{r f ( z , v ) } { 1 + \lambda ( c - v ) } dv \leq 1\,,\end{aligned}\ ] ] where @xmath163 is chosen such that @xmath507 . thus we can define the smallest @xmath566 $ ] such that @xmath567 . [ eigens ] let @xmath267 be defined as @xmath568 and @xmath569 be the operator associated with the symmetric kernel @xmath570 . the operator @xmath569 is nonpositive . we shall prove that @xmath64 is the perron eigenvalue of the operator @xmath569 . for that purpose we shall exhibit a positive eigenvector in the nullset of @xmath569 . the equation @xmath571 reads @xmath572 plugging the formula for @xmath573 into this expression we get , @xmath574 from the definitions - we have , @xmath575 the weight @xmath97 and the function @xmath267 are chosen such that @xmath576 therefore the equation @xmath577 is equivalent to @xmath578 where the macroscopic quantities @xmath579 and @xmath580 are defined as follows , @xmath581 to resolve this eigenvalue problem , we seek proper values for @xmath579 and @xmath580 . from we deduce that they are solution of a @xmath582 closed linear system , namely @xmath583 this system simplifies thanks to the choice of @xmath561 . indeed we have @xmath584 we are reduced to the following eigenvalue problem , @xmath585 clearly , @xmath586 is the unique solution up to multiplication . we obtain eventually that @xmath587 is given ( up to a multiplicative factor ) by @xmath588^{1/2}}{1 + \lambda ( z ) ( c - v ) } > 0 \,.\end{aligned}\ ] ] as a consequence , we have found that the symmetric operator @xmath569 , which is nonnegative out of the diagonal @xmath589 , possesses a positive eigenvector @xmath587 associated with the eigenvalue 0 . therefore it is a nonpositive operator . this ends the proof of the lemma . we can now conclude the proof of proposition [ linstability ] . lemma [ eigens ] claims that for all @xmath590 such that @xmath591 is solution to the linearized equation , we have @xmath592 which proves the proposition . [ rk : phi not optimal ] we believe that the weight @xmath593 proposed in definition [ eq : def weight phi ] is not optimal with respect to the spectral property of the linearized operator . indeed the dissipation factor @xmath594 is equivalent in the diffusion limit ( @xmath595 ) to @xmath596 , although we expect @xmath597 @xcite . the missing factor @xmath598 is responsible for the restriction @xmath599 in our nonlinear stability result , theorem [ nonlinstability ] . let us recall how to derive the spectral properties of the linearized equation in the diffusive limit , namely the linearized fisher - kpp equation , @xmath600 where @xmath601 is the profile of the travelling wave in the frame @xmath602 . applying the same procedure as in the proof of proposition [ linstability ] , we shall derive an equation for the weighted perturbation @xmath603 , and optimize the dissipation with respect to the weight @xmath97 ( see also @xcite ) , as follows @xmath604 the best choice is achieved when @xmath605 is constant and minimizes @xmath606 , _ i.e. @xmath607_. in the case of the minimal speed @xmath608 , we obtain the following dissipation formula for the linearized operator , @xmath609 notice the factor @xmath598 which is apparently missing in the dissipation term . a systematic way to find the correct weight is to derive the eigenvectors of the operator and its dual , then to use the framework of relative entropy ( see @xcite for a general presentation ) . this was done by kirchgssner @xcite who derived the so - called _ eichform _ for . the linearized operator @xmath610 possesses obviously the nonpositive eigenvector @xmath611 , @xmath612 . the dual operator @xmath613 possesses the nonpositive eigenvector @xmath614 , @xmath615 , as can be checked by direct calculation . therefore the relative entropy identity for the convex function @xmath616 writes for the linearized system as follows , @xmath617 which is equivalent to after straightforward computation ( recall @xmath618 ) . a similar strategy could be performed here : the linearized operator @xmath619 possesses the nonpositive eigenvector @xmath620 , @xmath612 ( recall @xmath621 is nonincreasing ) . to derive the corresponding relative entropy identity , we should find an eigenvector @xmath622 in the nullset of the dual operator @xmath623 existence of such an eigenvector would follow from the krein - rutman theorem . however we were not able to find an explicit formulation of @xmath622 , and thus of the dissipation , which is necessary to derive a quantitative nonlinear stability estimate such as proposition [ nonlinstability ] . this is the reason why we stick to the weight proposed in definition [ eq : def weight phi ] although we believe it is not the optimal one . first , the comparison principle of proposition [ prop - mp ] and yield @xmath624 now , we write the nonlinear equation verified by the weighted perturbation @xmath603 , @xmath625 and as for the linear stability problem we test against @xmath547 : @xmath626 using we deduce @xmath627 this last equation is very similar to . following the same steps as in the proof of proposition [ linstability ] , we find that using the same weight @xmath97 and setting @xmath628 we obtain the estimate . in this section , we provide some numerical simulations to illustrate our results , and to introduce the last section about accelerating fronts . we used a very basic explicit numerical scheme for approximating . the transport part is approximated by a upwind scheme . we show in figure [ fig : vborne ] the expected asymptotic behavior when the velocity space is bounded , namely convergence towards a travelling front with minimal speed . next we have investigated the case of @xmath41 . of course , numerical simulations require that the support of @xmath8 is truncated . we opted for the following strategy : we truncated the velocity set to a finite interval @xmath629 $ ] and we renormalized the distribution @xmath8 . for any @xmath255 we observed the asymptotic regime of a travelling front with finite speed , as expected . however , as @xmath630 , we observed no stabilization of the asymptotic spreading speed . in fact , we observed that the envelope of the spreading speed scales approximately as @xmath631 . hence the front is accelerating with the approximate power law @xmath119 . we expect that the scaling strongly depends on the decay of @xmath8 as @xmath632 . [ fig : vnonborne ] we assume in this section that @xmath116 and that @xmath112 for all @xmath633 . we prove superlinear spreading . we deduce as a corollary that there can not exist a travelling wave solution of . we also give some quantitative features about the spreading of the density in two cases : the case where @xmath8 is a gaussian , and the case where it is a cauchy s law . we construct explicit supersolutions from which we estimate the spreading from above . we expect those estimates to be sharp . in fact they are in accordance with numerical simulations . as we are not able to construct suitable subsolutions , we leave it as an open problem to prove the exact spreading rate , at least for these two specific cases ( gaussian and cauchy s law ) . before we go to the proof , let us give some heuristics concerning the superlinear spreading rate . reaction - diffusion fronts with kpp nonlinearity are _ pulled fronts _ : the spreading rate is determined by the dynamics of small populations at the far edge of the front . in the kinetic model with unbounded velocities , individuals with arbitrary large speeds go at the far edge of the front . no matter their low density , they yield exponential growth of the population and pull the accelerating front . of course we expect the acceleration to depend on the specific tail of the distribution @xmath30 . actually we conjecture that the spreading rates for a gaussian and a cauchy s law scales respectively as @xmath634 , and as @xmath635 . last , let us emphasize that the diffusive limit of leads to the classical fisher - kpp equation , under the assumption that @xmath8 has some finite moments which is obviously the case for a gaussian . the fisher - kpp equation exhibits linear spreading whereas the kinetic equation may exhibit superlinear spreading . the point is that small populations does not show the same scaling at the far edge of the front . to summarize we shall say that asymptotics of large deviations are very much different in the two cases : reaction - diffusion as opposed to kinetic transport - reaction . let @xmath636 so that @xmath637 . for all @xmath638 we define the renormalized truncated kernel and the associated growth rate , @xmath639}(v)}{\int_{-a}^a m } m(v ) \quad \hbox { and } \quad r_a = ( 1+r)\int_{-a}^a m(v)dv -1 \in ( 0 , r)\ , .\ ] ] as @xmath640 is compactly supported , we can apply the results proved when @xmath7 is bounded in order to construct appropriate subsolutions . before we proceed with subsolutions we investigate the dispersion relation in the limit @xmath630 . define for all @xmath641 and @xmath642 : @xmath643 and @xmath644 the corresponding minimal speed defined in lemma [ minlambda ] . [ infspeed ] one has @xmath645 . for all @xmath646 , let @xmath647 such that @xmath648 if @xmath644 does not diverge to @xmath153 as @xmath630 , then it is bounded along a sequence @xmath649 and one has @xmath650 simply by comparison @xmath651 . applying fatou s lemma to , one gets @xmath652 a contradiction . let @xmath653 the solution of @xmath654,\smallskip\\ g_a ( 0,x , v ) = g^0 ( x , v ) \quad & \hbox{in } { \mathbb{r}}\times [ -a , a ] , \end{array}\right.\ ] ] and @xmath655 . clearly , @xmath656 for all @xmath5 . hence , multiplying by @xmath657 , we get @xmath658 extending @xmath659 by @xmath64 outside of @xmath660 $ ] , as @xmath661 , we get that @xmath659 is a subsolution of and it follows from the maximum principle stated in proposition [ prop - mp ] that @xmath662 for all @xmath663 . on the other hand , we know from proposition [ prop : spreadingbounded ] that for all @xmath664 : @xmath665 hence , as @xmath666 and @xmath667 for all @xmath668 $ ] , one gets for all @xmath668 $ ] : @xmath669 . therefore we conclude @xmath670 the conclusion follows from the fact that @xmath671 . we construct below supersolutions for when @xmath41 and the distribution @xmath8 is a gaussian . [ supersolunbounded ] let @xmath116 and @xmath120 . for @xmath672 define @xmath673 let @xmath82 be defined by @xmath674 then @xmath675 is a supersolution of , that is : @xmath676 we shall prove that @xmath82 is a supersolution of . indeed , it will follow that @xmath677 is a supersolution since it is the minimum of two supersolutions . from the duhamel formula , we deduce that @xmath678 to prove that @xmath82 is a subsolution we must prove in fact that @xmath679 this is sufficient to prove that the inequality @xmath680 holds true . computing the expression of @xmath681 we obtain @xmath682 we first deal with the estimate of @xmath683 . we claim the following inequality holds true : for all @xmath684 and @xmath685 $ ] , @xmath686 in fact one has @xmath687^{\frac12 } } \exp \left ( - \frac{1}{2 \sigma^2 } \frac{x^2}{(s+a)^{2 } + ( t - s)^2 } \right)\\ & \leq \displaystyle \frac{1}{\sqrt{2\pi } \sigma}\exp \left ( - \frac{1}{2 \sigma^2 } \frac{x^2}{(t+a)^{2 } } \right),\end{aligned}\ ] ] since @xmath688 , \quad ( t+a)^2 \geq ( s+a)^2 + ( t - s)^2 \geq ( s+a)^2.\ ] ] this yields @xmath689 to estimate @xmath267 , we plug in the formula for @xmath690 , @xmath691 we compute the last integral using the formula for the initial condition @xmath117 , @xmath692 thus , for all @xmath693 : @xmath694 \right ) \exp \left ( -(1+r ) t \right ) \leq \exp \left ( -(1+r ) t \right)\ ] ] as long as @xmath695 and @xmath696 , that is @xmath697 . this concludes the proof . let @xmath330 . for all @xmath698 , we define the zone @xmath699 . from the definition of @xmath677 , we deduce that @xmath677 is a supersolution such that @xmath700 . however , for all @xmath147 and @xmath701 , we have @xmath702 it yields that @xmath703 computations are made easier above since the gaussian distribution is stable by convolution . this is also the case for the cauchy distribution . therefore we are able to derive an inequality similar to . let us comment that specific case before we give the spreading estimate . because the distribution @xmath8 has an infinite variance , we learn from @xcite that the correct macroscopic limit leads to a nonlocal fractional laplacian operator . on the other hand , we expect from @xcite an exponentially fast propagation in the diffusion regime . similarly as for our previous results , we can reasonably expect that the spreading rate is faster in the kinetic model than in the macroscopic limit . therefore we expect a spreading rate faster than exponential . in fact the supersolution that we are able to derive confirms this expectation . [ supersolunbounded - cauchy ] let @xmath116 and @xmath704 . for @xmath705 and @xmath706 $ ] , define @xmath707 let @xmath82 be defined by @xmath674 then @xmath675 is a supersolution of , that is : @xmath676 the proof is the same as for proposition [ supersolunbounded ] . we just show the main computations in the case of the cauchy distribution . to prove we use the residue method as follows , @xmath708 the analog computation for proving goes as follows . first we have @xmath709 thanks to the expression of the initial condition , we compute the latest integral : @xmath710 thus , @xmath711 which holds true if @xmath712 and @xmath713 . we give in this section the proofs of propositions [ cauchypbm ] and [ prop - mp ] . well - posedness relies on a fixed point argument which is also used for the comparison principle . we first state two lemmas . [ lem : eulin ] let @xmath714 and @xmath715 . then there exists a unique function @xmath716 such that @xmath717 in the sense of distributions . this solution also satisfy the duhamel formula : @xmath718 moreover , if @xmath719 and @xmath720 , then @xmath721 in @xmath722 . for @xmath723 we define the operator @xmath724 where @xmath725 take @xmath726 and define @xmath727 and @xmath728 . assume that @xmath729 over @xmath730 . for all @xmath731 , one has : @xmath732 hence , there exists @xmath733 such that for all @xmath734 , @xmath735 is a contraction over @xmath736 . if @xmath737 on @xmath730 , then such an estimate can be derived similarly . hence , @xmath735 admits a unique fixed point , which satisfies over @xmath730 . this gives the local existence and uniqueness of the solution of . moreover , as @xmath738 does not depend on the initial datum @xmath117 , the global existence follows . [ lem : posa_t ] assume that @xmath739 is everywhere positive and that @xmath7 is an interval . then if @xmath740 is nonnegative and if there exists @xmath741 such that @xmath742 , letting @xmath82 the unique solution of , one has @xmath743 for all @xmath141 such that @xmath455 . first , assume by contradiction that there exists @xmath693 such that @xmath744 , with @xmath745 . then integrating over @xmath7 , one gets @xmath746 hence , @xmath747 for all @xmath5 and @xmath748 . letting @xmath749 , one gets @xmath750 for all @xmath5 . as @xmath455 and @xmath7 is an interval , one can take @xmath5 such that @xmath751 , leading to @xmath752 . this is a contradiction since , as @xmath82 is continuous , nonegative and @xmath753 , one has @xmath754 . hence @xmath755 for all @xmath756 such that @xmath745 . define @xmath760 . as in the proof of lemma 6 in @xcite , we first remark that this function satisfies @xmath761 with @xmath762 for all @xmath285 . we define @xmath763 and @xmath764 . writing the integral formulation as in the proof of lemma [ lem : eulin ] gives @xmath765 and thus @xmath766 in @xmath730 for some operator @xmath735 which is monotone and contractive when @xmath767 is small enough . it follows that @xmath768 for all @xmath769 . since @xmath735 is contractive the sequence @xmath770 converges to @xmath64 . we conclude that @xmath771 , meaning that @xmath772 . next , assume that @xmath494 , @xmath7 is an interval , and that there exists @xmath138 such that @xmath139 . we can follow the proof of lemma [ lem : posa_t ] , where @xmath739 defined above is positive everywhere . we deduce that @xmath773 as soon as @xmath142 . aronson , h.f . weinberger , _ nonlinear diffusion in population genetics , combustion , and nerve pulse propagation_. partial differential equations and related topics . lecture notes in math . 446 , springer , berlin , 1975 . e. bouin , v. calvez , n. meunier , s. mirrahimi , b. perthame , g. raoul , and r. voituriez . _ invasion fronts with variable motility : phenotype selection , spatial sorting and wave acceleration_. c. r. math . paris , 350(15 - 16):761766 , 2012 . dunbar , h.g . othmer , _ on a nonlinear hyperbolic equation describing transmission lines , cell movement , and branching random walks_. nonlinear oscillations in biology and chemistry . lecture notes in biomath . 66 , springer , berlin , 1986 . kolmogorov , i.g . petrovsky , n.s . piskunov , _ etude de lquation de la diffusion avec croissance de la quantit de matire et son application un problme biologique _ , moskow univ . math . bull . 1 * ( 1937 ) , 125 .	in this paper , we study the existence and stability of travelling wave solutions of a kinetic reaction - transport equation . the model describes particles moving according to a velocity - jump process , and proliferating thanks to a reaction term of monostable type . the boundedness of the velocity set appears to be a necessary and sufficient condition for the existence of positive travelling waves . the minimal speed of propagation of waves is obtained from an explicit dispersion relation . we construct the waves using a technique of sub- and super- solutions and prove their stability in a weighted @xmath0 space . in case of an unbounded velocity set , we prove a superlinear spreading and give partial results concerning the rate of spreading associated to particular initial data . it appears that the rate of spreading depends strongly on the decay at infinity of the stationary maxwellian . * key - words : * kinetic equations , travelling waves , dispersion relation , superlinear spreading .
revealing the nature of dark matter has received much more attention since the successive results of the wmap cosmic microwave background all - sky observation@xcite . many dark matter direct search experiments have been concluded@xcite , are currently being performed@xcite , or are in the planning stage@xcite . weakly interacting massive particles ( wimps ) are very plausible candidates of dark matter . most dark matter search experiments are designed to measure only the energy transferred to the nucleus through wimp - nucleus scatterings . in these experiments , the distinct signals of wimps are the annual modulation of the energy spectrum . the halo is isotropic . because the amplitude of an annual modulation signal is only a few @xmath1 of the rate , positive signatures of wimps are very difficult to detect with these detectors@xcite . on the other hand , the motion of the solar system with respect to the galactic halo is considered to provide a much larger asymmetry in the direction - distribution of the wimp velocity observed at earth@xcite . a detection method of a positive signature of wimps by measuring the directions of the nuclear recoils have been studied@xcite . if the distribution of the local halo were not isotropic perhaps quite clumpy direction - sensitive methods would then potentially provide information on it . gaseous detectors are one of the most appropriate devices to detect this wimp - wind@xcite . the drift project has performed underground measurements for more than two years with 1 m@xmath2 time projection chambers ( tpc ) filled with a low - pressure @xmath3 gas aiming to detect the mainly wimp - wind via spin - independent(si ) interactions@xcite . since former studies suggest to perform the wimp - search both via si and spin - dependent(sd ) interactions@xcite , we proposed a sd - sensitive wimp - wind search , named newage ( new generation wimp - search with an advanced gaseous tracking device experiment)@xcite . we developed a @xmath4-tpc with a detection volume of 23 @xmath5 28 @xmath5 30 @xmath6 and performed a direction - sensitive wimp - search experiment in a surface laboratory . in this section , we describe the detector performance , which is also described in ref . we mainly measured the detector response in the energy range that we used for the wimp - search analysis ( 100 - 400kev , dm energy range ) . the lower threshold ( 100kev ) was set at the lowest energy for an effective gamma - ray rejection with a 150 torr @xmath0 gas , and the higher threshold ( 400kev ) was set at the expected highest recoil energy by the wimps at the escape velocity . a @xmath4-tpc is a gaseous time projection chamber with a micro pixel chamber ( @xmath4-pic ) readout . a @xmath4-pic is a gaseous two - dimensional position - sensitive detector manufactured by printed circuit board ( pcb ) technology . pcb technology realizes an economical mass production , which is one of the most important requirements to fabricate a dark matter detector . after a general - purposed detector r&d , we studied the performance of a @xmath4-tpc with a detection volume of @xmath7 using a 150 torr @xmath0 gas as our first step towards a dark matter experiment@xcite . in this study , we demonstrated the tracking of nuclear recoils and the gamma - ray rejection with a @xmath0 gas of 150 torr . we also found that we need some more studies to operate the @xmath4-tpc with a @xmath0 gas at a design value of 30 torr for more than a few days . we then developed a large - sized ( 31 @xmath5 31 @xmath8 ) @xmath4-pics@xcite , and subsequently developed a large - volume @xmath4-tpc with a detection volume of 23 @xmath5 28 @xmath5 30 @xmath6 . we studied the fundamental properties of this large - volume @xmath4-tpc with a standard ar-@xmath9 gas mixture at normal pressure@xcite . the size is of the @xmath4-pic is restricted by the working size of the pcb , and we plan to realize a larger readout area by placing many pieces , for instance 4 @xmath10 4 pieces for a 1 m@xmath11 , of @xmath4-pics . before we take the next technology steps , we performed a first direction - sensitive spin - dependent dark matter search experiment to demonstrate that a gaseous detector and a direction - sensitive method can actually set limits on the wimps . we report on the results of a direction - sensitive dark matter search experiment using this large - volume @xmath4-tpc with a 150 torr @xmath0 gas ( 9.0 g effective mass ) in a surface laboratory . we used the @xmath4-tpc described in ref . @xcite , and thus most parts of the detector were unchanged . herein , we describe only the significant properties and the modified parts of the detector system . figure [ fig : tpc ] is a schematic drawing of the @xmath4-tpc . a @xmath4-pic with a detection area of 31 @xmath5 31 @xmath8 ( toshiba / dnp , sn060222 - 3 ) was used as the main gas - multiplier and as readout electrodes . a gas electron multiplier ( gem@xcite ) with an effective area of 23 @xmath5 28 @xmath8 , which was manufactured by a japanese company ( scienergy co. ltd . ) , was used as a pre - amplifier . the drift length was 31 cm and the detection volume was 23 @xmath5 28 @xmath5 31 @xmath6 . the electric field of the drift volume was formed with a field - shaping pattern on a fluorocarbon circuit board surrounding the detection volume . the detector was placed in a 3 mm - thick stainless - steel vacuum chamber filled with 150 torr @xmath0 . the data - acquisition system , which is described in ref . @xcite , was triggered by the signals of the @xmath4-tpc itself . hence , the absolute z position was not measured , but only the relative positions . typical operation parameters were @xmath12 , @xmath13 , @xmath14 , and @xmath15=8.25kv . @xmath16 is the voltage supplied to the @xmath4-pic , @xmath17 is the voltage supplied to the top of the gem , @xmath18 is the voltage supplied to the bottom of the gem , and @xmath19 is the voltage supplied to the drift plane . these parameters were optimized to realize stable operation at a combined ( @xmath4-pic @xmath5 gem ) gas gain of 2300 , which is high enough to maintain the detection efficiency of the recoil nucleus . we still suffer from discharge problems when operating a @xmath4-pic alone . we calibrated and monitored the energy with alpha particles generated by the @xmath20 ( q=2.7mev ) reaction @xcite . we set a glass plate coated with a thin 0.6 @xmath4 m @xmath21 layer in the @xmath4-tpc , and irradiated the @xmath4-tpc with thermalized neutrons . alpha particles ( 5.6 mev , 6.1 mev , and 7.2 mev ) from the decays of the radon progeny were also used . the track - length and the energy - deposition correlation down to the dm energy range agreed with a calculation by srim@xcite . the measured energy resolution in 58mev was 50@xmath1(fwhm ) , which was dominated by the gain inhomogeneity of the @xmath4-pic . this would be the worst value when the gain inhomogeneity of the @xmath4-pic dominates the energy resolution . the energy resolution could be dominated by the the statistics of the primary electrons . in this case , the energy resolution at 100 kev with a cf@xmath22 gas was estimated from a measurement with an ar-@xmath23 gas mixture , because it is not very easily measured due to a small photo - absorption cross - section and a long range of the electron . the energy resolution at 60 kev , measured with an ar-@xmath23 gas mixture , was 60@xmath1 ( fwhm)@xcite . we caliculataed with @xmath24 . here , @xmath25 and @xmath26 are the energy resolution at 100 kev with a @xmath0 gas and that at 60 kev with an ar-@xmath23 gas mixture , respectively . @xmath27=54ev and @xmath28=26ev are the energies needed to created an ion - electron pair in a @xmath0 gas and an ar-@xmath23 gas mixture , respectively . thus , the estimated energy resolution was 70@xmath1 ( fwhm ) at 100 kev with a cf@xmath22 gas , if the statistics of the primary electron dominates the energy resolution . the energy resolution in the dm energy range was therefore estimated to be better than 70@xmath1 . the nuclear - recoil detection efficiency was measured by irradiating the @xmath4-tpc with fast neutrons from a @xmath29cf source of 0.72 mbq placed at several postitions . we selected the tracks within the fiducial volume ( 21.5 @xmath5 22 @xmath5 31 @xmath6 ) . we also required at least three digital hit points for a track . a typlical track - length and energy - deposition correlation is shown in the left panel of figure [ fig : scat252cf ] . a nuclear recoil band is seen along the x - axis . we set the upper limit of the `` nuclear recoil band '' at 1 cm . we compared the spectrum after the selection with a simulated one , and the calculated ratio was the overall ( detection and the selection ) efficiency . the efficiency curve is shown in the right panel of figure [ fig : scat252cf ] . the efficiency was about 40@xmath1 at 100 kev , and the efficiency dependence on the recoil energy was phenomenologicaly fitted with an error function , @xmath30 . the best - fit function was @xmath31 , where @xmath32 is the error function and @xmath33 is the energy in kev . we demonstrated that the track - length and the energy - deposition correlation measured with a gaseous tpc provided a strong gamma - ray rejection in our previous work@xcite . we performed quantitive measurements with a large - volume @xmath4-tpc for a wimp - search measurement in a surface laboratory . we irradiated the @xmath4-tpc with gamma - rays from a @xmath34 source of 0.80 mbq , placed at 25 cm from the center of the detection volume . the measured track - length and energy - deposition correlations with and without the gamma - ray source are shown in the left and the right panels of figure [ fig : scat137cs ] , respectively . the live time was 0.44 days for both measurements . a band along the y - axis labeled `` electron recoil band '' is seen only in the results with a gamma - ray source , which is obviously due to tracks of the recoil electrons . another band along the x - axis , seen in both results , is due to the tracks of the carbon and fluorine nuclei by the neutron background , and are labbeled `` nuclear recoil band '' . the spectra of these two measurements are shown in the left panel of figure [ fig : spec137cs ] . the spectrum of the no - source run was subtracted from the that of the @xmath34 run . the obtained " subtracted spectrum is shown in the right panel of fig [ fig : spec137cs ] . the subtracted spectrum is consistent with zero in the dm energy range , within the statistical error . we then derived the gamma - ray rejection factor by dividing the subtracted spectrum by a simulated electron recoil specrum . the measured gamma - ray rejection factor in the dm energy ranege was @xmath35 . the error is statistical , which is dominated by the background in the surface laboratory . the 90@xmath1 c.l . upper limit of the gamma - ray rejection factor was @xmath36 . the @xmath4-pic has an anisotropic response with regard to the three - dimensional track directions . this non - uniform response is due to a difference in the detection method of the z - axis ( time information ) and the other two axes ( 2-dimensional image taken by the @xmath4-pic ) . we measured this detection efficiency dependence on the track direction , or a response map , by irradiating the @xmath4-tpc with fast neutrons from several positions that generated uniform recoils . we fitted the digital hit points with a straight line , and determined the directions of each nuclear recoil track . because we did not measure the track senses in this measurement , we fixed the sign of the x component to positive and determined the ( x , y , z ) vector of the track in the coordinates shown in fig [ fig : tpc ] . the measured response map is shown in fig [ fig : responsemap ] . the map was normalized by a mean value . it can be seen that the correction factor is quite large because the fabrication technology of the large - sized @xmath4-pic is still in an r@xmath37d phase , and the gain inhomogeniety is still large . the detector showed a relatively high efficiency for tracks in the x - z and y - z planes , but showed a relatively low efficiency for tracks in the x - y plane . we have not overcome intrinsic weak points of the tpc that the efficiency to the tracks parallel to the readout plane is relatively low . it should be noted that the absolute detection efficiency was measured by a method described in the pervious sections , and this response map was used to correct the relative efficiency . we measured the position resolution using the tracks of the alpha particles measured in the energy calibration . we evaluated the position resolution using the same method as described in our previous work @xcite . the measured position resolution was 0.8 mm ( rms ) . the angular resolution of the carbon and fluorine nuclei in the dm energy range was estimated in the following manner . we calculated the sampling pitch of the carbon and fluorine tracks from real data . we then generated the hit points by a simulation based on the measured position resolution and the sampling pitch . simulated hit points were fitted with a straight line , and the angle between the fitted line and the real direction was calculated . the distribution of the calculated angle was fitted with a lorentzian and @xmath38 , where @xmath39 is a parameter of a lorentzian that corresponds to hwhm . we used this value for a direction - sensitive analysis . a dark matter search measurement was performed in a surface laboratory at kyoto university ( @xmath40 @xmath41 n , @xmath42 @xmath43 e , third floor in a five - story building ) . the measurement was performed from november 1st , 2006 to november 27th , 2006 . we set the @xmath4-pic plane horizontally and re - aligned the x- and y - axis of the @xmath4-tpc twice ( three different @xmath44 indicated in figure [ fig : tpc ] ) so as to cancel any remaining static errors . table [ table : tpcoperation ] lists the measurement properties . the total live time was 16.71 days and the exposure was 0.151 kg@xmath45 days . the sum of the calibration and maintenance time was 7.51 days , and the dead time due to the data acquisition was 1.0 days . the @xmath4-tpc was filled with 150 torr @xmath0 , and the vacuum chamber was sealed without any gas circulation .. the drift velocity , total gas gain , and the energy resolution were monitored every two or three days during the measurement . the drift velocity was 7.6 cm/@xmath4s at the beginning of the measurement and decreased to 6.9 cm/@xmath4s at the end . this decrease is thought to be due to the out - gas from the detector components . we used the measured velocities for the analysis . the mean gas gain in the entire detection area was 2300 , and a time - dependent variation was not observed within a measurement error of 4@xmath46 ( rms ) . the energy resolution was also stable within a measurement error of 7@xmath46 ( rms ) . the time - dependence was small compared to the energy resolution , itself , and the angular resolution in this measurement . .properties of the dark matter measurements . [ cols="<,<,<",options="header " , ] we next performed a direction - sensitive analysis . we created a sky map of the north sky with the direction of the three - dimensional nuclear recoil tracks . because we did not detect the sense of each track , the map was restricted to half of the sky larger and the southern part of the sky was folded . we plotted every nuclear event ( 1686 events in a 0.151 kg@xmath45days exposure ) in figure [ fig : skymap ] . the larger mark , labled " dm expected , is the direction of the solar motion or the cygnus from which the wimp - wind is expected . the wimp - wind direction has a diurnal motion around the polestar . we calculated @xmath47 , the angle between the wimp - wind direction and the recoil direction for each event , and drew the the @xmath48 distribution shown in figure [ fig : costheta ] . we compared this measured @xmath48 distribution with the expected wimp - signal . the expected wimp - signal was prepared by the following method . we first created a @xmath48 distribution for a given wimp mass and a recoil energy while considering the energy resolution . the dm energy range was divided into 15 bins , and the energy and angular resolution of the detector were taken into account . we used @xmath49 ( fwhm ) for the energy resolution and an angular resolution of @xmath38 . we followed ref . @xcite for the angular distribution and ref.@xcite for the count rate . we then created a sky - map expected by the wimp events with this @xmath48 distribution while considering the derection - dependence(figure [ fig : responsemap ] ) and the angular resolution . we used this sky map to create the expected @xmath48 distribution for a given wimp mass and energy bin . figure [ fig : costheta ] shows one of these @xmath48 distributions . here , the data was fitted with the distribution for @xmath50=100gev and 100120 kev bin , and the cross section that gave the minimum @xmath51 was 1.36 @xmath5 @xmath52pb and @xmath53 . this best - fitted wimp signal was rejected at the 90@xmath1 confidence level by @xmath51 tests . the cross section that would give the minimum @xmath51 was calculated for each energy bin , and the smallest cross section was taken as the limit for the given wimp mass . the upper and lower panels in figure [ fig : limit ] show the cross section limits and the corresponding @xmath51 values as a function of the wimp mass , respectively . the limits are shown with a thick - solid line ( labeled direction - sensitive ) and the @xmath51 values are shown by filled - square marks . the @xmath51 values at the 90@xmath1 c.l . upper limit are shown by a dotted line in the lower panel in figure [ fig : limit ] . the wimp signals were rejected at the 90@xmath1 confidence level because the best - fit @xmath51 values were all above the 90@xmath1-@xmath51 line . the data was also fitted with a flat @xmath48 distribution and @xmath54 was obtained independent of the wimp mass . the flat @xmath48 was not rejected at the 90@xmath1 confidence level . it was demonstrated that the direction - sensitive method gave a limit slightly better than the conventional spectrum - only method . the difference between this result and other experiments is explained by the neutron background in the surface laboratory . this is the first limit set by a dark matter search experiment with a gaseous tracking device . as for the spin - dependent limit , this is the first result by a direction - sensitive experiment . it should be stressed that the dark matter search experiment with a gaseous tracking device can actually set a limit on the wimps . these results are very convincing and encouraging towards developing a future wimp - wind detection with large - volume detectors . we have shown that wimp - proton cross section limits can be obtained by a direction - sensitive method . however , much work is necessary to reach the sensitivity achieved by other experiments performed with solid and liquid detectors . the sensistivity is restricted by the neutron background in a surface run , therefore , we are going to perform a dark matter search experiment in an underground laboratory . the fast neutron flux in kamioka observatory ( ( 1.9 @xmath55 0.21 ) @xmath56 ) is more than 3 orders of magnitude smaller than a typical fast neutron flux in a surface laboratory . it it expected that the background sources , such as radioactive isotopes in the detector components , would limit the sensitivity in the first stage of an underground run . we are going to replace the detector components with low - activity materials . the angular resolution is not good enough to take full advantage of the direction - sensitive method over the `` spestrum only '' method . we are going to investigate the detector response using a neutron source of a known energy . we plan to study the operation of the @xmath4-tpc at a lower pressure of 30 torr to lower the energy threshold . we are going to develop a larger readout area by placing many pieces of the of @xmath4-pics of 31 @xmath5 31 @xmath8 . after these technological breakthroughs are achieved , we expect to reach the susy predictions with an exposure of more than one hundred kg@xmath45days , which we have calculated in our previous work@xcite . we developed a three - dimensional tracking device , and performed a direction - sensitive dark matter search measurement in a surface laboratory . by using 150 torr carbon - tetrafluoride ( @xmath0 ) gas , we set the first limit on the spin - dependent wimp - proton cross section by a direction - sensitive method . thus , we have demonstrated that a dark matter search experiment with a gaseous tracking device can actually set a limit on wimps . furthermore , we have demonstrated that this method will potentially play a certain role in revealing the nature of dark matter when a low - background large - volume detector is developed . this work was supported by a grant - in - aid in scientific research of the ministry of education , culture , sports , science , technology of japan ; research - aid program of toray science foundation ; and grant - in - aid for the 21st century coe center for diversity and universality in physics . d. n. spergel _ et al . , _ astrophys . * 148 * ( 2003 ) 175 ; d. n. spergel _ et al . , _ astro - ph/0603449 . p. belli , r. bernabei , a. incicchitti , d. prosperi , nuovo cimento c * 15 * ( 1992 ) 475 ; r. bernabei , p. belli , f. nozzoli , a. incicchitti , eur . j. c 28 ( 2003)203 . r. bernabei _ et al . b * 436 * ( 1998 ) 379 ; r. bernabei _ et al . b * 424 * ( 1998 ) 195 ; * 450 * ( 1999 ) 448 ; * 480 * ( 2000 ) 23 ; astro - ph/0307403 . b. ahmed , _ et al . , _ astropart . . h. sekiya _ et al . , _ _ proceedings of the fourth international workshop on `` the identification of dark matter '' , edinburgh , uk , 2004 _ , edited by n. j.c . spooner and v. kudryavtsev , ( world scientific , 2005 ) , p. 378 . y. shimizu , _ et al . b * 633 * ( 2006 ) 195 g. angloher , _ et al . , _ astropart * 23 * ( 2005 ) 325 . et al . , _ astropart . phys . * 23 * ( 2005 ) 444 . girard , _ et al . b * 621 * ( 2005 ) 233 m. barnab@xmath57-heider , _ et al . b * 624 * ( 2005 ) 186 cdms collaboration , phys . d * 72 * 052009(2005 ) ; edelweiss collaboration , phys . d * 71 * ( 2005 ) 122002 . et al . , _ nucl . methods a * 555 * ( 2005 ) 173 . d. n. spergel , phys . d * 37 * ( 1988 ) 1353 . a. m. green and b. morgan , astropart . * 27 * ( 2007 ) 142 . j. rich and m. spiro , saclay preprint dphpe 88 - 04(1988 ) ; g. gerbier , j. rich , m. spiro , and c. tao _ proceedings of the workshop on `` particle astrophysics '' _ , edited by e. b. norman , ( world scientific , 1989 ) , p. 43 . g. masek , k. buckland , and m. mojaver , _ proceedings of the workshop on `` particle astrophysics '' _ , edited by e. b. norman , ( world scientific , 1989 ) , p. 41 . j. i. collor , _ et al . j. of phys . * 2 * , ( 2000 ) , 14.1 . ta girard and f. giuliani , phy . d * 75 * , ( 2007 ) 043512 . t. tanimori _ et al . b * 578 * ( 2004 ) 241 . k. miuchi _ et al . , _ _ proceedings of the fifth international workshop on `` the identification of dark matter '' , rhodes , greece , 2006 _ , physics/0701118 h. sekiya _ et . al . , _ proceedings of the conference of the 7th international conference on position sensitive detectors , liverpool , england . to appear in nucl . methods , doi:10.1016/j.nima.2006.10.322 , astro - ph/0601568 . a. takada _ , _ proceedings of the conference of the 7th international conference on position sensitive detectors , liverpool , england . to appear in nucl . methods , doi:10.1016/j.nima.2006.10.283 . k. miuchi _ et al . , _ _ proceedings of the 8th international workshop on radiation imaging detectors ( iworid8 ) _ pisa , italy , 2 - 6 july 2006 . to appear in nucl . methods a ( 2007),doi:10.1016/j.nima.2007.01.117 , , physics/0701085 . f. sauli and a. sharma , ann . nucl . part . 49 ( 1999 ) 341 . h. kubo _ et . al . , _ ieee nuclear science symposium conference record n14 - 30 ( 2005 ) . j. f. ziegler , j. p. biersack and u. littmark , srim - the stopping and range of ions in matter , pergamon press , new york , 1985 . . two bands along the y- and x - axises are seen in the @xmath34 run , while the electron recoil band is not seen in the no - source run . several events out of both bands are thought to be the tracks of protons from the wall of the drift volume.,title="fig : " ] . two bands along the y- and x - axises are seen in the @xmath34 run , while the electron recoil band is not seen in the no - source run . several events out of both bands are thought to be the tracks of protons from the wall of the drift volume.,title="fig : " ] c.l . upper limits on the wimp - proton spin - dependent cross section ( upper ) and @xmath51 values ( lower ) as functions of the wimp mass . the thick solid and dotted lines show the limits obtained with and without the direction information , respectively . limits from other experiments ( dama(xe)@xcite , dama(nai)@xcite , naiad@xcite , tokyo @xmath59@xcite , simple 2005@xcite , picasso@xcite , and cdms@xcite ) are shown for comparison . the filled squares show the @xmath51 minimum values of the best - fit wimp distribution , the filled - circles show the best - fit flat @xmath48 distribution , and the dotted line show the @xmath51 values at the 90@xmath1 c.l . upper limit . ]	we developed a three - dimensional gaseous tracking device and performed a direction - sensitive dark matter search in a surface laboratory . by using 150 torr carbon - tetrafluoride ( @xmath0 ) gas , we obtained a sky map drawn with the recoil directions of the carbon and fluorine nuclei , and set the first limit on the spin - dependent wimp ( weakly interacting massive particles)-proton cross section by a direction - sensitive method . thus , we showed that a wimp - search experiment with a gaseous tracking device can actually set limits . furthermore , we demonstrated that this method will potentially play a certain role in revealing the nature of dark matter when a low - background large - volume detector is developed . , , , , , , , , , , , , time projection chamber , micro pattern detector , dark matter , wimp , direction - sensitive , 14.80.ly 29.40.cs , 29.40.gx , 95.35.+d
the radiative capture reaction @xmath0 ( @xmath3 mev ) takes place during stellar core helium burning @xcite , where @xmath4c is produced by the triple - alpha process . the capture cross section @xmath5 at the relevant gamow energy , @xmath6 mev for @xmath7 k , determines together with the convection mechanism at the edge of the stellar core @xcite the abundances of carbon and oxygen at the end of helium burning . this , in turn , strongly influences the nucleosynthesis of elements up to the iron region for massive stars @xcite and the composition of co white dwarfs , whose progenitors are intermediate and low mass stars @xcite . a recent experiment confirmed that the reaction rate of the triple - alpha process is known with a precision of about 10% for temperatures near 10@xmath8 k @xcite . a similar precision is needed for the rate of @xmath4c(@xmath9)@xmath1 to provide an adequate input for stellar models @xcite . the remarkable experimental efforts over the last decades @xcite focused on the observation of the capture @xmath10-rays , including one experiment @xcite that used the coincident detection of the @xmath1 recoils . due to the low cross section and various background sources depending on the exact nature of the experiments , @xmath10-ray data with useful but still inadequate precision were limited to center - of - mass energies @xmath11 to 3.2 mev . at the low - energy range , the data were limited e.g. by cosmic - ray background and at the high - energy range the data were limited by intense background reactions such as @xmath12c(@xmath13,n)@xmath14o ( for an @xmath13-beam experiment ) and @xmath4c(@xmath4c , n)@xmath15 mg ( for a @xmath4c - beam experiment ) . the cross section @xmath5 is expected to be dominated by p - wave ( e1 ) and d - wave ( e2 ) capture to the ( @xmath16 ) @xmath1 ground state . two bound states , at 6.92 mev ( @xmath17 ) and 7.12 mev ( @xmath18 ) , which correspond to subthreshold resonances at @xmath19 and @xmath20 kev , appear to provide the bulk of the capture strength @xmath5 through their finite widths that extend into the continuum . r - matrix analyses are performed , in order to model the energy dependence of the cross section . in these analyses the contribution of each amplitude to the total cross section is expressed in terms of a small number of resonances and a direct capture contribution . the parameters of the model are determined by a fit to the experimental data . the extrapolation to @xmath21 mev is sensitive to the properties of the nearby levels , but it is sensitive also to the properties of the high lying resonances , since their tails extend to low energy . the effect of these resonances is usually included by a single high energy `` background '' resonance one for each amplitude ; these resonances are needed to obtain a good fit to the data . analyses of the available capture data together with data from the @xmath22c elastic scattering and the @xmath23-delayed @xmath13-decay of @xmath14n still lead to large uncertainties in the extrapolation to e@xmath24 @xcite . this is partly caused by large errors , both statistical and systematic , affecting the low energy capture data , and partly due to the weak experimental constrains of the background resonances . clearly , new measurements at low energies are needed . however , of equal importance are also new measurements at significantly higher energies , well above the range of the recent experiments , which may improve the experimental characterization of the background resonances and may thus reduce the uncertainty in the extrapolated astrophysical s factor s(e@xmath24 ) . there may exist even an s - wave capture ( monopole e0 ) and significant capture amplitudes to excited @xmath1 states . since the various capture amplitudes have different energy dependences , a better knowledge of the individual amplitudes requires new data at both low and high energies to provide an improved basis for their extrapolation to e@xmath24 . schematic diagram of the recoil separator erna . wf = wien filter , qs = quadrupole singlet , qd = quadrupole doublet , qt = quadrupole triplet , st = steerer , fc = faraday cup , ss = slit system , ap = aperture . at the end of the separator , there is a @xmath25e - e telescope for particle identification.,width=302 ] a new experimental approach ( fig . [ setup ] ) has been developed at the 4 mv dynamitron tandem accelerator in bochum , called erna = european recoil separator for nuclear astrophysics @xcite . in this approach , the reaction is initiated in inverse kinematics , @xmath26he(@xmath4c,@xmath10)@xmath1 , i.e. a @xmath4c ion beam with a particle current of up to 10 @xmath27a is guided into a windowless @xmath26he gas target . a beam contamination in the order of @xmath28 @xmath14o ions per incident @xmath4c projectile was found , by far too high for a direct detection of the @xmath14o recoils . therefore , for the purpose of beam purification , there is one wien filter before the analysing magnet ( wf1 , not shown in fig . [ setup ] ) and one before the gas target ( wf2 ) ; with this combination a beam purity better than 10@xmath29 could be achieved @xcite . the windowless gas target - entrance and exit aperture of the gas cell with a diameter of 6 mm - includes an ar post - target stripping system . after the gas target , the separator consists sequentially of the following elements : a quadrupole triplet ( qt ) , a wien filter ( wf3 ) , a quadrupole singlet ( qs ) , a 60@xmath30 dipole magnet , a quadrupole doublet ( qd4 ) , and a wien filter ( wf4 ) . the recoil separator suppresses the intense @xmath4c beam ; the @xmath1 recoils in a selected charge state are then counted in a @xmath25e - e telescope placed at the end of the beam line . the ratio between leaky @xmath4c events detected in the telescope and the number of @xmath4c projectiles is in the range of @xmath31 at the high energy limit and @xmath32 at the lower limit . additionally , a beam suppression factor of the telescope alone of 10@xmath33 - ratio of @xmath14o and leaky @xmath4c in the @xmath25e - e matrix ( fig . [ matrix ] ) - can be achieved leading to a total suppression factor of better than @xmath34 . erna is designed to study the reaction over the energy range @xmath35 to 5.0 mev . the detection of the @xmath1 recoils allows , for the first time , a direct measurement of the total cross section of @xmath36 , including possible non radiative transitions . the number n@xmath37 of recoils collected in the telescope is given by the relation @xmath38 where n@xmath39 and n@xmath40 represent , respectively , the target number density and the number of projectiles impinging on the target , @xmath41 is the cross section at the effective interaction energy e@xmath42 , t@xmath43 is the transmission of erna for the recoils in the selected charge state of probability @xmath44 , and @xmath45 is the detection efficiency of the telescope . therefore @xmath41 can be determined with high precision , if all quantities in eq . 1 are known with high accuracy . identification matrix of the @xmath25e - e telescope at @xmath46 mev for the standard @xmath26he target and ar post - stripper density ( upper panel ) . the different components are labeled . the lower panel shows the background at the same energy measured only with the ar post - stripper after removing the @xmath26he target gas . the running time for both spectra is the same.,width=264 ] the thickness of the extended windowless he gas target was determined using the @xmath26he(@xmath47li,@xmath10)@xmath48b reaction at @xmath49 mev @xcite , energy loss measurements of different ions @xcite , and the @xmath26he(@xmath47li,@xmath26he)@xmath47li@xmath50 reaction at @xmath51 mev @xcite . the weighted average of the results yields n@xmath52 atoms/@xmath53 with an effective target length of @xmath54 mm , that corresponds to an energy loss , e@xmath55 , for the @xmath4c ions smaller than 25 kev in the investigated energy range . the small target thickness leads to a nearly constant cross section along the target , thus e@xmath42 = ( e@xmath56/2)*m@xmath39/(m@xmath57 ) . the number of projectiles n@xmath40 is determined through the detection of the elastically scattered @xmath26he nuclei in two collimated silicon detectors located in the target chamber at 75@xmath30 from the beam axis . calibration runs performed at each energy before and after the measurement runs allowed to relate the observed scattering rate to the concurrent beam current . the scattering rate was measured in short runs of typically 60 s to achieve a statistical precision of better than 1% . the beam current without target gas was monitored in a faraday cup ( fc2 ) located after the quadrupole triplet before and after the determination of the scattering rate . this procedure was found to be independent of the beam focussing and reproducible within the statistical error . a 100% transmission of the incident beam through the gas target is a requirement of the separator @xcite and was verified by the full transmission through a retractable focussing aperture in front of the gas cell with a diameter of 3 mm . .[table1]astrophysical s(e ) factor of @xmath0 . the quoted errors are statistical only . the data in the energy range of the narrow resonances are left out . [ cols="^,^,^,^",options="header " , ] the charge state distribution @xmath44 of the @xmath1 recoils produced in the @xmath26he gas target depends on the geometric origin in the target : @xmath1 recoils produced in the upstream part of the target will most likely reach an equilibrium charge state distribution in the passage of the remaining target length , while those produced near the downstream end of the target will not , i.e. they will keep memory of the charge state at the moment of their formation . since this effect is not accurately predictable @xcite , the charge state distribution of the @xmath1 recoils after passing the he gas is affected by a significant uncertainty . to remove this uncertainty , an ar stripper was installed after the @xmath26he gas target with a number density @xmath58 atoms/@xmath53 . this density is sufficient for all ions produced at different locations in the @xmath26he target to reach the same charge state distribution . finally , the observed @xmath1 charge state distribution in the combination of @xmath26he and ar gas was measured over the full energy range @xcite : for each charge state the separator was set properly and the resulting current observed at the end of the separator in fc4 ( fig . [ setup ] ) . the measured charge state distribution differs from the equilibrium charge state distribution of @xmath14o ions in ar gas @xcite due to the effect of charge exchange in the @xmath26he rest gas in the downstream pumping stages of the gas target after the post - target stripper . this effect is energy and charge state dependent and amounts to a chance of the equilibrium charge state distribution of 20 % at most . the transmission of the separator t@xmath43 essentially depends on its acceptance compared to the emittance of the recoils , which in turn depends on @xmath10-ray emission , target effects , and the beam emittance . the angular acceptance of erna has been measured using an @xmath1 beam and an electrostatic deflection unit @xcite , which can deflect the beam at any position within the target region in order to simulate the recoil ion angular opening at the different geometrical locations where @xmath1 recoils are produced . the energy acceptance was measured by varying the beam energy from the accelerator . for both quantities , erna turned out to fulfill the requirements for measuring @xmath41 in the energy range @xmath59 to 5.0 mev within a region of @xmath60 mm around the target center , that includes 96% of the target nuclei . the angular straggling of the recoils in the gas target results in a loss of recoils less than 0.5 % , which gives a final value of @xmath61 @xcite . the detector efficiency @xmath62 is reduced by the transparency t@xmath63 of the grids of a parallel grid avalanche counter at the entrance window of the @xmath25e - e telescope leading to a total detection efficiency @xmath64 . it is important to note that the 100 % transmission of the recoils in the selected charge state is a key requirement for measuring reliably the absolute capture cross section @xmath41 using a recoil mass separator . indeed , this is a condition for determining @xmath41 without a detailed knowledge of the distribution of the recoils in the phase space , which depends on the @xmath10-ray energy and @xmath10-ray angular distribution , on target effects and beam emittance . finally , background runs ( fig . [ matrix ] lower panel ) were performed at each energy using the ar post - target stripper only . the measurements showed the presence of an @xmath1 background , where at energies above 2.0 mev a signal - to - background ratio of typically 20 was observed . the observed background rate in the region of interest was normalized to the number of projectiles and subtracted at each energy . an investigation of this @xmath1 background excluded as possible sources a beam contamination , which is suppressed by the erna beam purification system , as well as elastic scattering on rest gas . there are strong indications that the background is mostly due to the fraction of the @xmath4c beam impinging on the plates of the first wien filter , and possibly other beamline components , producing @xmath1 via @xmath4c+@xmath4c fusion , e.g. the @xmath4c(@xmath4c,@xmath8be)@xmath14o reaction . since the fraction of the beam impinging on the plates increases with decreasing energy , measurements at energies lower than @xmath65 mev are affected by a signal - to - background ratio significantly poorer than at higher energy . [ sfactor ] and table [ table1 ] show the total s(e ) factor at @xmath66 to 4.9 mev obtained with erna ; the statistical error is of the order of 4% or less at energies above 2.2 mev and the systematic error is 6.5% ( 3% n@xmath67 , 2% n@xmath68 , 1% t@xmath69 , 2% @xmath70 , 0.5% @xmath45 ) . in fig . [ sfactor ] we present also a comparison of our data with the sum of the different amplitudes as reported in @xcite , i.e. the ground state transition ( e1 and e2 ) plus cascade transitions . note the good agreement in absolute s(e ) values between previous work and the present data on top of the @xmath71 , 3.2 , and 4.3 mev resonances . the r - matrix calculation represents the best fit to the available experimental data at that time including resonance information from @xcite . one should note that in this analysis the interference effects of the cascade transitions are neglected . there is a clear disagreement at energies around @xmath72 and 4.0 mev , where the calculation of @xcite underestimates the total cross section , while it slightly overestimates the cross section on the high energy tail of the broad j@xmath73=1@xmath74 resonance at @xmath71 mev . a preliminary analysis indicates that these discrepancies may be partly caused by a wrong choice of the interference pattern of the different amplitudes , especially of the e2 amplitude . the analyses @xcite show that the @xmath75-fit is hardly sensitive to different e2 interference patterns leading to a broad range of extrapolated @xmath76 values . a detailed study of the influence of the present data on s(e@xmath24 ) is in progress , that requires an r - matrix code including a fit to the different amplitudes ( see for example @xcite ) as well as to the total cross section . finally , a new resonance was found in @xmath0 at @xmath77 mev , corresponding to a known 0@xmath78 state ( @xmath79 kev ) reported in @xcite . the analysis of the recoil data results in a resonance strength @xmath80 mev . the authors thank the technical staff of the mechanical workshop and of the dynamitron tandem laboratorium in bochum for their support to the project . the project is supported in part by deutsche forschungsgemeinschaft ( ro429/35 - 2 ) and istituto nazionale di fisica nucleare ( erna ) , and daad - crui vigoni .	the total cross section of @xmath0 was measured for the first time by a direct and ungated detection of the @xmath1 recoils . this measurement in inverse kinematics using the recoil mass separator erna in combination with a windowless he gas target allowed to collect data with high precision in the energy range @xmath2 to 4.9 mev . the data represent new information for the determination of the astrophysical s(e ) factor .
a riemann ellipsoid is a relative equilibrium for a dynamical model of a rotating self - gravitating fluid mass that remains an ellipsoid at all times , and for which the fluid velocity is a linear function of the coordinates . this model , first studied and formulated by dirichlet , is nowadays known as dirichlet s model . the linearity assumption on the allowed motions makes the study of these deformable bodies very attractive since it implies that their dynamics is governed by a system of ordinary differential equations with a finite number of degrees of freedom . these bodies are also known as affine - rigid bodies or pseudo - rigid bodies . dirichlet s model can be viewed as a first order model for the study of the shape of the earth , and the study of the stability of its solutions can be used in planetary stability research . the study of self - gravitating fluid masses has a long history which can be traced back to newton s times with many contributions from dirichlet , maclaurin , jacobi , dedekind , riemann , liapunov , poincar and cartan , just to name a few . we can distinguish two classical approaches to the study of the stability of riemann ellipsoids . one initiated independently in the latter part of the nineteenth century by poincar @xcite and liapunov @xcite who used expansions in spherical and ellipsoidal harmonics to study the stability of maclaurin ellipsoids and jacobi ellipsoids , not only under dirichlet s assumptions but also under more general conditions ( allowing perturbations not preserving the ellipsoidal shape ) . the other approach occurred in the middle of the twentieth century with the works of chandrasekhar and collaborators who developed the so - called virial method by applying it to the study of the linear stability of riemann ellipsoids . these works are collected in the book of chandrasekhar @xcite which constitutes a comprehensive survey on the subject and related problems with many historical facts on this model . in recent times the subject has had the attention of several researchers , in particular in what respects to the application of new formulations and methods ( @xcite ) to study rotating deformable bodies , not only subject to the self - gravitating potential but also for other potentials modelling nuclei ( see for instance @xcite ) , or elastic bodies ( see @xcite and @xcite ) . our aim is to use geometric methods not only to obtain a complete characterization of the conditions for the existence of riemann ellipsoids having configurations with at least two equal axes but also to obtain the complete description of their nonlinear stability . these geometric methods exploit the geometry and symmetries of the problem and its hamiltonian structure . some works using the same philosophical approach to dirichlet s model are available in the literature for studying several aspects of the problem such as in @xcite which obtains riemann s theorem as a consequence of the symmetry alone , or @xcite where some results on the formal stability and bifurcations of maclaurin spheroids are obtained . we do not address the problem of the stability for riemann ellipsoids with three distinct axes for which the self - gravitating potential is an elliptic integral . in @xcite this has been studied employing numerical analysis techniques . we view dirichlet s model as a hamiltonian system where hamilton s function , @xmath0 , has the form kinetic plus potential energy and is defined on the cotangent bundle , @xmath1 , of the set of all @xmath2 matrices of determinant 1 . furthermore , @xmath0 is invariant for the action of @xmath3 on the phase space . the @xmath4 symmetry reflects the existence of two conserved vector quantities ( by noether s theorem ) : the angular momentum and circulation . the @xmath5 symmetry reflects the reciprocity theorem of dedekind : interchanging the angular velocity and vorticity vectors one obtains another ( physically different ) solution for dirichlet s model with the same geometric configuration . it is well known that noether s conserved quantities are organized as the components of the momentum map . in the case of relative equilibria with configurations having at least two equal axes ( symmetric configurations ) the corresponding momentum map value can be singular . recently , one of the authors developed in @xcite a method appropriated for the study of nonlinear stability of this kind of relative equilibria . this construction extends the so - called reduced energy - momentum method of simo , lewis and marsden @xcite to the case of singular relative equilibria and so we will refer to the method @xcite as the singular reduced energy - momentum method . this is the approach we use in this work to study the nonlinear stability of riemann ellipsoids with symmetric configurations . the preliminary sections of the paper are organized as follow : in section [ sec geometric formulation ] we give the geometric formulation of dirichlet s model , in section [ sec singular reduced ] the singular reduced energy - momentum method is briefly reviewed and in section [ sec potential ] we compute the augmented potential energy for symmetric configurations , as a necessary step towards the stability analysis . the main results of the paper are in the following two sections . in section [ sec existence ] , theorem [ theorem existence ] , we give the complete characterization of all the possible riemann ellipsoids with symmetric configurations . we prove that for dirichlet s model the only relative equilibria with configurations having at least two equal axes are : the spherical configuration which has zero angular momentum and circulation , the _ maclaurin spheroids _ which are oblate spheroids rotating around the ( shortest ) symmetry axis and have angular momentum and circulation aligned with it , and the _ transversal spheroids _ , which have prolate spheroidal configurations that rotate around an axis , say @xmath6 , perpendicular to the ( longest ) symmetry axis and have angular momentum and circulation aligned with @xmath6 . we also prove that there are no symmetric relative equilibria for which the angular velocity and vorticity are not aligned with a principal axis of the relative equilibrium configuration . that is , there are not symmetric configurations which are not of type s in chandrasekhar s terminology . in section [ sec stability ] we apply the singular reduced energy - momentum method to the study of the nonlinear stability of the relative equilibria found in the previous section . the main results of this section are theorems [ spherical stability ] , [ maclaurin stability ] and [ transversal stability ] giving respectively necessary and sufficient conditions for the nonlinear stability of the spherical equilibrium , maclaurin spheroids and transversal spheroids . our results on the eccentricity range for the nonlinear @xmath7-stability of the maclaurin spheroids agree with those already obtained by liapunov and poincar ( see remark [ rem maclaurin ] ) . in the works of these authors there is no reference to the transversal spheroids , however their existence is acknowledged in pag . 143 of chandrasekhar s book @xcite . it is not clear for us what are the results obtained by chandrasekhar with respect to the stability of the transversal spheroids , however his use of the virial method only gives linear stability . to the best of our knowledge , our result on the nonlinear stability of the transversal spheroids is new . in conclusion , this work presents , from a purely geometric point of view , a self - contained and complete study of the nonlinear stability of all symmetric relative equilibria for dirichlet s model . at the same time , the richness of the model helps to clarify the applicability of the singular reduced energy - momentum method , providing also a methodology for its application to other models like . * acknowledgements . * we would like to thank to centro de anlise matemtica , geometria e sistemas dinmicos of the ist , lisbon , for the portuguese translation of original riemann s paper @xcite made by c.e . the work of esd has been supported by the fundao para a cincia e a tecnologia through the program poci 2010/feder . let @xmath8 be a riemannian manifold ( the configuration manifold ) , @xmath9 a lie group that acts by isometries on @xmath10 ( the symmetry group ) and @xmath11 a @xmath9-invariant function ( the potential energy ) . with these ingredients we construct a symmetric hamiltonian system on @xmath12 ( which is a manifold equipped with a natural symplectic structure ) as follows : the potential energy @xmath13 can be lifted to @xmath12 with the pullback of the cotangent bundle projection @xmath14 . we denote this lifted function also by @xmath13 . the riemannian metric on @xmath10 induces an inner product on each cotangent fiber @xmath15 , @xmath16 . then the hamiltonian is defined as @xmath17 the @xmath9-action on @xmath10 induces a cotangent - lifted hamiltonian action on @xmath12 with associated equivariant momentum map @xmath18 defined by @xmath19 where @xmath20 is the fundamental vector field on @xmath10 associated to the generator @xmath21 , defined by @xmath22 this momentum map is @xmath23-equivariant in the sense that @xmath24 for every @xmath25 , @xmath26 . the hamiltonian @xmath0 is @xmath9-invariant for this lifted action ( this follows from the invariance of the metric and of @xmath13 ) . therefore , due to noether s theorem , the components of @xmath27 are conserved quantities for the hamiltonian dynamics associated to @xmath0 . the quadruple @xmath28 is called a _ symmetric simple mechanical system_. let @xmath28 be a simple mechanical system with symmetry . a _ relative equilibrium _ is a point in phase space @xmath29 such that its hamiltonian orbit lies inside a group orbit for the cotangent - lifted action . this amounts to the existence of a generator @xmath30 such that the evolution of @xmath31 is given by @xmath32 . the element @xmath21 is called a _ velocity _ for the relative equilibrium and is defined up to addition of elements in @xmath33 , where @xmath34 is the stabilizer of @xmath31 under the cotangent - lifted action . a useful criterion for finding relative equilibria in simple mechanical systems is given by the following theorem : [ rethm]a point @xmath29 of a symmetric simple mechanical system @xmath28 is a relative equilibrium with velocity @xmath30 if and only if the following conditions are verified : 1 . 2 . @xmath36 is a critical point of @xmath37 , where @xmath38 is defined by @xmath39 . moreover , the momentum @xmath40 of the relative equilibrium is given by @xmath41 note that , in virtue of the above theorem , any relative equilibrium is characterized by a configuration - velocity pair @xmath42 satisfying @xmath43 . the map @xmath44 is called the _ locked inertia tensor _ , while the function @xmath45 is called the _ augmented potential_. we indicate for later use that the kernel of @xmath44 is precisely @xmath46 , the lie algebra of @xmath47 , the stabilizer of @xmath36 for the @xmath9-action on @xmath10 . the knowledge of the pair @xmath48 allows us to compute the stabilizer of the corresponding relative equilibrium @xmath49 with the formula ( see @xcite ) : @xmath50 dirichlet s model is a symmetric simple mechanical system for the motion of a homogenous and incompressible fluid mass of density @xmath51 having as reference configuration the unit ball centered at the origin in @xmath52 and subject to the self - gravitating potential . the only allowed configurations for this model are linear embeddings of the reference ball into @xmath52 preserving volume and orientation . the configuration manifold @xmath10 for a self - gravitating fluid mass under dirichlet s conditions is then @xmath53 , the group of all @xmath2 matrices with determinant equal to 1 , which is equivalent to the space of orientation and volume preserving linear automorphisms of @xmath52 . in what follows we review the geometric formulation of dirichlet s model as a symmetric simple mechanical system on @xmath54 , the group of all @xmath2 matrices with positive determinant , with a symmetric holonomic constraint . the singular value decomposition of any linear automorphism of @xmath52 allows to decompose ( non - uniquely ) any matrix @xmath55 as @xmath56 where @xmath57 , and @xmath58 is a diagonal matrix with positive entries called singular values ( the square roots of the eigenvalues of @xmath59 ) . it follows from this decomposition that the reference unit ball is mapped by @xmath60 into a solid ellipsoid of equation @xmath61 , having principal axes half - lengths equal to the entries of @xmath58 . physically the matrix @xmath62 describes the rigid rotation of the body in space relative to an inertial frame and @xmath63 is related to the rigid internal motion of the fluid with respect to a moving frame . then @xmath58 is an orientation preserving dilatation of the original reference body into an ellipsoid with principal axes aligned with the eigenvectors of @xmath58 . the condition on the volume preservation of the total embedding corresponds to impose the holonomic constraint @xmath64 ( or equivalently @xmath65 ) , which in turn amounts to consider our system as defined on @xmath53 . the tangent space at @xmath66 is isomorphic to @xmath67 , the vector space of @xmath2 matrices . we can define a riemannian metric on @xmath68 as : @xmath69 for @xmath70 , and @xmath71 is a constant depending on the density of the reference body and other physical parameters of the system . in the case of interest here , the reference body is a homogeneous unit ball of constant density @xmath51 , and @xmath71 in is @xmath72 the symmetry group @xmath9 of our model is the semi - direct product @xmath3 , where @xmath73 . several actions of @xmath9 of interest in this paper are : 1 . the @xmath9-action on @xmath9 : if @xmath74 then @xmath75 where for the nontrivial element @xmath76 , @xmath77 , and @xmath4 acts on itself by the direct product of left matrix multiplications . 2 . the adjoint representation of @xmath9 : we identify the lie algebra of @xmath78 , the set @xmath79 of skew - symmetric @xmath2 matrices , with @xmath52 via the usual isomorphism @xmath80 : @xmath81 this is an isomorphism of lie algebras , i.e @xmath82)$ ] is isomorphic by to @xmath83 , where @xmath84 $ ] denotes the commutator of matrices and @xmath85 denotes the vector product of vectors in @xmath52 . + the lie algebra of @xmath9 is then @xmath86 and the adjoint action is given by @xmath87 where @xmath88 and @xmath89 is the rotation of @xmath90 by @xmath91 ( and similarly for @xmath92 ) . + using the standard inner product in @xmath52 we also identify @xmath93 with @xmath94 . under this identification it follows easily that the coadjoint representation has the expression @xmath95 3 . the @xmath9-action on @xmath68 : @xmath96 note that the @xmath5 transposition symmetry on @xmath53 ( first noticed by dedekind ) maps a rigidly rotating configuration without internal motion into one that is stationary in space but with rigid fluid internal motions . that is , for a given ellipsoid there is an adjoint one , obtained by transposition . these adjoint type of ellipsoids are called dedekind ellipsoids . more generally , the transposition symmetry interchanges external rotations and internal motions for any solution of dirichlet s model . any @xmath9-invariant function @xmath97 on @xmath68 can be written as @xmath98 where @xmath99 , and @xmath100 and @xmath101 are the three principal invariants of a @xmath2 matrix , given by @xmath102 with @xmath103 ( also valid interchanging @xmath104 with @xmath59 ) . note that @xmath105 are @xmath9-invariant and that @xmath106 is a regular value of @xmath101 . hence we have @xmath107 as a @xmath9-invariant submanifold of @xmath68 . the condition @xmath108 is the holonomic constraint of the model . note also that the restriction of a @xmath9-invariant function @xmath109 to @xmath110 is given by @xmath111 for @xmath112 . therefore any @xmath9-invariant function on @xmath110 can be written as @xmath113 , with @xmath114 . any @xmath9-invariant function on @xmath53 can be then extended invariantly to @xmath68 by declaring it to be independent of @xmath115 . from now on we will drop the tildes and hats from the corresponding functions unless there is risk of confusion . since @xmath9 acts on @xmath68 by isometries with respect to , the induced metric on @xmath53 ( which we will denote by the same symbol ) is also @xmath9-invariant . for later use we recall that tangent vectors to @xmath53 can be seen as tangent vectors to @xmath68 satisfying the linearization of the constraint @xmath108 . in other words , @xmath116 dirichlet s model is governed by a hamiltonian function of the form kinetic plus potential energy on the phase space @xmath117 given by @xmath118 here @xmath119 is the norm of the covector @xmath120 ( seen as a @xmath2 matrix ) relative to the metric on @xmath53 . the potential energy @xmath13 for a self - gravitating body of homogeneous density @xmath51 under dirichlet s assumptions is given by restricting the function @xmath121 where @xmath122 , @xmath9 is the gravitational constant and @xmath123 the quadruple @xmath124 defines a symmetric simple mechanical system on @xmath53 . the infinitesimal generator for the @xmath9-action on @xmath125 ( and on @xmath110 ) corresponding to @xmath126 is : @xmath127 the vectors @xmath90 and @xmath92 are respectively the angular velocity and vorticity of the fluid motion . we denote the momentum value of @xmath120 by @xmath128 . the components @xmath129 and @xmath130 are respectively the angular momentum and circulation of the instantaneous state @xmath120 ( see for instance @xcite or @xcite ) . a riemann ellipsoid ( a.k.a . an ellipsoidal figure of equilibrium ) is a solution of the hamiltonian system defined by dirichlet s model with angular velocity , vorticity and principal axes lengths all constant . in our setting , riemman ellipsoids correspond exactly to relative equilibria of the underlying symmetric simple mechanical system . therefore , a riemann ellipsoid is represented by a triple @xmath131 , where @xmath132 is the configuration matrix and the lie algebra element @xmath133 is the angular velocity - vorticity pair . in the last years , several are the works studying the stability of relative equilibria of hamiltonian systems by exploiting their symmetry and the geometric properties of their phase space ( see for instance @xcite , @xcite , and @xcite for a overview ) . most of these methods can be used to test the stability of relative equilibria lying in singular level sets of the momentum map , for instance @xcite and @xcite under the hypotesis of @xmath7 compact and @xcite , @xcite using topologic properties . this observation is important since , as we will see , the class of symmetric riemann ellipsoids known as maclaurin spheroids corresponds precisely to non - trivial relative equilibria for dirichlet s model having singular momentum values . we refer the reader to @xcite for a comparison of the applicability of several existing methods . the generally adopted notion of stability for relative equilibria of symmetric hamiltonian systems is that of @xmath7-stability , introduced in @xcite and that we now review in the context of symmetric simple mechanical systems . this notion is closely related to the liapunov stability of the induced hamiltonian system on the reduced phase space . [ patrick ] let @xmath28 be a symmetric simple mechanical system and @xmath29 a relative equilibrium with momentum value @xmath134 . we say that @xmath31 is @xmath7 stable if for every @xmath7-invariant neighborhood @xmath135 of the orbit @xmath136 there exists a neighborhood @xmath137 of @xmath31 such that the hamiltonian evolution of @xmath137 lies in @xmath138 for all time . in the early 1990 s ( @xcite ) a tool known as the reduced energy - momentum method has been developed , providing sufficient conditions for the stability of relative equilibria of a simple mechanical system under the hypothesis that its momentum is a regular value of the momentum map . this method is especially well - suited for simple mechanical systems since in incorporates all of their distinguishing characteristics with respect to general hamiltonian systems . recently , based on the characterization @xcite of the so - called symplectic normal space @xmath139 for a cotangent - lifted action , the reduced energy - momentum method was generalized in @xcite to cover also the case of singular momentum values . in this section we outline the implementation of this singular reduced energy - momentum method following @xcite . our setup will be as in definition [ patrick ] and section 2 . in particular we will fix a relative equilibrium @xmath31 with configuration - velocity pair @xmath48 and momentum @xmath140 . we will also assume that the @xmath9-action on @xmath10 is proper and that there exists a @xmath7-invariant complement to @xmath141 in @xmath142 . these last two conditions are always satisfied for any relative equilibrium in dirichlet s model due to the compactness of @xmath9 . we start by stating some key observations : first , by equivariance of @xmath143 and @xmath14 , one has @xmath144 and @xmath145 . in fact , it is not difficult to prove the characterization @xmath146 we remark that the above formula is not valid in general for covectors @xmath31 other than relative equilibria . second , also by equivariance of @xmath147 together with the bifurcation lemma ( see @xcite ) , if @xmath140 is a singular momentum value then @xmath148 , in which case @xmath149 . third , the properness of the @xmath9-action implies that @xmath34 is compact . this , together with allows to define the following @xmath34-invariant splittings : @xmath150 which by duality induce the splittings @xmath151 here @xmath152 and @xmath153 must be chosen in such a way that @xmath154 . note , from the definition of the locked inertia tensor in theorem [ rethm ] , that @xmath155 , so the restriction @xmath156 is a @xmath34-equivariant isomorphism . then the condition on the above splitting is that @xmath152 and @xmath153 must be orthogonal with respect to the inner product on @xmath157 induced by @xmath158 . we will denote generically the linear projections associated to the splittings and by the letter @xmath159 with an appropriate subindex . for instance @xmath160 or @xmath161 . it is a consequence of noether s theorem that @xmath162 and so we will denote by @xmath163 ( the orthogonal velocity of the relative equilibrium ) . in this work we only need a particular version of the singular reduced energy - momentum method . for , consider the following definitions : * let @xmath164 be the linear orthogonal slice for the @xmath9-action on @xmath10 at @xmath36 , i.e. @xmath165 * define the subspace @xmath166 as @xmath167=0\right\}.\ ] ] * define the space of internal variations @xmath168 * for any @xmath169 , the correction term is the bilinear form on @xmath170 defined by @xmath171\ , , \ , { \widehat{\mathbb{i}}_0}^{-1}\left({\mathbb{p}}_{\mathfrak{p}^\oplus\mathfrak{t}^}\left[\left({\mathbf{d}}{\mathbb{i}}\cdot v_2\right ) ( \xi)\right]\right ) \right\rangle.\ ] ] * the arnold form @xmath172 is defined by : @xmath173 \right\rangle.\ ] ] the following theorem ( corollary 6.2 of @xcite ) is the synthesis of the singular reduced energy - momentum method . [ energy miguel ] let @xmath29 be a relative equilibrium with configuration - velocity pair @xmath174 and momentum @xmath175 such that @xmath176 . let @xmath177 be its orthogonal velocity . if the arnold form is non - degenerate at @xmath31 and @xmath178 is positive definite , then the relative equilibrium is @xmath7-stable . when the arnold form is non - degenerate it is also shown in @xcite that the symplectic matrix of the symplectic normal space at @xmath31 has a particularly simple block - diagonal expression . we quote this result which will be essential in the proof of the linear unstability of some riemann ellipsoids . [ symplectic mat ] if the arnold form is non - degenerate at the relative equilibrium @xmath31 with configuration - velocity pair @xmath48 and momentum @xmath140 , the symplectic normal space at @xmath31 , is symplectomorphic to @xmath179 equipped with the symplectic matrix @xmath180 \end{array}\ ] ] where @xmath181 and @xmath182 is the one - form defined by @xmath183 , for all @xmath184 . [ orbitre ] the @xmath9-invariance of @xmath13 and @xmath185 imply the following property . if @xmath48 is a relative equilibrium , then the orbit @xmath186 for every @xmath26 consists of relative equilibria with the same stability or unstability properties . [ augmentations ] the reason why in the previous section we looked at dirichlet s model as a simple mechanical system holonomicaly constrained is that the unconstrained space @xmath68 is an open domain of the vector space @xmath67 , and then the implementation of the reduced energy - momentum method is easier than if one is working directly on @xmath53 . in view of the survey of the method , the strategy will be to use the trivial extension of the self - gravitating potential to @xmath68 and consider its augmented potential with respect to the locked inertia tensor corresponding to the original riemannian metric on @xmath68 . then we further augment this augmented potential with the constraint function @xmath101 and lagrange multiplier @xmath187 . denoting by @xmath188 the resulting twice augmented potential , we have : 1 . relative equilibria for dirichlet s model correspond to triples @xmath131 with @xmath189 such that the following two equations hold : @xmath190 2 . the stability test now follows from the following observation . if we call @xmath191 and @xmath192 the spaces of internal variations for @xmath68 and @xmath53 associated to the triple @xmath131 , according to , we notice that @xmath193 therefore , according to the general method , and standard lagrange multiplier theory , to conclude stability it suffices to study the eigenvalues of the bilinear form @xmath194 where @xmath192 is seen as a vector subspace of @xmath191 . from now on we will omit the superindex @xmath110 for the space of internal variations . in this section we compute the augmented potential @xmath195 in the unconstrained configuration space @xmath68 and collect some results for the self - gravitating potential @xmath13 . the potential @xmath196 at a typical configuration is an elliptic integral except for symmetric configurations ( i.e. with at least two equal singular values ) for which it can be integrated by elementary functions . the extension to the unconstrained space of the potential @xmath13 depends on @xmath55 through the two principal invariants @xmath197 and @xmath198 defined by and respectively . in the study of the existence and stability of relative equilibria of simple mechanical systems will be necessary to compute the first and second derivatives of @xmath199 and the results of this section are essential to this end . in the following we will restrict ourselves to diagonal configurations only . there is no loss of generality with this assumption since , according to the singular value decomposition , every matrix @xmath55 belongs to the @xmath9-orbit of some diagonal configuration @xmath200 by some element @xmath201 . hence by remark [ orbitre ] the qualitative properties of a relative equilibrium @xmath202 are the same as those of @xmath203 . let @xmath204 , with @xmath205 be the following family of integrals : @xmath206 and denote by @xmath207 ( @xmath208 ) the partial derivative of @xmath13 with respect to @xmath209 and by @xmath210 the partial derivative of @xmath207 with respect to @xmath211 for @xmath212 . using , elementary calculus computations give : @xmath213 note that the integrals @xmath204 are all positive as well as @xmath214 and @xmath215 . next proposition gives the values of @xmath204 in the case of spheroidal ( two equal axes ) configurations . [ integrals]let @xmath216 be a spheroidal configuration with @xmath217 . * the integral @xmath204 for the oblate spheroid @xmath60 ( @xmath218 ) and eccentricity @xmath219 is given by @xmath220 * the integral @xmath204 for the prolate spheroid @xmath60 ( @xmath221 ) and eccentricity @xmath222 is given by @xmath223 note that for the diagonal configuration @xmath224 the value of @xmath225 in definition is @xmath226^{1/2}.\ ] ] for ( i ) : making the change of variables @xmath227 we get @xmath228 since the eccentricity is @xmath229 then @xmath230 and @xmath231 because @xmath217 . then , from the above expression for @xmath232 one gets @xmath233 making @xmath234 the result follows . for ( ii ) : the change of variables @xmath235 gives @xmath236 the eccentricity @xmath237 of the prolate spheroid is such that @xmath238 . as @xmath217 then @xmath239 and @xmath240 and the result follows for @xmath234 . as stated in the previous section , in order to find critical points of a @xmath9-invariant function defined in @xmath110 we will work with its extension to @xmath68 subject to the constraint @xmath64 . for , since any such function can be written as @xmath241 , in order to compute the critical points we use the augmented function @xmath242 subject to the condition @xmath243 . for the differentiation of @xmath244 consider the pairing between vectors @xmath245 and covectors @xmath246 to be @xmath247 . then using the chain rule we get that critical points must verify the following set of equations : @xmath248\\ \nonumber & = & 0.\\ \nonumber \det(f ) & = & 1\end{aligned}\ ] ] since @xmath249\\ \label{traceformula3}{\mathbf{d}}\det ( f)\cdot\delta f & = & \det ( f)\mathrm{tr}(f^{-1}\delta f ) .\end{aligned}\ ] ] ( see for instance @xcite or @xcite ) . next proposition gives the expression for the locked inertia tensor . [ lockedprop]the locked inertia tensor for the @xmath9-action on @xmath250 , at a configuration @xmath55 , is defined by @xmath251\ ] ] where @xmath252 and @xmath253 for @xmath208 . under the isomorphism , the locked inertia tensor is also equivalent to : @xmath254 where @xmath103 , @xmath59 and @xmath255 ( @xmath256 denotes the identity matrix ) . by the locked inertia tensor definition in proposition [ rethm ] , the expression for the infinitesimal generators of the @xmath9-action on @xmath68 and the definition for the riemannian metric , we have @xmath257.\end{aligned}\ ] ] using the fact that @xmath258 and @xmath259 are skew - symmetric matrices and the cyclic property of the trace of a matrix , it is straightforward to obtain expression . for the expression we need some standard properties of the isomorphism . in particular , @xmath260 where @xmath261 denotes the standard inner product on @xmath52 . let us compute some terms of the expression since the other are done similarly @xmath262&\quad\text{(by \eqref{eqisosecond})}\\ & = \frac{1}{2 } { \operatorname{tr}\,}\left(\widehat{\xi_l}^t \widehat{\mathbf{i}_s\eta_l}\right)&\quad\text{(by \eqref{eqisothird})}\\ & = \xi_l\cdot \mathbf{i}_s\eta_l&\quad\text{(by \eqref{eqisofirst})}\\\end{aligned}\ ] ] @xmath263&\quad\text{(by \eqref{eqisofour})}\\ & = 2\det ( f ) \xi_r\cdot f^{-1 } \eta_l & \quad\text{(by \eqref{eqisofirst})}\\\end{aligned}\ ] ] as a straightforward consequence we can obtain the momentum of a relative equilibrium for dirichlet s model , that is its angular momentum and circulation . [ momcirc ] the momentum for a relative equilibrium with configuration @xmath60 and velocity - vorticity pair @xmath264 is @xmath265 that is , the angular momentum and circulation of a riemann ellipsoid with configuration by @xmath60 , and angular velocity - vorticity pair @xmath266 are given , respectively , by @xmath267 the expression for the twice augmented potential @xmath268 follows now easily from proposition [ lockedprop ] . in this section we classify symmetric relative equilibria for dirichlet s model . we will treat the spherical case ( i.e. a configuration having three equal principal axes ) as a particular case of a symmetric configuration . from the singular value decomposition and the definition of the action of @xmath270 on @xmath125 ( or on @xmath271 ) it follows that the stabilizer of a symmetric configuration @xmath60 is conjugate to the stabilizer of a diagonal configuration . that is , conjugate to @xmath272 or @xmath273 if @xmath60 has 2 or 1 different singular values , respectively ( see @xcite for a derivation of this result ) . actually , if the configurations are diagonal , these are exactly their stabilizers . here , @xmath274 denotes the diagonal embedding of @xmath275 in @xmath4 and @xmath276 is the subgroup of @xmath277 generated by all the rotations @xmath278 around a given axis @xmath279 in @xmath52 and a rotation , @xmath280 , by @xmath281 around an axis @xmath282 perpendicular to @xmath279 . in case of the diagonal configuration @xmath283 , @xmath284 is the rotation matrix by and angle @xmath285 around @xmath286 and @xmath280 can be chosen to be @xmath287 . we introduce the following subgroups : * @xmath288 , generated by elements @xmath289 , with @xmath290 and @xmath291 , * @xmath292 , generated by elements @xmath293 , with @xmath294 and @xmath291 , * @xmath295 , the cyclic group isomorphic to @xmath5 generated by the element @xmath296 . * more generally , if @xmath297 is a subgroup of @xmath4 , we denote also by @xmath297 the subgroup of @xmath298 generated by elements @xmath299 , with @xmath300 . note that since we are going to impose the constraint @xmath301 , we will consider only two kinds of symmetric configurations , especifically : * spherical : @xmath302 , * spheroidal : @xmath303 , with @xmath217 . to find all the possible riemann ellipsoids with symmetric configurations , we will have to solve with @xmath60 of the above forms and different pairs @xmath266 . the possible solutions are summarized in the following theorem . [ theorem existence ] the relative equilibria for dirichlet s model of a self - gravitating fluid mass are : * the * spherical equilibrium * with spherical configuration @xmath304 , velocity - vorticity pair @xmath305 and lagrange multiplier @xmath306 . its corresponding momentum and isotropy groups are @xmath307 * the family of * maclaurin spheroids * which have oblate spheroidal configurations @xmath283 ( with @xmath308 ) and angular velocity and vorticity parallel to the axis of symmetry @xmath309 . in terms of the parameter @xmath310 defined by @xmath311 , this family is characterized by @xmath312 and the following constraint between @xmath310 and the eccentricity @xmath237 : @xmath313 its corresponding momentum and isotropy groups are : @xmath314 @xmath315 * two branches of * transversal spheroids * which have prolate spheroidal configurations @xmath283 ( with @xmath316 ) . we distinguish the two branches of this family with the signs @xmath317 and @xmath318 . these branches are characterized by the lagrange multiplier @xmath319 , the velocity - vorticity pair @xmath320 with @xmath321 ( where @xmath6 is a unit vector perpendicular to @xmath322 ) and the following constraints between @xmath323 and the eccentricity @xmath237 : @xmath324 the corresponding momentum and isotropy groups are : @xmath325 @xmath326 before proving the theorem we remark that formula has already been obtained by maclaurin in 1742 , as it is claimed in page 4 of chandrasekhar s book @xcite . first , using , and , it is easy to see that the general conditions are equivalent to @xmath327\\ \label{recond2 } 1 & = & \det(f)\end{aligned}\ ] ] [ [ spherical - case ] ] spherical case : + + + + + + + + + + + + + + + if @xmath328 , then , are simply @xmath329\right ) \mathbf{i}+t\left[\xi_l\otimes\xi_l+\xi_r\otimes\xi_r-2\xi_l\otimes\xi_r\right]=0\ ] ] taking @xmath330 and the same sorte of notation for @xmath92 , the off - diagonal terms of this expression are independent of @xmath331 and @xmath187 , and equivalent to the following 6 equations : @xmath332 it follows then that @xmath333 . recall from theorem [ rethm ] that the momentum of a relative equilibrium with configuration @xmath36 and velocity @xmath21 is given by @xmath334 . then , from we have @xmath335 . therefore , @xmath336 , @xmath337 , @xmath338 . now , noting that @xmath339 if @xmath333 , and that the velocity of a relative equilibrium is defined only up to addition of elements in @xmath340 , the relative equilibrium @xmath341 is the same as @xmath342 . then the relative equilibrium conditions are satisfied with @xmath306 . [ [ spheroidal - case ] ] spheroidal case : + + + + + + + + + + + + + + + + we now consider @xmath303 with @xmath217 . since in the @xmath343 plane all directions are equivalent , we can assume without loss of generality that @xmath344 . now , conditions , are equivalent to the system @xmath345 note that these equations imply that @xmath346 . indeed , if @xmath347 then implies @xmath348 and so and imply @xmath346 which is a contradiction . we will now proceed systematically considering four main cases : ( i ) @xmath349 , ( ii ) @xmath350 , ( iii ) @xmath351 and @xmath352 , and ( iv ) @xmath353 and @xmath352 . \(i ) if @xmath349 , then , and are the only non - trivial conditions , which are equivalent to @xmath354 as @xmath214 and @xmath215 are positive and @xmath355 it follows from the last equation that there is no solution . \(ii ) if @xmath350 , then from and we have @xmath356 and so and give @xmath357 . let @xmath358 . then the remaining non - trivial equations , , and , give @xmath359 as @xmath214 and @xmath215 are positive , then last equality implies that the spheroidal configuration @xmath60 is oblate , that is @xmath218 . the eccentricity of the spheroid is @xmath360 and @xmath361 . using the relations for the partial derivatives of the self - gravitating potential , @xmath214 and @xmath215 , the expression @xmath362 for the integrals @xmath204 in the oblate case given by and @xmath363 , then the equation is equivalent to @xmath364 one can easily compute the definite integrals @xmath365 and @xmath366 , although we avoid to display their expressions since they are quite lengthy . however the expression @xmath367 is @xmath368 from which follows . from corollary [ momcirc ] it is trivial to obtain that the momentum of this relative equilibrium . hence , using , the appropriate isotropy groups are also straightforward . \(iii ) if @xmath351 and @xmath369 , it follows from and that @xmath370 . then and give @xmath371 we can set @xmath372 and @xmath373 . so , substituting the above value of @xmath187 into and these equations are @xmath374 from we obtain the solutions @xmath375 , from which follows that the spheroids must be prolate @xmath376 . in terms of the eccentricity @xmath377 we have @xmath378 therefore , the equation gives @xmath379 substituting in this expression @xmath380 and @xmath381 , as well as @xmath382 gives . as before , using the expression of the locked inertia tensor and the remaining results follow . \(iv ) in this case we have @xmath369 , @xmath353 and @xmath383 . note that from and one should also have @xmath384 and @xmath385 . so , let @xmath386 and @xmath387 for some reals @xmath388 . then , using also , the equations and reduce to @xmath389 these equations have solutions @xmath390 comparing and using @xmath386 , @xmath387 and we obtain @xmath391 and so @xmath392 must be @xmath393 or @xmath394 , but this is a contradiction with . so there is no solution for the above system . [ transversal adjoint ] note that the functions @xmath395 and @xmath396 appearing in the characterization of transversal spheroids are inverse of each other . therefore the two families of transversal spheroids belong to a single orbit of the symmetry group . indeed , the @xmath5 symmetry interchanges the @xmath317 and @xmath318 families , since @xmath397 , as it follows from their definitions that @xmath398 . theorem [ theorem existence ] is in agreement with riemann s classification of ellipsoidal figures of equilibrium for dirichlet s model of self - gravitating fluid masses . that is , these ellipsoidal figures of equilibrium must lie in one of the following categories : ( a ) the case of a uniform rotation with no internal motion ( or uniform vorticity and no rotation ) ; ( b ) the case when the directions of the angular velocity @xmath90 and vorticity @xmath92 are the same and coincide with a principal axis of the ellipsoid ( also known as ellipsoids of type s ) ; ( c ) the case when the angular velocity and vorticity are not parallel but lie in the same principal plane . in particular , we show that for dirichlet s model it is not possible to obtain relative equilibria with spheroidal configurations belonging to category ( c ) . the existence of transversal spheroids is referred in chandrasekhar s book @xcite ( see for instance page 143 ) , however their study is not present in the classical works of liapunov @xcite and poincar @xcite . in this section we apply the singular version of the reduced energy - momentum method introduced in @xcite in order to deduce the stability of the symmetric relative equilibria obtained in theorem [ theorem existence ] . in order to apply this method it is essential to compute the second derivative of the twice augmented potential @xmath399 . next lemma gives that result . [ second derivative augmented ] if @xmath60 is a critical point of the twice augmented potential @xmath400 for @xmath401 and @xmath402 as in proposition [ lockedprop ] and @xmath403 , then @xmath404 where @xmath405 and @xmath406 then , applying standard properties of the trace , we get @xmath413.\ ] ] the expression for @xmath411 stated follows now easily . finally , using to differenciate the expression @xmath414 we get @xmath415.\end{aligned}\ ] ] we now study the stability of the spherical equilibrium . notice from the outline of the method in section 3 that for this equilibrium whe have that @xmath416 , the correction term and the arnold form are all trivial , as well as the velocity - vorticity pair @xmath266 . as a consequence , @xmath417 , the orthogonal complement to the @xmath9-orbit at the identity in @xmath110 . hence , to conclude stability of the spherical equilibrium we need to study the definiteness of @xmath418 recall that @xmath419 is the space of traceless matrices . also , the infinitesimal action of @xmath142 on @xmath68 at @xmath256 is @xmath420 . then @xmath421 therefore @xmath422 is the space of traceless symmetric matrices . that is , matrices of the form @xmath423 we fix a basis for @xmath422 with respect to which the components of @xmath58 are @xmath424 . by lemma [ second derivative augmented ] , the expression of @xmath425 , with @xmath426 as given in theorem [ theorem existence ] , reduces to : @xmath427 where we have applied the fact that @xmath58 and @xmath428 are traceless symmetric matrices . therefore in this basis we have @xmath429 as @xmath430 is positive , the eigenvalues of this matrix are @xmath431 and @xmath432 with multiplicities @xmath433 and @xmath434 respectively . these are all positive , therefore the spherical equilibrium is @xmath9-stable . we now study the nonlinear stability of maclaurin spheroids in the setup of previous sections ( theorem [ theorem existence ] ) . as it has been stated , a maclaurin spheroid has an oblate configuration which , with no loss of generality , we suppose diagonal . this configuration is uniquely characterized by the eccentricity @xmath435 . in order to apply theorem [ energy miguel ] one needs first to split @xmath436 according to , that is as @xmath437 recall that for the maclaurin spheroid one has @xmath438 and @xmath439 . one can then choose the following ordered orthonormal bases ( with respect to the euclidean product in @xmath440 ) for each of the spaces of the above splitting : if we define @xmath441 , then @xmath442 it is straightforward to check that these subspaces are invariant for @xmath443 . the orthogonal velocity @xmath444 for the maclaurin relative equilibrium is the orthogonal projection of the velocity @xmath21 onto @xmath152 . then , @xmath445 where @xmath310 must satisfy . as already defined , @xmath446 is the restriction of @xmath44 to @xmath157 . the locked inertia matrix for the configuration @xmath303 is , according to , @xmath447 where @xmath448 and @xmath449 . it is now straighforward to obtain the @xmath446 matrix with respect to the basis @xmath450 . that is @xmath451 or , in terms of the eccentricity @xmath452{1-e^2 } } & 0 & 0 & 0 & 0\\ 0 & \frac{2-e^2}{\sqrt[3]{1-e^2 } } & -2\sqrt[6]{1-e^2 } & 0 & 0\\ 0 & -2\sqrt[6]{1-e^2 } & \frac{2-e^2}{\sqrt[3]{1-e^2 } } & 0 & 0 \\ 0 & 0 & 0 & \frac{2-e^2}{\sqrt[3]{1-e^2 } } & -2\sqrt[6]{1-e^2 } \\ 0 & 0 & 0 & -2\sqrt[6]{1-e^2 } & \frac{2-e^2}{\sqrt[3]{1-e^2 } } \end{bmatrix}\ ] ] we can use @xmath158 and @xmath453 to compute the momentum of a maclaurin spheroid . indeed @xmath454 which is of course the same as the value obtained in theorem [ theorem existence ] under the identification @xmath455 induced by the euclidean product in @xmath440 . first , we compute the space @xmath416 . notice the following relations for the adjoint representation of @xmath9 : @xmath459 and @xmath460 also , recall that under our identification @xmath455 we have @xmath461 , for @xmath462 , and where @xmath51 in the right hand side is identified with an element of @xmath142 . therefore @xmath463 for @xmath464 , hence @xmath456 . as @xmath465 then @xmath466 . then , from we obtain @xmath467 the inverse of the matrix is not difficult to compute . here we just state the values of @xmath468 on vectors @xmath469 of the fixed basis : @xmath470 from these expressions it follows easily that @xmath471 finally , the entries of the arnold matrix are given by @xmath472 using the arnold matrix is then given by @xmath473 for ( i ) : as the arnold form is non - degenerate , @xmath7-stability will follow whenever @xmath478 is positive definite . for , recall from that @xmath479 where @xmath422 is the linear slice for the @xmath9-action on @xmath53 at the oblate configuration , @xmath303 . matrices @xmath480 must verify @xmath481,\quad\forall \xi,\eta\in{\mathbb{r}}^3,\quad\text{and}\\ 0 & = & { \operatorname{tr}\,}(f^{-1}a)\end{aligned}\ ] ] because @xmath58 must belong , respectively to the orthogonal complement to the tangent space to the group orbit through @xmath60 and @xmath482 . these two conditions give that @xmath58 must be of the form @xmath483 therefore we can describe @xmath422 as the set of matrices of the form @xmath484 with @xmath485 . let the vector @xmath486 be @xmath487 with respect to the basis @xmath488 . therefore , using , we have @xmath489 the set @xmath490 is precisely the set of matrices @xmath491 for which @xmath492 where @xmath493 denotes the annihilator of @xmath494 . using @xmath495 and @xmath496 and differentiating , the computation of @xmath497 gives , in terms of the eccentricity @xmath237 : @xmath498\\ \langle t_2,\left({\mathbf{d}}{\mathbb{i}}(f ) \cdot \left(\gamma_{\mathrm{sl}(3)}(f)+a\right)\right ) ( \xi_l,\xi_r)^\perp\rangle&= \frac{t\omega}{2 ( 1-e^2)^{1/3 } } \left[2\sqrt{1-e^2 } \gamma_3- ( 2+e^2 ) \gamma_4\right]\\ \langle t_3,\left({\mathbf{d}}{\mathbb{i}}(f ) \cdot \left(\gamma_{\mathrm{sl}(3)}(f)+a\right)\right ) ( \xi_l,\xi_r)^\perp\rangle&= \frac{t\omega}{2 ( 1-e^2)^{1/3 } } \left[-(2+e^2 ) \gamma_1 + 2\sqrt{1-e^2 } \gamma_2\right]\\ \langle t_4,\left({\mathbf{d}}{\mathbb{i}}(f ) \cdot \left(\gamma_{\mathrm{sl}(3)}(f)+a\right)\right ) ( \xi_l,\xi_r)^\perp\rangle&= \frac{t\omega}{2 ( 1-e^2)^{1/3 } } \left[-2\sqrt{1-e^2 } \gamma_1 + ( 2+e^2 ) \gamma_2\right]\end{aligned}\ ] ] it follows from the above expressions that @xmath499 if and only if @xmath500 . this is equivalent to @xmath501 . let us now compute the correction term restricted to @xmath501 . for any @xmath480 one has @xmath502 and so @xmath503 = \left(\frac{4t\omega \sqrt{2}}{(1-e^2)^{1/6 } } a_1,0,0,0,0\right).\ ] ] from the expression of @xmath504 it is straightforward to obtain @xmath505\right)= \sqrt{2 } \omega ( 1-e^2)^{1/6 } b_1 p,\ ] ] where @xmath506 is the entry of @xmath507 playing the same role of @xmath508 in @xmath58 . then , from @xmath509 the plots of @xmath521 and @xmath522 are shown in figure [ plots maclaurin ] . they show that @xmath521 is always positive in @xmath523 while @xmath522 has a root @xmath524 being positive for @xmath475 and negative for @xmath477 . we used the mathematica programing system for the numerical computation of @xmath525 to obtain @xmath526 . therefore maclaurin spheroids are @xmath7-stable for @xmath475 . for ( ii ) : when @xmath527 we study the spectrum of the linearized hamiltonian vector field @xmath528 , where in our case @xmath139 and @xmath529 are as in theorem [ symplectic mat ] also , from proposition 6.2 in @xcite we have @xmath530 let us start by computing the block @xmath531 of the above matrix , where @xmath531 is the inner product on the dual of @xmath422 induced by the riemannian metric on @xmath422 . for , let as before fix the ordered basis @xmath532 on the slice @xmath422 where @xmath533 recalling that @xmath534 then the matrix that represents @xmath535 in the fixed basis is @xmath536 let @xmath537 be the dual basis of @xmath532 under the identification of @xmath538 with @xmath539 using the inner product @xmath540 . in this basis the induced inner product @xmath531 is represented by @xmath541 . recall that @xmath456 for a maclaurin spheroid . then , from theorem [ symplectic mat ] , for @xmath546 we have : @xmath547 using , the matrix @xmath548 is given by @xmath549 since for a maclaurin spheroid @xmath501 it is immediate from the definition of @xmath550 in theorem [ symplectic mat ] that @xmath550 is the zero matrix . start by considering for @xmath554 the corresponding left - invariant extensions @xmath555 . we have @xmath556 for every @xmath55 . recall that @xmath557 . then according to the definition given in theorem [ symplectic mat ] we have @xmath558 also , since @xmath560 are left - invariant vector fields , the identity @xmath561=x_{uv - vu}$ ] holds and we have , from @xmath562)(f)= -\frac{\omega t}{2}{\operatorname{tr}\,}\left(\widehat{\mathbf{e_3}}(f^tf(uv - vu)+f(uv - vu)f^t)\right).\ ] ] in order to compute @xmath551 let @xmath283 and @xmath563 . the unique left - invariant vector fields extending @xmath58 and @xmath428 are @xmath564 and @xmath565 respectively . then , using together with and it is immediate to obtain @xmath566 . therefore , from theorem [ symplectic mat ] the symplectic matrix @xmath567 and its inverse are @xmath568 so the linearized vector field is @xmath569 where in our basis @xmath570 is given in . as @xmath576 and @xmath521 are positive for @xmath577 , then @xmath578 is always imaginary in that range . however , @xmath579 becomes real if @xmath522 becomes negative , which happens exactly at @xmath525 as seen from the previous stability analysis . hence , if @xmath477 the maclaurin spheroid becomes linearly unstable , therefore unstable . this loss of stability corresponds exactly to a collision at 0 of @xmath579 for @xmath580 , which passes from being pure imaginary to be real . in riemann s work @xcite some conclusions were made concerning the stability of maclaurin spheroids by studying the existence of a minimum of a certain function @xmath9 . the existence of this minimum was not done by studying the second variation of @xmath9 . riemann even says in page 188 : `` the direct analysis of the second variation of @xmath9 when the first variation vanishes would be very complicated ; we can however decide if this function has a minimum by the following form : ... '' . he follows with the analysis of the behavior of @xmath9 . his final conclusion on the stability ends the paragraph 9 of his paper and is the following : from this study it follows that the case of a rotation of an oblate ellipsoid , around its shortest axis , case already known to maclaurin , can only be unstable if the relation between the shortest axis with the others is less than 0.303327 ... " . we note that if the relation between the shortest axis of the maclaurin spheroid and the others is @xmath581 this is equivalent to say that the eccentricity @xmath582 . the value @xmath583 obtained by riemann follows from his study on the existence of oblate spheroids in pages 184185 of @xcite , namely as the root of the last displayed equation in page 184 . we remark that this equation is equivalent to the equation @xmath584 where @xmath522 is as in the proof of ( i ) in theorem [ maclaurin stability ] . indeed @xmath585 taking @xmath586 and @xmath587 one has @xmath588 the equation is the same one appearing in riemann s paper . with respect to the computation of the root of this equation riemann just says : `` ... this equation has , for @xmath589 between @xmath590 and @xmath591 , the unique root @xmath592 ... '' . indeed , this is equivalent to say that @xmath593 \2 . concerning the stability of maclaurin spheroids , it is shown in liapunov s work @xcite that under the hypothesis of the preservation of the ellipsoidal shape ( setup we used ) the value @xmath525 of loss of stability is the same as in riemann s work , but if this hypothesis is dropped then the maclaurin spheroid is only stable for @xmath594 with @xmath595 . the point @xmath596 is exactly the point where the family of maclaurin spheroids bifurcates into a branch of ellipsoids with 3 distinct axes lengths ( jacobi ellipsoids ) . we refer the reader to pages 52 and 61 - 63 of @xcite . all the qualitative properties , including stability , of two relative equilibria lying in the same orbit of the symmetry group are the same . therefore , in view of remark [ transversal adjoint ] , in this subsection we analyze the nonlinear stability of the @xmath317 family of transversal spheroids and the main result , theorem [ transversal stability ] , will follow for both families . we will set @xmath597 , @xmath598 and @xmath599 for notational simplicity . also , to keep the notation consistent with the proof of theorem [ theorem existence ] , we will set @xmath372 . in this case the computation of the splitting is simplified due to the fact that @xmath600 and therefore @xmath601 . introducing the vectors @xmath602 , we choose @xmath603 these subspaces are obviously invariant under the action of @xmath604 . with respect to the basis @xmath605 for @xmath157 we have , @xmath606.\ ] ] or in terms of the eccentricity of a prolate spheroid , @xmath607.\ ] ] it follows immediately that @xmath152 and @xmath153 are orthogonal with respect to @xmath158 , so our choice of the splitting @xmath608 is correct . in this basis the orthogonal velocity is @xmath609 with @xmath610 and @xmath97 as in the @xmath317 family in theorem [ theorem existence ] . recall that the momentum of a transversal spheroid is @xmath612 where @xmath613 $ ] and @xmath614 $ ] . from lemma [ transadjoint ] , and recalling that @xmath461 , we have , for @xmath615 : @xmath616 this is zero iff @xmath617 , with @xmath618 and therefore @xmath619 . a basis for @xmath416 is given by @xmath620 . now , proceeding as for the maclaurin spheroid , we compute @xmath621 @xmath622 differentiating we find @xmath630 in order for @xmath631 all the above expressions must vanish , which happens if and only if @xmath632 and @xmath633 , with @xmath634 . therefore we can chose a basis for the space of internal variations @xmath625 such that any element @xmath635 belonging to it has components @xmath636 with the parametrization @xmath637.\ ] ] it is straightforward to obtain @xmath638 then the correction term , for @xmath639 , is given by : @xmath640 for the restriction of the hessian of @xmath641 at @xmath60 we compute first the following second variations putting all the contributions together , and substituting the integral expressions for the derivatives of the potential and the values of @xmath187 and @xmath610 given in theorem [ theorem existence ] , we find that the matrix representing @xmath646 in the fixed basis of @xmath625 is block diagonal , with a @xmath647 symmetric block in the first two components , and the other block given by the coefficient of @xmath648 , say @xmath649 , which has the following expression : @xmath650.\ ] ] it is clear that @xmath649 is positive since @xmath651 is positive in @xmath523 and @xmath214 and @xmath215 are always positive . the matrix @xmath138 has real eigenvalues . the trace and the determinant of @xmath138 are : @xmath654\ ] ] and @xmath655.\end{aligned}\ ] ] these two quantities are positive in the interval @xmath656 . their plots are displayed in figure [ prolateplots ] . therefore , both families of transversal spheroids are stable for all eccentricities .	we apply geometric techniques to obtain the necessary and sufficient conditions on the existence and nonlinear stability of self - gravitating riemann ellipsoids having at least two equal axes .
with our rapidly expanding knowledge of the galactic center ( see , e.g. , melia 1999 ) , we now have a real hope of unraveling how the blrs in agns are produced , and how they evolve spatially and temporally . an understanding of the structure of the blr in turn is important in determining how the gaseous environment behaves near the central engine , presumably a supermassive black hole ( smbh ) , in these sources . unlike the galactic center , however , where we know the mass very accurately ( e.g. , @xcite ; @xcite ) , we do not yet have a reliable estimate of @xmath0 for these objects . but since all of the relevant length scales depend on the central mass , one may reasonably expect that a careful modeling of these environments with temporally resolved line profiles can provide us with a tightly constrained range of values for @xmath0 . reverberation methods ( e.g. , @xcite ; @xcite ) have become popular tools for estimating the blr size in agns . the cross - correlation function ( ccf ) centroid @xmath4 for a given line , typically measured in days , represents a delay between a change in the continuum flux and a corresponding ( presumably causal ) change in the line flux . from it , we can infer the responsivity - weighted mean radius , @xmath5 , of the line - emitting gas , assuming the delay is due to light - travel time effects ( @xcite ) . if we make the very reasonable assumption that the central mass dominates the motion of the blr gas , the velocity dispersion @xmath6 of this line can then be used to estimate the central mass , @xmath7 , where @xmath8 is a corrective factor of order unity representing our uncertainty in the geometry and isotropy of the blr . an example of how this works in a simplified case , where the velocity distribution of the gas is such that the line profile in the root mean square ( rms ) spectrum assumes a roughly gaussian shape , has recently been studied by peterson & wandel ( 1999 ) . using the uv and optical data for ngc 5548 from the extensive monitoring campaigns conducted by the agn watch consortium ( see , e.g. , @xcite ; @xcite ) , they computed the rms spectrum from the set of individual spectra . the line profiles in the rms spectrum represent those portions of the emission lines that are actually varying in response to the continuum fluctuations , and therefore correspond to the ccf measurements . using this method , the fwhm of each line is measured directly from the line profile , and a velocity dispersion is calculated according to the relation @xmath9 . using the reported values of @xmath4 , they obtained a central mass in this source of @xmath10 . however , although estimating the velocity dispersion from the fwhm is valid in the case of a gaussian or other regular ( e.g. , laplacian ) line profile , this method can not be used in the general case where the line profile is highly irregular . in this paper , we introduce a more generally applicable method to calculate the velocity dispersion directly , and show that our results agree with the fwhm procedure in the limit of a gaussian profile . we avoid the simplifying assumptions concerning the line shape , which allows us to significantly broaden the class of objects for which the central mass may be estimated using reverberation methods . we have assembled and analyzed the uv data obtained with agn watch observations of five agns : 3c 390.3 ( @xcite ) , fairall 9 ( @xcite ) , ngc 3783 ( @xcite ) , ngc 5548 during the @xmath11 fos campaign ( @xcite ) , and ngc 7469 ( @xcite ) . the first object in this list is classified as a broad - line radio galaxy , whereas the others are seyfert 1 galaxies . the mean and rms spectra corresponding to each data set have been determined according to the procedure described in rodriguez - pascual , et al . ( 1997 ) , and are shown for each of the five sources individually in figures 1 and 2 . these spectra have already appeared in the references cited above , but we reproduce them here as a class to demonstrate the range of line profiles one typically encounters with this process , and also to show the wavelength band used for each of the lines in our analysis ( see below ) . we have selected the ly@xmath12 and @xmath3 lines for our analysis because they typically have the strongest signal - to - noise ratios . in those cases where the ( redshifted ) ly@xmath1 line has been contaminated by geocoronal emission , we have attempted to set the lower bound of the wavelength bins beyond the contamination region . however , the ly@xmath1 lines in ngc 5548 and ngc 7469 still show significant contamination ( see fig . [ fig:2spectra ] ) . in all these objects , the ly@xmath1 line is further contaminated by @xmath13 , but no attempt to deblend these two lines has been made here . the net effect of this is to increase our estimate of the ly@xmath1 velocity dispersion . these contaminants make the ly@xmath1 measurements much less reliable than the measurements , which are only slightly affected by , e.g. , @xmath14 . these broad line profiles are understood to be produced by the composite emissivity of a broad line region ( blr ) comprised of distinct clouds , which constitute a distribution in velocity around the central mass concentration . along any given line - of - sight , say the @xmath15-axis , the velocity dispersion of these emitters is defined as @xmath16 where clearly @xmath17 denotes the @xmath15-component of the total velocity vector @xmath18 . then , assuming that @xmath19 , we can approximate the doppler - shift formula as @xmath20 , where @xmath21 is the wavelength in the emitter s frame . thus , as long as the velocity field is isotropic ( i.e. , @xmath22 ) , this relation can be used to replace @xmath17 in equation ( [ eq : sigma_z ] ) , and thereby to obtain an expression for @xmath6 in terms of observable quantities : @xmath23 the mean values of @xmath24 and @xmath25 appearing on the right hand side of this expression are often calculated from the data assuming that the line profiles are gaussian , for which a determination of the fwhm is then sufficient . to keep this procedure as general as possible , we here instead calculate the expectation ( or mean ) value @xmath26 of any given function @xmath27 of @xmath25 using a flux - weighted sum : @xmath28 where @xmath29 is the ( continuum - subtracted ) line flux at wavelength @xmath25 and the summation is taken over the appropriate wavelength bins @xmath30 . the uncertainty in the calculated value of @xmath31 is then estimated by assigning the highest and lowest plausible values for the underlying continuum . in our calculation of @xmath0 , we have used the values of @xmath4 published in the references cited above for the 5 sources considered here . the uncertainty in the value of @xmath4 is estimated using the monte carlo techniques described in white & peterson ( 1994 ) ; however , rather than using a thin shell model for the blr , we have instead assumed an extended blr consistent with the observations and the general inferences drawn from modeling ( see , e.g. , @xcite ) . we chose the line response function to be @xmath32 over the range of radii @xmath33 , where @xmath34 is the cloud emission anisotropy and @xmath35 and @xmath36 are the inner and outer radial extents , respectively , of the blr . for example , a @xmath37 that is fairly constant over all blr radii is consistent with the density and velocity fields required to model the @xmath3 line in fairall 9 ( @xcite ) . the inner extent of the blr is set to correspond with the @xmath14 cross - correlation lag ( typically having the shortest lag of any of the broad emission lines ) , and the outer extent is optimized such that a histogram of modeled line delays peaks at the observed line lag ( see @xcite ) . we used the values of @xmath38 for ly@xmath39 and @xmath40 for @xmath3 calculated by ferland , et al . ( 1992 ) . table [ tbl-1 ] summarizes the results of these calculations . our estimates of the errors associated with @xmath4 are much higher than those obtained using a thin - shell blr model . this is intuitively correct , since we would expect an extended blr to show a much broader , flatter ccf . in most cases , our extended blr model also yields larger error estimates for @xmath4 than the model - independent method described in peterson , et al . ( 1998 ) , which estimates errors based on uncertainties due to the observational sampling and the flux measurements only . this suggests that the latter method underestimates the actual uncertainty if the blr is indeed spatially extended . the sources of error in @xmath31 include the uncertain fitting of the continuum level , the blending of the broad line wings with the continuum , and line contamination ( as discussed above ) . since only the former is included in our error estimates , the @xmath31 errors we quote here are probably underestimated . we also restate that the values of @xmath0 listed in table [ tbl-1 ] are subject to corrective factors of order unity , stemming from our assumption of an isotropic velocity field . the last column in table [ tbl-1 ] lists the masses estimated by wandel , et al . ( 1999 ) for the same objects using h@xmath41 reverberation data . their results assume a gaussian line profile ( usually a good approximation for the h@xmath41 lines ) and use the fwhm to estimate the velocity dispersion . they use the model - independent method of peterson , et al . ( 1998 ) to estimate the uncertainty in @xmath4 , so their reported mass errors may be underestimated ( as discussed above ) . perhaps the most striking inference we can draw from an inspection of table [ tbl-1 ] is that the masses calculated here are considerably lower than those estimated using older reverberation data . for example , koratkar & gaskell ( 1991b ) predicted masses of @xmath42 , @xmath43 , and @xmath44 for ngc 3783 , ngc 5548 , and fairall 9 , respectively . the main discrepancy seems to be that the higher ( and regularly - spaced ) sampling rate adopted by the agn watch has facilitated the detection of shorter time scale variations and cross - correlation lags for these objects . an additional effect stems from our use of the rms spectrum rather than the mean spectrum to estimate @xmath45 or @xmath31 . a cursory glance at figures [ fig:3spectra ] and [ fig:2spectra ] suggests that the velocity dispersions of the rms spectral lines are _ smaller _ than those of the mean spectra . the physical interpretation of this effect is not yet clear , though it may be signaling that the emission from the innermost blr gas does not vary strongly in response to continuum changes . our mass estimates agree with the wandel , et al . ( 1999 ) results within error bars in all cases except the ly@xmath1 results of 3c 390.3 and ngc 3783 . the ngc 5548 mass estimate is also consistent with the result obtained by peterson & wandel ( 1999 ) using the same spectral data , though with the assumption that the lines are gaussian . collier et al . ( 1998 ) estimated the mass of ngc 7469 using the spectral variability in the optical regime combined with a measurement of @xmath45 for the h@xmath39 and h@xmath41 lines to obtain a mass for this object of @xmath46 , which is again consistent with our result . these cases provide some support for the validity of our method , since they show a consistent mass determination for sources that have `` normal '' ( i.e. , gaussian ) line profiles . our method is clearly more general , since it appears to be applicable also to sources in which the velocity distribution of the blr does not naturally produce a gaussian line shape . a caveat one must consider with this procedure is that the error associated with the ly@xmath1 fit can be significant . the mass estimates made using the ly@xmath1 data are higher than those based on the line in all cases except fairall 9 . as we discussed above , this is probably due to a contamination of the line , which suggests that the results may be more reliable . it is possible that this problem may be alleviated by line deblending techniques , which we have not incorporated into our analysis here . our results support the argument made by peterson & wandel ( 1999 ) that a very large mass must be present within a distance of a few light - days from the centers of many agns . for the class of @xmath47 objects in our sample , this mass is at least @xmath48 , which is rather compelling evidence for the presence of a point - like object . this mass density is even higher than that in the galactic center , where we are reasonably sure that the enclosed mass within @xmath49 pc of the nucleus is close to @xmath50 . alternative models for the latter that invoke a distribution of dark objects , such as neutron stars , white dwarfs , or @xmath51 black holes , can now be ruled out on the basis that in equilibrium , such a distribution would have a core radius significantly larger than the observed limiting size of @xmath49 pc . in the sample of agns considered here , this exclusion of a distributed dark matter contribution to the enclosed mass appears to be even stronger ( cf . maoz 1998 ) . one of the most important results presented here is the relatively accurate determination of the central point mass in these objects . with the exclusion of 3c 390.3 , which appears to belong to a different mass class , the masses of the other four agns ( all seyfert i s ) are very close to each other and are near the low end of the mass scale for agns this characteristic , and the fact that the smbh at the galactic center has a mass within a factor of @xmath52 or so of this range , hints at a possibly similar formation and evolutionary history for these nuclei . in the galactic center , the power generated by the supermassive black hole seems to be the quasi - steady accretion of plasma ejected by wind - producing stars in the circumnuclear environment , with a rate @xmath53 yr@xmath54 ( coker & melia 1997 ) . thus , given a central mass @xmath55 , the characteristic time @xmath56 billion years suggests that the black hole sgr a * may have grown to its current size gradually over a hubble time via the steady accretion of gas from its environment . we speculate that the nuclei in the 4 seyfert i s we have considered here may themselves have formed and evolved in a similar fashion , and that the range in masses may therefore be due principally to a variation of the nuclear gas density from which the black holes accrete . this work was supported by an nsf graduate fellowship at the university of arizona , by a sir thomas lyle fellowship for distinguished overseas visitors at the university of melbourne , and by nasa grant nag58239 . we thank bradley peterson for helpful discussions .	we here estimate the mass @xmath0 of the central object in five active galactic nuclei ( agns ) using the most recent reverberation data obtained by the agn watch consortium . the cross - correlation function ( ccf ) centroids of the broad ly@xmath1 @xmath2 and @xmath3 lines are used to estimate the size of the broad line region ( blr ) in these sources . we calculate the velocity dispersions of these lines in the root mean square ( rms ) spectra , and then use our results to estimate @xmath0 . we argue that our technique of calculating the velocity dispersion should work in the general case of an arbitrary line profile , unlike methods that depend on the measurement of the full - width at half - maximum ( fwhm ) of the broad line . we also show that our results agree with the fwhm method in the limit of a normal ( gaussian ) line profile . the masses calculated here are considerably smaller than those calculated with the previous generation of reverberation data . submitted to the astrophysical journal
the main feature of a lvy - type density distribution @xmath3 is the slow , power - law , decay of its tail . more precisely , for large @xmath5 , @xmath6 with @xmath7 . note that the second moment of @xmath3 diverges for all @xmath8 and if @xmath9 also the first moment diverges . this kind of distributions are also known as @xmath8-stable distributions @xcite . random processes characterized by probability densities with a long tail ( lvy - type processes ) have been found and studied in very different phenomena and fields such as biology , economy , and physics . among many of recently studied systems showing lvy - type processes we can mention : animal foraging @xcite , human mobility @xcite , earthquake statistics @xcite , mosquitoes flight under epidemiological modeling @xcite , and light transmission through a disordered glass @xcite . see also @xcite for a compilation of systems displaying lvy flights . in particular , to help us to introduce later the main model system of this study , i.e. the _ lvy map _ , we want to describe in some detail a simple dynamical model characterized by lvy processes : the _ ripple billiard_. the ripple billiard , see for example chapter 6 of @xcite , consists of two walls : one flat at @xmath10 and a rippled one given by the function @xmath11 ; here @xmath12 is the average width of the billiard and @xmath13 the ripple amplitude , see fig . [ fig1 ] . an attractive feature of the ripple billiard is that its classical phase space undergoes the generic transition to global chaos as the amplitude of the cosine function increases . then , results from the analysis of this system are applicable to a large class of systems , namely non - degenerate , non - integrable hamiltonians @xcite . moreover , the dynamics of classical particles inside the ripple billiard can be well approximated by a two - dimensional ( 2d ) poincar map @xmath14 between successive collisions with the rippled boundary @xcite @xmath15 , where @xmath16 is the angle the particle s trajectory makes with the @xmath17-axis just before the @xmath18th bounce with the rippled boundary at @xmath19 . map @xmath14 can be easily derived and , after the assumptions @xmath20 and @xmath21 , it gets the simple form @xmath22 as an example , in fig . [ fig2](a ) we plot the poincar map @xmath14 for @xmath23 and @xmath24 . it is clear from this plot that this combination of geometrical parameters produces ergodic dynamics ( also known as global chaos ) . notice that in this figure we have plotted the variable @xmath17 modulus @xmath25 , as usual for this kind of 2d maps ; with this , we can globally visualize the map dynamics in a single plot but we may loose important information . , see eq . ( [ m]).,width=264 ] among the dynamical information which is lost when applying @xmath26 to a map such as @xmath14 , we can mention the length of paths between successive map iterations @xmath27 , i.e. the length between two successive collisions with the rippled boundary of the billiard . in fact , in fig . [ fig2](c ) we present @xmath3 for the same parameters used to construct fig . [ fig2](a ) . from this figure we can clearly see that : ( i ) even though most of the paths @xmath5 produced by map @xmath14 are short ( i.e. @xmath3 is highly peaked at @xmath28 ) , there is a non - negligible probability for very large values of @xmath5 to occur : notice that the values of @xmath29 at the edges of fig . [ fig2](c ) mean that a particle has traveled about 160 periods of the rippled billiard between two successive collisions with the rippled boundary and ; ( ii ) @xmath3 decays as a power - law similar to eq . ( [ levy ] ) . these two facts are explicit evidences of lvy flights in the dynamics of map @xmath14 . thus , the following question becomes pertinent : can we provide an analytic expression for the shape of @xmath3 given the simple form of map @xmath14 ? fortunately , the answer is positive as we will show it below . if we consider the dynamics of map @xmath14 to be in the regime of full chaos then a single trajectory can explore the full available phase space homogeneously , as shown in fig . [ fig2](a ) , so @xmath30 is constant and equal to @xmath31 , as verified in fig . [ fig2](b ) . also , from the second equation in map @xmath14 we obtain @xmath32 . thus , using @xmath33 , we can write @xmath34 which is in fact a lvy - type probability distribution function with @xmath1 ; compare with eq . ( [ levy ] ) . then , in fig . [ fig2](c ) we plot eq . ( [ levym ] ) ( as the red full line ) together with the numerically obtained @xmath3 and observe a very good correspondence making clear the existence of lvy processes , characterized by the power - law decay @xmath1 , in the dynamics of the rippled billiard . in fact , the origin of the lvy - type probability distribution of eq . ( [ levym ] ) for the lengths @xmath5 in the ripple billiard is the existence of lvy flights . since a typical chaotic trajectory fills the available phase space uniformly , see fig . [ fig2](a ) , then all angles @xmath35 are equally probable ; however , different angles produce quite different lengths @xmath5 . for example , an angle @xmath36 very close to @xmath37 corresponds to a very short length @xmath38 , see fig . [ fig1 ] . while @xmath36 tending to zero or @xmath39 produce trajectories which are nearly parallel to the @xmath17-axis that may travel very long distances between successive collisions with the ripple boundary : in such case @xmath40 . these _ grazing _ trajectories are indeed lvy flights , known to produce heavy - tailed distribution functions @xcite ; for the ripple billiard the lvy flights produce eq . ( [ levym ] ) as derived above . moreover , grazing trajectories in the ripple billiard have been found to play a prominent role when defining the classical analogs of the quantum _ structure of eigenstates _ and _ local density of states _ @xcite . equation ( [ levym ] ) is already an interesting result on the dynamics of the rippled billiard ( and of general chaotic extended billiards with infinite horizon ) that deserves additional attention , however our goal here is different . since now we know that map @xmath14 produces lvy flights characterized by @xmath1 we ask ourselves : can we propose a general 2d non - linear map where @xmath8 can be included as a parameter ? more generally , can we construct the map @xmath2 able to produce lvy flights characterized by @xmath4 ? indeed , in the following section we elaborate on these questions . we introduce the _ lvy map _ @xmath2 by following the opposite procedure we used above to obtain the distribution function of eq . ( [ levym ] ) from map @xmath14 . let us * consider the 2d map @xmath41 to have the same iteration relation for the angle @xmath36 as map @xmath14 , see eq . ( [ m ] ) , * assume the map @xmath2 to be in a regime of global chaos , such that @xmath42 and * demand the variable @xmath43 from map @xmath2 to be characterized by the lvy - type density distribution function @xmath44 where @xmath4 and @xmath45 is a normalization constant . then , @xmath46 provides @xmath47 . therefore , we define the _ lvy map _ as @xmath48 where @xmath13 , @xmath49 , and @xmath8 are the map parameters . we have introduced the absolute value in the second equation of @xmath2 to avoid fractional powers of negative angles . this , in turn , makes all _ lengths _ @xmath5 to be positive . notice that for @xmath1 and @xmath50 , where @xmath51 , we recover map @xmath14 from @xmath2 ( with @xmath52 ) . we also note that @xmath2 has a similar form than the maps studied in refs . @xcite in the sense that the function @xmath53 , in the second line of the map @xmath2 , is inverse proportional to @xmath54 to a non - integer power . below we will focus our attention on map @xmath2 with the parameter @xmath8 into the interval @xmath4 because our motivation is to construct a map able to produce pseudo - random variables distributed according to @xmath8-stable distributions . however , the parameter @xmath8 may also take values outside this interval . in general , depending on the values of the parameters @xmath55 , the dynamics of the lvy map may be integrable , mixed ( where the phase space contains periodic islands surrounded by chaotic seas which may be limited by invariant spanning curves ) , or ergodic . that is , the classical phase space of map @xmath2 develops the generic transition to global chaos ( not shown here ) . however , for eq . ( [ pofteta ] ) to be valid the map dynamics must be ergodic . therefore , below , by applying chirikov s overlapping resonance criteria we shall identify the onset of global chaos as a function of the parameters of the lvy map . following @xcite we linearize @xmath2 around the period - one fixed points @xmath56 , which are defined through @xmath57 this condition provides @xmath58 then , for an angle close to @xmath59 we can write @xmath60 getting @xmath61 thus , @xmath62 in addition , for @xmath17 we have @xmath63 \nonumber \\ & = & x_n + 2\pi m [ 1-(\gamma/2\pi m)^{-\alpha}\delta\theta_{n+1 } ] \nonumber \\ & = & x_n - \gamma^{-\alpha } ( 2\pi m)^{\alpha+1}\delta\theta_{n+1 } \ . \label{xn1}\end{aligned}\ ] ] finally , by substituting the new angle @xmath64 in ( [ deltateta ] ) and ( [ xn1 ] ) we can write the linearized map @xmath65 where @xmath66 and @xmath17 , respectively , play the role of action - angle variables in the _ standard map _ @xcite with @xmath67 being the stochasticity parameter . chirikov s overlapping resonance criteria predicts the transition to global chaos for @xmath68 , where @xmath69 @xcite . global chaos means that chaotic regions are interconnected over the whole phase space ( stability islands may still exist but are sufficiently small that the chaotic sea extends throughout the vast majority of phase space ) . this criteria for the lvy map reads as @xmath70 in fact , to get eq . ( [ caosg ] ) from eq . ( [ k ] ) we have applied the resonance criteria to the period - one fixed point corresponding to @xmath71 , see eq . ( [ fp ] ) , which is the fixed point having the largest @xmath36 ( i.e. it is located highest in phase space ) and the one closer to the last invariant spanning curve bounding the diffusion of the variable @xmath36 . indeed , we have verified that the phase space of @xmath2 is ergodic if condition ( [ caosg ] ) is satisfied ( not shown here ) . moreover , in figs . [ fig4](a - d ) we plot the phase distribution functions @xmath30 for the lvy map with @xmath72 corresponding to @xmath73 , 1/2 , 1 , and 3/2 . from these figures , it is clear that @xmath30 is certainly a constant distribution . in particular , note that with @xmath72 condition ( [ caosg ] ) is satisfied for any @xmath8 , so we shall use this set of parameter values in all figures below . and ( e - h ) length distributions @xmath74 for the lvy map @xmath2 with @xmath72 . ( a , e ) @xmath73 , ( b , f ) @xmath75 , ( c , g ) @xmath1 , and ( d , h ) @xmath76 . to construct each histogram a single initial condition @xmath77 was iterated @xmath78 times . the red full curve in ( e - h ) is @xmath79.,width=302 ] now we would like to verify that once @xmath80 , @xmath2 must produce lengths @xmath81 distributed according to eq . ( [ levy2 ] ) . however , we noticed that for @xmath82 the lvy map produces huge values of @xmath5 . to show this , in fig . [ fig5 ] we plot the typical value of @xmath5 , @xmath83 as a function of @xmath8 ; where we can observe that for @xmath73 the typical @xmath5 is larger than @xmath84 ( in fact , from the data we used to construct the @xmath30 of fig . [ fig4](a ) we obtained several lengths @xmath5 of the order of @xmath85 ! ) . thus , it is not practical to construct @xmath3 to test the validity of eq . ( [ levy2 ] ) itself . instead , we make the change of variable @xmath86 which leads to @xmath87 then in figs . [ fig4](e - h ) we show length distribution functions @xmath74 for the lvy map with @xmath73 , 1/2 , 1 , and 3/2 ( histograms ) . as clearly seen , the agreement between the histograms and @xmath79 ( shown as red thick lines ) is indeed excellent . it is relevant to stress that since the phase space of the lvy map is ergodic when condition ( [ caosg ] ) is satisfied , the sequences @xmath81 can then be considered as lvy - distributed pseudo - random numbers . in fact , in the next section we will show through a specific application that the lengths @xmath5 can be used in practice as random numbers . , @xmath88 , for the lvy map @xmath2 as a function of @xmath8 . @xmath72 were used . the average was performed over @xmath78 values of @xmath5 obtained by iterating @xmath2 from the single initial condition @xmath77.,width=264 ] there is a good deal of work devoted to the use of non - linear maps as pseudo - random number generators , see some examples in refs . therefore , in a similar way , we would like to use the lvy map to generate pseudo - random numbers particularly distributed according to the lvy - type probability distribution function of eq . ( [ levy2 ] ) . however , instead of analyzing the randomness of the sequences @xmath81 produced by @xmath2 , here we will show that these numbers can be successfully used already in a specific application : we shall compute transmission through one - dimensional ( 1d ) disordered wires . recently , the electron transport through 1d quantum wires with lvy - type disorder was studied in refs . there , it was found that the average ( over different disorder realizations ) of the logarithm of the dimensionless conductance @xmath89 behaves as @xmath90 where @xmath91 is a _ length _ that depends on the wire model . for example , for a wire represented as a sequence of potential barriers with random lengths , @xmath92 @xcite ; where @xmath93 is the length of the @xmath18th barrier in the wave propagation direction . while for a wire represented by the 1d anderson model with off - diagonal disorder , @xmath94 @xcite ; where @xmath95 is the hopping integral between the sites @xmath18 and @xmath96 . here , we use the 1d anderson model with off - diagonal disorder to represent 1d quantum wires ( see details in the appendix ) where the hopping integrals @xmath95 are , in fact , the pseudo - random lengths @xmath5 generated by our lvy map . then , in fig . [ fig6 ] we plot @xmath97 as a function of @xmath91 for the 1d anderson model with lvy - type disorder characterized by @xmath73 , 1/2 , 1 , and 3/2 . we have computed the dimensionless conductance by the use of the effective hamiltonian approach ( see details in the appendix ) . in fig . [ fig6 ] we are using @xmath98 to normalize @xmath91 to be able to show curves corresponding to different values of @xmath8 into the same figure panel . also , in fig . [ fig6 ] we are including fittings of the curves @xmath97 vs. @xmath91 with eq . ( [ g ] ) , see red dashed lines , which certainly show the _ anomalous _ conductance behavior predicted in refs . @xcite . therefore , validating in this way the use of the lvy map as a pseudo - random number generator . as a function of @xmath99 for the 1d anderson model with off - diagonal lvy - type disorder characterized by @xmath8 . we used an incoming wave with energy @xmath100 . the dashed lines are fittings of the data with eq . ( [ g ] ) . each symbol was calculated using @xmath84 wire realizations . each wire realization is constructed from a single sequence of lengths @xmath81 generated by map @xmath2 having random initial conditions uniformly distributed in the intervals @xmath101 and @xmath102.,width=226 ] in this paper we have introduced the so - called _ lvy map _ : a two - dimensional nonlinear map characterized by tunable lvy flights . indeed it is described by a 2d nonlinear and area preserving map with a control parameter driving two important transitions : ( i ) integrability ( @xmath103 ) to non - integrability ( @xmath104 ) and ; ( ii ) local chaos with @xmath105 to globally chaotic dynamics with @xmath106 . we have applied chirikov s overlapping resonance criteria to identify the onset of global chaos as a function of the parameters of the map therefore reaching to condition ( ii ) as described on the line above . in this way we stated the requirements under which the lvy map could be used as a lvy pseudo - random number generator and confirmed its effectiveness by computing scattering properties of disordered wires . in sect . [ conductance ] we have considered 1d tight - binding chains of size @xmath107 described by the hamiltonian @xmath108 where @xmath109 are on - site potentials that we set to zero and @xmath110 are hopping amplitudes connecting nearest sites . here , @xmath111 . we open the 1d chains by attaching two single - mode semi - infinite leads to the opposite sites on the 1d samples . each lead is described by the 1d semi - infinite tight - binding hamiltonian @xmath112 then , following the _ effective hamiltonian approach _ , the scattering matrix ( @xmath113-matrix ) has the form @xcite @xmath114 where @xmath115 , @xmath116 , @xmath117 , and @xmath118 are transmission and reflection amplitudes ; @xmath119 is the @xmath120 unit matrix , @xmath121 is the wave vector supported in the leads , and @xmath122 is an effective energy - dependent non - hermitian hamiltonian given by @xmath123 above , @xmath124 is a @xmath125 matrix with elements @xmath126^{1/2}(\delta_{1,1}+\delta_{l,2})$ ] . * acknowledgments . * j.a.m .- b is grateful to fapesp ( 2013/14655 - 9 ) brazilian agency ; partial support form viep - buap grant mebj - exc14-i and fondo institucional pifca 2013 ( buap - ca-169 ) is also acknowledged . b also thanks the warm hospitality at departamento de fsica at unesp rio claro , where this work was mostly developed . j.a.o thanks prope / fundunesp / unesp . thanks to fapesp ( 2012/23688 - 5 ) , cnpq , and capes , brazilian agencies . g. a. luna - acosta , k. na , l. e. reichl , and a. krokhin , phys . e * 53 * , 3271 ( 1996 ) ; a. j. martnez - mendoza , j. a. mndez - bermdez , g. a. luna - acosta , and n. atenco - analco , rev . mex . fis . s * 58 * , 6 ( 2012 ) . c. mahaux and h. a weidenmller , _ shell model approach in nuclear reactions _ , ( north - holland , amsterdam,1969 ) ; j. j. m. verbaarschot , h. a. weidenmller , and m. r. zirnbauer , phys . rep . * 129 * , 367 ( 1985 ) ; i. rotter , rep . . phys . * 54 * , 635 ( 1991 ) .	once recognizing that point particles moving inside the extended version of the rippled billiard perform lvy flights characterized by a lvy - type distribution @xmath0 with @xmath1 , we derive a generalized two - dimensional non - linear map @xmath2 able to produce lvy flights described by @xmath3 with @xmath4 . due to this property , we name @xmath2 as the _ lvy map_. then , by applying chirikov s overlapping resonance criteria we are able to identify the onset of global chaos as a function of the parameters of the map . with this , we state the conditions under which the lvy map could be used as a lvy pseudo - random number generator and , furthermore , confirm its applicability by computing scattering properties of disordered wires .
the present talk is devoted to the analysis of the family of one - dimensional quasi - periodic schrdinger operators acting on @xmath2 defined by @xmath3 we assume that ( h1 ) : : @xmath4 is a non constant , locally square integrable , @xmath5-periodic function ; ( h2 ) : : @xmath6 is a small positive number ; ( h3 ) : : @xmath7 is a real parameter ; ( h4 ) : : @xmath8 is a potential that is real analytic in a conic neighboring the real axis that admits a limit at @xmath0 and at @xmath1 ; the precise assumption is stated in section [ sec : assumptions ] . as @xmath6 is small , the operator is a slow perturbation of the periodic schrdinger operator @xmath9 acting on @xmath2 . + when @xmath10 , the operator @xmath11 is independent of @xmath7 ( up to a unitary equivalence , a translation actually ) and becomes a semi - classical schrdinger operator ( by a simple rescaling of the variable @xmath12 ) . the resonances for such operators have been the subject of a vast literature over the last 20 years ; a detailed review and many references can be found in the lecture notes @xcite and in the papers @xcite . + the problem of computing the resonances for slowly varying perturbations of a periodic schrdinger operator has been much less studied . the papers @xcite deal with the multi - dimensional case ; asymptotic formulas are obtained for the real part of the resonances . this is done using trace formulas and the imaginary parts of the resonances have not been computed . the papers @xcite are devoted to the study of stark - wannier resonances in a small electric field . it is well known that the eigenvalues and resonances for @xmath11 near an energy @xmath13 depends very much on the region of energy one is studying . les us assume @xmath8 tends to @xmath14 at both ends of the real axis . then , the absolutely continuous spectrum of @xmath11 is the spectrum of @xmath15 . in this case , we compute the resonances of @xmath11 near energies inside the spectrum of @xmath15 . actually , for @xmath8 , we also consider the case when the limits of @xmath8 at @xmath0 and @xmath1 differ . we obtain that the real parts of the resonances are given by a quantization condition interpreted naturally in the adiabatic phase space ( see section [ sec : heuristics ] ) . the imaginary parts of the resonances are then given by an exponentially small tunneling coefficient . though the results depend crucially on the case dealt with , from the technical point of view , both cases can be dealt with in a similar fashion . to study , we use the asymptotic method developed in @xcite for the analysis of slow perturbations of one - dimensional periodic equations . we first recall some of elements of the spectral theory of one - dimensional periodic schrdinger operator @xmath15 ; then , we present our results . for more details and proofs we refer to section [ s3 ] and to @xcite . the spectrum of the operator @xmath15 defined in is a union of countably many intervals of the real axis , say @xmath16 $ ] for @xmath17 , such that @xmath18 this spectrum is purely absolutely continuous . the points @xmath19 are the eigenvalues of the self - adjoint operator obtained by considering the differential polynomial acting in @xmath20)$ ] with periodic boundary conditions ( see @xcite ) . the intervals @xmath16 $ ] , @xmath17 , are the _ spectral bands _ , and the intervals @xmath21 , @xmath22 , the _ spectral gaps_. when @xmath23 , one says that the @xmath24-th gap is _ open _ ; when @xmath25 $ ] is separated from the rest of the spectrum by open gaps , the @xmath24-th band is said to be _ isolated_. from now on , to simplify the exposition , we suppose that ( o ) : : all the gaps of the spectrum of @xmath15 are open . let @xmath26 be a non trivial solution to the periodic schrdinger equation @xmath27 such that , for some @xmath28 and all @xmath29 , @xmath30 . this solution is called a _ bloch solution _ to the equation , and @xmath31 is the _ floquet multiplier _ associated to @xmath32 . one may write @xmath33 ; then , @xmath34 is the _ bloch quasi - momentum _ of the bloch solution @xmath32 . the mapping @xmath35 is an analytic multi - valued function ; its branch points are the points @xmath36 , @xmath37 , @xmath38 , @xmath39 , @xmath40 , @xmath39 . they are all of `` square root '' type . the dispersion relation @xmath41 is the inverse of the bloch quasi - momentum . we refer to section [ s3 ] for more details on @xmath34 . we now make assumption ( h4 ) more precise . we introduce the following notation : for @xmath42 , define @xmath43 to be the cone @xmath44 we assume ( h4a ) : : there exists @xmath42 such that @xmath45 is real analytic , non constant ; ( h4b ) : : there exist @xmath46 and @xmath47 such that , for any @xmath48 @xmath49 + \sup_{\substack{z\in\mathcal{c}_{c_1}\\{{\rm re}\,}z<0 } } \left[\|z\|^s\|w(z)-w_-\|\right]<+\infty\ ] ] so @xmath8 is short range in cones neighboring @xmath0 and @xmath1 . it is well known that , when @xmath15 is the free laplace operator , this assumption guarantees that the resolvent of @xmath50 can be analytically continued from the upper half - plane through the spectrum of @xmath15 as a mapping from @xmath51 to @xmath52 ( for some @xmath53 ) ( see @xcite and references therein ) . here , for @xmath54 , we have defined @xmath55 for @xmath56 and @xmath57 , the resolvent of @xmath11 at energy @xmath13 , that is @xmath58 is a function valued in the bounded operators on @xmath2 ; moreover , under assumption ( h4 ) , a simple resolvent expansion shows that , for any @xmath59 , it is analytic in a neighborhood of @xmath60 . we prove [ thr:2 ] assume ( h1 ) ( h4 ) are satisfied . pick @xmath61 such that either @xmath62 , or @xmath63 , or both hold . + then , there exist @xmath64 , @xmath53 , @xmath65 and a complex valued function @xmath66 defined on @xmath67 such that , for @xmath68 , * the mapping @xmath69 does not vanish on @xmath70 ; * the mapping @xmath69 is analytic on @xmath71 ; * the mapping @xmath72 can be continued analytically from @xmath73 to @xmath74 as a non - vanishing bounded operator from @xmath75 to @xmath76 ; * for any @xmath77 , the mapping @xmath78 is @xmath6-periodic . following the now classical definition ( see @xcite ) , we set [ def:2 ] fix @xmath7 real and @xmath79 as in theorem [ thr:2 ] . an energy @xmath80 is a resonance for @xmath11 if it is a pole of @xmath81 i.e. if it is a zero of @xmath82 . our goal is to describe the resonances for @xmath11 near @xmath79 , an energy satisfying the assumptions of theorem [ thr:2 ] . our main interest is in computing the imaginary part of the resonances . it is well known that the existence and the positions of resonances will depend crucially on the relative position of the spectral window @xmath83 with respect to @xmath84 ( see @xcite ) . therefore , one introduces the set @xmath85 . it is a subset of the domain @xmath43 where @xmath8 is analytic . pick @xmath61 such that either @xmath62 , or @xmath63 , or both hold . a simple weyl sequence argument then shows that @xmath86 ( for any @xmath57 and @xmath87 ) . + we consider two types of situations and therefore introduce two different assumptions ( h5 ) : : the set @xmath88 can be decomposed into @xmath89 where @xmath90 and @xmath91 are two - by - two disjoint and they satisfy : + * @xmath90 and @xmath91 are either empty or semi - infinite intervals , respectively neighborhoods of @xmath1 and @xmath0 ; * the finite edges of @xmath90 and @xmath91 ( when they exist ) are not critical points of @xmath8 . and ( h6 ) : : the set @xmath88 can be decomposed into @xmath92 where @xmath90 , @xmath93 and @xmath91 are two - by - two disjoint and they satisfy : + * @xmath93 is a compact interval not reduced to a single point ; * @xmath90 and @xmath91 are as in ( h5 ) . * the finite edges of @xmath90 , @xmath91 and @xmath93 ( when they exist ) are not critical points of @xmath8 . 1.5 cm in fig . [ fig : exwh6 ] , [ fig : exwh6sec ] and [ fig : exwh6prim ] , we provide some examples of potential profiles ; we graphed @xmath94 , and , on the vertical axis represented the relevant spectral intervals for @xmath15 . 1.5 cm let us now shortly discuss these assumptions . one easily convinces oneself that , if ( h5 ) or ( h6 ) holds for some energy @xmath79 , it will hold for all real energies in a neighborhood of @xmath79 . moreover , the fact that the sets @xmath95 or @xmath96 in the decomposition of @xmath97 are empty will not depend on @xmath13 ( in this neighborhood ) . [ fig : exwh6 ] provides examples where assumption ( h5 ) holds . under assumption ( h6 ) , fig . [ fig : exwh6 ] provides some examples where both @xmath95 and @xmath96 are non empty , and fig . [ fig : exwh6prim ] , examples where either @xmath95 or @xmath96 is empty . the two assumptions ( h5 ) and ( h6 ) correspond to the simplest possible cases : in the general case , i.e. when the sets @xmath98 and @xmath84 are in a general position , there can be more than one compact connected component to the set @xmath99 . 1.5 cm + we now state our results on the resonances of @xmath11 . we begin our results with the energies a neighborhood of which do not carry resonances , namely the energies satisfying ( h5 ) . indeed , we prove [ thr:3 ] fix @xmath79 satisfying ( h5 ) . there exist @xmath65 , @xmath100 , @xmath101 , a neighborhood of @xmath79 such that , for @xmath68 , for all real @xmath7 , the set @xmath102 contains no resonance of @xmath11 . this is quite analogous to what is found in the case when @xmath15 is the free laplace operator ( see e.g. @xcite ) . to study the real energies close to which one finds resonances , we introduce a few auxiliary functions necessary to describe the resonances . the complex momentum @xmath103 is defined by @xmath104 as @xmath34 , @xmath105 is analytic and multi - valued . as the branch points of @xmath34 are the points @xmath106 , the branch points of @xmath105 satisfy @xmath107 as @xmath13 is real , the set of these points is symmetric with respect to the real axis , to the imaginary axis and it is @xmath108-periodic in @xmath7 . more details are given in section [ sec : kappa ] . + fix @xmath79 satisfying assumption ( h6 ) . by , for @xmath13 real close to @xmath79 , the ends of @xmath109 and the finite ends of @xmath95 and @xmath96 ( when they are not empty ) are branch points of @xmath105 . to fix ideas , let us define @xmath110,\quad u_-(e)=(-\infty,\zeta^-(e ) ] \quad\text{and}\quad u_+(e)=[\zeta^+(e),+\infty)\ ] ] when @xmath96 ( resp . @xmath95 ) is empty , we set @xmath111 ( resp . @xmath112 ) . + one shows that there exists a determination of the complex momentum , say @xmath113 , and a real neighborhood of @xmath79 , say @xmath114 , such that , for @xmath115 , @xmath116 $ ] for @xmath117 . one defines @xmath118.\ ] ] one shows [ le:1 ] the constant @xmath119 belongs to @xmath120 and is independent of @xmath13 in @xmath114 . the function @xmath121 can be extended to a function analytic in a complex neighborhood of @xmath79 . moreover , there exists @xmath122 such that , in this neighborhood , one has @xmath123 assume @xmath95 or @xmath96 or both are not empty . one then proves that the imaginary part of the determination @xmath113 keeps a fixed sign in the interval between @xmath95 and @xmath109 , and between @xmath109 and @xmath96 , i.e. on the intervals @xmath124 and @xmath125 . we define @xmath126 where @xmath127 is chosen so that @xmath128 and @xmath129 are positive . one proves [ le:2 ] pick @xmath130 . the function @xmath131 can be extended to a function analytic in a complex neighborhood of @xmath79 . when @xmath132 ( resp . @xmath133 ) is empty , it stays empty for @xmath13 close to @xmath79 , and one sets @xmath134 ( resp . @xmath135 ) . define the tunneling coefficients @xmath136 we prove [ thr:4 ] fix @xmath79 satisfying ( h6 ) . there exist @xmath65 , @xmath100 , @xmath101 , a neighborhood of @xmath79 , and a real analytic function @xmath137 defined on @xmath138 $ ] satisfying the uniform asymptotics @xmath139 such that , if one defines the finite sequence of points in @xmath140 , say @xmath141 , by @xmath142 then , for @xmath68 , for all real @xmath7 , the resonances of @xmath11 in @xmath102 are contained in the union of the disks @xmath143 more precisely , each of these disks contains exactly one simple resonance , say @xmath144 , that satisfies @xmath145 where @xmath146 when @xmath147 and @xmath148 is a constant . let us discuss the resonances obtained in theorem [ thr:4 ] , in particular , their behavior as functions of @xmath7 . by theorem [ thr:2 ] , the function @xmath78 is @xmath6-periodic and by theorem [ thr:4 ] , the resonances are simple , the resonances being the zeroes of @xmath82 , they are @xmath6-periodic functions of @xmath7 . theorem [ thr:4 ] says that their imaginary part does not oscillate to first order . it also gives some information about the oscillations of the real part . therefore one has to distinguish the two cases @xmath149 and @xmath150 . when @xmath149 , the quantization condition and equations and show that the real part of the resonances is monotonous in @xmath7 . but , as they come in a sequence , renumbering the resonances , they can also be seen as oscillating with amplitude roughly @xmath151 ( for @xmath122 ) . of course , to see this oscillation phenomenon , one needs to renumber the resonances . the situation is similar to that encountered when studying stark - wannier resonances ( see @xcite ) ; in this case also , the resonances still exhibit oscillations of frequency @xmath152 but the amplitude is @xmath6 ( the stark - wannier ladders ) . when @xmath149 , theorem [ thr:4 ] shows that the oscillations of the real part are at most exponentially small . under more precise assumption than those made in the present talk , one can compute the amplitude of the oscillations of the resonances ( @xcite ) . an analogous behavior was found for the eigenvalues of @xmath11 outside the essential spectrum in @xcite ; they oscillate with a frequency @xmath152 and an amplitude that is exponentially small in @xmath6 ; the amplitude was given by a complex tunneling coefficient . moving the energy @xmath13 ( for a fixed @xmath8 ) , one sees that , in some cases , one can pass continuously from assumption ( h5 ) to ( h6 ) and vice - versa . it would be interesting to study what happens to the resonances uncovered in theorem [ thr:4 ] when one crosses this transition . such a study has been done in @xcite in the case when @xmath10 . we now discuss the heuristics explaining these results . in the figures below , in the part indexed a ) , we represented the local configurations of the potential to show the assumption made on the energy @xmath13 in the different cases . in the part indexed b ) , we represented the phase space picture of the iso - energy curve i.e. the sets @xmath153 . it is @xmath108-periodic in the @xmath105-direction ( i.e. the vertical direction ) so we depicted only two periods . 1.5 cm on the same picture , we also show some loops that are dashed . these are special loops in @xmath154 that join the connected components of @xmath153 . + assumption ( h5 ) just means that the real iso - energy curve is empty , in which case there are no resonances as we saw . we now assume ( h6 ) holds . in fig . [ fig : esp1 ] , the compact connected components of the iso - energy curve consist of a single torus per periodicity cell . the torus gives rise to the quantization condition : the corresponding phase is obtained ( to first order ) by integrating the canonical one - form @xmath155 along this torus . the fact that this corresponds to a resonance rather than to a eigenvalue is due to the fact that the iso - energy curve has components leading to infinity . the dashed curve represents the instanton linking the two . to compute the action determining the lifetime of the resonance , one integrates the canonical one - form @xmath155 along this curve . 1.5 cm in fig . [ fig : esp2 ] , the compact connected components of the iso - energy curve are absent . nevertheless , one can compute a phase ( to first order ) by integrating the canonical one - form @xmath155 along a period of the central curve i.e. the one that is extended in the @xmath105-direction . when the connected component extended in the @xmath7-directions are absent , one can show that such a phase does not give rise to eigenvalues . the presence of these components leading to infinity in the @xmath7-direction gives rise to resonances . again , to compute the action determining the lifetime of the resonance , one integrates the canonical one - form @xmath155 along the dashed curve . in @xcite and @xcite , we have developed a new asymptotic method to study solutions to an adiabatically perturbed periodic schrdinger equation i.e. , to study solutions of the equation @xmath156 in the limit @xmath157 . the function @xmath158 is an analytic function in a neighborhood of the real axis . the main idea of the method is to get the information on the behavior of the solutions in @xmath12 from the study of their behavior on the complex plane of @xmath7 . the natural condition allowing to relate the behavior in @xmath7 to the behavior in @xmath12 is the consistency condition @xmath159 one can construct solutions to both and that have a simple asymptotic behavior on certain domains of the complex plane of @xmath7 . + the ideas underlying this result are simple . for solutions of satisfying , it is equivalent to know their behavior for large @xmath12 or for large @xmath7 . hence , the idea is to study their behavior for large @xmath7 . keeping @xmath12 fixed to some compact interval , say @xmath160 $ ] , to first order in @xmath6 , a solution to is a solution to @xmath161 for @xmath162 . hence , the bloch solutions are special solutions to this equation . from these solution , multiplying them by a suitably chosen function of @xmath7 ( i.e. independent of @xmath12 ) , one can construct solution to both and at the same time . the standard form of the solutions is given below in . we first describe the standard asymptotic behavior of consistent solutions ; in order to do this , we recall some fact on periodic schrdinger operators on @xmath163 . then , we review the theory developed in @xcite on the existence of consistent solutions with these asymptotics . we next describe some new results used to control the behavior of the solutions at infinity . we conclude this section with the proof of theorem [ thr:2 ] . in this section , we discuss the periodic schrdinger operator where @xmath164 is a @xmath5-periodic , real valued , @xmath165-function . first , we collect well known results needed in the present paper ( see @xcite ) . in the second part of the section , we introduce a meromorphic differential defined on the riemann surface associated to the periodic operator . this object plays an important role for the adiabatic constructions ( see @xcite ) . let @xmath32 be a nontrivial solution of the equation satisfying the relation @xmath166 for all @xmath29 with @xmath167 independent of @xmath12 . such a solution is called a _ bloch solution _ , and the number @xmath168 is called the _ floquet multiplier_. we now discuss properties of bloch solutions ( see @xcite ) . as in section [ sec : periodic - operator ] , we denote the spectral bands of the periodic schrdinger equation by @xmath169 $ ] , @xmath170 $ ] , @xmath39 , @xmath16 $ ] , @xmath39 . consider @xmath171 , two copies of the complex plane @xmath172 cut along the spectral bands . paste them together to get a riemann surface with square root branch points . we denote this riemann surface by @xmath173 . in the sequel , @xmath174 is the canonical projection . one can construct a bloch solution @xmath175 of equation meromorphic on @xmath176 . for any @xmath177 , we normalize it by the condition @xmath178 . then , the poles of @xmath179 are projected by @xmath180 either in the open spectral gaps or at their ends . more precisely , there is exactly one simple pole per open gap . the position of the pole is independent of @xmath12 ( see @xcite ) . let @xmath181 be the canonical transposition mapping : for any point @xmath182 , the point @xmath183 is the unique solution to the equation @xmath184 different from @xmath185 outside the branch points . + the function @xmath186 is one more bloch solution of . except at the edges of the spectrum ( i.e. the branch points of @xmath173 ) , the functions @xmath187 and @xmath188 are linearly independent solutions of . in the spectral gaps , they are real valued functions of @xmath12 , and , on the spectral bands , they differ only by the complex conjugation ( see @xcite ) . consider the bloch solution @xmath175 . the corresponding floquet multiplier @xmath189 is analytic on @xmath173 . represent it in the form @xmath190 . the function @xmath191 is the _ bloch quasi - momentum_. + the bloch quasi - momentum is an analytic multi - valued function of @xmath177 . it has the same branch points as @xmath175 ( see @xcite ) . + let @xmath192 be a simply connected domain containing no branch point of the bloch quasi - momentum @xmath34 . on @xmath193 , fix @xmath194 , a continuous ( hence , analytic ) branch of @xmath34 . all other branches of @xmath34 that are continuous on @xmath193 are then given by the formula @xmath195 all the branch points of the bloch quasi - momentum are of square root type : let @xmath196 be a branch point , then , in a sufficiently small neighborhood of @xmath196 , the quasi - momentum is analytic as a function of the variable @xmath197 ; for any analytic branch of @xmath34 , one has @xmath198 with constants @xmath199 and @xmath200 depending on the branch . + let @xmath201 be the upper complex half - plane . there exists @xmath202 , an analytic branch of @xmath34 that conformally maps @xmath201 onto the quadrant @xmath203 cut along compact vertical intervals , say @xmath204 where @xmath205 and @xmath206 , ( see @xcite ) . the branch @xmath202 is continuous up to the real line . it is real and increasing along the spectrum of @xmath15 ; it maps the spectral band @xmath207 $ ] on the interval @xmath208 $ ] . on the open gaps , @xmath209 is constant , and @xmath210 is positive and has exactly one maximum ; this maximum is non degenerate . + we call @xmath202 the _ main _ branch of the bloch quasi - momentum . + finally , we note that the main branch can be analytically continued on the complex plane cut only along the spectral gaps of the periodic operator . on the riemann surface @xmath212 , consider the function @xmath213 where @xmath34 is the bloch quasi - momentum of @xmath32 , and the dot denotes the partial derivative with respect to @xmath214 . this function was introduced in @xcite ( the definition given in that paper is equivalent to ) . in @xcite , we have proved that @xmath215 has the following properties : 1 . the differential @xmath216 is meromorphic on @xmath212 ; its poles are the points of @xmath217 , where @xmath218 is the set of poles of @xmath219 , and @xmath220 is the set of points where @xmath221 ; 2 . all the poles of @xmath211 are simple ; 3 . @xmath222 , @xmath223 ; @xmath224 , @xmath225 ; @xmath226 , @xmath227 . 4 . if @xmath228 belongs to a gap , then @xmath229 ; 5 . if @xmath228 belongs to a band , then @xmath230 . we now discuss in more detail two analytic objects central to the complex wkb method , the complex momentum defined in and the canonical bloch solutions defined below . for @xmath231 , the domain of analyticity of the function @xmath8 , we define @xmath232 the complex momentum and the canonical bloch solutions are the bloch quasi - momentum and particular bloch solutions of the equation @xmath233 considered as functions of @xmath7 . for @xmath231 , the domain of analyticity of the function @xmath8 , the complex momentum is given by the formula @xmath234 where @xmath34 is the bloch quasi - momentum of . clearly , @xmath105 is a multi - valued analytic function ; a point @xmath7 such that @xmath235 is a branch point of @xmath105 if and only if it satisfies . all the branch points of the complex momentum are of square root type . a simply connected set @xmath236 containing no branch points of @xmath105 is called _ regular_. let @xmath237 be a branch of the complex momentum analytic in a regular domain @xmath193 . all the other branches analytic in @xmath193 are described by @xmath238 to describe the asymptotic formulae of the complex wkb method , one needs bloch solutions of equation analytic in @xmath7 on a given regular domain . we build them using the 1-form @xmath239 introduced in section [ sec : omega ] . pick @xmath240 , a regular point i.e. a point that is not a branch point of @xmath105 . let @xmath241 . assume that @xmath242 ( the sets @xmath218 and @xmath220 are defined in section [ sec : omega ] ) . in @xmath243 , a sufficiently small neighborhood of @xmath244 , we fix @xmath34 , a branch of the bloch quasi - momentum , and @xmath245 , two branches of the bloch solution @xmath175 such that @xmath34 is the bloch quasi - momentum of @xmath246 . also , in @xmath243 , consider @xmath247 , the two corresponding branches of @xmath211 , and fix a branch of the function @xmath248 . assume that @xmath102 is a neighborhood of @xmath240 such that @xmath249 . for @xmath250 , we let @xmath251 the functions @xmath252 are the _ canonical bloch solutions normalized at the point @xmath240_. the quasi - momentum associated to these solutions is @xmath253 . the properties of the differential @xmath211 imply that the solutions @xmath252 can be analytically continued from @xmath102 to any regular domain @xmath193 containing @xmath102 . + one has ( see @xcite ) @xmath254 as @xmath255 , the wronskian @xmath256 does not vanish . fix @xmath257 . let @xmath193 be a regular domain . fix @xmath258 so that @xmath259 . let @xmath105 be a branch of the complex momentum continuous in @xmath193 , and let @xmath252 be the canonical bloch solutions defined on @xmath193 , normalized at @xmath240 and indexed so that @xmath105 be the quasi - momentum for @xmath260 . + we recall the following basic definition from @xcite [ def:1 ] fix @xmath261 , @xmath258 , @xmath262 , @xmath102 , a complex neighborhood of some energy @xmath79 and @xmath193 a domain in the @xmath7-plane . we say that @xmath263 , a solution of , has standard asymptotics ( or standard behavior ) @xmath264 in @xmath265 if * @xmath263 is defined and satisfies and for any @xmath266 ; * @xmath263 is analytic in @xmath267 and in @xmath268 ; * for any @xmath269 , compact subset of @xmath193 , there exists @xmath270 , a neighborhood of @xmath79 , such that , for @xmath271 , @xmath263 has the uniform asymptotics @xmath272 * this asymptotic can be differentiated once in @xmath12 . let @xmath273 be two solutions of having standard behavior @xmath274 in @xmath265 . one computes @xmath275 by , for @xmath7 in any fixed compact subset of @xmath193 and @xmath6 sufficiently small , the solutions @xmath273 are linearly independent . a basic and important example of a domain where one can construct a solution with standard asymptotic behavior is a canonical domain . let us define canonical domains and formulate the basic result of the adiabatic complex wkb method . we recall that a curve @xmath276 is vertical if it intersects the lines @xmath277 at non - zero angles @xmath278 , @xmath279 . vertical lines are naturally parametrized by @xmath280 . + a set @xmath193 in the @xmath7-plane is said to be bounded regular at energy @xmath79 if @xmath281 , the closure of @xmath193 , is compact and does not contain any branch point of @xmath282 . * if a vertical curve is bounded regular at energy @xmath79 , it is bounded regular at any energy @xmath13 in a neighborhood of @xmath79 ; * if a vertical curve is bounded regular and canonical with respect to @xmath105 at energy @xmath79 , it is also canonical with respect to @xmath105 at any energy @xmath13 in a neighborhood of @xmath79 . let @xmath269 be a bounded regular domain at energy @xmath79 i.e. @xmath286 is compact and does not contain any branch point of @xmath282 . on @xmath269 , fix a continuous branch of the complex momentum , say @xmath105 . [ def:4 ] the domain @xmath269 is called _ canonical _ ( with respect to @xmath105 at energy @xmath79 ) if there exists two points @xmath287 and @xmath288 located on @xmath289 , the boundary of @xmath269 , such that @xmath269 is the union of curves connecting @xmath287 and @xmath288 that are canonical ( with respect to @xmath105 at energy @xmath79 ) . [ t5.1 ] let @xmath269 be a bounded domain canonical with respect to @xmath105 at energy @xmath79 . fix @xmath262 and @xmath290 . then , there exists @xmath102 , a neighborhood of @xmath79 such that , for sufficiently small positive @xmath6 , there exist @xmath291 , two solutions of , having the standard behavior in @xmath292 that is @xmath293 for any fixed @xmath29 , the functions @xmath294 are analytic in @xmath295 where @xmath296 is the smallest horizontal strip containing @xmath269 . in addition to this result , we need analogues for domains that are neighborhoods of infinity . such results were developed in @xcite ; they are not sufficient for the purpose of the present paper . hence , we present two new results in the next section . [ thr:1 ] assume ( h1)-(h4 ) are satisfied . fix @xmath261 and @xmath297 . pick @xmath61 such that either @xmath298 . fix @xmath262 and @xmath48 ( where @xmath299 is defined in assumption ( h4 ) , section [ sec : assumptions ] ) . then , there exist @xmath122 and @xmath300 such that , for any @xmath301 , the cone @xmath302 is a canonical domain . more precisely , fix @xmath303 . then , there exist a branch of the quasi - momentum @xmath304 that satisfies @xmath305 and a function @xmath306 defined on @xmath307 that satisfies : * it is a consistent solution to equation * for @xmath308 , @xmath309 is analytic on @xmath310 ; * the function @xmath263 admits the asymptotics @xmath311 where @xmath312 is the canonical bloch solutions @xmath260 associated to @xmath105 ( see ) and @xmath313 * the asymptotics can be differentiated once in @xmath12 ; * define the function @xmath314 ; then , @xmath315 form a basis of consistent solutions and satisfy @xmath316 [ thr:5 ] assume ( h1)-(h4 ) are satisfied . fix @xmath261 and @xmath297 . pick @xmath61 such that @xmath318 . fix @xmath262 and @xmath48 ( where @xmath299 is defined in assumption ( h4 ) , section [ sec : assumptions ] ) . then , there exist @xmath122 and @xmath300 such that , for any @xmath319 , the cone @xmath320 is a canonical domain . more precisely , fix @xmath303 . then , there exist a branch of the quasi - momentum @xmath304 that satisfies , for some @xmath321 , @xmath322 and a function @xmath306 defined on @xmath307 that satisfies : * @xmath263 is a consistent solution to equation * for @xmath308 , @xmath309 is analytic on @xmath310 ; * the function @xmath263 admits the asymptotics @xmath323 where @xmath324 are the canonical bloch solutions associated to @xmath105 ( see ) and @xmath325 satisfy ; * the asymptotics can be differentiated once in @xmath12 ; * the functions satisfy @xmath326 now to achieve the goal described in section [ sec : complex - wkb - method ] , namely , to construct consistent solutions to with known asymptotics in large complex domains of @xmath7 , as we have the form of standard asymptotics locally both at a finite point in @xmath7 and near infinity , we only need patch together the various domain on which we found consistent basis with standard asymptotics . an alternative way to obtain global asymptotics was developed in @xcite . it consists in the proof of quite general continuation lemmas that describe geometric situations in the complex plane of @xmath7 under which one can prove that , starting from a canonical domain and solutions with standard asymptotics in this domain , one can continue them . the main obstacles to continuation ( i.e. to the validity of standard asymptotics ) are the nodal lines of @xmath327 and the branching points of @xmath105 . we will not discuss this further and refer to @xcite for the details . m. zworski . quantum resonances and partial differential equations . in _ proceedings of the international congress of mathematicians , iii ( beijing , 2002 ) _ , pages 243252 , beijing , 2002 . higher ed . press .	in this talk , we report on results about the width of the resonances for a slowly varying perturbation of a periodic operator . the study takes place in dimension one . the perturbation is assumed to be analytic and local in the sense that it tends to a constant at @xmath0 and at @xmath1 ; these constants may differ . modulo an assumption on the relative position of the range of the local perturbation with respect to the spectrum of the background periodic operator , we show that the width of the resonances is essentially given by a tunneling effect in a suitable phase space . .5 cm rsum . dans cet expos , nous dcrirons le calcul de la largeur des rsonances de perturbations lentes doprateurs de schrdinger priodiques . cette tude est uni - dimensionnelle . les perturbations lentes considres sont analytiques et locales au sens o elles tendent vers une constante en @xmath0 et en @xmath1 ; ces deux constantes peuvent toutefois tre diffrentes . sous des hypothses adquates sur la position relative de limage de la perturbation locale par rapport au spectre de loprateur de schrdinger priodique , nous dmontrons que la largeur des rsonances est donne par un effet tunnel dans un espace de phase adquat .
recently topologically non - trivial superconductivity has been theoretically predicted to occur in two classes of systems with spin - orbit ( so ) coupling . they are : ( a ) so coupled semiconductors in which @xmath0-wave superconducting pair potential is induced by the proximity effect @xcite and ( b ) so coupled systems with superconductivity due to intrinsic @xmath0-wave pairing interaction @xcite . a third system - surface of a 3d strong topological insulator ( ti ) - can also support topological superconductivity when the latter is proximity induced @xcite . in this paper we will ignore the latter and concentrate only on ( a ) and ( b ) . in both classes ( a ) and ( b ) the so coupling is a consequence of the breakdown of the structural space inversion ( si ) symmetry . we will take the resultant so coupling to be of the rashba type . a system in class ( a ) can be artificially grown as a heterostructure consisting of a so coupled semiconductor in proximity contact with a @xmath0-wave superconductor @xcite . an example of class ( b ) is a non - centrosymmetric superconductor @xcite or an @xmath0-wave feshbach resonant system with an accompanying spin - orbit coupling and zeeman splitting both of which can be created in a cold fermion atomic system @xcite . there is also another class ( class ( c ) ) of topological superconductors with non - abelian statistics , which have been studied extensively in the recent literature . here the topological nature arises entirely from intrinsic chiral @xmath10-wave superconductivity without having any underlying @xmath0-wave superconductivity in the system . some examples of this class ( c ) are the even - denominator 5/2 fractional quantum hall effect @xcite , superconducting strontium ruthenate @xcite , a - phase of superfluid he-3 @xcite , and superfluid ultracold fermionic gases based directly on the @xmath10-wave feshbach resonance @xcite . these class ( c ) non - abelain superconductors are effectively equivalent to being spinless , i.e. completely spin - polarized , and are therefore immune to any zeeman splitting to the leading order . we do not discuss the class ( c ) systems in this paper since the main conceptual issue being addressed in this paper does not apply to these systems . the topologically non - trivial superconducting systems mentioned above are characterized by order parameter defects , such as a vortex and a sample edge , which carry a unique bound state at zero excitation energy @xcite . these bound zero energy modes , called majorana fermion modes after e. majorana @xcite , can actually be thought of as particles which are their own anti - particles @xcite . in other words , they are represented by second quantized operators @xmath11 which satisfy the hermiticity condition @xmath12 . this is strikingly different from the regular fermionic modes for which the second quantized operators @xmath13 . because of the existence of the zero - energy majorana modes , a 2d non - abelian topological superconductor with @xmath14 vortices each carrying a single majorana mode constitutes a system with a @xmath15-fold ground state degeneracy protected by an excitation gap @xcite . when a majorana fermion mode in a given order parameter defect is adiabatically moved ( braided ) around another majorana mode , the initial state transforms into another one which is a different linear combination in the same ground state manifold . these unitary transformations within the ground state manifold are manifestations of the non - abelian statistics of the majorana fermions . the non - abelian statistics of the majorana fermions has recently come under intense focus because of its potential application in fault tolerant topological quantum computation ( tqc ) @xcite . both 2d spin - orbit coupled superconducting films and its 1d version as quantum nanowires in a wire - network have been proposed as potential platforms for tqc @xcite . in this paper we will not discuss the majorana fermion modes or their potential application to tqc . instead we will focus on the bulk superconducting phases and the quantum phase transitions ( qpt ) by exploring the interplay of the spin - orbit coupling , zeeman splitting , and the @xmath0-wave pairing interactions . both systems in classes ( a ) and ( b ) are in an ordinary ( non - topological ) superconducting phase in the absence of an externally imposed zeeman splitting . the zeeman splitting , which can be applied either by a parallel magnetic field @xcite or by the proximity effect of a nearby magnetic insulator @xcite , creates a gap between the two spin - orbit bands as shown in fig . [ fig0 ] . from mean field calculations @xcite it is clear that when such a zeeman splitting @xmath1 crosses a critical value , @xmath4 , where @xmath3 is the @xmath0-wave superconducting pair potential and @xmath16 is the chemical potential , the system makes a transition to a topological non - abelian superconducting phase . the critical zeeman splitting corresponds to the value at which the underlying fermi surface in the absence of superconductivity shifts from being at both so bands to occupying only the lower so band ( fig . [ fig0 ] ) . thus , at this value of the zeeman splitting , the underlying system changes from having two fermi surfaces ( one in each band ) to having just one . since @xmath16 usually far exceeds @xmath3 , the critical value of the zeeman splitting @xmath2 also far exceeds @xmath3 . therefore , in the absence of so coupling , @xmath2 far exceeds the zeeman splitting ( @xmath8 ) above which an @xmath0-wave pair potential should decay to zero . the loss of an @xmath0-wave pair potential due to a strong zeeman splitting is due to the fact that for @xmath17 forming a spin - singlet pair potential with zero net momentum is impossible ( for the discussion of topological phase transition induced by zeeman splitting we will ignore the states with non - zero values of the cooper pair momentum ) . now let us emphasize the fact that in both cases we consider in this paper ( superconducting pair potential proximity induced , and superconducting pair potential due to an intrinsic on - site pairing interaction ) , the pair potential is strictly @xmath0-wave in both the ordinary and the topologically non - trivial superconducting phases . the fact that the pair potential is @xmath0-wave when it is proximity induced from a @xmath0-wave superconductor is self - evident . that it remains @xmath0-wave ( and is not a mixture of @xmath0- and @xmath10-wave due to the so coupling ) even when the pair potential is due to an intrinsic on - site pairing interaction is not so obvious . in this case , the pure @xmath0-wave symmetry of the pair potential follows from the fact that the intrinsic pairing interaction we consider is spatially local , and thus forming a @xmath10-wave component of the pair potential is forbidden by the fermion anticommutation relation ( for a more detailed discussion see sec . since the pair potential is purely @xmath0-wave in both classes ( a ) and ( b ) and in both phases ( ordinary and topological ) in each , how an @xmath0-wave pair potential survives a strong zeeman splitting @xmath9 to realize the topologically non - trivial phase is the central conceptual question we address in this paper . the basic conceptual issue being discussed here is the topic often alluded to as the chandrasekhar - clogston ( cc ) limit @xcite in ordinary @xmath0-wave superconductivity , which states , in effect , that an @xmath0-wave superconductor , where the cooper pairing is between spin - up and spin - down electrons near the fermi surface , can not withstand a zeeman splitting larger than the superconducting gap . this is because then spin splitting exceeds the superconducting gap energy , making it impossible for a superconducting ground state to develop . on first sight , it appears that the condition on the zeeman splitting needed for superconductivity in refs . [ 1 - 4 ] far exceeds this limit , thus destroying all superconductivity ! this has caused some confusion about the very existence of the topological superconducting phase using either a heterostructure @xcite where @xmath0-wave superconductivity is induced by proximity effect or using so - coupled systems with intrinsic @xmath0-wave pairing interactions @xcite . the mean field calculations of ref . [ 1 - 4 ] are not enough to resolve this question . this is because a mean field theory is not just a postulate to assume the existence of a mean field pair potential in the hamiltonian h as is done in these works ; one is also required to establish the finiteness of the pair potential by satisfying the self - consistent gap equation . in other words , we need to satisfy the bardeen - cooper - schrieffer ( bcs ) self - consistent gap equation with a strong zeeman potential ( @xmath18 ) to check if a non - zero @xmath0-wave pair potential @xmath3 gives a consistent solution . this will ensure that the mean field h in refs . [ 1 - 4 ] is not flawed to begin with , and our bdg solution of the majorana fermion is not a spurious mathematical result with no physical connection . in this paper we perform this study by self - consistently solving the appropriate bcs gap equations in the presence of attractive @xmath0-wave pairing interaction , so coupling , and an externally applied zeeman splitting . _ note that satisfying such a gap equation is a requirement for superconducting pair potential when it is derived from the microscopic pairing interactions . however , when the pair potential is proximity induced on a spin - orbit coupled system by a nearby @xmath0-wave superconductor , the gap equation need not be satisfied . in this case , the so coupled system simply ` inherits ' the pair potential of the nearby superconductor . _ for the case of intrinsic pairing interactions , we show that the @xmath0-wave pair potential indeed remains non - zero even beyond the zeeman splitting above which it would be lost in the absence of so coupling . in fact , in the non - abelian phase the non - zero value of the pair potential crucially depends on and _ increases _ with the so coupling constant , which is consistent with the fact that it is zero in the absence of the so coupling . a simple intuitive way to understand this starts by recalling how @xmath0-wave superconductivity is destroyed by a zeeman splitting . in the absence of so coupling , the two spin bands are shifted by an energy gap proportional to the zeeman splitting @xmath1 . with increasing @xmath1 it becomes increasingly difficult for the system to create @xmath0-wave spin - singlet pairs with zero net momentum . finally , when @xmath1 crosses a value @xmath19 the @xmath0-wave pair potential vanishes . the critical zeeman splitting , @xmath4 , needed for the topological phase transition is thus squarely beyond the acceptable zeeman splitting the pair potential can sustain . it is now important to realize that , in the presence of the so coupling , the two spin - orbit bands can not simply be viewed as ` spin - up ' and ` spin - down ' bands . instead , they both have a non - zero minority spin amplitude coexisting with the majority spin component . therefore , even when the zeeman splitting is large enough to make the fermi surface lie only in the lower band , spin - singlet @xmath0-wave pairs can not be completely lost . if the superconductivity is due to an intrinsic pairing interaction the gap equation shows that the pair potential , though always non - zero , decays with increasing zeeman couplings . however , the magnitude of the pair potential in the non - abelian phase can be increased by increasing the magnitude of the spin - orbit coupling , which , therefore , enables a stable non - abelian phase in the phase diagram . on the other hand , if the superconductivity is due to the proximity effect , there is no requirement of satisfying the self - consistent gap equation . in this case , which applies to the heterostructure geometry , the superconducting pair potential is simply ` inherited ' from the adjacent @xmath0-wave superconductor ( sec . we assume that the quasi-2d electron system is described by the model hamiltonian @xmath20 where @xmath21 describes the bulk conduction electrons , @xmath22 is the spin - orbit interaction term , @xmath23 represents the zeeman coupling , and @xmath24 represents the electron - electron interaction . explicitly , we have @xmath25 where @xmath26 is the bulk spectrum ( measured relative to the chemical potential @xmath16 ) , @xmath27 is the strength of the rashba spin - orbit coupling , @xmath1 represents the zeeman field , @xmath28 is the short - ranged interaction potential ( we will later restrict ourselves only to an on - site pairing interaction , @xmath28 independent of @xmath29 , natural for @xmath0-wave order ) , @xmath30 are pauli matrices , and @xmath31 ( @xmath32 ) is the creation ( annihilation ) operator corresponding to the single - particle state with momentum @xmath33 and spin @xmath34 . it is convenient to work in the spinor basis provided by the eigenfunctions @xmath35 of the single - particle hamiltonian @xmath36 , @xmath37 where @xmath38 and @xmath39 . the corresponding eigenvalues are @xmath40 the electron @xmath41-operators can be expressed in terms of the annihilation operators @xmath42 associated with the spinor eigenstates as @xmath43 using the spinor representation , the hamiltonian becomes @xmath44 where @xmath45 is the scalar product of two spinors @xmath46 and @xmath47 . to derive the gap equations , we first introduce the regular and anomalous green functions as @xmath48 where @xmath49 is the time ordering operator and the operators @xmath50 are in the heisenberg representation . the correlation functions @xmath51 have the properties @xmath52 the definitions ( [ flbd ] ) and ( [ fxlbd ] ) of the anomalous correlation functions follow the convention used by gorkov and rashba @xcite . introducing the definition of the gap function in eq . ( [ eqmotion1 ] ) , we have @xmath55{\cal g}_{\lambda\lambda^\prime}({\bm p } , \tau-\tau^\prime ) - \gamma { \cal g}_{-\lambda\lambda^\prime}({\bm p } , \tau-\tau^\prime ) + \sum_{\lambda_1}\delta_{\lambda\lambda_1}({\bm p } ) { \cal f}_{\lambda_1\lambda^\prime}^+(-{\bm p } , \tau-\tau^\prime ) = \delta_{\lambda\lambda^\prime } \delta(\tau-\tau^\prime ) , \nonumber \\ & ~&[-\partial_{\tau } + \epsilon_{\lambda}({\bm p})]{\cal f}_{\lambda\lambda^\prime}^{+}(-{\bm p } , \tau-\tau^\prime)-\gamma { \cal f}_{-\lambda\lambda^\prime}^{+}(-{\bm p } , \tau-\tau^\prime ) + \sum_{\lambda_1 } \delta_{\lambda_1\lambda}^ * { \cal g}_{\lambda_1\lambda}({\bm p } , \tau-\tau^\prime ) = 0 . \label{eqmotion2}\end{aligned}\ ] ] defining the fourier transforms of the correlation functions in the usual way , @xmath57 , the set of equations of motion can be expressed in a matrix form as @xmath58 where the arguments of the green functions have been omitted for simplicity . a similar set of equations , which can be obtained from ( [ meqns ] ) by switching the @xmath59 and @xmath60 labels , couples @xmath61 , @xmath62 , @xmath63 , and @xmath64 . the superconducting spectrum can be obtained from the condition that the determinant of the @xmath65 matrix in eq . ( [ meqns ] ) vanish taking @xmath66 . also , by solving the system of equations of motion for @xmath67 and introducing the solutions in eq . ( [ delta ] ) we obtain the self - consistent gap equations . in general , we have @xmath68 $ ] , where the symmetric and antisymmetric components of the diagonal gap functions are @xmath69 e^{i\theta_{\bm q } } , \label{del0s } \\ \delta_{0a}({\bm p } ) & = & -i \sum_{\bm q } \frac{v({\bm p}-{\bm q})-v({\bm p}+{\bm q } ) } { 2 } \left[({\cal f}_{++}-{\cal f}_{--})\cos(\theta_{\bm p}-\theta_{\bm q})e^{i\theta_{\bm q } } + i({\cal f}_{+-}-{\cal f}_{-+})\sin(\theta_{\bm p}-\theta_{\bm q})e^{i\theta_{\bm q } } \right].\nonumber\\ \label{del0a}\end{aligned}\ ] ] similarly , the off - diagonal gap functions can be expressed as @xmath70 $ ] , with @xmath71 e^{i\theta_{\bm q } } , \label{del1s } \\ \delta_{1a}({\bm p } ) & = & -i \sum_{\bm q } \frac{v({\bm p}-{\bm q})-v({\bm p}+{\bm q } ) } { 2 } \left[i ( { \cal f}_{++}-{\cal f}_{--})\sin(\theta_{\bm p}-\theta_{\bm q})e^{i\theta_{\bm q } } + ( { \cal f}_{+-}-{\cal f}_{-+})\cos(\theta_{\bm p}-\theta_{\bm q})e^{i\theta_{\bm q } } \right].\nonumber\\ \label{del1a}\end{aligned}\ ] ] note that @xmath72 and @xmath73 , i.e. , @xmath74 and @xmath75 represent the singlet and triplet components of the gap functions , respectively . instead of solving the complicated coupled set of gap equations above , we simplify matters by considering the case of strictly local interactions . in other words , we neglect the momentum dependence of the interaction potential , @xmath76 . then the only non - vanishing component of the superconducting gap is the singlet component , @xmath77 , and it becomes momentum - independent . since by a zeeman splitting the singlet component of the gap function will be the most affected , we can make this approximation to examine the fate of the superconducting condensate with increasing zeeman potential . for a strictly local attractive interaction , the superconducting spectrum is given by , @xmath78 where @xmath79 . solving the kinetic equations for @xmath80 and @xmath63 and using eq ( [ del0s ] ) we obtain the gap equation for the strictly local attractive interaction , @xmath81=1.\ ] ] taking the zero temperature limit and performing the summation over the frequencies we obtain @xmath82 . \label{gapeq}\ ] ] to obtain a deeper understanding of the singlet triplet mixing in superconductors with spin orbit coupling @xcite , it is useful to determine the expressions of the anomalous correlation functions and of the gap functions in terms of the original electron operators . we first express the @xmath41operators in terms of @xmath53operators , @xmath83 and @xmath84 , and we obtain for the singlet and triplet anomalous correlation functions the expressions @xmath85 , \\ \langle t_{\tau } c_{-{\bm p } \sigma}(\tau ) c_{{\bm p } \sigma}(\tau^{\prime } ) \rangle & = & \frac{1}{2 } e^{(1-\sigma)i\theta_{\bm p } } \left [ { \cal f}_{++}(-{\bm p } , \tau-\tau^{\prime } ) - { \cal f}_{--}(-{\bm p } , \tau-\tau^{\prime } ) + \sigma { \cal f}_{+-}(-{\bm p } , \tau-\tau^{\prime } ) -\sigma { \cal f}_{-+}(-{\bm p } , \tau-\tau^{\prime})\right ] , \nonumber\end{aligned}\ ] ] where the @xmath86 anomalous functions are given by equations ( 14 ) and ( 15 ) . in the limit of local interactions we can determine the explicit dependence of the @xmath86 functions on the parameters of the model using eq . ( 21 ) , and we have @xmath87 and @xmath88 . consequently , in the @xmath41operator representation both the singlet and the triplet components of the anomalous correlation function are proportional to the @xmath0wave gap , @xmath89 and @xmath90 , respectively . we emphasize that , in the limit of local pairing interaction , the anomalous correlation function in the @xmath41operator representation has both singlet and triplet components , but the corresponding gap function is purely @xmath0wave . to show this property explicitly , we can re - derive the gap equations in the @xmath41operator representation and , instead of eq . ( 20 ) , we obtain @xmath91 where @xmath92 is the normal green function and the gap functions are defined as @xmath93 the equal time anomalous correlation @xmath94 can be expressed in terms of @xmath95 . explicitly , we have @xmath96 -\sum_{\bm q}\frac{v({\bm p}-{\bm q})- v({\bm p}+{\bm q})}{2}\frac{i}{2}e^{i\theta_{\bm q}}\left[{\cal f}_{+-}+{\cal f}_{-+}\right ] , \nonumber \\ \delta_{\uparrow \uparrow}({\bm p } ) & = & -\sum_{\bm q}\frac{v({\bm p}-{\bm q})- v({\bm p}+{\bm q})}{2}\frac{1}{2}\left[{\cal f}_{++}-{\cal f}_{--}\right ] -\sum_{\bm q}\frac{v({\bm p}-{\bm q})+ v({\bm p}+{\bm q})}{2}\frac{1}{2}\left[{\cal f}_{+-}-{\cal f}_{-+}\right].\nonumber \end{aligned}\ ] ] in the limit of strictly local interactions @xmath97 and @xmath98 and , consequently , the triplet component vanishes , @xmath99 . on the other hand , the expression for the singlet component of the gap becomes identical with the right hand side of eq . ( 22 ) , hence we have @xmath100 . we conclude that in a superconductor with spin orbit coupling and on - site pairing interactions the anomalous correlation function is characterized by a mixture of singlet and triplet components , yet the gap function has purely @xmath0wave symmetry . a @xmath10wave component of the gap can develop only in the presence of non local pairing interactions . as a consequence , in a system with strictly local pairing interaction , if the singlet anomalous correlation vanishes , the superconducting gap as well as all the other components of the anomalous correlation function will vanish . we first analyze the gap equation , eq . ( [ gapeq ] ) , in some special cases for which the solutions are well known . this will serve as a test for the validity of our analytical calculations . by putting @xmath101 and @xmath102 , which corresponds to the standard bcs case of a local attractive interaction with no spin - orbit coupling and zeeman splitting , we find @xmath103 . in this case , from eq . ( [ gapeq ] ) we recover the standard bcs gap equation , @xmath104 where the summation over @xmath105 should be performed over states satisfying @xmath106 , with @xmath107 some cut - off debye energy scale . as is well known @xcite , since the integral on the right hand side ( r.h.s . ) diverges in the limit @xmath108 , in this case a non - zero solution for @xmath3 exists for any @xmath109 . next we take the system with @xmath101 , @xmath110 . in this case the gap equation now becomes , @xmath112 \label{caseii}\ ] ] as in the previous case , the integrals on the r.h.s . ( [ caseii ] ) diverge when @xmath108 , hence a non - vanishing solution for @xmath3 exists for any @xmath113 . to establish the familiar result that @xmath0-wave superconductivity is destroyed by zeeman splitting ( in the absence of spin - orbit coupling ) , we consider the special case , @xmath114 , and @xmath102 . in this case @xmath115 , and the gap equation becomes @xmath116 where the summation over @xmath105 is done over states satisfying @xmath106 and @xmath117 . the second constraint results from the cancelations of two terms from eq . ( [ gapeq ] ) that diverge in the limit @xmath108 . since the integral on the r.h.s . of the gap equation no longer diverges , a non - zero solution for @xmath3 exists only for @xmath118 larger that a critical value . this implies that for a given strength of the attractive potential @xmath118 , no non - zero solution for @xmath3 can be found above a critical value of the zeeman potential @xmath1 . finally we consider the most general case of a non - zero zeeman potential as well as a non - zero spin - orbit coupling , @xmath114 , @xmath110 . the gap equation is then given by eq . ( [ gapeq ] ) . the exact cancelation of the divergent terms that characterizes the @xmath102 case does no longer hold and the r.h.s . of eq . ( [ gapeq ] ) becomes arbitrarily large in the limit @xmath108 . consequently , a non - vanishing solution for @xmath3 exists for _ any _ negative value of @xmath119 . this implies that @xmath3 does not vanish for any value of @xmath1 , or , in other words , the pair potential can not be completely destroyed by a zeeman splitting in the presence of a non - zero spin - orbit coupling . this is in agreement with a similar result derived earlier in a different context @xcite . nonetheless , at large values of @xmath1 the superconducting pair potential decreases exponentially with the strength of the zeeman splitting as we show in the next section . next , we determine the general solution of the gap equation by solving eq . ( [ gapeq ] ) numerically . we address two distinct cases : 1 ) high carrier concentration regime , when the chemical potential @xmath16 ( i.e. , the fermi energy in our zero temperature limit ) represents the largest energy scale in the problem , @xmath120 , and 2 ) low carrier concentration , when @xmath121 . here @xmath107 is the analog of the debye frequency ( i.e. , the characteristic energy cut - off for the intrinsic pairing interaction ) , @xmath122 is the strength of the spin - orbit interaction at the fermi wave vector , and @xmath123 is the value of the superconducting gap at zero zeeman splitting . the zero field gap is a measure of the pairing interaction strength , and in fact one can use @xmath119 , instead of @xmath123 , as an independent parameter . the debye frequency acts as a cutoff in eq . ( [ gapeq ] ) , i.e. , the summation over @xmath105 of a function @xmath124 is restricted to the values of the wave vector satisfying @xmath125 . ( [ theta ] ) ) with @xmath3 . the values of @xmath3 at which @xmath126 are the solutions of the self - consistent gap equation . ( a ) : @xmath127 with @xmath3 for various values of the zeeman splitting @xmath1 for a small value of the spin - orbit coupling constant @xmath27 ( @xmath128 ) and large @xmath129 . initially for smaller values of the zeeman splitting there is just one solution for @xmath3 . with increasing values of @xmath1 , there are three solutions for @xmath3 , one of which ( the one in the middle ) is unstable . the lower ( higher ) solution for @xmath3 is the high ( low ) field solution which co - exist in a region of co - existence . with @xmath1 going up , the high field solution becomes smaller but is never zero . consequently , there is no true first order transition in the presence of a small @xmath27 , even though there is a first - order - like precipitous drop of @xmath3 at some value of @xmath1 , which is a remnant of the first order transition for @xmath102 . ( b ) : @xmath127 with @xmath3 for increasing values of @xmath1 with a fixed large @xmath130 and _ a larger spin orbit coupling _ @xmath131 . @xmath127 is now monotonically decreasing and the resultant unique solution for @xmath3 decreases for increasing values of @xmath1 . the rate of decrease of this solution with @xmath1 is much slower and continuous . as with the case of smaller @xmath27 , there is no quantum phase transition.,scaledwidth=45.0% ] to obtain the self - consistent numerical solution for the gap , we define the function @xmath132 - 1 . \label{theta}\ ] ] with this notation , eq . ( [ gapeq ] ) becomes @xmath133 . this equation is characterized by two qualitatively different regimes that are controlled by the relative strength of the spin - orbit interaction and the zero - field gap . if @xmath134 , @xmath127 is a monotonically decreasing function that starts from large positive values at @xmath135 and eq . ( [ theta ] ) has always one non - vanishing solution . by contrast , when @xmath136 , the function @xmath127 becomes non - monotonic for certain values of the zeeman field @xmath1 , which means that eq . ( [ theta ] ) can have multiple non vanishing solutions for a given set of parameters . to illustrate this situation , we show in fig . [ fig1 ] the function @xmath127 for a system with large carrier concentration ( @xmath129 ) and extremely low spin - orbit interaction ( @xmath137in panel ( a ) and @xmath138in panel ( b ) , i.e. , @xmath128 and @xmath139 , respectively ) . the debye frequency is @xmath140 and the zero field gap is @xmath141 . at low zeeman splitting , @xmath127 vanishes at a single point @xmath142 , but increasing @xmath1 leads to a local minimum in @xmath127 that goes to zero for @xmath143 . further increasing the zeeman splitting leads to three non - vanishing solutions @xmath144 ( see fig . [ fig1 ] ) , where @xmath145 and @xmath146 are the `` low field '' and `` high field '' solutions , respectively , and @xmath147 is an unstable solution . the `` low- '' and `` high - field '' solutions coexist in some range of zeeman field strengths suggesting that the system undergoes a precipitous drop in @xmath3 akin to a field - tuned first order phase transition . the coexistence region shrinks as the strength of the spin - orbit coupling increases ( see fig . [ fig1]b ) and vanishes at a value @xmath148 . we note that in real systems such as non - centrosymmetric superconductors the strength of the spin - orbit coupling is usually larger than this critical value and , consequently , the first - order - like precipitous drop in @xmath3 may not be observable . in cold fermion systems the spin - orbit coupling constant can be used as a tuning parameter to interpolate between these two behaviors . ) plotted against the zeeman splitting @xmath1 for various values of the spin - orbit coupling constant @xmath27 ( in @xmath149 ) . for small values of @xmath27 , the superconducting gap falls discontinuously with @xmath1 but it is never strictly zero in the presence of the so coupling . consequently , there is only a first - order - like crossover which is a remnant of the true first order phase transition with @xmath1 for @xmath102 . for larger values of @xmath27 ( black curve ) , the decay of the superconducting pair potential with the zeeman splitting is much slower and continuous . it falls exponentially ( but is never strictly zero ) only for higher fields @xmath150.,scaledwidth=45.0% ] the dependence of the solution of the gap equations on @xmath1 is shown in fig . the coexistence region can be easily seen for @xmath151(green line in fig [ fig2 ] ) , corresponding to the ( stable ) solutions of the equation @xmath126 for the @xmath152 function shown in fig . note the exponential decay of the `` high field '' solution @xmath153 with increasing @xmath1 . practically the superconducting gap is negligible ( @xmath154 ) for @xmath155 . the coexistence region shrinks as we approach the critical spin - orbit coupling ( red line in fig . [ fig2 ] and fig . [ fig1]b ) , then , for @xmath156 the gap equation has a continuous solution @xmath157 that decreases monotonically with the zeeman field ( black line in fig [ fig2 ] ) . note that at high fields , @xmath150 , the gap decreases exponentially . however , the energy scale for the spin - orbit coupling , @xmath158 , can be significant in realistic systems ( tens of mev ) and the high field regime may not be attainable , i.e. , the gap will not vanish for any realistic value of the zeeman field . the existence of a first - order - like drop in @xmath3 that ends at a critical value of the so coupling @xmath159 is generic , i.e. , this feature is present at any value of the carrier density . however , to realize a topologically nontrivial non - abelian regime , it is necessary to satisfy the condition @xmath160 . consequently , we study the solutions of the gap equation in the low density regime , when the chemical potential , the zeeman field , the spin - orbit interaction , and the superconducting order parameter are comparable . _ in particular we address the following question : is it possible to realize the condition for the existence of a topologically nontrivial non - abelian phase while maintaining a reasonable superconducting gap ? _ before presenting the results , we note that in the low - density regime the zero - field gap has a strong dependence on the chemical potential . more precisely , for a given set of parameters @xmath119 , @xmath107 , and @xmath27 , the zero - field gap @xmath123 decreases with @xmath16 . in our calculations we fix @xmath119 at a value that corresponds to @xmath161 at @xmath162 and , at lower carrier densities ( i.e. , lower values of @xmath16 ) , we calculate the zero - field gap using the gap equation . also , we note that , as we vary the zeeman splitting @xmath1 , the chemical potential of a system with fixed carrier density @xmath163 remains constant as long as the high - energy band @xmath164 has non - zero occupation . for higher values of @xmath1 , i.e. , when the bands split , we determine the chemical potential @xmath165 corresponding to the fixed carrier density . the values of @xmath16 provided below represent zero field values . for three different values of @xmath16 for a fixed value of the spin - orbit coupling constant @xmath166 . the value of @xmath27 is large enough so that @xmath3 remains continuous with @xmath1 . for larger values of @xmath16 ( black and green curves ) , @xmath3 becomes inappreciably small for @xmath167 . for the red curve , however , @xmath3 is appreciable ( @xmath168 mev ) for @xmath169 mev . since to the right of this @xmath1 it is possible to satisfy @xmath170 , the system is in a topologically non - trivial phase in this region . therefore , somewhere above @xmath171 mev ( shown with an arrow ) , there is a topological quantum phase transition from a regular superconducting phase ( to the left of the arrow ) to a topologically non - trivial non - abelian phase ( to the right of the arrow).,scaledwidth=45.0% ] shown in fig . [ fig3 ] is the dependence of the solution of the gap equation on the zeeman splitting for three different values of the chemical potential ( i.e. , three carrier densities ) @xmath172 . the debye frequency is taken as @xmath173 and the rashba coupling is @xmath174 , i.e. , the system is characterized by a strong spin - orbit coupling . for these parameters the system is above the critical value of @xmath27 for the discontinuous fall of @xmath3 and hence @xmath3 is now a continuous function of @xmath1 . before analyzing the plots in fig . [ fig3 ] , let us remind ourselves that in order to satisfy the conditions for the non - abelian @xmath0-wave phase ( @xmath160 ) , we need an appreciable @xmath3 when @xmath1 has crossed @xmath175 . coming back to fig . [ fig3 ] we note that , similar to fig . [ fig2 ] ( black curve ) , @xmath3 falls with increasing values of @xmath1 . for the black and the green curves ( higher @xmath16 ) @xmath3 becomes inappreciably small ( though it is never zero ) by the time @xmath1 becomes @xmath175 however , for the red curve ( @xmath176 ) there is a residual superconducting pair potential @xmath177 for @xmath178 , i.e. , the system is in a topologically nontrivial non - abelian phase . moreover , as shown in fig . [ fig4 ] , the magnitude of this residual @xmath0-wave pair potential increases with @xmath27 , and thus can be increased by increasing the value of the spin - orbit coupling . therefore , for these parameter values , there is topological quantum phase transition ( tqpt ) when @xmath1 crosses the critical value @xmath179 ( shown with an arrow in fig . [ fig3 ] ) . the tqpt separates a regular ( non - topological ) superconducting phase ( @xmath180 ) from a topological non - abelian superconducting phase ( @xmath7 ) . from our self - consistent mean field theory we find this tqpt to be continuous , that is , there is no change in @xmath3 at the critical value of the zeeman splitting . in the topological phase on the strength of rashba spin - orbit coupling @xmath27 . the system is characterized by a chemical potential at zero magnetic field @xmath181 ( see the red curve in fig . [ fig3 ] for the dependence of @xmath3 on @xmath1 ) . the zeeman splitting is fixed at @xmath182 . with these parameters the system is in a non - abelian phase for @xmath166 . the superconducting pair potential increases monotonically with @xmath27 . for @xmath183(black arrow ) the condition @xmath184 is no longer satisfied and the system undergoes another topological quantum phase transition to an ordinary superconducting phase ( at large @xmath27 ) . the inset shows the two excitation energies , eq . ( [ e12 ] ) , of which the smaller one ( blue curve ) vanishes at @xmath185 at this @xmath27-tuned tqpt.,scaledwidth=45.0% ] an alternative and perhaps more robust way to create a topologically non - trivial non - abelian superconductor is to induce a superconducting pair potential in a spin - orbit coupled semiconductor by proximity effect @xcite . ideally , for the proximity effect induced superconductivity the pairing interaction resides in a parent @xmath0-wave superconductor such as al or nb while the quasiparticles of interest are confined to a two- or one - dimensional semiconductor layer on the surface of the superconductor . the proximity effect has been shown to create a topological superconductor similar to the ones discussed above on the surface state of a topological insulator @xcite and also in a 2d semiconductor layer @xcite . physically , the proximity effect arises from multiple andreev reflections of electrons in a semiconductor that is connected to a superconductor by tunneling . for most realistic cases , there is no pairing interaction in the semiconductor . thus , strictly speaking the superconducting pair potential vanishes in the semiconductor and at first glance it appears that there is no superconductivity induced in the semiconductor . however , the superconducting order parameter defined by @xmath186 is found to remain non - zero in the semiconductor layer . furthermore , the multiple andreev reflections open a gap in the spectrum of quasiparticles that are localized in the semiconductor layer . the spectra of such quasiparticles can be shown to be identical to quasiparticles with an _ effective _ pairing potential in the semiconductor layer @xcite . therefore from the point of the quasiparticle spectrum , which is the only property that is relevant to the definition of a topological superconductor , the proximity to a superconductor induces a superconducting quasiparticle gap in the semiconductor . versus zeeman coupling @xmath1 for various values of spin - orbit interaction @xmath27 . the strength of the spin - orbit coupling in the inset is such that @xmath187 corresponds to @xmath188 ev - . the proximity induced pair potential and chemical potential are taken to be @xmath189 and @xmath190 . the quasiparticle gap vanishes at the critical value @xmath191 . above the critical point spin - orbit coupling opens a quasiparticle gap that is proportional to @xmath27 in the small @xmath27 limit . , scaledwidth=45.0% ] the proximity effect can be induced by even weak tunneling between the semiconductor and the superconductor . therefore , the quasiparticle spectrum in the semiconductor does not affect the pairing potential in the superconductor significantly . _ specifically , for the proximity induced superconductivity case , the self - consistency effects that were important in the discussions in the previous sections become insignificant . _ furthermore , if the zeeman potential is also induced by proximity effect from a magnetic insulator on the other surface of the semiconductor , there is no direct tunneling between the superconductor and the magnetic insulator and therefore no suppression of the order parameter in the superconductor @xcite . thus , in contrast to the discussions in the previous sections , where the zeeman potential induced topological phase transition was accompanied by significant changes in the pair potential @xmath3 , the pair potential @xmath3 in the proximity induced case remains unaffected by the zeeman splitting . the tqpt in both cases ( the proximity induced case and the case when the pair potential is due to an intrinsic pairing interaction ) can be characterized by the closing of the superconducting quasiparticle gap ( shown in fig . [ fig5 ] ) as the zeeman potential is raised from @xmath101 past the critical value @xmath192 . in the proximity induced case , @xmath3 is the proximity induced effective pair potential and @xmath16 is the fermi energy in the semiconductor . the quasiparticle gap @xmath193 ( minimum of @xmath194 in eq . ( [ e12 ] ) ) closes at @xmath185 exactly when @xmath1 passes through @xmath2 ( fig . [ fig5 ] ) , indicating the existence of a qpt even though the superconducting pair potential @xmath3 remains perfectly continuous . the quasiparticle gap for @xmath7 shows a linear dependence on the spin - orbit coupling strength @xmath27 at small @xmath27 @xcite . here it is appropriate to mention a caveat for the case where the zeeman potential is not proximity induced and instead induced by a magnetic field @xcite . in this case , the zeeman potential also suppresses the superconducting pair potential in the parent @xmath0-wave superconductor . however , this effect can be small provided the @xmath195-factor in the semiconductor is much larger than in the superconductor as is often the case in 2d electron systems . topologically non - trivial non - abelian superconductivity can be realized in two different classes of systems . in class ( a ) , superconductivity is proximity induced on a semiconductor ( in the form of a film or a wire ) which has a strong spin - orbit coupling . in class ( b ) , superconductivity arises from intrinsic attractive pairing interaction in a system which also has a sizable spin - orbit coupling . in both cases a firm requirement for a phase transition from an ordinary superconducting phase to a topologically non - trivial superconducting phase is an externally imposed zeeman splitting . the zeeman splitting creates a gap in the spin orbit bands ( fig . [ fig0 ] ) . when this gap is large ( zeeman splitting is comparable to the chemical potential ) so the fermi surface lies in only the lower band , it triggers a qpt at which the system goes from a regular superconducting phase ( small zeeman splitting ) to a topolgical superconducting phase ( large zeeman splitting ) . this value of zeeman splitting far exceeds the value at which an ordinary @xmath0-wave superconductor is known to lose superconductivity due to its inability to form spin - singlet zero - momentum cooper pairs . as we have shown above , this is where the requirement of a sizable spin - orbit coupling is important to stabilize a topological superconducting phase . below we recapitulate and discuss the main results first for the case where the superconductivity arises from an intrinsic pairing interaction followed by the much simpler case of superconductivity arising from proximity effect . to discuss the various phases and the quantum phase transitions we have divided the parameter space in two distinct regimes by the relative magnitude of @xmath16 with respect to all other energy scales in the problem . in the large @xmath16 regime the underlying system always has two fermi surfaces irrespective of the magnitude of the zeeman splitting @xmath1 . in the absence of the spin - orbit coupling @xmath27 , with increasing values of @xmath1 it becomes increasingly difficult for the system to create spin - singlet @xmath0-wave pair potential at zero net momentum . ignoring the possibility of cooper pairs with non - zero net momentum , we find that when @xmath1 crosses a critical value @xmath19 the system becomes non - superconducting at a first order qpt . at this transition the pair potential drops discontinuously to zero . by including a non - zero @xmath27 we find that , surprisingly , there is always a non - zero solution of the gap equation eq . ( [ gapeq ] ) . this is because with @xmath196 the individual bands can no longer be viewed as carrying a single spin component . rather , both bands now carry a minority spin amplitude along with the majority component , which allows @xmath0-wave superconducting pairing even for large values of @xmath1 . if @xmath27 is small , @xmath197 where @xmath123 is the value of the order parameter for zero zeeman splitting , there is still a precipitous drop of @xmath3 at zeeman splitting @xmath198 ( fig . [ fig2 ] ) . however , this is not a qpt , since , as already mentioned , @xmath3 is never strictly zero in the presence of a non - zero @xmath27 . when @xmath27 itself crosses a threshold value , @xmath199 , the first - order - like drop of @xmath3 as a function of @xmath1 turns into a slower continuous decay ( black curve in fig . [ fig2 ] ) . for high values of @xmath200 , @xmath3 again decays exponentially with @xmath1 . however , this high field scale , comparable to the spin - orbit strength at the fermi surface , may not be attainable in real systems . consequently , the @xmath0-wave superconducting gap may never vanish with a zeeman coupling in the presence of strong spin - orbit coupling . the regime of small @xmath16 is particularly important because of the possibility of a topological phase transition . in this case , the behavior of @xmath3 with @xmath1 for @xmath102 ( first order qpt ) and small @xmath27 ( precipitous drop of @xmath3 with @xmath1 ) remain unchanged from the case with large @xmath16 . for small @xmath16 , however , @xmath201 itself is small . consequently @xmath27 is always in the large spin - orbit coupling regime , @xmath202 . therefore , for realistic values of @xmath27 , @xmath3 falls only gradually with @xmath1 and strictly speaking is never zero ( fig . [ fig3 ] ) . let us now recall that for a tqpt from a regular @xmath0-wave superconductor to a topologically non - trivial superconductor the parameters need to satisfy the condition @xmath203 . this implies that , for a robust non - abelian phase , we require an appreciable @xmath3 when @xmath1 becomes @xmath204 . from the red curve in fig . [ fig3 ] , we note that for @xmath205 , @xmath3 is still appreciable , @xmath206 , and thus a stable non - abelian phase is in principle allowed . moreover , as shown in fig . [ fig4 ] , the value of @xmath3 for large @xmath1 ( i.e. , @xmath3 in the non - abelian phase ) is directly related to the spin - orbit strength @xmath27 and increases appreciably if @xmath27 can be increased ( as in a cold fermion system ) . conversely , there is no non - abelian phase ( @xmath207 ) if the system has no spin - orbit coupling . from our self - consistent mean field theory , we find the tqpt at @xmath208 to be continuous . by this we mean that the magnitude of @xmath3 is continuous across this transition . at @xmath209 the underlying system shifts from having two fermi surfaces ( @xmath180 ) to just one in the lower band ( @xmath7 ) . as shown in ref . [ ] , for @xmath7 a defect in the superconducting order parameter ( e.g. , vortex , sample edge ) traps a unique zero energy bound state majorana mode . such a non - degenerate bound state solution is absent for @xmath180 . the emergence of the topological bound state majorana solution for @xmath7 makes the transition a topological one . the exact location of the topological transition is indicated by the quasiparticle excitation energy @xmath193 ( minimum of @xmath194 in eq . ( [ e12 ] ) ) passing through zero . this happens at @xmath185 exactly when @xmath1 passes through @xmath2 ( fig . [ fig5 ] ) , indicating the existence of a qpt even though the superconducting order parameter @xmath3 remains perfectly continuous . when @xmath0-wave superconductivity is proximity induced on a semiconductor , there is no self - consistent gap equation to be satisfied in the semiconductor . thus there is no self - consistency effects that suppress the pair potential with the zeeman splitting as discussed above . in this case , the semiconductor simply ` inherits ' the superconducting pair potential and its quasiparticle spectrum is modified accordingly . for weak tunneling between the semiconductor and the superconductor layers , the quasiparticles in the semiconductor can not significantly influence the pair potential in the host superconductor . therefore , the self - consistency requirement as in the discussions above can be neglected . if the zeeman potential is also induced by the proximity effect of a magnetic insulator from the opposite side of the semiconductor , there will be minimal effect of the magnetic insulator on the @xmath0-wave superconductor . if the zeeman potential is induced by a parallel magnetic field , then the effect on the host superconductor will again be minimal provided the @xmath195-factor in the semiconductor is larger than that in the superconductor . to conclude , we have considered spin - orbit coupled systems with superconductivity arising from either intrinsic on - site @xmath0-wave pairing interactions or from the proximity effect of an adjacent superconductor . in both cases , using bdg analysis of a postulated mean field hamiltonian with an @xmath0-wave pair potential @xmath3 , it has been shown @xcite that when an externally imposed zeeman splitting crosses a critical value , there is a majorana fermion mode at a vortex core . the required zeeman splitting , @xmath210 , seems to far exceed the value ( @xmath8 ) above which an @xmath0-wave pair potential @xmath3 is known to vanish . this gives rise to the conceptual question if the postulated pair potential in ref . [ ] and all the subsequent works on this system is spurious for @xmath7 . if true this will indicate that the bdg result of the majorana fermion at a vortex core for @xmath7 , based on the postulated mean field h @xcite , is a spurious mathematical result with no physical connection . in this paper we have resolved this question by showing that in the presence of spin - orbit coupling the @xmath0-wave pair potential can never be made strictly zero by the application of a zeeman potential . this is in agreement with a similar result derived previously in a different context @xcite . when the @xmath0-wave pair potential arises from an intrinsic local pairing interaction , our self - consistent analysis of the gap equation reveals that the decay of the pair potential with zeeman splitting is more gradual in the presence of spin - orbit coupling , although for large enough zeeman splitting the decay is exponential . thus there can be a small but finite region in the parameter space ( rhs of arrow , which indicates a tqpt , on fig . 4 ) where a topologically non - trivial superconducting phase can be realized . when the @xmath0-wave pair potential is proximity induced on a semiconductor , there is no requirement of satisfying the self - consistent gap equations . in this case , the pair potential is simply ` inherited ' from the adjacent superconductor . thus in this case the topologically non - trivial phase is much more robust than the other case where it is due to intrinsic pairing interactions . two final comments are in order here . for long- but finite - range pairing interactions ( as opposed to local interactions as in this paper ) it is well known that the spin - orbit interaction mixes @xmath0-wave and @xmath10-wave pair potentials @xcite . in this case it may appear that superconductivity can evade the cc limit merely because the @xmath10-wave part of the pair potential can survive the strong zeeman field , even if the @xmath0-wave part can not . it is , however , incorrect to ascribe the existence of the topological superconducting state at large @xmath1 to this effect . as we have shown in detail in ref . [ ] , the topological state owes its existence solely to the survival of the @xmath0-wave part of the pair potential . ( the pfaffian topological invariant discussed in ref . [ ] is completely insensitive to the @xmath10-wave part . ) it is precisely to isolate and eliminate the effect of the mixed @xmath10-wave pair potential that in this paper we confined ourselves to a strictly local pairing interaction . the existence of the topological state at high zeeman fields is strictly due to the survival of the @xmath0-wave pair potential , the physics of which is discussed in this paper and also summarized in the concluding paragraph of the introduction . but this work is not just an academic resolution of the question of the survival of an @xmath0-wave pair potential in the presence of a strong zeeman field . it also proves that all properties of the topological state when superconductivity is proximity induced continue to hold even when superconductivity is due to local @xmath0-wave pairing interactions . this result is directly relevant to the case of an @xmath0-wave feshbach cold atom system . note that in this case @xmath3 can not just be assumed in the bdg equations ( as in the case of the proximity effect framework @xcite ) , but has to be calculated from the gap equations as in the present paper .	topologically non - trivial superconductivity has been predicted to occur in superconductors with a sizable spin - orbit coupling in the presence of an external zeeman splitting . two such systems have been proposed : ( a ) @xmath0-wave superconductor pair potential is proximity induced on a semiconductor , and ( b ) pair potential naturally arises from an intrinsic @xmath0-wave pairing interaction . as is now well known , such systems in the form of a 2d film or 1d nano - wires in a wire - network can be used for topological quantum computation . when the external zeeman splitting @xmath1 crosses a critical value @xmath2 , the system passes from a regular superconducting phase to a non - abelian topological superconducting phase . in both cases ( a ) and ( b ) we consider in this paper the pair potential @xmath3 is strictly @xmath0-wave in both the ordinary and the topological superconducting phases , which are separated by a topological quantum critical point at @xmath4 , where @xmath5 is the chemical potential . on the other hand , since @xmath6 , the zeeman splitting required for the topological phase ( @xmath7 ) far exceeds the value ( @xmath8 ) above which an @xmath0-wave pair potential is expected to vanish ( and the system to become non - superconducting ) in the absence of spin - orbit coupling . we are thus led to a situation that the topological superconducting phase appears to set in a parameter regime at which the system actually is non - superconducting in the absence of spin - orbit coupling . in this paper we address the question of how a pure @xmath0-wave pair potential can survive a strong zeeman field to give rise to a topological superconducting phase . we show that the spin - orbit coupling is the crucial parameter for the quantum transition into and the robustness of the topologically non - trivial superconducting phase realized for @xmath9 .
quantum gravity with a non - vanishing cosmological constant formulated in ashtekar s spin - connection variables @xcite has interesting physical states given by the exponential of the chern - simons functional @xcite and appropriate transformations thereof . in order to elucidate the physical meaning of such states it is interesting to consider their restrictions to spatially homogeneous cosmological models . in a recent paper @xcite , henceforth quoted as [ i ] , we considered the diagonal bianchi ix models with positive cosmological constant from this point of view and found that five linearly independent physical states in the metric representation could be derived from the chern - simons functional . this set of solutions was found to be in one to one correspondence with the set of topologically different integration contours which exist for the generalized fourier - transformation from the ashtekar - representation to the metric representation . due to the positivity of the cosmological constant the quantum states found in [ i ] describe an expanding ( or collapsing ) classically interpretable lorentzian universe at large scale parameters . on the other hand , at sufficiently small scale parameters the action defined by the exponent of the wavefunction becomes imaginary and can be associated only with a quantum mechanically allowed euclidean universe . the two `` phases '' are separated by a caustic surface in minisuperspace . it was found that only one of the five linearly independent states defines a normalizable probability distribution on this caustic , that this state satisfies the no - boundary condition of hartle and hawking @xcite semi - classically for @xmath2 ( which means on scales large compared to the planck scale ) , and that , again for @xmath2 , it picks out an initial condition which evolves into a classically interpretable lorentzian universe . for details and further literature we refer to [ i ] . it is now of interest to consider also what happens for negative cosmological constant , even though it seems very unlikely that our universe has @xmath3 ( the age - problem resulting from recently measured high values of the hubble - parameter @xcite and the measured large age of globular star - clusters and also the observed high density of galaxies with large red - shifts seem to call for @xmath1 ) . our motivation is rather to try the method of [ i ] for a model - universe which recollapses , i.e. for which quantum - mechanically a classically interpretable lorentzian evolution phase is bounded for small _ and _ large values of the scale parameter by euclidean evolution phases . therefore , we have to expect the appearance of two caustic surfaces in the minsuperspace of these models , one at small and the other at large scale parameter . is there still a wavefunction , or are there even several , which give a normalizable probability distribution on these surfaces , and how are these wavefunctions related to the no - boundary state ? to answer these questions we apply in section ii the method of [ i ] to obtain expressions for again five linearly independent physical states and identify the caustic surfaces in minisuperspace . in section iii we determine the behavior of the absolute square of the wavefunctions on the caustic and identify a single physical state which gives a normalizable probability distribution in this way . in section iv we summarize our results . within the narrow class of models we consider here they seem to rule out , with high probability , a classically evolving universe with @xmath3 . in this section we want to construct solutions of the wheeler - dewitt equation for the diagonal bianchi type ix model with a cosmological constant @xmath3 , @xmath4 @xmath5 + in this notation @xmath6 and @xmath7 denote derivatives with respect to the variables @xmath8 and @xmath9 , respectively . by writing the wheeler - dewitt equation in the form ( [ 1.0 ] ) we have assumed a specific factor - ordering , which is suggested by a supersymmetric extension of the model @xcite . a different factor - ordering is obtained by considering ( [ 1.0 ] ) with @xmath10 replaced by @xmath11 . in the present paper , as in [ i ] , we will restrict ourselves to the factor - ordering as in ( [ 1.0 ] ) , while a brief comment on the solutions in the second case @xmath12 is given in appendix a. if the expression ( [ 1.0 + ] ) for @xmath10 is inserted into the wheeler - dewitt equation ( [ 1.0 ] ) , the following more explicit form is obtained @xmath13 + where we have introduced the mean scale factor @xmath14 . as in the case @xmath15 , solutions of ( [ 1.1 ] ) can be obtained by a transformation to the ashtekar representation , where the chern - simons functional , restricted to the bianchi type ix case , turns out to be an exact solution . two of the fourier integrals which occur in the transformation back to the metric representation can be carried out analytically without any loss of generality and afterwards the same one dimensional integral representation as in [ i ] is obtained : , the _ total _ action , including the part which effects the similarity transformation between ashtekar and metric variables , has been defined as the exponent of the integrand . ] @xmath16 @xmath17 + here we have introduced the new variable @xmath18 and parameter @xmath19 @xmath20 + thereby effectively reducing the number of parameters occuring in ( [ 1.2 ] ) , and we shall also make use of the variables @xmath21 defined by @xmath22 + where the @xmath23 are the entries of the diagonal anisotropy matrix @xmath24 . the integration contour @xmath25 in the integral - representation ( [ 1.2 ] ) can be chosen quite freely , as long as a sufficiently strong fall - off for the integrand and its @xmath26-derivatives at the borders @xmath27 of @xmath25 is guaranteed . the proportionality factor left open in ( [ 1.2 ] ) may depend on @xmath19 and will be fixed later . while in the case @xmath15 the curves of steepest _ descent _ of @xmath28 were of interest , now , due to the different sign of @xmath0 , the curves of steepest _ ascent _ lead to suitable integration contours . moreover , new possibilities for the location of the saddle - points occur , which we classify as follows : * by choosing @xmath29 sufficiently small at fixed @xmath30 , it is always possible to make all five saddle points of @xmath31 lie on the real axis of the complex @xmath32-plane , defining the _ euclid i_-region of minisuperspace . note , however , that the corresponding points in the @xmath26-plane of fig.[grcsd- ] ( here @xmath33 ) are real - valued only for @xmath34 , whereas real @xmath32-values with @xmath35 are mapped into complex conjugate pairs of points on the axes @xmath36 and periodic repetitions thereof . * except for the case @xmath37 , where all five saddle - points are on the real @xmath32-axis , there is the possibility for two of the saddle - points to become complex in the @xmath32-plane , which defines the _ lorentzian regime_. * for large values of @xmath38 one always enters the _ euclid ii _ region , where again all five saddle - points of @xmath31 become real - valued . some typical locations of the saddle - points in these different regimes of minisuperspace and the corresponding curves of steepest ascent are presented in fig.[grcsd- ] . 0 cm by passing from one of these regions to another , a _ situation occurs , where two of the saddle - points confluate . we will refer to the corresponding hypersurface in minisuperspace as the _ caustic _ ; it has been calculated and is plotted in fig.[grkaustik- ] . in contrast to the case @xmath15 the caustic obtained here consists of an upper and a lower branch , which are connected just by a single _ point _ at @xmath39 . furthermore , there are _ kinks _ at @xmath40 and also at the other half - rays of the @xmath30-plane , related to the former by the typical @xmath30-symmetries of diagonal bianchi ix . obviously , an exactly isotropic universe @xmath41 has to stay purely euclidean throughout its evolution . on the other hand , `` large '' universes with lorentzian geometry must become very anisotropic . apart from the possibility of a negative cosmological constant very close to zero , which would allow for large scale parameters even at @xmath29-values of order one , it seems impossible for the model under investigation to describe the universe observed today . 0 cm nevertheless , let us now construct a basis of solutions to the wheeler - dewitt equation ( [ 1.1 ] ) by choosing topologically independent integration contours @xmath25 in the representation ( [ 1.2 ] ) . 0 cm using the curves defined in fig.[grintc ] we introduce the following solutions @xmath42 @xmath43 + which , by definition , are real - valued and normalized in accordance with @xmath44 + the functions @xmath45 and @xmath46 occuring in ( [ 1.5 ] ) are the usual modified bessel functions with index @xmath47 . it will be of some advantage to replace the solutions @xmath48 and @xmath49 by @xmath50 + then saddle - point expansions of the integrals ( [ 1.5 ] ) in the limit @xmath51 , @xmath52 and @xmath30 fixed , reveal @xmath53 + where the upper index `` @xmath47 '' denotes the solutions of the @xmath54-model given in [ i ] . without proof we mention that @xmath55 , are three asymmetric solutions which generate each other by cyclic permutations of the @xmath21 , so consequently the sum of these states , @xmath56 + besides @xmath57 and @xmath58 , turns out to be symmetric with respect to arbitrary @xmath21-permutations . up to different normalization factors , the asymptotic behavior in the limit @xmath59 can immediately be extracted from the corresponding expansions in [ i ] by taking account of the negative sign of @xmath0 . the only difficulty lies in the determination of the saddle - points , which give the dominating contribution to the different solutions in this limit . a detailed calculation finally yields @xmath60 @xmath61 + at @xmath41 , i.e. while @xmath57 falls of rapidly for @xmath62 , the wavefunction @xmath58 is strongly divergent in the same limit . moreover , since @xmath58 _ always _ gets its dominant contribution from the real saddle - point @xmath63 ( corresponding to the points @xmath64 and @xmath65 in the complex @xmath26-plane of fig.[grcsd- ] via @xmath66 ) , just euclidean geometries are described by this state , so we will reject @xmath58 as a physically relevant solution . note , however , that it is the only state which satisfies the _ no - boundary _ condition in the limit @xmath67 , hence the name of this wavefunction . to give the asymptotic behavior in the limit @xmath59 for the states @xmath55 , it will be helpful to consider @xmath68 + instead . for these solutions the asymptotic expansions @xmath69 + hold , so they fall off very rapidly for @xmath62 ( remember the negative sign of @xmath19 ! ) . by considering additional asymptotic expansions for large anisotropy it is possible to show that the four states @xmath70 , are all normalizable on minisuperspace in the distribution sense ( see [ i ] for a discussion of this point for @xmath1 ) , i.e. so far we are left with a still four dimensional space of physically interesting solutions . however , while in the lorentzian regime @xmath57 receives saddle - point contributions exclusively from the saddle - points at _ complex _ @xmath32 and thus describes a lorentzian universe in this part of minisuperspace , the states @xmath55 , get additional euclidean contributions of similar order of magnitude from _ real _ saddle - points and are therefore hard to interpret . the classical trajectories which are generated by @xmath57 in the semi - classical limit @xmath2 _ in the lorentzian regime _ can be computed by solving the equations @xmath71 + where we have chosen the lapse - function to be @xmath72 . while the complex saddle - point @xmath73 occuring in ( [ 1.13 ] ) is intended to correspond to the point @xmath74 of fig.[grcsd- ] , the complex conjugate saddle - point @xmath75 , which describes the time - reversed classical evolution , can be considered with the same right . the corresponding second branch of the classical evolution of the universe is actually _ needed _ to define the continuation of a classical trajectory which has reached the caustic : in approaching the caustic the saddle - points @xmath73 and @xmath76 confluate and become real - valued , so that , in accordance with ( [ 1.13 ] ) , the time - derivatives of @xmath77 and @xmath30 vanish . to continue such a trajectory in time , the time - reversed version of ( [ 1.13 ] ) has to be considered . since the universe is `` reflected '' in this way whenever it meets the caustic , and since in the generic case the classical trajectories have both of their endpoints on the caustic , _ oscillating _ universes are described by @xmath57 . the numerical results for the classical trajectories which are obtained in the plane @xmath78 of the minisuperspace are presented in fig.[class.traj ] . 0 cm we should stress one important difference in the pictures which are obtained for the different signs of @xmath8 : while for @xmath79 all trajectories run to infinite anisotropy ( which is , indeed , a peculiarity of the special @xmath30-direction , corresponding to a kink on the caustic , cf . fig.[grkaustik- ] ) , in the case @xmath80 the trajectories meet the lower branch of the caustic again at a finite @xmath8-value , representing the general situation . this feature gives rise to the existence of a special trajectory with coinciding start- and endpoints , hence describing a universe that never really becomes lorentzian . the corresponding points in minisuperspace can be calculated analytically , requiring the solution of ( [ 1.13 ] ) to be _ tangential _ to the caustic , with the result @xmath81{2}\,\left ( \frac{2}{5 } \right ) ^ { \frac{4}{3}}\ , \qquad \beta_{+}+i\,\beta_{-}=\frac{1}{6 } ( \ln 5 - 5 \ln 2 ) \ , e^{\mbox{{\footnotesize $ \frac{2 \pi i n}{3}$}}}\ , \qquad n\,\epsilon\,\{-1,0,1\}\ .\ ] ] + these points will play an important role in the following section . since the classical lorentzian evolution of the universe described by the wavefunctions @xmath70 , is bounded by the caustic in minisuperspace , the value of @xmath82 _ on the caustic _ predicted by the different solutions is of particular interest . in fact @xmath82 governs the realization of the different possible histories of the universe and may thus be interpreted as the `` initial '' value distribution for the classical evolution . however , at this stage a new problem arises due to the different branches of the caustic . since the semi - classical trajectories allways can be passed through in both directions , it is impossible to distinguish between their start- and endpoints . the distributions of @xmath82 on the upper and lower branch of the caustic may therefore be considered with the same right , and we will always discuss them together in the following . the numerical results obtained for @xmath83 and @xmath84 on the lower caustic are given in fig.[grkaustl ] , and fig.[grkaustu ] shows the behavior on the upper caustic , which is very similar for the two different solutions . in the following the additional indices `` @xmath26 '' and `` @xmath85 '' denote the upper and lower branch of the caustic , respectively . 0 cm 1.5 cm 0 cm while on the upper caustic both distributions fall off rapidly with increasing @xmath30 and may be shown to be integrable over the @xmath30-plane , there are @xmath30-directions on the lower caustic , in which @xmath86 approaches a finite value at infinity . consequently , the wavefunctions on the lower branch are _ not _ square - integrable , and hence difficult to interpret as probability distributions . nevertheless , as in the case @xmath1 , one may construct a new wavefunction as a linear combination of the two symmetric wavefunctions @xmath57 and @xmath87 : by normalizing @xmath57 and @xmath87 to approach unity in the critical @xmath30-directions , the difference of these new functions is square - integrable on the _ full _ caustic . to give an explicit expression for the quantum state obtained in this way , we introduce the integrals @xmath88 @xmath89 + which , as far as we know , have no simple representation in terms of tabulated functions . defining now @xmath90 + the new state can be written in the form @xmath91 + where the overall normalization factor has again been chosen to make @xmath92 at @xmath93 . the behavior of @xmath94 on the caustic has been computed for fig.[grkausth ] . taking account of the full distribution , three maxima on the lower branch of the caustic pick out special initial values for the classical evolution of the universe . the general representation ( [ 1.2 ] ) easily reveals that the wavefunction becomes arbitrarily sharply concentrated about these maxima in the limit @xmath95 , i.e. in particular in the limit @xmath2 at fixed @xmath0 . consequently , in the semi - classical limit there are just three histories of the universe which occur with significant probability . 0 cm in the following we shall be interested in the special points of minisuperspace where the maxima of @xmath96 arise . using the saddle - point method for @xmath95 the asymptotic behavior of the integrals defined in ( [ 2.1 ] ) after some calculation yields @xmath97 + and with this result the relation @xmath98 + can be shown to hold at least on the lower caustic . since @xmath94 is a real - valued , non - vanishing wavefunction , the maxima of @xmath99 coincide with the maxima of @xmath94 , and using ( [ 2.5 ] ) they may also be calculated from @xmath57 in the semi - classical limit . by performing again a saddle - point expansion for @xmath95 , now in the integral representation ( [ 1.5 ] ) of @xmath57 , the maxima of @xmath57 on the caustic can be calculated analytically with the result given exactly by ( [ 1.14 ] ) . consequently , within the class of solutions considered here , the _ only _ quantum state that is square - integrable on the full caustic turns out to predict a universe which never becomes lorentzian in the classical limit ( albeit classical lorentzian solutions of the bianchi type ix model with negative cosmological constant actually do exist , cf . fig.[class.traj ] ) . in the present paper we constructed exact quantum states for the diagonal bianchi type ix model with a negative cosmological constant . we found that the method presented in [ i ] for @xmath1 is indeed perfectly applicable to the model with @xmath3 . as for @xmath1 it gives five linearly independent solutions , which are generated by the chern - simons state using topologically different integration contours in the generalized fourier - transformation to the metric representation . imposing the condition that the wavefunction be normalizable on the caustic , just one wavefunction remains , that turns out to have some nice additional properties : it is found to be normalizable in minisuperspace in the distribution sense and it respects the symmetries of the bianchi type ix model . however , this state does _ not _ satisfy the no - boundary condition in the semi - classical limit in contrast to the case @xmath1 , and it turns out to predict a universe that never becomes lorentzian , after all . hence we obtain the result , that , _ if _ one allows for a non - zero cosmological constant at all , it should be positive , at least as far as the chern - simons functional related states of the quantized bianchi ix model are concerned . support of this work by the deutsche forschungsgemeinschaft through the sonderforschungsbereich `` unordnung und groe fluktuationen '' is gratefully acknowledged . for completeness , and in order to obtain an important argument for the factor - ordering chosen for the wheeler - dewitt equation ( [ 1.1 ] ) , we shall make some comments on a further class of solutions , which can again be discussed by using the methods of [ i ] . considering the wheeler - dewitt equation in the form ( [ 1.0 ] ) one may ask , why we have not chosen a different factor - ordering , which is obtained by changing @xmath100 . this choice , of course , would not have affected the classical hamiltonian , but the quantum correction @xmath101 in ( [ 1.1 ] ) would have changed its sign .- representation introduced by kodama in @xcite , in contrast to the @xmath102-representation , which we have considered up to now . ] since the coordinate transformation @xmath103 has exactly the same effect as the above mentioned change of the factor - ordering , it is possible to discuss the solutions of the wheeler - dewitt equation in the new factor - ordering by considering still equation ( [ 1.1 ] ) , but substituting formally @xmath104 in the solutions . in the following it will be more convenient to use the coordinates @xmath21 and @xmath19 introduced in ( [ 1.4 ] ) and ( [ 1.4 + ] ) , which transform like @xmath105 under this substitution . it should be clear that the solutions of the wheeler - dewitt equation ( [ 1.1 ] ) are still of the form ( [ 1.2 ] ) , but while we looked at the cases @xmath106 in [ i ] and @xmath107 in the present paper , now the remaining sectors @xmath108 and @xmath109 are of interest , which , because of the formal substitution @xmath110 just mentioned , describe solutions for _ positive _ and _ negative _ cosmological constant in the _ new _ factor - ordering , respectively . it is easily checked that the location of the saddle - points , and therefore the caustic , depends only on @xmath111 defined in eq . ( [ 1.3 ] ) . this means that , irrespective of the sign of @xmath19 , we deal with the caustic of [ i ] in the case @xmath112 , and with the caustic fig.[grkaustik- ] in the case @xmath113 . on the other hand , it is just the sign of @xmath19 which decides whether the integration curves of [ i ] ( for @xmath114 ) or of fig.[grintc ] ( for @xmath115 ) give suitable integration contours . however , constructing the solutions for the new factor - ordering in this manner and applying the saddle - point method to the integral representation ( [ 1.2 ] ) in the limit of large anisotropy @xmath30 , it finally turns out , that _ any _ solution to the wheeler - dewitt equation in the new factor - ordering _ diverges _ for @xmath116 , at least in some @xmath30-sectors . in other words , in the new factor - ordering there is no solution which is normalizable in minisuperspace , not even in the distribution sense . furthermore , if the behavior of the wavefunctions on the caustic is considered , actually none of these solutions is found to be square - integrable with respect to @xmath30 . comparing these results with the nice normalizability properties of the solutions of the wheeler - dewitt equation in the factor - ordering of ( [ 1.0 ] ) presented in [ i ] and the present paper , we believe to have a compelling argument to rule out the new factor - ordering . it would be interesting if this argument could be extended even to the general , inhomogeneous case of quantum gravity .	quantum states of the diagonal bianchi type ix model with _ negative _ cosmological constant @xmath0 are obtained by transforming the chern - simons solution in ashtekar s variables to the metric representation . we apply our method developed earlier for @xmath1 and obtain five linearly independent solutions by using the complete set of topologically inequivalent integration contours in the required generalized fourier - transformation . a caustic in minisuperspace separates two euclidean regimes at _ small _ and _ large _ values of the scale parameter from a single classically interpretable lorentzian regime in between , corresponding to the fact that classically these model - universes recollapse . just one particular solution out of the five we find gives a normalizable probability distribution on both branches of the caustic . however , in contrast to the case of positive cosmological constant , this particular solution neither satisfies the semi - classical no - boundary condition , nor does the special initial condition it picks out for @xmath2 evolve into a classically interpretable universe .
in the paper @xcite l. de branges initiated the study of hilbert spaces of entire functions , which satisfy specific additional axioms . these spaces can be viewed as a generalization of the classical paley wiener spaces @xmath2 , which consist of all entire functions of exponential type at most @xmath3 whose restriction to the real line is square - integrable . the theory of de branges spaces can be viewed as a generalization of classical fourier analysis . for example , their structure theory gives rise to generalizations of the paley wiener theorem , which identifies @xmath2 as the fourier image of all square - integrable functions supported in the interval @xmath4 $ ] . de branges spaces also appear in many other areas of analysis , like the theory of volterra operators and entire operators in the sense of m.g . kren , the shift operator in the hardy space , v.p . potapov s @xmath5-theory , the spectral theory of schrdinger operators , stieltjes or hamburger power moment problems , or prediction theory of gaussian processes , cf . @xcite , @xcite , @xcite , @xcite , @xcite , @xcite . the present paper is the first part of a series , in which we investigate the aspect of majorization in de branges spaces . such considerations have a long history in complex analysis , going back to the beurling malliavin multiplier theorem , cf . @xcite . in recent investigations by v. havin and j. mashreghi , results of this kind were proven in the more general setting of shift - coinvariant subspaces of the hardy space , cf . @xcite , @xcite . all these considerations , as well as our previous work @xcite , deal with majorization along the real line . having these concepts in mind , a general notion of majorization in de branges spaces evolves : a1 let @xmath6 be a de branges space , and let @xmath7 where @xmath8 . set @xmath9 and define @xmath10 it turns out that , provided @xmath11 and @xmath12 satisfies a mild regularity condition , the space @xmath13 is a de branges subspace of @xmath6 , i.e. is itself a de branges space when endowed with the inner product inherited from @xmath6 . the following questions related to this concept come up naturally . + * _ which de branges subspaces @xmath14 of a given de branges space @xmath6 can be realized as @xmath15 with some majorant @xmath12 ? _ + * _ if @xmath14 is of the form @xmath13 with some @xmath12 , how big or how small can @xmath12 be chosen such that still @xmath15 ? _ + let us point out the two aspects of the second question . if @xmath15 , we have available a dense linear subspace of @xmath14 which consists of functions with limited growth on the domain @xmath0 of @xmath12 , namely @xmath16 . this knowledge becomes stronger , the smaller @xmath12 is . on the other hand , the equality @xmath15 also says that an element of @xmath6 already belongs to @xmath14 if it is majorized by @xmath12 . this knowledge becomes stronger , the bigger @xmath12 is . answers to these questions will , of course , depend on the set @xmath0 where majorization is permitted . up to now , only majorization along @xmath17 has been considered . for this case , the first question has been answered completely in @xcite . the `` how small''part of the second question is related to the deep investigations in @xcite , @xcite . in this paper we give some answers to the first question , and to the `` how big''part of the second question . as domains @xmath0 of majorization we consider , among others , rays contained in the closed upper half - plane , lines parallel to the real axis contained in the closed upper half - plane , or combinations of such types of sets . for example , it turns out that each de branges subspace @xmath14 of any given de branges space @xmath6 can be realized as @xmath13 , when majorization is allowed on @xmath18 . even more , one can choose for @xmath12 a majorant which is naturally associated to @xmath14 , does not depend on the external space @xmath6 , is quite big , and actually gives @xmath19 . it is an interesting and , on first sight , maybe surprising consequence of de branges theory , that the main strength of majorization is contributed by boundedness along the imaginary half - line , and not along @xmath17 . in fact , if we permit majorization only on some ray @xmath20 where @xmath21 , then all de branges subspaces subject to an obvious necessary condition can be realized in the way stated above . similar phenomena , where growth restrictions on the imaginary half - axis imply a certain behaviour along the real line , have already been experienced in the classical theory , see e.g. ( * ? ? ? * theorem 2 ) or ( * ? ? ? * theorem 26 ) . let us close this introduction with an outline of the organization of this paper . in order to make the presentation as self - contained as possible , we start in section 2 with recalling some basic definitions and collecting some results which are essential for what follows , among them , the definition of de branges spaces of entire functions , their relation to entire functions of hermite biehler class , and the structure of de branges subspaces . in section 3 , we make precise under which conditions on @xmath12 the space @xmath13 becomes a de branges subspace of @xmath6 , and discuss some examples of majorants . sections 4 and 5 contain the main results of this paper . first we deal with representation of de branges subspaces by majorization along rays not parallel to the real axis . then we turn to spaces @xmath13 obtained when majorization is required on a set close to the real axis , for example a line parallel to @xmath17 . the paper closes with two appendices . in the first appendix we prove an auxiliary result on model subspaces generated by inner functions , which is employed in section 5 . we decided to move this theorem out of the main text , since it is interesting on its own right and independent of the presentation concerning de branges spaces . in the second appendix , we are summing up the representation theorems for de branges subspaces obtained in sections 4 and 5 in tabularic form . * acknowledgements . * we would like to thank alexei poltoratski who suggested the use of weak type estimates for the proof of theorem [ tha60 ] . the first author was partially supported by the grants mk-5027.2008.1 and nsh-2409.2008.1 . * i. mean type and zero divisors * we will use the standard theory of hardy spaces in the half - plane as presented e.g. in @xcite or @xcite . in this place , let us only recall the following notations . we denote by 1 . @xmath22 the set of all functions of _ bounded type _ , that is , of all functions @xmath23 analytic in @xmath24 , which can be represented as a quotient @xmath25 of two bounded and analytic functions @xmath26 and @xmath27 . 2 . @xmath28 the _ smirnov class _ , that is , the set of all functions @xmath23 analytic in @xmath24 , which can be represented as @xmath25 with two bounded and analytic functions @xmath26 and @xmath27 where in addition @xmath26 is _ outer_. 3 . @xmath29 the _ hardy space _ , that is , the set of all functions @xmath23 analytic in @xmath24 which satisfy @xmath30 if @xmath31 , the _ mean type of @xmath23 _ is defined by the formula @xmath32 then @xmath33 , and the radial growth of @xmath23 is determined by the number @xmath34 in the following sense : for every @xmath35 and @xmath36 , there exists an open set @xmath37 with finite logarithmic length , such that @xmath38 uniformly for @xmath39 $ ] . if , for some @xmath40 , the angle @xmath41 $ ] does not contain any zeros of @xmath42 , then one can choose @xmath43 . here we understand by the logarithmic length of a subset @xmath44 of @xmath45 the value of the integral @xmath46 . when speaking about logarithmic length of a set @xmath44 , we always include that @xmath44 should be measurable . a7 let @xmath47 be a function defined on some subset @xmath0 of the complex plane . 1 . by analogy with we define the _ mean type of @xmath12 _ as @xmath48 \,,\ ] ] where the infimum is taken over those values @xmath35 , @xmath49 , and those sets @xmath50 of infinite logarithmic length , for which @xmath51 . thereby we understand the infimum of the empty set as @xmath52 . we associate to @xmath12 its _ zero divisor _ @xmath53 . if @xmath54 , then @xmath55 is defined as the infimum of all numbers @xmath56 , such that there exists a neighbourhood @xmath57 of @xmath58 with the property @xmath59 note that in general @xmath60 may take the values @xmath61 . however , the above definition ensures that @xmath60 coincides with the classical notion in case @xmath62 . a similar remark applies to @xmath63 . if @xmath0 is open , and @xmath12 is analytic , then @xmath64 is just the usual zero divisor of @xmath12 , i.e. @xmath55 is the multiplicity of the point @xmath58 as a zero of @xmath12 whenever @xmath65 . moreover , note that the definition of @xmath63 is made in such a way that @xmath66 whenever @xmath67 . axiomatics of de branges spaces of entire functions * our standard reference concerning the theory of de branges spaces of entire functions is @xcite . in this and the following two subsections we will recall some basic facts about de branges spaces . our aim is not only to set up the necessary notation , but also to put emphasis on those results which are significant in the context of the present paper . we start with the axiomatic definition of a de branges space . a14 a _ de branges space _ is a hilbert space @xmath68 , @xmath69 , with the following properties : 14 mm the elements of @xmath6 are entire functions , and for each @xmath54 the point evaluation @xmath70 is a continuous linear functional on @xmath6 . if @xmath71 , also @xmath72 belongs to @xmath6 and @xmath73 . if @xmath74 and @xmath71 , @xmath75 , then @xmath76 by ( db1 ) a de branges space @xmath6 is a reproducing kernel hilbert space . we will denote the kernel corresponding to @xmath54 by @xmath77 or , if it is necessary to be more specific , by @xmath78 . a particular role is played by the norm of reproducing kernel functions . we will denote @xmath79 this norm can be computed e.g. as @xmath80 let us explicitly point out that every element of @xmath6 is majorized by @xmath81 : by the schwarz inequality we have @xmath82 a29 let @xmath6 be a de branges space . for a subset @xmath83 we define @xmath84 as @xmath85 due to the axiom ( db3 ) , we have @xmath86 , @xmath74 . in fact , if @xmath71 and @xmath58 is a nonreal zero of @xmath87 , then @xmath88 . this need not be true for real points @xmath58 . however , one can show that , if @xmath89 and @xmath90 , then @xmath88 . a30 let @xmath6 be a de branges space , and let @xmath47 be a function defined on some subset @xmath0 of the complex plane . we define the _ mean type of @xmath12 relative to @xmath6 _ by @xmath91 if @xmath92 is a subset of @xmath6 , the _ mean type of @xmath92 relative to @xmath6 _ is @xmath93 note that , by , we have @xmath94 . for each @xmath95 the set @xmath96 is closed , cf . this implies that always @xmath97 . a31 for a de branges space @xmath6 let @xmath98 denote the operator of multiplication by the independent variable . that is , @xmath99 the relationship between de branges spaces and entire operators in the sense of m.g . kren is based on the fact that @xmath98 is a closed symmetric operator with defect index @xmath100 for which every complex number is a point of regular type . a24 taking up the operator theoretic viewpoint , the role played by _ functions associated to @xmath6 _ can be explained neatly . for a de branges space @xmath6 , the set of functions associated to @xmath6 can be defined as @xmath101 clearly , @xmath102 is a linear space which contains @xmath6 . the space @xmath102 can be used to describe the extensions of @xmath98 by means of difference quotients . we have @xmath103 moreover , for each @xmath104 and @xmath54 , @xmath105 , the difference quotient operator @xmath106 is a bounded linear operator of @xmath6 into itself , actually , the resolvent of some extension of @xmath98 . let us note that , if @xmath87 is not only associated to @xmath6 but belongs to @xmath6 , we have @xmath107 , @xmath105 . * iii . de branges spaces and hermite biehler functions * it is a basic fact that a de branges space @xmath6 is completely determined by a single entire function . a22 we say that an entire function @xmath108 belongs to the _ hermite biehler class @xmath109 _ , if @xmath110 if @xmath111 , define @xmath112 and @xmath113 instead of @xmath114 one could , equivalently , require that @xmath115 and @xmath116 are of bounded type and nonpositive mean type in the upper half - plane , and that @xmath117 . this is , in fact , the original definition in @xcite . the relation between de branges spaces and hermite - biehler functions is established by the following fact : a28 for every function @xmath111 , the space @xmath118 is a de branges space , and conversely every de branges space can be obtained in this way . the function @xmath111 which realizes a given de branges space @xmath68 as @xmath118 is not unique . however , if @xmath119 and @xmath120 , then there exists a constant @xmath121-matrix @xmath44 with real entries and determinant @xmath122 , such that @xmath123 here , and later on , we use the generic decomposition of a function @xmath111 as @xmath124 with @xmath125 for each two function @xmath119 with @xmath126 , there exist constants @xmath127 such that @xmath128 the notion of a phase function is important in the theory of de branges spaces . for @xmath111 , a _ phase function _ of @xmath108 is a continuous , increasing function @xmath129 with @xmath130 , @xmath131 . a phase function @xmath132 is by this requirement defined uniquely up to an additive constant which belongs to @xmath133 . its derivative is continuous , positive , and can be computed as @xmath134 where @xmath135 are zeros of @xmath108 listed according to their multiplicities , and @xmath136 . a25 let @xmath68 be a de branges space , and let @xmath111 be such that @xmath137 . then all information about @xmath6 can , theoretically , be extracted from @xmath108 . in general this is a difficult task , however , for some items it can be done explicitly . for example : 1 . the reproducing kernel @xmath77 of @xmath6 is given as @xmath138 in particular , this implies that @xmath139 . 2 . we have @xmath140 . this equality even holds if we only assume that @xmath141 as sets , i.e. , without assuming equality of norms . the function @xmath81 is given as @xmath142 in particular , we have @xmath143 . we have @xmath144 this follows from the estimates ( with @xmath145 fixed ) @xmath146 which are deduced from the inequality @xmath147 and . * iv . structure of db - subspaces * the , probably , most important notion in the theory of de branges spaces is the one of de branges subspaces . a43 a subset @xmath14 of a de branges space @xmath6 is called a _ db - subspace _ of @xmath6 , if it is itself , with the norm inherited from @xmath6 , a de branges space . we will denote the set of all db - subspaces of a given space @xmath6 by @xmath148 . if @xmath149 , we set @xmath150 since db - subspaces with @xmath151 appear quite frequently , we introduce the shorthand notation @xmath152 . it is apparent from the axioms ( db1)(db3 ) of definition [ dea14 ] that a subset @xmath14 of @xmath6 is a db - subspace if and only if the following three conditions hold : 1 . @xmath14 is a closed linear subspace of @xmath6 ; 2 . if @xmath153 , then also @xmath154 ; 3 . if @xmath153 and @xmath155 is such that @xmath156 , then @xmath157 . a26 some examples of db - subspaces can be obtained by imposing conditions on real zeros or on mean type . if @xmath149 , @xmath158 , is a function such that @xmath159 for some @xmath160 , then @xmath161 we have @xmath162 . if @xmath95 is such that @xmath163 for some @xmath160 , then @xmath164 and we have @xmath165 . those db - subspaces which are defined by mean type conditions will in general not exhaust all of @xmath166 . however , sometimes , this also might be the case . trivially , the set @xmath148 , and hence also each of the sets @xmath167 , is partially ordered with respect to set - theoretic inclusion . one of the most fundamental and deep results in the theory of de branges spaces is the _ ordering theorem for subspaces of @xmath6 _ , cf . * theorem 35 ) where even a somewhat more general version is proved . a35 let @xmath6 be a de branges space and let @xmath149 . then @xmath167 is totally ordered . the chains @xmath167 have the following continuity property : for a db - subspace @xmath14 of @xmath6 , set @xmath168 @xmath169 then @xmath170 a27 let us explicitly mention two examples of de branges spaces , which show in some sense extreme behaviour . 1 . consider the _ paley wiener space _ @xmath2 where @xmath171 . this space is a de branges space . it can be obtained as @xmath172 with @xmath173 . the chain @xmath174 is equal to @xmath175 apparently , we have @xmath176 , and hence in this example all db - subspaces are obtained by mean type restrictions . 2 . in the study of the indeterminate hamburger moment problem de branges spaces occur which contain the set of all polynomials @xmath177 $ ] as a dense linear subspace , see e.g. @xcite , @xcite , @xcite + if @xmath6 is such that @xmath178 $ ] , then the chain @xmath166 has order type @xmath179 . in fact , @xmath180_n:\,n\in{{\mathbb{n}}}_0\big\}\cup\{{{\mathcal{h}}}\ } \,,\ ] ] where @xmath177_n$ ] denotes the set of all polynomials whose degree is at most @xmath181 . examples of de branges spaces @xmath6 for which the chain @xmath166 has all different kinds of order types can be constructed using canonical systems of differential equations , see e.g. ( * ? ? ? * theorems 37,38 ) , @xcite , or @xcite . with help of the estimates , it is easy to see that @xmath182 this implies that for any db - subspace @xmath14 of @xmath6 the supremum in the definition of @xmath183 is attained ( e.g. on the reproducing kernel functions @xmath184 ) . moreover , we obtain @xmath185 . also the fact whether a given de branges space @xmath14 is contained in @xmath6 as a db - subspace can be characterized via generating hermite biehler functions . choose @xmath186 with @xmath141 and @xmath187 . then @xmath188 if and only if there exists a @xmath121-matrix function @xmath189 such that the following four conditions hold : 1 . the entries @xmath190 of @xmath191 are entire functions , satisify @xmath192 , and @xmath193 . the kernel @xmath194 is positive semidefinite . 3 . write @xmath124 and @xmath195 according to . then @xmath196 4 . denote by @xmath197 the reproducing kernel hilbert space of @xmath198-vector functions generated by the kernel @xmath199 . then there exists no constant function @xmath200 with @xmath201 . assuming that @xmath188 , the orthogonal complement of @xmath14 in @xmath6 can be described via the above matrix function @xmath191 . in fact , the map @xmath202 is an isometric isomorphism of @xmath197 onto @xmath203 . let us remark in this context that a function @xmath204 can also be written in the form @xmath205 and vice versa . a detailed discussion of the situation when such functions belong to @xmath172 and a criterion in terms of the zeros of @xmath108 can be found e.g. in @xcite . let us discuss in a bit more detail the particular situation that @xmath188 and @xmath206 , cf . * theorem 29 , problem 87 ) . in this case choose @xmath186 with @xmath141 and @xmath187 . then the matrix @xmath191 introduced in the above item is a linear polynomial of the form @xmath208 where @xmath209 , @xmath210 , and @xmath44 is a constant @xmath121-matrix with real entries and determinant @xmath122 . the space @xmath197 is spanned by the constant function @xmath211 . we see that @xmath212 in what follows we will work with functions @xmath12 which are defined on subsets of the closed upper half - plane and take nonnegative real values . to simplify notation we will often drop explicit notation of the domain of definition of the function @xmath12 . if @xmath213 , we write @xmath214 if there exists a positive constant @xmath215 , such that @xmath216 , @xmath217 . moreover , @xmath218 stands for `` @xmath214 and @xmath219 '' . using this notation , we can write @xmath220 compare with definition [ dea1 ] . our first aim is to show that @xmath13 will become a de branges subspace of @xmath6 , whenever @xmath221 , and @xmath12 satisfies an obvious regularity condition . a4 let @xmath6 be a de branges space , and let @xmath7 be a function with @xmath8 . then @xmath222 if and only if @xmath12 satisfies 16 mm @xmath223 ; @xmath16 contains a nonzero element . in this case we have @xmath224 necessity of the conditions ( adm1 ) and ( adm2 ) is easy to see . in the proof of sufficiency we will employ the following elementary lemma . a3 let @xmath6 be a de branges space and let @xmath92 be a nonzero linear subspace of @xmath6 such that : 1 . if @xmath225 , then also @xmath226 ; 2 . if @xmath225 and @xmath155 with @xmath156 , then also @xmath227 then @xmath228 . the mapping @xmath229 is continuous on @xmath6 . we have @xmath230 , and hence @xmath231 . let @xmath232 , @xmath155 with @xmath156 , be given . we have to show that @xmath233 or , equivalently , that @xmath234 . choose an element @xmath235 with @xmath236 . such a choice is possible by @xmath237 . the mapping @xmath238 is continuous . moreover , by @xmath237 , we have @xmath239 . thus @xmath240 in particular , if @xmath232 and @xmath156 , then @xmath234 . together , we conclude that @xmath228 . theorem [ tha4 ] assume first that @xmath222 . then , clearly , @xmath221 , i.e. ( adm2 ) holds . let @xmath54 , and choose @xmath241 with @xmath242 . by analyticity , we have for some disk @xmath57 centred at @xmath58 , @xmath243 since @xmath244 , @xmath245 , we obtain that @xmath246 . it follows that @xmath63 takes only finite values and that @xmath247 is a discrete subset of @xmath17 . in particular ( adm1 ) holds . moreover , we see that @xmath248 for the converse assume that @xmath12 satisfies the conditions and . we will apply lemma [ lea3 ] with @xmath249 . the hypothesis @xmath250 of lemma [ lea3 ] is satisfied by the definition of @xmath16 . let @xmath241 , @xmath54 , and assume that @xmath251 . then @xmath252 . let @xmath57 be a compact neighbourhood of @xmath58 such that @xmath253 for @xmath254 , we have @xmath255 , and hence @xmath256 , @xmath257 . the function @xmath258 is analytic , and hence bounded , on @xmath57 . it follows that @xmath259 and hence @xmath260 the same argument will show that @xmath261 , @xmath217 , and hence @xmath262 . since , by ( adm1 ) , @xmath263 for @xmath264 , we conclude that @xmath16 satisfies the hypothesis @xmath237 of lemma [ lea3 ] . moreover , @xmath265 hence @xmath266 , and @xmath267 . we have proved the asserted equivalence and equality of divisors in . let @xmath241 . then @xmath244 , @xmath217 , and hence @xmath268 . this proves the assertion concerning mean types . theorem [ tha4 ] justifies the following definition . a2 let @xmath6 be a de branges space . a function @xmath7 where @xmath8 , is called an _ admissible majorant for @xmath6 _ if it satisfies the conditions ( adm1 ) and ( adm2 ) of theorem [ tha4 ] . the set of all admissible majorants is denoted by @xmath269 . for the set of all those admissible majorants which are defined on a fixed set @xmath0 , we write @xmath270 . a5 1 . we allow majorization on a subset of the closed upper half - plane . of course , the same definitions could be made for majorants @xmath12 defined on just any subset of the complex plane . however , due to the symmetry with respect to the real line which is included into the definition of @xmath271 , this would not be a gain in generality . majorants defined on bounded sets give only trivial results : if @xmath8 is bounded and @xmath272 , then @xmath273 . in view of this fact , we shall once and for all exclude bounded sets @xmath0 from our considerations . a59 as we have already noted after the definition of @xmath63 , we have @xmath66 whenever @xmath67 . the first formula in hence gives @xmath274 it is worth to notice that this statement can also be read in a slightly different way : assume that @xmath275 is represented as @xmath15 with some @xmath272 . then @xmath276 . assume that @xmath277 and set @xmath278 unless @xmath279 , the subspace @xmath280 will be a db - subspace of @xmath6 . it follows from the above notice that no subspace @xmath275 with @xmath281 can be realized as @xmath13 with some @xmath272 . let us provide some standard examples of admissible majorants . we will mostly work with these majorants . a8 an obvious , but surprisingly important , example of admissible majorants is provided by the functions @xmath282 , @xmath275 . since always @xmath283 , ( adm2 ) is satisfied . also , it follows that @xmath284 and this yields ( adm1 ) . thus @xmath285 . the function @xmath286 is actually for several reasons a distinguished admissible majorant . this will be discussed in more detail later ( see , also , the forthcoming paper @xcite ) . a58 other examples of admissible majorants can be constructed from functions associated to the space @xmath6 . let @xmath287 , and assume that @xmath288 does not vanish identically and does not satisfy @xmath289 , @xmath290 . moreover , let @xmath8 be such that @xmath291 for all @xmath292 . define @xmath293 then @xmath294 . in fact , we have @xmath295 and , if @xmath296 is such that @xmath297 , then the function @xmath298 belongs to @xmath16 . provided @xmath0 containes a part of some ray in @xmath24 with positive logarithmic density , we have @xmath299 a50 let @xmath172 be a de branges space , and let @xmath300 . choose @xmath301 with @xmath187 . then @xmath302 and the conditions required in example [ exa58 ] are fullfilled for @xmath303 . since each kernel function @xmath184 of the space @xmath14 belongs to @xmath304 , we always have @xmath305 . note that the space @xmath304 does not depend on the particular choice of @xmath303 with @xmath187 . in fact , if @xmath303 and @xmath306 both generate the space @xmath14 , then we will have @xmath307 throughout @xmath308 , and hence also @xmath309 . let us state explicitly how the majorants @xmath310 and @xmath286 are related among each other and with the space @xmath14 . by we have @xmath311 . moreover , @xmath312 & \parbox{2mm}{\hspace{1mm}\begin{rotate}{40 } $ \subseteq$ \end{rotate } } & & \hspace{-1mm}\parbox{2mm}{\vspace{-3mm}\begin{rotate}{-40 } $ \subseteq$\end{rotate } } & \\[-3pt ] { { \mathcal{l } } } & & & & \hspace{-2mm}{{\mathcal{r}}}_{\nabla_{\!{{\mathcal{l}}}}}({{\mathcal{h}}})\\[-3pt ] & \parbox{2mm}{\hspace{0mm}\begin{rotate}{-40 } $ \subseteq$ \end{rotate } } & & \hspace{-1mm}\parbox{2mm}{\vspace{3mm}\begin{rotate}{40 } $ \subseteq$\end{rotate } } & \\[-3pt ] & & r_{\nabla_{\!{{\mathcal{l}}}}}({{\mathcal{h } } } ) & & \\ \end{array}\ ] ] in this section we consider subspaces which are generated by majorization along rays being not parallel to the real axis . the following two statements are our main results in this respect . a12 consider @xmath314 where @xmath21 . let @xmath6 be a de branges space , and let @xmath188 . then @xmath315 . a37 consider @xmath316 where @xmath21 and @xmath317 . let @xmath6 be a de branges space , assume that each element of @xmath6 is of zero type with respect to the order @xmath318 , and let @xmath188 . then @xmath315 . a55 1 . in both theorems we have @xmath319 . hence , the requirement @xmath188 is necessary in order that @xmath14 can be represented in the form @xmath13 with some @xmath272 , cf . remark [ rea59 ] . 2 . with the completely similar proof , the analogue of theorem [ tha37 ] for rays @xmath0 contained in the second quadrant holds true . in any case @xmath320 . hence , in order to establish the asserted equality @xmath315 in either theorem [ tha12 ] or theorem [ tha37 ] , it is sufficient to show that @xmath321 . for the proof of this fact , we will employ the same method as used in the proof of ( ? ? ? * theorem 26 ) . let us recall the crucial construction : let @xmath104 and @xmath322 be given . if @xmath323 , we may consider the function @xmath324 in the proof of ( * ? ? ? * theorem 26 ) it was shown that this function does not depend on the particular choice of @xmath323 , is entire , and of zero exponential type . first we treat a particular situation . a52 consider @xmath316 where @xmath21 and @xmath325 $ ] . let @xmath6 be a de branges space , let @xmath188 , and assume that @xmath326 . then @xmath315 . let @xmath186 be such that @xmath141 and @xmath187 . assume for definiteness that the choice of @xmath108 and @xmath303 is made such that @xmath327 and @xmath328 . then we have @xmath329 . since @xmath320 , the assertion @xmath330 is equivalent to @xmath331 . assume on the contrary that @xmath332 , @xmath217 . we have @xmath333 it follows that @xmath334 , @xmath217 , and hence @xmath335 since @xmath336 , we know from the proof of ( * ? ? ? * theorem 22 ) that @xmath337 and have obtained a contradiction . the general case will be reduced to this special case with the help of the next lemma . recall the notation : @xmath338 a17 let @xmath6 be a de branges space , and let @xmath188 , @xmath339 . then @xmath340 set @xmath341 . first we show that @xmath342 . since both @xmath6 and @xmath343 are invariant with respect to @xmath344 and with respect to division by blaschke factors , also @xmath345 has this property . the crucial point is to show that @xmath345 is closed in @xmath6 . to this end , let @xmath346 be a sequence of elements of @xmath345 which converges to some element @xmath71 in the norm of @xmath6 . choose @xmath347 and @xmath348 , @xmath349 . since the difference quotient operator @xmath350 is continuous , we have @xmath351 in the norm of @xmath6 . however , @xmath352 and @xmath353 , and therefore @xmath354 . since @xmath14 is closed in @xmath6 , we obtain @xmath355 . the relation @xmath356 gives @xmath357 . we conclude that @xmath358 . the fact that @xmath359 yields in particular that @xmath342 . since we chose @xmath353 and @xmath359 , we have @xmath360 . however , since @xmath361 , @xmath350 maps @xmath345 into @xmath14 and it follows that @xmath362 . by ( * ? ? ? * theorem 29 ) , @xmath363 and hence @xmath364 . in case @xmath365 , this implies that @xmath366 , and hence also the asserted equality @xmath367 holds true . assume that @xmath368 . then there exist @xmath369 and a number @xmath210 , such that @xmath370 moreover , with these choices , we have @xmath371 . since certainly @xmath372 , we conclude that @xmath373 . it follows that also in this case @xmath374 . theorem [ tha12 ] fix @xmath186 such that @xmath141 and @xmath187 . let @xmath375 and @xmath322 be given , and consider the function @xmath376 defined in . the basic estimate for our argument is obtained by writing out the inner product in the definition of @xmath376 as an @xmath377-integral , and applying the schwarz inequality in @xmath378 : for @xmath323 and @xmath74 with @xmath379 we have @xmath380 note that the integrals on the right side of this inequality converge , since @xmath71 and @xmath381 . + _ step 1 : @xmath376 vanishes identically . let us use the estimate with @xmath382 and @xmath383 , @xmath384 . from and @xmath375 , we obtain @xmath385 . by the bounded convergence theorem both integrals in tend to @xmath386 if @xmath387 . in total , @xmath388 similar reasoning applies with @xmath389 and @xmath390 , @xmath384 , in . it follows that also @xmath391 . since @xmath376 is of zero exponential type , we may apply the phragmn lindelf principle in the left and right half - planes separately , and conclude that the function @xmath376 vanishes identically . + _ step 2 : end of proof . _ since @xmath375 and @xmath322 were arbitrary , we conclude that @xmath392 this just says that @xmath393 , and lemma [ lea17 ] gives @xmath394 . if @xmath395 , we are already done . otherwise , applying lemma [ lea52 ] with the spaces @xmath396 and @xmath397 , and using the already established fact @xmath394 , gives @xmath398 theorem [ tha37 ] again fix @xmath186 such that @xmath141 and @xmath187 . let @xmath375 and @xmath322 be given , and consider the function @xmath376 . + _ step 1 : @xmath376 is of order @xmath399 and zero type . _ the function @xmath303 belongs to @xmath109 and is of order @xmath400 . hence @xmath401 is a nondecreasing function of @xmath402 . let @xmath40 be given , then there exists @xmath403 with @xmath404 using with @xmath382 and @xmath383 , @xmath405 , we obtain @xmath406 @xmath407 using @xmath389 instead of @xmath303 , gives the analogous estimate @xmath408 for @xmath409 . in total , we have @xmath408 , @xmath410 . applying the phragmn lindelf principle to the left and right half - planes separately , yields that @xmath376 is of order @xmath399 . since @xmath40 was arbitrary , @xmath376 is of zero type with respect to this order . + _ step 2 : @xmath376 vanishes identically . _ let @xmath403 be such that @xmath411 , @xmath217 . we obtain from and the estimate used with @xmath382 and @xmath65 , that @xmath412 since @xmath413 , and @xmath317 , both integrals tend to zero if @xmath414 within @xmath0 . we obtain that @xmath415 the similar argument , applying with @xmath389 and @xmath416 , will give @xmath417 . consider the region @xmath418 which is bounded by the ray @xmath0 , its conjugate ray , and the line segment connecting these rays . moreover , let @xmath419 . ( 0,0 ) # 1#2#3#4#5 @font ( 2291,1848)(6838,-2305 ) ( 8278,-530)(0,0)[b ] ( 7312,-878)(0,0)[b ] ( 8394,-1084)(0,0)[b ] the opening of @xmath418 is @xmath420 . the opening of @xmath421 is @xmath422 . since @xmath376 is of order @xmath399 zero type , we may apply the phragmn lindelf principle to the regions @xmath418 and @xmath421 separately , and conclude that @xmath376 vanishes identically . + _ step 3 : end of proof . _ we repeat word by word the same reasoning as in step 2 of the proof of theorem [ tha12 ] , and obtain the desired assertion . a39 consider @xmath316 where @xmath21 and @xmath317 . we do not know at present , and find this an intriguing problem , whether theorem [ tha37 ] remains valid without any assumptions on the growth of elements of @xmath6 . it is clear where the argument in the above proof breaks : if we merely know that @xmath376 is of zero exponential type , its smallness on the boundary of @xmath418 does not imply that @xmath423 . on the other hand , we were not able to construct a counterexample . one reason for this will be explained later , cf . remark [ rea53 ] . having in mind theorem [ tha12 ] and the formula for zero divisors , it does not anymore come as a surprise that all db - subspaces of a given de branges space can be realized by majorization on @xmath424 . a13 consider @xmath424 . let @xmath6 be a de branges space , and let @xmath275 . then @xmath315 . we have @xmath425 , and hence @xmath426 . since @xmath427 , we may apply theorem [ tha12 ] , and obtain @xmath428 another statement which fits the present context can be proved by a more elementary argument . also this result is not much of a surprise , when thinking of ( * ? ? ? * theorem 3.4 ) ( see theorem [ tha10 ] below ) and the estimate for mean type . a54 consider @xmath429 where @xmath325 $ ] . let @xmath6 be a de branges space , and let @xmath430 . then @xmath431 . in any case , we have @xmath432 , cf . example [ exa50 ] . hence , in order to establish the asserted equality , it suffices to show that @xmath433 . let @xmath434 be given , so that @xmath435 , @xmath436 . since @xmath71 , the quotient @xmath437 is of bounded type in @xmath24 . since @xmath87 is majorized by @xmath310 on the ray @xmath438 , we have @xmath439 . the same arguments apply to @xmath440 . finally , since @xmath87 is majorized along the real axis , we have @xmath441 . it follows that @xmath442 . a32 we would like to point out that , although seemingly very similar , proposition [ pra54 ] differs in some essential points from the previous results theorem [ tha12 ] , theorem [ tha37 ] and corollary [ coa13 ] . the obvious differences are of course that on the one hand @xmath443 , but on the other hand also in case @xmath317 there are no growth assumptions on @xmath6 . the following two notices are not so obvious . first , in proposition [ pra54 ] we obtain only @xmath444 , and taking the closure is in general necessary . in the previous statements , we had @xmath315 and hence actually @xmath445 , cf . the chain of inclusions in example [ exa50 ] . secondly , the argument used to prove proposition [ pra54 ] relies mainly on majorization along @xmath17 ; majorization along the ray is only used to control mean type . contrasting this , the argument used to deduce corollary [ coa13 ] from theorem [ tha12 ] relies mainly on majorization on the ray ; majorization along @xmath17 is only used to control @xmath446 . in this section , we focus on majorization on sets @xmath0 which are close to the real axis . if majorization is permitted only on @xmath17 itself , we already know precisely which subspaces can be represented . recall : let @xmath108 be such that @xmath141 , pick @xmath241 , @xmath455 , and set @xmath456 . if @xmath457 , we are already done , hence assume that @xmath458 . let @xmath459 , and let @xmath460 be a positive and even function which is increasing on @xmath451 , satisfies @xmath461 , @xmath462 , and is such that still @xmath463 it is a well - known consequence of the beurling malliavin theorem , that there exists a nonzero function @xmath464 with @xmath465 see e.g. @xcite or @xcite . by the phragmn lindelf principle the functions @xmath466 and @xmath467 are bounded by @xmath122 throughout @xmath24 . consider the function @xmath472 then @xmath473 , @xmath474 , and @xmath475 we conclude that @xmath476 . next we show that @xmath477 . indeed , if @xmath217 or @xmath478 , @xmath479 , then @xmath480 . since @xmath481 , @xmath462 , we may estimate @xmath482 finally , by the phragmn lindelf principle , @xmath483 . using , it follows that @xmath484 and so @xmath485 . since @xmath459 was arbitrary , we conclude that @xmath454 . theorem [ tha11 ] can , of course , be viewed as a necessary condition for representability of a db - subspace @xmath14 as @xmath13 , where @xmath12 is defined on some set @xmath0 close to @xmath17 . we turn to the discussion of sufficient conditions for representability . to start with , let us discuss representability with the standard majorant @xmath286 restricted to @xmath17 . a15 consider @xmath447 . let @xmath6 be a de branges space , and let @xmath430 . if @xmath448 and @xmath486 , then @xmath487 thereby @xmath488 , if and only if @xmath368 and for some @xmath489 we have @xmath490 let @xmath491 be given . since @xmath448 , we have @xmath492 . thus @xmath493 . due to our present assumption on @xmath494 , we have @xmath495 the smirnov maximum principle implies that @xmath496 throughout the half - plane @xmath24 . it follows that @xmath357 . using lemma [ lea17 ] , we obtain @xmath497 . a51 let @xmath6 be a de branges space , let @xmath275 , and write @xmath187 . if @xmath514 , then @xmath515 . this follows since the condition @xmath514 certainly implies that every linear combination @xmath516 is majorized by @xmath286 on the real axis . as we see from the proof of theorem [ tha15 ] , it will be a good start to construct the space @xmath187 such that @xmath520 grows very fast . let @xmath521 , @xmath522 , @xmath523 . since @xmath524 , there exists a function @xmath301 with @xmath525 and @xmath526 , which has the points @xmath135 as simple zeros and does not vanish at any other point . for this function , we have @xmath527 } \,.\ ] ] let @xmath528 be given , and choose @xmath529 , @xmath530 , such that @xmath531 . then we have @xmath532 & \leq k\log^2k\bigg[\big(\underbrace{\log(k+1)-\log k}_{=\log(1+\frac 1k ) \leq\frac 1k}\big)^2+\frac 1{k^2\log^4k}\bigg ] \\ & \leq\frac{\log^2k}k\bigg[1+\frac 1{\log^4k}\bigg]\leq\frac{\log^2k}k\bigg[1+\frac 1{\log^42}\bigg ] \ , . \end{aligned}\ ] ] it follows that @xmath533^{-1}\frac k{\log^2k}\geq \frac{e^x-1}{x^2}\big[1+\frac 1{\log^42}\big]^{-1},\qquad x\geq\ln 2 \,.\ ] ] next , choose an entire matrix function @xmath189 , @xmath534 , of zero exponential type , with @xmath192 , @xmath535 , @xmath193 , such that the kernel @xmath199 is positive semidefinite and the reproducing kernel space @xmath197 does not contain a constant vector function . examples of such matrix functions can be obtained easily using the theory of canonical systems . define a function @xmath124 by @xmath536 then @xmath111 and @xmath537 . moreover , the space @xmath197 is isomorphic to the orthogonal complement @xmath538 via the map @xmath539 in particular , @xmath540 . note here that @xmath197 is certainly infinite dimensional , since it does not contain any constant . finally , note that all elements @xmath541 of @xmath197 are of zero exponential type . in view of and the symmetry of @xmath303 , this implies that @xmath542 i.e. @xmath543 . we conclude that @xmath544 . our next aim is to obtain some information about majorization on lines @xmath545 , @xmath21 , parallel to the real axis . in order that a subspace @xmath275 can be represented by majorization on @xmath0 , it is not only necessary to have @xmath448 , but also that @xmath188 . the function @xmath553 is analytic on a domain containing the closed half - plane @xmath554 and is of bounded type in the open half - plane @xmath555 . also , we have @xmath556 thus it belongs to the smirnov class @xmath557 in the half - plane @xmath558 . the smirnov maximum principle hence applies , and we obtain that @xmath553 is bounded throughout @xmath558 . the same argument applies with @xmath559 in place of @xmath288 . applying ( * ? ? ? * theorem 26 ) with the measure @xmath560 gives @xmath552 . + _ step 2 : _ let @xmath375 . we have @xmath561 and hence @xmath562 and @xmath563 are bounded on @xmath0 . by step 1 , it follows that @xmath357 . lemma [ lea17 ] implies that @xmath564 , and we conclude that @xmath565 . + _ step 3 : _ let @xmath434 . the function @xmath566 is associated to @xmath6 , and @xmath553 as well as @xmath553 are bounded on @xmath0 . by step 1 , @xmath552 , and thus @xmath153 . we see that @xmath567 . the reverse inclusion holds in any case , cf . example [ exa50 ] . a20 let @xmath545 where @xmath21 . consider the space @xmath569 generated by the function @xmath570 the choice of @xmath108 is made such that @xmath571 thus @xmath172 contains @xmath572 as a db - subspace with codimension @xmath122 , and @xmath573 . since for @xmath574 , @xmath575 , the function @xmath576 belongs to @xmath577 . thus @xmath578 . a48 consider @xmath579 where @xmath580 and @xmath581 . let @xmath6 be a de branges space , assume that each element of @xmath6 is of zero type with respect to the order @xmath582 , and let @xmath546 . then @xmath583 . + _ step 1 : the case @xmath584 . _ we proceed similar as in the proof of theorem [ tha37 ] . fix @xmath186 such that @xmath141 and @xmath187 . let @xmath434 and @xmath322 be given , and consider the function @xmath376 defined as in . the argument which was carried out in step 1 of the proof of theorem [ tha37 ] , yields that @xmath376 is of zero type with respect to the order @xmath585 . let @xmath403 be such that @xmath586 , @xmath217 . moreover , let @xmath587 be a zero of @xmath303 . the basic estimate , used with @xmath588 and @xmath65 , gives @xmath589 + c\frac{\|w - z_0\|}{\|w+i\|}\big(\int_{{{\mathbb{r}}}}\big\|\frac{g(t)}{e(t)}\big\|^2\frac{dt}{\|t - w\|^2}\big)^{1/2 } \cdot\\|h\\|_{{{\mathcal{h}}}},\qquad w\in d \ , . \end{gathered}\ ] ] however , since @xmath71 and @xmath590 , we have @xmath591 . moreover , for @xmath65 , @xmath592 . hence , we may apply the bounded convergence theorem to obtain @xmath415 since @xmath376 is of order @xmath585 and zero type , the phragmn lindelf principle implies that @xmath376 vanishes identically . in order to complete the proof , assume on the contrary that @xmath595 and @xmath596 . let @xmath597 be such that @xmath598 note that , in particular , @xmath599 . since @xmath600 , we can find a function @xmath601 and a constant @xmath602 , such that @xmath603 set @xmath604 and consider the associated model subspace @xmath605 of the hardy space . recall that the mapping @xmath606 is a unitary transform of @xmath607 onto @xmath608 . then @xmath609 theorem [ tha60 ] implies that @xmath610 . this contradicts the fact that @xmath599 . let @xmath434 be given . consider the function @xmath613 . then @xmath23 is of bounded type in @xmath24 . since @xmath87 and @xmath303 are entire functions of order @xmath585 , we certainly have @xmath614 . moreover , @xmath23 has an analytic continuation to some domain which contains the closure of @xmath24 . we conclude that @xmath23 belongs to the smirnov class @xmath615 . hence @xmath469 is majorized throughout the half - plane @xmath24 by the poisson integral of its boundary values : @xmath616 since @xmath434 , the function @xmath23 is bounded on @xmath617 . again , since @xmath87 and @xmath303 are of order @xmath585 , we have @xmath618 cf . * , theorem ) . hence we may estimate @xmath619 since @xmath23 is bounded on @xmath620 , the first summand is bounded independently of @xmath621 . the second and third summands are , for each fixed @xmath402 , nonincreasing and nonnegative functions of @xmath622 . in particular , they are bounded on each ray @xmath612 , @xmath402 . it follows that , for each fixed positive value of @xmath623 , we have @xmath624 , @xmath625 . a38 there exists a space @xmath141 with @xmath108 of order @xmath585 and finite type , and a subspace @xmath626 , such that @xmath627 . to show this we construct a matrix @xmath191 with components of order @xmath585 such that the space @xmath197 exists and such that one of its rows is bounded on @xmath628 consider the two auxiliary functions @xmath629 @xmath630 let @xmath631 , @xmath632 . then we can write @xmath633 and it is easy to see that @xmath634 next , note that @xmath635 combining this , we conclude that @xmath636 set @xmath637 . then @xmath638 is of order @xmath639 and finite type , we have @xmath640 , @xmath641 , and @xmath642 , @xmath643 . therefore @xmath644 . let @xmath645 . changing slightly the function @xmath638 we may assume that @xmath646 , @xmath647 . then by ( * ? ? ? * theorems 27 , 28 ) , there exist real entire functions @xmath648 , with @xmath649 , @xmath650 , such that for the matrix @xmath651 the space @xmath197 exists . thus also the space @xmath652 exists , where @xmath653 let @xmath607 be an arbitrary de branges space , and set @xmath124 , where @xmath654 . define @xmath655 then @xmath656and so @xmath657 . however , this function is majorized by @xmath310 on @xmath628 . thus @xmath658 . a41 there exist spaces @xmath141 , @xmath659 , with functions @xmath108 , @xmath303 , of arbitrarily small order such that @xmath275 and , for @xmath660 , @xmath661 thus , in the statement of theorem [ tha48 ] we can not replace @xmath662 by @xmath663 . to show this , let @xmath664 , let @xmath665 , @xmath666 , and let @xmath667 . put @xmath668 the series converges since @xmath669 there exist real entire functions @xmath670 and @xmath671 such that @xmath672 and @xmath607 exists . we show that for @xmath187 and @xmath584 we have @xmath673 put @xmath674 . then is equivalent to @xmath675 also , @xmath676 and so @xmath677 it is easy to see that @xmath678 , which implies . indeed , let @xmath679 $ ] , @xmath680 . then @xmath681 with constants independent of @xmath682 . hence @xmath683 a53 we return to the comment made in remark [ rea39 ] . seeking an example that the growth assumption in theorem [ tha37 ] is necessary , the first idea would be to proceed in the same way as in example [ exa38 ] . but this is not possible . the reason is that there exists no function @xmath684 , such that @xmath685 and @xmath645 is bounded on @xmath0 . indeed , if @xmath638 would have these properties , then @xmath686 , which implies that @xmath687 , @xmath474 . also since @xmath638 is not a polynomial , @xmath688 , @xmath387 , for any fixed @xmath689 . applying the poisson formula in the angle @xmath690 to @xmath691 ( `` two constant theorem '' ) we conclude that @xmath638 is unbounded on @xmath0 . startsection section1@ -3.5ex plus -1ex minus -.2ex2.3ex plus.2ex * appendix * estimates of inner functions on horizontal rays in this appendix we prove a theorem about asymptotic behavior of inner functions along horizontal rays . it is well known that for any ray @xmath692 with @xmath693 , the estimate @xmath694 is equivalent to @xmath695 . if @xmath696 is a meromorphic inner function the latter condition means that @xmath697 . for the de branges space setting see ( * ? ? ? * theorem 22 ) , the case of general inner functions is discussed , e.g , in @xcite . we show that an analogous and even stronger statement is true for the rays @xmath698 , @xmath699 . 1 . the lebesgue measure on @xmath17 is denoted by @xmath705 . moreover , @xmath706 denotes the poisson measure on @xmath17 , that is @xmath707 . the set of all functions @xmath708 which are defined and analytic in @xmath24 and have nonnegative real part throughout this half - plane is denoted by @xmath709 . 3 . recall that a function @xmath708 belongs to @xmath709 if and only if it has an integral representation of the form @xmath710 where @xmath711 the data @xmath712 and @xmath713 in this representation are uniquely determined by the function @xmath708 ( see , e.g. , @xcite ) . note that , if the function @xmath708 has a continuous extension to the closed half - plane @xmath308 , then the measure @xmath713 is absolutely continuous with respect to @xmath705 and @xmath714 4 . two subclasses of @xmath709 are defined as @xmath715 @xmath716 recall that @xmath717 if and only if in we have @xmath718 i.e. , if and only if @xmath708 can be represented in the form @xmath719 with a finite positive borel measure @xmath713 , see e.g. ( * ? ? ? * theorem 6.4 ) . in this case we have @xmath720 weak type estimates enter the discussion in the form lemma [ lea63 ] below , and can be used to conclude that @xmath610 , cf . lemma [ lea64 ] . let @xmath725 , and consider the function @xmath726 then @xmath727 is continuous in @xmath308 and belongs to @xmath709 . in particular , @xmath728 and it is easy to see that @xmath729 . moreover , @xmath730 ( or @xmath731 ) if and only if @xmath708 has the respective property . for the proof of @xmath250 , assume that @xmath721 . then we have @xmath732 by kolmogorov s theorem on the harmonic conjugate , to be more specific by ( * ? ? ? * , corollary ) , we have @xmath733 since also @xmath734 the desired limit relation follows . for the proof of @xmath237 , assume that @xmath717 . then we have @xmath735 , and @xmath736 this shows that @xmath737 is the standard hilbert transform of @xmath738 . thus , by ( * ? ? ? * v , lemma 2.8 ) , we have the weak type estimate @xmath739 where @xmath740 is the euler number . since @xmath741 we obtain the desired estimate , e.g. with the constant @xmath742 . a64 let @xmath700 be an inner function in @xmath24 , and let @xmath584 . assume that there exist positive constants @xmath743 , and @xmath744 , such that @xmath745:\,\|1-\theta(x+iy_0)\|\leq\frac c{\|x+iy_0\|}\big\}\big)\geq c'r,\qquad r\geq r_0 \,.\ ] ] then @xmath610 . consider the function @xmath746 for @xmath747 set @xmath748:\,\|1-\theta(x+iy_0)\| \leq\frac c{\|x+iy_0\|}\big\ } \,.\ ] ] then , by our hypothesis , we have @xmath749 , @xmath750 . assume that @xmath751 and let @xmath752 with @xmath753 . then @xmath754 since @xmath755 . thus , we have @xmath756 it follows that for @xmath757 , @xmath758 , @xmath759 @xmath760 if @xmath761 , then @xmath762 , and we may use this particular value of @xmath763 in . it follows that , for @xmath764 , @xmath765 where the constant @xmath766 depends only on @xmath767 and @xmath768 . the function @xmath708 is analytic in @xmath24 and has nonnegative real part throughout this half - plane . by lemma [ lea63 ] , @xmath250 , it can not belong to the subclass @xmath769 , i.e. , we have @xmath770 however , this property of @xmath708 is , e.g. by the discussion in @xcite , equivalent to @xmath771 belonging to @xmath608 . theorem [ tha60 ] assume on the contrary that @xmath772 . our aim is to show that , under the assumptions of the theorem , the relation holds for some appropriate values of @xmath743 and @xmath744 . once this has been achieved , lemma [ lea64 ] implies that @xmath610 , and we have derived a contradiction . the function @xmath771 does not belong to @xmath608 if and only if @xmath773 is in @xmath769 , that is , @xmath774 in . thus , @xmath775 the measure @xmath713 , called the clark measure , has many important properties ( see , e.g. , ( * ? ? ? 2 , part d , chapter 4 ) ) . in particular , it was shown in @xcite that each function @xmath776 has radial boundary values @xmath713-a.e . and the restriction operator @xmath777 is a unitary operator from @xmath608 onto @xmath778 . note that @xmath779 @xmath713-a.e . on @xmath780 . now let @xmath786 be a function as in the hypothesis of theorem [ tha60 ] . note that , for each @xmath787 , there exists a positive constant @xmath788 such that @xmath789 } \frac{f(t)}{t - x - iy_0}\ , d\mu(t ) \bigg\| \leq\frac{c_m}{\|x+iy_0\|},\qquad x\in{{\mathbb{r } } } \,.\ ] ] the representation of the function @xmath800 may be rewritten as @xmath801 let @xmath802 and let @xmath803 , @xmath804 . set @xmath805 then @xmath717 and @xmath806 using lemma [ lea63 ] , @xmath237 , we obtain @xmath807 the same argument applies when we take @xmath808 , as well as @xmath809 and @xmath810 for @xmath811 , instead of @xmath812 in the definition of @xmath708 . altogether , we conclude that @xmath813 let @xmath747 and @xmath814 $ ] be given , then @xmath815 it follows that , for any @xmath747 and @xmath171 , @xmath816:\,\|\gamma_{\epsilon}(x+iy_0)\|>\epsilon+ ( 2r+y_0)a\big\}\big)\leq\\ \leq m\bigg(\big\{x\in [ r,2r]:\,\big\| \int_{{{\mathbb{r } } } } \frac{f(t)}{t}\cdot \frac{d\mu_\epsilon(t)}{t - z}\big\| > a\big\}\bigg ) \leq\frac { 16a}{a}\epsilon \ , , \end{gathered}\ ] ] and hence @xmath817:\,\|\gamma_{\epsilon}(x+iy_0)\|\leq \epsilon+(2r+y_0 ) a \big\}\big)\geq r-\frac{16a}{a}\epsilon\ ] ] assume that @xmath818 , and use this inequality for the particular value @xmath819 of @xmath3 . then it follows that @xmath820:\,\|\gamma_{\epsilon}(x+iy_0)\| \leq \epsilon+3\sqrt\epsilon\big\}\big)\geq r(1 - 16a\sqrt\epsilon ) \,.\ ] ] at this point we make a particular choice of @xmath821 , namely , we take @xmath40 so small that @xmath822 and @xmath823 . then gives @xmath824:\,\|\gamma_{\epsilon}(x+iy_0)\|\leq \frac 12\big\}\big)\geq\frac 12r,\qquad r\geq y_0 \,.\ ] ] however , if @xmath814 $ ] is such that @xmath825 , then by the hypothesis and the relation we obtain that @xmath826 we conclude that , for @xmath827 , @xmath828:\,\|1-\theta(x+iy_0)\|\leq\frac{2c}{\|x+iy_0\|}\big\}\big)\geq \\ \geq m\big(\big\{x\in[r,2r]:\,\|\gamma_{\epsilon}(x+iy_0)\|\leq \frac 12\big\}\big ) \geq\frac 12r \ , , \end{gathered}\ ] ] i.e. holds . @xmath829 \hline \parbox[t]{60mm}{${\operatorname{im}}z\leq \psi({\operatorname{re}}z)$ , $ z\in d$,\\[1 mm ] $ \psi$ positive , even , increasing on $ [ 0,\infty)$,\\[1 mm ] $ \int_0^\infty(t^2 + 1)^{-1}\psi(t)\,dt<\infty$ } \rule{0pt}{5 mm } & \operatorname{mt}_{{{\mathcal{h}}}}{{\mathcal{l}}}=0 \\[13 mm ] \hline \end{array}\ ] ] @xmath830 \hline\hline { { \mathbb{r}}}\cup i[0,\infty ) \rule{0pt}{5 mm } & { { \mathcal{l}}}={{\mathcal{r}}}_{\nabla_{\!{{\mathcal{l}}}}\|_d}({{\mathcal{h } } } ) & \raisebox{1mm}{\rule{10mm}{0.5pt } } & \raisebox{1mm}{\rule{10mm}{0.5pt } } \\[1 mm ] \parbox[t]{28mm}{${{\mathbb{r}}}\cup e^{i\pi\beta}[0,\infty)$\\ $ \beta\in(0,1)$ } \rule{0pt}{5 mm } & { { \mathcal{l}}}={{\mathcal{r}}}_{{{\mathfrak{m}}}_{e_1}\|_d}({{\mathcal{h } } } ) & \raisebox{1mm}{\rule{10mm}{0.5pt } } & \raisebox{1mm}{\rule{10mm}{0.5pt } } \\[6 mm ] \hline i[h,\infty),\ h>0 \rule{0pt}{5 mm } & { { \mathcal{l}}}={{\mathcal{r}}}_{\nabla_{\!{{\mathcal{l}}}}\|_d}({{\mathcal{h } } } ) & { { \mathfrak{d}}}_{{{\mathcal{l}}}}={{\mathfrak{d}}}_{{{\mathcal{h } } } } & \raisebox{1mm}{\rule{10mm}{0.5pt } } \\[2 mm ] \parbox[t]{28mm}{$e^{i\pi\beta}[h,\infty),\ h>0,$\\ $ \beta\in(0,\frac 12)\cup(\frac 12,1)$ } \rule{0pt}{5 mm } & { { \mathcal{l}}}={{\mathcal{r}}}_{\nabla_{\!{{\mathcal{l}}}}\|_d}({{\mathcal{h } } } ) & { { \mathfrak{d}}}_{{{\mathcal{l}}}}={{\mathfrak{d}}}_{{{\mathcal{h } } } } & \parbox[t]{25mm}{order $ ( 2 - 2\beta)^{-1}$ zero type } \\[6 mm ] \hline { { \mathbb{r } } } \rule{0pt}{5 mm } & { { \mathcal{l}}}={{\mathcal{r}}}_{{{\mathfrak{m}}}_{e_1}\|_d}({{\mathcal{h } } } ) & \operatorname{mt}_{{{\mathcal{h}}}}{{\mathcal{l}}}=0 & \raisebox{1mm}{\rule{10mm}{0.5pt}}\\[2 mm ] & { { \mathcal{l}}}\!\subseteq{{\mathcal{r}}}_{\nabla_{\!{{\mathcal{l}}}}\|_d}({{\mathcal{h}}})\subseteq\!\breve{{{\mathcal{l } } } } & \parbox[t]{21mm}{$\operatorname{mt}_{{{\mathcal{h}}}}{{\mathcal{l}}}=0$,\\ $ \sup_{{{\mathbb{r}}}}\varphi_{e_1}'\!<\!\infty$ } & \raisebox{1mm}{\rule{10mm}{0.5pt}}\\[7 mm ] & { { \mathcal{l}}}={{\mathcal{r}}}_{\nabla_{\!{{\mathcal{l}}}}\|_d}({{\mathcal{h } } } ) & { { \mathfrak{d}}}_{{{\mathcal{l}}}}={{\mathfrak{d}}}_{{{\mathcal{h } } } } & \parbox[t]{25mm}{$\sup_{{{\mathbb{r}}}}\varphi_e'\!<\!\infty$,\\ $ \forall{{\mathcal{k}}}:{{\mathcal{k}}}=\breve{{{\mathcal{k}}}}$ } \\[6 mm ] \hline { { \mathbb{r}}}+ih,\ h>0 \rule{0pt}{5 mm } & { { \mathcal{l}}}={{\mathcal{r}}}_{{{\mathfrak{m}}}_{e_1}\|_d}({{\mathcal{h } } } ) & \parbox[t]{17mm}{${{\mathfrak{d}}}_{{{\mathcal{l}}}}={{\mathfrak{d}}}_{{{\mathcal{h}}}}$,\\ $ \operatorname{mt}_{{{\mathcal{h}}}}{{\mathcal{l}}}=0 $ } & \raisebox{0mm}{\rule{10mm}{0.5pt } } \\[6 mm ] & { { \mathcal{l}}}\!\subseteq{{\mathcal{r}}}_{\nabla_{\!{{\mathcal{l}}}}\|_d}({{\mathcal{h}}})\subseteq\!\breve{{{\mathcal{l } } } } & \parbox[t]{17mm}{${{\mathfrak{d}}}_{{{\mathcal{l}}}}={{\mathfrak{d}}}_{{{\mathcal{h}}}}$,\\ $ \operatorname{mt}_{{{\mathcal{h}}}}{{\mathcal{l}}}=0 $ } & \raisebox{0mm}{\rule{10mm}{0.5pt } } \\[6 mm ] \hline \parbox[t]{29mm}{$iy_0{\scriptstyle+}[h,\infty),\ h\in{{\mathbb{r}}}$,\\ $ y_0\geq 0 $ } \rule{0pt}{5 mm } & { { \mathcal{l}}}={{\mathcal{r}}}_{{{\mathfrak{m}}}_{e_1}\|_d}({{\mathcal{h } } } ) & { { \mathfrak{d}}}_{{{\mathcal{l}}}}={{\mathfrak{d}}}_{{{\mathcal{h } } } } & \parbox[t]{25mm}{order $ \frac 12$\\ zero type } \\[6 mm ] \hline \end{array}\ ] ] a.d . baranov : _ isometric embeddings of the spaces @xmath831 in the upper half - plane , _ j. math . 105(5 ) ( 2001 ) , 23192329 . a.d . baranov : _ polynomials in the de branges spaces of entire functions , _ ark . mat . 44 ( 2006 ) , 1638 . a. baranov , h. woracek : _ subspaces of de branges spaces generated by majorants , _ canad . j. 61 ( 3 ) ( 2009 ) , 503517 . a. baranov , h. woracek : _ majorization in de branges spaces ii . banach spaces generated by majorants , _ submitted . preprint available online as asc preprint series 13/2009 , http://asc.tuwien.ac.at . a. beurling , p. malliavin : _ on fourier transforms of measures with compact support , _ acta math . 107 ( 1962 ) , 291309 . a. borichev , m. sodin : _ weighted polynomial approximation and the hamburger moment problem , _ complex analysis and differential equations , proceedings of the marcus wallenberg symposium in honor of matts essn , uppsala university , 1998 . l. de branges : _ some hilbert spaces of entire functions , _ proc . 10 ( 5 ) ( 1959 ) , 840846 . l. de branges : _ hilbert spaces of entire functions , _ prentice - hall , london , 1968 . h. dym , h. mckean : _ gaussian processes , function theory , and the inverse spectral problem , _ academic press , new york , 1976 . garnett : _ bounded analytic functions , _ academic press , new york , 1981 . i. gohberg , m.g . kren : _ theory and applications of volterra operators in hilbert space , _ translations of mathematical monographs , ams . providence , rhode island , 1970 . l. golinskii , i. mikhailova : _ hilbert spaces of entire functions as a @xmath5-theory subject , _ oper . theory adv . 95 ( 1997 ) , 205251 . gorbachuk , v.i . gorbachuk : _ m.g . krein s lectures on entire operators , _ oper . theory adv . 97 , birkhuser verlag , basel , 1997 . s. hassi , h.s.v . de snoo , h. winkler : _ boundary - value problems for two - dimensional canonical systems , _ integr . theory 36(4 ) ( 2000 ) , 445479 . v. havin , b. jricke : _ the uncertainty principle in harmonic analysis , _ springer verlag , berlin , 1994 . havin , j. mashreghi : _ admissible majorants for model subspaces of @xmath702 . part i : slow winding of the generating inner function , _ can . j. math . 55 ( 6 ) ( 2003 ) , 12311263 . havin , j. mashreghi : _ admissible majorants for model subspaces of @xmath702 . part ii : fast winding of the generating inner function , _ 55 ( 6 ) ( 2003 ) , 12641301 . m. kaltenbck , h. woracek : _ de branges spaces of exponential type : general theory of growth , _ math ( szeged ) 71 ( 1/2 ) ( 2005 ) , 231284 . koosis : _ the logarithmic integral i , _ cambridge stud . adv . 12 , cambridge university press , 1988 . koosis : _ the logarithmic integral ii , _ cambridge stud . adv . 21 , cambridge university press , 1992 . n. nikolski : _ treatise on the shift operator , _ springer verlag , berlin heidelberg , 1986 . n. nikolski : _ operators , functions , and systems : an easy reading . vol . 12 , _ math . surveys monogr . , 9293 , ams , providence , ri , 2002 . poltoratski : _ on the boundary behavior of pseudocontinuable functions , _ petersburg math . j. 5 ( 2 ) ( 1994 ) , 389406 . c. remling : _ schrdinger operators and de branges spaces , _ j. funct . 196 ( 2 ) ( 2002 ) , 323394 . m. rosenblum , j. rovnyak : _ topics in hardy classes and univalent functions , _ birkhuser verlag , basel , 1994 . e. stein , g. weiss : _ introduction to fourier analysis on euclidean spaces , _ princeton university press , new jersey , 1971 .	in this series of papers we study subspaces of de branges spaces of entire functions which are generated by majorization on subsets @xmath0 of the closed upper half - plane . the present , first , part is addressed to the question which subspaces of a given de branges space can be represented by means of majorization . results depend on the set @xmath0 where majorization is permitted . significantly different situations are encountered when @xmath0 is close to the real axis or accumulates to @xmath1 . * * majorization in de branges spaces i. representability of subspaces anton baranov , harald woracek 46e20 , 30d15 , 46e22 + * keywords : * de branges subspace , majorant , beurling - malliavin theorem
self - similar sets are often constructed as the invariant set of iterated function systems @xcite , @xcite , @xcite . many other examples , such as the inflation rule of the penrose tilings , show that a graph directed generalization of iterated function systems is also interesting and has been developed as mauldin - williams graphs @xcite , @xcite , @xcite . since we can regard graph directed iterated function systems as dynamical systems , we expect that there exist fruitful connections between mauldin - williams graphs and @xmath2-algebras . in @xcite the first - named author defined a @xmath2-correspondence @xmath5 for a mauldin - williams graph @xmath0 and showed that the cuntz - pimsner algebra @xmath6 ( see @xcite , @xcite ) is isomorphic to the cuntz - krieger algebra @xmath7 ( @xcite ) associated with the underlying graph @xmath4 . in particular , if the mauldin - williams graph has one vertex and @xmath8 edges , the cuntz - pimsner algebra @xmath6 is isomorphic to the cuntz algebra @xmath9 ( @xcite ) , which recovers a result in @xcite . the construction is useful because it gives many examples of different non - self - adjoint algebras which are not completely isometrically isomorphic , but have the same @xmath2-envelope ( @xcite ) as shown by the first - named author in @xcite . on the other hand , kajiwara and the second - named author introduced @xmath2-algebras associated with rational functions in @xcite including the singular points ( i.e. branched points ) . they developed the idea to associate @xmath2-algebras with self - similar sets considering the singular points in @xcite . in this paper we associate another @xmath2-algebra @xmath3 with a mauldin - williams graph @xmath0 and its invariant set @xmath1 putting emphasis on the singular points as above . we show that the associated @xmath2-algebra @xmath3 is not isomorphic to the cuntz - krieger algebra @xmath7 for the underlying graph @xmath4 in general . this comes from the fact that the singular points cause the failure of the injectivity of the coding by the markov shift for @xmath4 . we assume that the underlying graph g has no sinks and no sources . we show that the associated @xmath2-algebra @xmath3 is simple and purely infinite if @xmath0 satisfies the open set condition in @xmath1 and if @xmath4 is irreducible and is not a cyclic permutation . we calculate the @xmath1-groups for some examples including the inflation rule of the penrose tilings . the @xmath2-algebras associated with tilings were first studied by connes in @xcite and also discussed by mingo @xcite and anderson - putnam @xcite . but we do not know the exact relatition between their constructions and ours . our construction has a common flavor with several topological generalizations of graph @xmath2-algebras @xcite by brenken @xcite , katsura @xcite , muhly and solel @xcite and muhly and tomforde @xcite . since graph directed iterated function systems are sometimes obtained as continuous cross sections of expanding maps , @xmath2-algebras associated with interval maps introduced by deaconu and shultz in @xcite are closely related with our construction . in a different point of view , bratteli and jorgensen studied a relation between iterated function systems and representation of cuntz algebras in @xcite . the authors would like to thank p. muhly for suggesting their collaboration and giving them the chance to discuss this work at a conference at the university of iowa . mauldin - williams _ _ graph _ we mean a system @xmath11 , where @xmath12 is a graph with a finite set of vertices @xmath13 , a finite set of edges @xmath14 , a _ range _ map @xmath15 and a _ source _ map @xmath16 , and where @xmath17 and @xmath18 are families such that : 1 . each @xmath19 is a compact metric space with a prescribed metric @xmath20 , @xmath21 . 2 . for @xmath22 , @xmath23 is a continuous map from @xmath24 to @xmath25 such that@xmath26 for some constants @xmath27 satisfying @xmath28 ( independent of @xmath29 ) and all @xmath30 . we shall assume , too , that the source map @xmath16 and the range map @xmath15 are surjective . thus , we assume that there are no sinks and no sources in the graph @xmath4 . in the particular case when we have only one vertex and @xmath8 edges , we obtain a so called _ iterated function system _ @xmath31 . an _ invariant list _ associated with a mauldin - williams graph @xmath11 is a family @xmath32 of compact sets such that @xmath33 for all @xmath21 , and such that@xmath34 since each @xmath23 is a contraction , @xmath0 has a unique invariant list ( see ( * ? ? ? * theorem 1 ) ) . we set @xmath35 and @xmath36 and we call @xmath1 the _ invariant set _ of the mauldin - williams graph . in the particular case when we have one vertex @xmath37 and @xmath38 edges , i.e. in the setting of an _ iterated function system , _ the invariant set is the unique compact subset @xmath39 of @xmath40 such that @xmath41 in this paper we forget about the ambient space @xmath42 . that is , we consider that the mauldin - williams graph is @xmath43 , with @xmath44 being the invariant set . we say that a graph @xmath12 is _ irreducible _ ( or _ totally connected _ ) if for every @xmath45 there is a finite path @xmath46 such that @xmath47 and @xmath48 . we will assume in this paper that the graph @xmath12 is irreducible and not a cyclic permutation . that is there exists a vertex @xmath49 and there exist two edges @xmath50 such that @xmath51 . for a natural number @xmath52 , we define @xmath53 . an element @xmath54 is called a path of length @xmath52 . we extend @xmath16 and @xmath15 to @xmath55 by @xmath56 and @xmath57 for all @xmath54 . we set @xmath58 and denote the length of a path @xmath46 by @xmath59 . we define also @xmath60 to be the set of paths of length @xmath52 starting at the vertex @xmath37 and @xmath61 to be the set of finite paths starting at @xmath37 . the infinite path space is @xmath62 and the space of infinite paths starting at a vertex @xmath37 is @xmath63 . on @xmath64 we define the metric @xmath65 if @xmath66 and @xmath67 otherwise , where @xmath68 is the longest common prefix of @xmath69 and @xmath70 ( see ( * ? ? ? * page 116 ) ) . then @xmath64 is a compact metric space , and , since @xmath71 equals the disjoint union of the spaces @xmath64 , @xmath71 is a compact metric space in a natural way . for @xmath54 , let @xmath72 and @xmath73 . then for any infinite path @xmath74 , @xmath75 contains only one point . therefore we can define a map @xmath76 by @xmath77 . since @xmath78 is also an invariant set , we have that @xmath79 . thus @xmath80 is a continuous , onto map . moreover , for any @xmath81 and any neighborhood @xmath82 of @xmath83 , there exist @xmath84 and @xmath85 such that@xmath86 we say that a mauldin - williams graph @xmath0 satisfies the _ open set condition in @xmath1 _ if there exists a family of non - empty sets @xmath87 , such that @xmath88 for all @xmath21 , and such that@xmath89 @xmath90 then @xmath91 is an open dense subset of @xmath1 . moreover , for @xmath84 and @xmath92 , if @xmath93 and @xmath94 then @xmath95 . for @xmath96 , we define the _ cograph of _ @xmath23 to be the set @xmath97 we shall consider the union@xmath98 consider the @xmath10-algebra @xmath99 and let @xmath100 . then @xmath101 is a @xmath10-correspondence over @xmath102 with the structure defined by the formulae:@xmath103 for all @xmath104 , @xmath105 , @xmath106 and @xmath107 . it is clear that the @xmath102-valued inner product is well defined . the left multiplication is given by the @xmath108-homomorphism @xmath109 such that @xmath110 for @xmath111 and @xmath112 . put @xmath113 . for any natural number @xmath38 , we define @xmath114 and a @xmath10-correspondence @xmath115 similarly . we also define a path space @xmath116 of length @xmath38 by@xmath117 then @xmath118 is a @xmath10-correspondence over @xmath102 with an @xmath102-valued inner product defined by@xmath119 for all @xmath120 and @xmath107 . let @xmath43 be a mauldin - williams graph . let @xmath1 be the invariant set . then @xmath100 is a full @xmath10-correspondence over @xmath99 without completion . the left action @xmath109 is unital and faithful . similar statements hold for @xmath121 . for any @xmath112 we have@xmath122 where @xmath8 is the number of edges in @xmath14 . therefore the norms @xmath123 and @xmath124 are equivalent . since @xmath125 is complete with respect to @xmath124 , it is also complete with respect to @xmath123 . let @xmath112 be defined by the formula@xmath126 then @xmath127 , hence @xmath128 contains the identity of @xmath102 . therefore @xmath101 is full . if @xmath111 is not zero , then there exists @xmath129 such that @xmath130 . since @xmath1 is the invariant set of the mauldin - williams graph , there exists @xmath96 and @xmath131 such that @xmath132 and @xmath133 . choose @xmath112 such that @xmath134 . then @xmath135 , hence @xmath136 is faithful . the statements for @xmath137 are similarly proved . let @xmath43 be a mauldin - williams graph with the invariant set @xmath1 . we associate a @xmath10-algebra @xmath3 to @xmath0 as the cuntz - pimsner algebra @xmath138 of the @xmath10-correspondence @xmath100 over the @xmath10-algebra @xmath99 . as in @xcite , we denote by @xmath139 the @xmath108-algebra generated algebraically by @xmath102 and @xmath140 with @xmath112 . the _ gauge action _ is @xmath141 defined by @xmath142 for all @xmath112 , and @xmath143 for all @xmath111 . [ pro : embeddings]let @xmath43 be a mauldin - williams graph . assume that @xmath1 is the invariant set of the graph . then there is an isomorphism @xmath144 as @xmath10-correspondences over @xmath102 such that@xmath145 @xmath146 for all @xmath147 , @xmath107 , and @xmath148 such that @xmath149 . moreover , let @xmath150 be an onto continuous map such that@xmath151 then @xmath152 is an embedding as a hilbert submodule preserving inner product . it is easy to see that @xmath153 is well - defined and a bimodule morphism . we show that @xmath153 preserves inner product . consider the case when @xmath154 for simplicity of the notation . let @xmath155 . we have@xmath156 since @xmath153 preserves the inner product , it is one to one . using the stone - weierstrass theorem , one can show that @xmath153 is also onto . the rest is clear . we let @xmath157 be the natural inner - product preserving embedding , for @xmath158 . consider a covering map @xmath159 defined by @xmath160 for @xmath106 . define the set@xmath161 the set @xmath162 will be described by the ideal @xmath163 of @xmath102 . we define a branch index @xmath164 at @xmath106 by@xmath165 hence @xmath166 if and only if there exists some @xmath107 with @xmath167 . for @xmath168 we define@xmath169 in the above situation , if @xmath170 , then there exists an open neighborhood @xmath171 of @xmath172 satisfying the following : 1 . 2 . if @xmath174 , then @xmath175 for @xmath176 , such that @xmath177 . 3 . if @xmath178 , then @xmath179 . let @xmath170 . let @xmath180 such that @xmath181 . since @xmath162 and @xmath182 are closed and @xmath172 is not in either of them , there exists an open neighbourhood @xmath183 of @xmath172 such that@xmath184 for each @xmath174 there exists a unique @xmath185 with @xmath186 , since @xmath187 . for @xmath188 , if @xmath177 and @xmath189 then @xmath190 . therefore there exists an open neighborhood @xmath191 of @xmath172 such that @xmath192 if @xmath193 and @xmath194 . let @xmath195 . then @xmath171 is an open neighborhood of @xmath172 and satisfies all the requirements . let @xmath43 be a mauldin - williams graph . assume that the system @xmath0 satisfies the open set condition in @xmath1 . then@xmath196 the proof requires only minor modifications from the proof of ( * ? ? ? * proposition 2.4 ) . @xmath197 . the closed set @xmath162 is empty if and only if @xmath198 is contained in @xmath199 if and only if @xmath101 is finitely generated projective right @xmath102 module . let @xmath43 be a mauldin - williams graph . let @xmath99 and @xmath100 . for @xmath96 define an endomorphism @xmath200 by@xmath201 for all @xmath111 and @xmath107 . we also define a unital completely positive map @xmath202 by@xmath203 for @xmath111 , @xmath107 . for the function @xmath204 defined by the formula@xmath205 we have@xmath206 let @xmath207 . in the above situation , for @xmath111 , we have that @xmath208 and in particular @xmath209 . the same with ( * lemma 3.1 ) . let @xmath43 be a mauldin - williams graph . we say that an element @xmath210 is @xmath211-invariant if@xmath212 if @xmath111 is @xmath211-invariant then @xmath213 is also @xmath214-invariant . then , similar like in ( * ? ? ? * definition on page 11 ) , if @xmath213 is @xmath211-invariant , we can define@xmath215 then , for any @xmath147 and @xmath111 @xmath211-invariant , we have the relation:@xmath216 let @xmath43 be a mauldin - williams graph such that @xmath217 is an irreducible graph . for any non - zero positive element @xmath111 and for every @xmath218 there exists @xmath84 and @xmath219 with @xmath220 such that@xmath221 let @xmath129 be such that @xmath222 . let @xmath180 such that @xmath223 . then there exists an open neighborhood @xmath224 of @xmath225 in @xmath226 such that for any @xmath227 we have @xmath228 . let @xmath229 an open neighborhood of @xmath225 in @xmath226 and @xmath230 compact such that @xmath231 . since the map @xmath76 is onto and continuous , there exists some @xmath232 and @xmath233 such that @xmath234 . for any vertex @xmath235 , since the graph @xmath4 is irreducible , there exists a path @xmath236 from @xmath237 to @xmath37 . then @xmath238 . hence , for each @xmath235 , there is @xmath239 , for some @xmath240 , such that @xmath241 . for each @xmath235 , define the closed subsets @xmath242 and @xmath243 of @xmath244 by@xmath245 since @xmath246 for all @xmath247 and @xmath248 if @xmath249 , there exists @xmath250 such that @xmath251 and@xmath252 since @xmath241 for each @xmath253 , there exists @xmath254 such that @xmath255 . therefore@xmath256 for all @xmath253 . let @xmath257 . we identify @xmath258 with @xmath259 as in proposition [ pro : embeddings ] . we denote @xmath260 also by @xmath261 for each @xmath235 . let @xmath262 . then @xmath263 for all @xmath107 . thus @xmath264 is positive and invertible . let @xmath265 . then@xmath266 for any @xmath107 and any @xmath267 such that @xmath268 , let @xmath269 . if @xmath227 , then @xmath270 , and , if @xmath271 , then@xmath272 therefore@xmath273 we also have that@xmath274 let @xmath43 be a mauldin - williams graph . assume that @xmath1 is the invariant set of the graph . for any non - zero positive element @xmath111 and for any @xmath218 with @xmath275 , there exists @xmath84 and @xmath276 such that@xmath277 the proof is identical with the proof of ( * ? ? ? * lemma 3.4 ) . let @xmath43 be a mauldin - williams graph . suppose that the graph @xmath4 has no sinks and no sources and it is irreducible and not a cyclic permutation . assume that @xmath1 is the invariant set of the graph and that @xmath0 satisfies the open set condition in @xmath1 . for any @xmath84 , for any @xmath278 , and for any @xmath218 there exists a positive element @xmath111 such that @xmath213 is @xmath211-invariant,@xmath279 and @xmath280 for @xmath281 with @xmath282 . let @xmath84 , let @xmath278 , and let @xmath218 . then there exists @xmath219 such that @xmath283 and @xmath284 . hence there exists @xmath285 for some @xmath286 such that@xmath287 then there exists an open neighborhood @xmath224 of @xmath288 in @xmath226 such that for any @xmath289@xmath290 since @xmath0 satisfies the open set condition in @xmath1 , there exists a family of non - empty sets @xmath87 , such that @xmath88 , for all @xmath21,@xmath89 @xmath90 then there exists @xmath291 and an open neighborhood @xmath229 of @xmath292 with @xmath293 . moreover , there is some @xmath294 and @xmath295 such that@xmath296 since the graph @xmath4 is not a cyclic permutation , there is a vertex @xmath49 and two edges @xmath297 such that @xmath298 . since the graph @xmath4 is irreducible , there exists a path from @xmath299 to @xmath300 . hence we have a path @xmath301 for some @xmath302 , @xmath303 , such that @xmath304 and @xmath305 then we can find a path @xmath306 such that @xmath307 if @xmath308 . to see this , let @xmath309 . if @xmath310 , take @xmath311 . if @xmath312 , since @xmath4 has no sinks , there is an edge @xmath96 such that @xmath313 . then @xmath314 . let @xmath315 . if @xmath316 , take @xmath317 ; if @xmath318 , take @xmath319 to be any edge such that @xmath320 . therefore @xmath321 . inductively , we obtain the path @xmath306 with the desired property . then@xmath322 there exist @xmath323 , an open neighborhood @xmath324 of @xmath325 in @xmath226 and a compact set @xmath326 such that@xmath327 let @xmath328 such that @xmath329 , @xmath330 and @xmath331 . for @xmath267 such that @xmath332 , we have@xmath333 moreover , for @xmath334 such that @xmath335 , by the open set condition,@xmath336 we define a positive function @xmath213 on @xmath1 by the formula@xmath337 then @xmath213 is continuous on @xmath1 , hence @xmath111 . by construction , @xmath213 is @xmath211-invariant . let @xmath338 be a natural number . let @xmath339 . if there is no path @xmath340 such that @xmath341 , then @xmath342 . if there is at least one path @xmath340 such that @xmath341 , then@xmath343 since @xmath344 we have that@xmath345 then , for @xmath346 we have that@xmath347 and@xmath348 since @xmath349-th subsuffixes are different as @xmath350 , we have that @xmath351 hence @xmath280 . moreover , we have@xmath352 because @xmath353 . thus @xmath354 . as in @xcite , we let @xmath355 be the @xmath10-subalgebra of @xmath356 generated by @xmath357 , @xmath358 and @xmath359 be the @xmath10-subalgebra of @xmath138 generated by@xmath360 we will also use the isomorphism @xmath361 such that@xmath362 in the above situation , let @xmath363 for some @xmath364 . we decompose @xmath365 with @xmath366 . for any @xmath218 there exists @xmath367 with @xmath368 satisfying the following : 1 . @xmath369 @xmath370 . 2 . @xmath371 . the proof requires only small modifications from the proof of ( * ? ? ? * lemma 3.6 ) . having proved the equivalent of the ( * ? ? ? * lemma 3.1-lemma 3.6 ) , we obtain , using the same proof as ( * ? ? ? * theorem 3.7 ) , the corresponding result for the mauldin - williams graph : let @xmath43 be a mauldin - williams graph . suppose that the graph @xmath4 has no sinks and no sources , it is irreducible , and is not a cyclic permutation . assume that @xmath1 is the invariant set of the mauldin - williams graph and that @xmath0 satisfies the open set condition in @xmath1 . then the associated @xmath10-algebra @xmath3 is simple and purely infinite . using the same argument as in ( * ? ? ? * proposition 3.8 ) one can show that @xmath3 is separable and nuclear , and satisfies the universal coefficient theorem . thus , by the classification theorem of kirchberg and phillips @xcite , the isomorphism class of @xmath3 is completely determined by the @xmath1-theory with the class of the unit . we will compute the @xmath1-groups of the @xmath10-algebra associated with a graph @xmath12 using the fact that @xmath372 is isomorphic to the kernel of @xmath373 , and @xmath374 is isomorphic to the cokernel of the same map , where @xmath375 is the vertex @xmath376 matrix defined by@xmath377 for the @xmath1-groups of the cuntz - pimsner algebras we will use the following six - term exact sequence due to pimsner @xcite ( see also ( * ? ? ? * section 4 ) ) first we give a condition , which is similar to the one in ( * ? ? ? * section 4 ) , that implies that the associated @xmath10-algebra @xmath3 is isomorphic to the @xmath10-algebra associated with the underlying graph . let @xmath43 be a mauldin - williams graph which satisfies the graph separation condition . then the associated @xmath10-algebra is isomorphic to the @xmath10-algebra associated with the underlying graph @xmath4 . let @xmath379 . let @xmath5 be the @xmath10-correspondence over @xmath99 associated in @xcite with a mauldin - williams graph . that is , @xmath380 with the operations given by the formulae:@xmath381 @xmath382 where @xmath383 and @xmath384 , and@xmath385 for @xmath386 . it was proved in ( * ? ? ? * theorem 2.3 ) that the cuntz - pimsner algebra associated with this @xmath10-correspondence is isomorphic to the cuntz - krieger @xmath10-algebra @xmath387 of the underlying graph @xmath4 . we will show that if the mauldin - williams graph satisfies the graph separation condition , then @xmath101 and @xmath5 are isomorphic as @xmath10-correspondences in the sense of @xcite . let @xmath388 defined by the formula@xmath389 then@xmath390 for all @xmath104 and @xmath112 . also@xmath391 for all @xmath105 . finally , let @xmath392 . for @xmath393 , since @xmath0 satisfies the graph separation condition , there is a unique @xmath96 such that @xmath394 . define @xmath112 by the formula@xmath395 then @xmath396 is well defined and @xmath397 , hence @xmath398 is onto . thus @xmath398 is a @xmath10-correspondence isomorphism . therefore @xmath3 is isomorphic to @xmath387 . let @xmath399 be two disjoint compact sets such that @xmath230 contains the origin , @xmath400 contains @xmath401 , and @xmath230 , @xmath400 are symmetric about @xmath172-axis . let @xmath402 be similarities , such that the map @xmath403 has ratio @xmath404 , fixed point @xmath405 , and rotation @xmath406 degrees counterclockwise . the map @xmath407 has ratio @xmath408 , fixed point @xmath409 , and rotation @xmath410 degrees clockwise . the map @xmath411 has ratio @xmath404 , fixed point @xmath405 and rotation @xmath412 degrees counterclockwise . the map @xmath413 has ratio @xmath414 , fixed point @xmath409 , and rotation @xmath415 degrees clockwise ( see ( * ? ? ? * page 167 ) for more details ) . then the mauldin - williams graph @xmath0 satisfies the graph separation condition , hence @xmath3 is isomorphic to @xmath387 . let @xmath12 be the graph with the vertex matrix @xmath375@xmath416.\ ] ] hence @xmath417 and @xmath418 . let @xmath230 and @xmath400 be two copies of the square of length one with the maps @xmath419 as in the following diagram : that is @xmath419 are similarities , @xmath420 , @xmath421 , @xmath422 , @xmath423 , such that @xmath424 , @xmath425 , @xmath426 , @xmath427 , @xmath428 , @xmath429 , @xmath430 , @xmath431 , @xmath432 , @xmath433 , @xmath434 , @xmath435 , @xmath436 , @xmath437 , @xmath438 . then @xmath439 is the invariant set of the mauldin - williams graph and @xmath440 . so @xmath441 , @xmath442 . then @xmath443 and @xmath444 . therefore @xmath445 is not isomorphic to @xmath446 . thus @xmath3 is not isomorphic to @xmath387 . ( penrose tiling ) the tiles of the penrose tiling are two triangles such that the angles of the first triangle are equal to @xmath447 and the angles of the second triangle are equal to @xmath448 . these triangles are cut from a rectangular pentagon by the diagonals issued from a common vertex . hence the ratio between the lengths of the shorter and the longer sides of the triangles is equal to @xmath449 . the mauldin - williams graph associated with the penrose tiling is given in the following diagram as shown in @xcite : then the invariant set @xmath1 of this mauldin - williams is the union of the two triangles . the vertex matrix of the graph is@xmath450,\ ] ] hence @xmath451 . since @xmath452 and @xmath453 , @xmath3 is not isomorphic to @xmath387 . 10 j. anderson and i. putnam , ` topological invariants for substitution tilings and their associated @xmath2-algebras ' , _ ergod . th . & dynam . ( 1998 ) _ , * 18 * , 509 - 537 . m. f. barnsley , _ fractals everywhere _ , second edition , academic press professional , boston , ma , 1993 . l. bartholdi , r. grigorchuk and v. nekrashevych , ` from fractal groups to fractal sets ' , _ trends in mathematics : fractals in graz 2001 _ , 25 - 118 , birkhauser . o. bratteli and p.e.t . jorgensen , ` iterated function systems and permutation representations of the cuntz algebra ' , _ mem . * 139 * ( 1999 ) , no . b. brenken , @xmath2-algebras associated with topological relations , _ j. ramanujan math . soc . _ * 19 * ( 2004 ) , no . 1 , 3555 . a. connes , _ non - commutative geometry _ , academic press , 1994 . j. cuntz , ` simple @xmath2-algebras generated by isometries ' , _ comm . math . phys _ , 57 ( 1977 ) , no . 2 , 173185 . j. cuntz and w. krieger , ` a class of @xmath2-algebras and topological markov chains ' , _ invent . math . _ 56 ( 1980 ) , no . 3 , 251268 . v. deaconu and f. shultz , @xmath2-algebras associated with interval maps , preprint oa/0405469 . g. a. edgar , _ measure , topology , and fractal geometry _ , undergraduate texts in mathematics . springer - verlag , new york , 1990 . g. a. edgar and r.d . mauldin , ` multifractal decompositions of digraph recursive fractals ' , _ proc . london math . _ 65(1992 ) , 604 - 628 . n. j. fowler , p. s. muhly and i. raeburn , ` representations of cuntz - pimsner algebras ' , _ indiana univ . j. _ 52 ( 2003 ) , no . 3 , 569605 . j. e. hutchinson , ` fractals and self - similarity ' , _ indiana univ . math . j. _ 30 ( 1981 ) , no . 5 , 713747 . m. ionescu , ` operator algebras and mauldin - williams graphs ' , preprint oa/0401408 , _ rocky mountain j. math . _ , to appear . m. ionescu , ` maudin - williams graphs , morita equivalence and isomorphisms ' , preprint oa/ 0408433 , _ proc . amer . math . _ , to appear . t. kajiwara and y. watatani , ` @xmath10-algebras associated with self - similar sets ' , preprint oa/0312481 . t. kajiwara and y. watatani , ` @xmath2-algebras associated with complex dynamical systems ' , preprint oa/0309293 , _ indiana univ . j. _ , to appear . t. katsura , ` a class of @xmath2-algebras generalizing both graph algebras and homeomorphism @xmath2-algebras , i , fundamental results ' , _ trans . soc . 356 _ ( 2004 ) , no . 11 , 42874322 . j. kigami , _ analysis on fractals _ , 2001 , cambridge university press . e. kirchberg , ` the classification of purely infinite @xmath10-algebras using kasparov s theory ' , preprint . a. kumjian , d. pask and i. raeburn , ` cuntz - krieger algebras of directed graphs ' , _ pacific j. math . _ 184 ( 1998 ) , no . 1 , 161174 . e. c. lance , _ hilbert @xmath2-modules . a toolkit for operator algebraists _ , london mathematical society lecture note series , 210 . cambridge university press , cambridge , 1995 . r. d. mauldin and s. c. williams , ` hausdorff dimension in graph directed constructions ' , _ trans . _ 309 ( 1988 ) , no . 2 , 811829 . j. mingo , ` @xmath2-algebras associated with one dimensional almost periodic tilings ' , _ commun . in math . phys_. * 183 * , 307 - 337 ( 1977 ) . p. s. muhly and b. solel , ` tensor algebras over @xmath2-correspondences : representations , dilations , and @xmath2-envelopes ' , _ j. funct . 158 ( 1998 ) , no . 2 , 389457 . p. s. muhly and b. solel , ` on the morita equivalence of tensor algebras ' , _ proc . london math . ( 3 ) 81 ( 2000 ) , no . 1 , 113168 . p. muhly and m. tomforde , ` topological quivers ' , preprint oa/0312109 . phillips , ` a classification theorem for nuclear purely infinite simple @xmath10-algebras ' , documenta math . ( 2000 ) , 113 - 168 . m. v. pimsner , ` a class of @xmath2-algebras generalizing both cuntz - krieger algebras and crossed products by @xmath454 ' , _ free probability theory _ ( waterloo , on , 1995 ) , 189212 , fields inst . , 12 , amer . soc . , providence , ri , 1997 . c. pinzari , y. watatani and k. yonetani , ` kms states , entropy and the variational principle in full @xmath2-dynamical systems ' , _ comm . _ 213 ( 2000 ) , no . 2 , 331379 .	a mauldin - williams graph @xmath0 is a generalization of an iterated function system by a directed graph . its invariant set @xmath1 plays the role of the self - similar set . we associate a @xmath2-algebra @xmath3 with a mauldin - williams graph @xmath0 and the invariant set @xmath1 laying emphasis on the singular points . we assume that the underlying graph @xmath4 has no sinks and no sources . if @xmath0 satisfies the open set condition in @xmath1 and @xmath4 is irreducible and is not a cyclic permutation , then the associated @xmath2-algebra @xmath3 is simple and purely infinite . we calculate the k - groups for some examples including the inflation rule of the penrose tilings .
kilohertz quasi - periodic oscillations ( khz qpos ) in the x - ray flux of low - mass x - ray binaries have drawn much attention since their discovery , about ten years ago . the reason for this continued interest is that since they most likely reflect the motion of matter very close to the neutron star ( or black hole ) primary in these systems , these qpos may provide one of the few direct ways of measuring effects that are unique to the strong gravitational - field regime in these systems . in general two simultaneous khz qpos are seen in the power density spectra of low - mass x - ray binaries , the lower and the upper khz qpo according to how they appear sorted in frequency . most ( but not all ) of the work on qpos in these years ( see * ? ? ? * for a review ) has focused on the frequencies of these qpos , because those frequencies provide insights into the dynamics of the system . several models have been proposed to explain the observed relation between both khz qpo frequencies @xcite , their relation with the spin of the neutron star ( e.g. , * ? ? ? * ; * ? ? ? * ) as well as with other low - frequency variability ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , both low - frequency qpos and broad - band variability . from the beginning there has been interest on the other two properties of the khz qpos , their amplitude and coherence ( e.g. , * ? ? ? * ; * ? ? ? * ) ; but systematic studies of those other qpo properties started to take off slightly later on @xcite . recently , @xcite carried out a systematic study of the khz qpo coherence and rms amplitude in three x - ray binaries , 4u 163653 , 4u 160852 , and 4u 173544 . they find that in all three sources the coherence and rms amplitude of the lower khz qpo increase slowly with frequency , and after they reach their maximum values they decrease abruptly as the qpo frequency keeps on increasing . ( the sudden drop is most noticeable in the coherence of the lower khz qpo ; in the case of the rms amplitude the drop is less abrupt . ) they propose that this behaviour is due to effects related to the innermost stable circular orbit , isco , around the neutron star in these systems . triggered by these results , in this paper i investigate the dependence of the maximum coherence and rms amplitude of both khz qpos in a large sample of low - mass x - ray binaries . in [ data ] i describe the data i use in the rest of the paper . since i collected most of the data from the literature , i spend some time describing how those data were obtained and the selection criteria . in [ results ] i present the results , and i discuss them in [ discussion ] . all the data that i use in this paper were obtained over the last 10 years with the proportional counting array , ( pca ; * ? ? ? * ) , on board the rossi x - ray timing explorer ( _ rxte _ ; * ? ? ? * ) . i collected most of these data from the literature . for this i searched , for as many sources as possible , all published values of the rms fractional amplitude and the coherence of the khz qpos . the coherence @xmath7 of a qpo is defined as @xmath8 , where @xmath9 and @xmath10 are the frequency and the full - width at half - maximum of the qpo . some authors report @xmath10 and @xmath11 instead of @xmath7 ; in those cases i calculate @xmath7 using the previous formula . the rms amplitude , @xmath12 , is calculated from @xmath13 , the integral from @xmath14 to @xmath15 of the fourier power under the qpo , and the source intensity , @xmath16 , as @xmath17 @xcite ; from this definition , @xmath12 is expressed as a percent of the total intensity of the source . from now on i will use the naming convention of the khz qpos introduced by @xcite , in which the lower khz qpo is called @xmath18 , and the upper khz qpo is called @xmath19 . the frequency , coherence and amplitude of each qpo carry a subscript or superscript @xmath20 or @xmath21 , respectively . a clarification is needed about what i call khz qpo in this paper . some authors ( e.g. , * ? ? ? * ) identify low - frequency features as possibly being @xmath18 on the basis of an extension of the qpo frequency - frequency correlation , @xmath22 , to low frequencies ( below @xmath23 hz ) . it remains still uncertain whether these identifications are correct ( cf * ? ? ? * ; * ? ? ? in this paper i use the expression `` kilohertz qpo '' to refer to features in the power spectra of neutron star systems that have frequencies @xmath24 hz , and that have not been identified as hectohertz qpo ( e.g. , * ? ? ? since the goal is to compare the maximum values of @xmath7 and @xmath12 of each khz qpo in different sources , one of the requirements in collecting the data is that there are sufficient measurements of both parameters to have confidence that the maximum values have been measured . this is in part facilitated by the fact that we know the general dependence of @xmath7 and @xmath12 on qpo frequency for both qpos . for instance , @xcite , @xcite , @xcite , @xcite , and @xcite show that both @xmath25 and @xmath26 increase with @xmath27 , they peak at intermediate to high values of @xmath27 , and then they decrease as @xmath27 increases ( see e.g. , figure 3 in * ? ? ? * and figure 3 in @xcite ) . on the other hand , @xmath28 is more or less flat at low values of @xmath29 where it peaks , and then decreases as @xmath29 increases , while @xmath30 is more or less constant with @xmath29 , although it increases towards higher @xmath29 values ( see e.g. , figure 3 in * ? ? ? * and figure 10 in @xcite ) . for this reason , here i compile data from papers that present systematic studies of the qpo properties , or that have enough measurements to deduce the systematic trends of those properties . i do not use result from papers reporting the discovery of khz qpos in a source , in which only one or two measurements of the qpo properties are available . [ cols="<,^,>,^,^,>,^ , > " , ] notes the @xmath31-@xmath32 errors are given . an aribtrary 25% error is used for the luminosity ( see text ) the parameters for gx 340 + 0 and cyg x-2 have been measured in the @xmath33 kev energy range . the parameters for gx 17 + 2 have been measured in the @xmath34 kev energy range . the parameters for sco x-1 have been measured over a variable energy range ( see text ) . all the other parameters have been measured in the full pca band . references ( 1 ) @xcite ; ( 2 ) @xcite ; ( 3 ) @xcite ; ( 4 ) @xcite ; ( 5 ) this paper ; ( 6 ) @xcite ; ( 7 ) @xcite ; ( 8) @xcite ; ( 9 ) @xcite ; ( 10 ) @xcite ; ( 11 ) @xcite ; ( 12 ) @xcite ; ( 13 ) @xcite ; ( 14 ) @xcite ; ( 15 ) @xcite ; ( 16 ) @xcite ; ( 17 ) @xcite ; ( 18 ) @xcite ; ( 19 ) @xcite ; ( 20 ) @xcite ; ( 21 ) @xcite a word must be said about the different ways in which @xmath7 and @xmath12 are measured in these sources , and how these different ways of measuring them can affect their values : 1 . the first qpo measurements were obtained over continuous _ rxte _ observations , usually stretches of up to @xmath35-s exposure time interrupted by the occultation of the source below the earth s horizon , or by the passage of the satellite by the south atlantic anomaly , when many instruments were switched off for safety reasons . because the qpo frequency typically changes by a few tens of hz within a thousand seconds @xcite , the qpos appear broader than they actually are in the average power spectra of those observations . since the width measured in the average power spectrum , @xmath36 , when a qpo of intrinsic width @xmath10 drifts by an amount @xmath37 is roughly @xmath38 , clearly intrinsically narrower peaks are more affected by the qpo frequency drift than intrinsically broader peaks . because in general @xmath18 is narrower than @xmath19 , the width of @xmath18 is the most affected . on the other hand , since the total power in the qpo is conserved in the averaging process , the rms amplitudes of the qpos are less affected by the qpo frequency drift . ( actually , the drift changes the shape of the qpo in the average power spectrum , and because the function used to fit them , usually a lorentzian , no longer represents them properly , also the rms amplitudes are affected , although to a lesser degree ) . 2 . when more data per source were collected , and it was realised that the qpo frequencies depended on other source parameters , it became clear that a much better way of studying the qpo properties was to average data taken at different epochs but having similar range of values of those other source parameters . the typical examples are data selection based on the colours of the source , or more specifically the position along the track traced out by the source in a colour - colour or colour - intensity diagram ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , or the frequency of one of the khz qpo , usually @xmath18 ( e.g. , * ? ? ? * ; * ? ? ? * ) , but sometimes also @xmath19 @xcite , or the frequency of a low - frequency qpo ( e.g. , * ? ? ? while these selections tend to reduce the spurious broadening of the qpos due to the qpo frequency drift , this effect may still be present in the results , especially when the parameter used in selection spans a large interval . in this paper i try to use as much as possible data collected according to the second method just described , especially for the values of @xmath7 , and in particular for @xmath26 , which would be the quantity that is most affected by the averaging . in a few cases one or more of the qpo parameters of a source are not available in the literature ; in some of those cases i myself have systematically analysed an published other qpo parameters ( e.g. , qpo frequency ) of those sources , and since i still have the data available , i analyse them myself for this paper to obtain those missing qpo parameters . i describe those cases individually , including a summary of the analysis procedures , below . in tables [ table_l ] and [ table_u ] i provide all the measurements of the maximum coherence , @xmath1 , and maximum rms amplitude , @xmath0 , for each qpo for each source , including the frequency of the corresponding qpo at which @xmath1 and @xmath0 occur ; in those tables i also provide references to the papers from which these values were collected . in most cases in the literature , the qpo properties reported are those measured over the full energy band of the pca ; nominally , this range is @xmath39 kev . in any case , the pca instrument is not very sensitive above @xmath40 kev because of the lower effective area at those energies , and because the background dominates the signal . in a few cases ( see tables [ table_l ] and [ table_u ] ) measurements were done over a different energy band to increase the sensitivity to khz qpos ; since the rms spectrum of the qpos increases steeply with energy ( see e.g. , * ? ? ? * ) , in those other cases the authors analysed the high - energy part of the data , above @xmath41 kev . on the other hand , the ageing of the pca units ( pcu ) , and changes applied to the gain voltage over the years to preserve the detectors affected the energy scale of the pca as well as its effective area as a function of energy . the largest impact occurred at the low - energy end , since the low - energy cut - off of the detectors is a strong function of the instrument s high - voltage ; from 1996 until present the lower energy boundary of the pca increased by @xmath42 due to the ageing of the instrument and the gain changes applied to it . because of the strong dependence of the rms spectrum of the qpos on energy , it is not straightforward to compare @xmath12 measurements over different energy bands : the higher the low - energy boundary of the band over which one measures @xmath12 , the higher the value of @xmath12 . on the other hand , @xmath7 measures the number of cycles of the oscillation which , at least to first order , does not depend on energy . therefore , one expects that the values of @xmath7 would not be much affected by the choice of the energy band over which they are measured . to confirm the above , and to assess the effect of the change of the energy band on the parameters , i proceed as follows : i use the data for 4u 0614 + 09 from @xcite who measured @xmath7 and @xmath12 of both khz qpos over the full pca band ( @xmath39 kev , nominally ) , and @xcite who measured the same properties in the @xmath43 kev range . in both cases they report @xmath7 and @xmath12 as a function of qpo frequency , therefore i compare @xmath7 and @xmath12 in both papers at the same qpo frequency . i find that @xmath7 is consistent with being the same in both energy bands . on the other hand , the rms fractional amplitudes in the @xmath44 kev band are @xmath45 ( @xmath31-@xmath32 error ) times higher than the rms fractional amplitudes in the full pca band . however , the energy spectrum of 4u 0614 + 09 is rather hard , and it may be that in other sources that have softer spectra , the rms amplitudes in different energy bands behave differently . to check this , i compiled rms amplitudes in the @xmath46 band for 5 sources : 4u 160852 @xcite , 4u 172834 @xcite , 4u 163653 @xcite , 4u 0614 + 09 @xcite , and 4u 182030 @xcite , and the corresponding rms amplitudes in the @xmath33 kev range @xcite . i find that the rms fractional amplitude in both bands are very well correlated ( the correlation coefficient is 0.97 ) , and that the rms amplitude in the @xmath33 kev band in these sources is @xmath47 times larger than the rms amplitude in the @xmath46 kev . this number is slightly higher , but still consistent ( within @xmath48-@xmath32 ) with the value obtained above for 4u 0614 + 09 . in summary , the above results show that : ( i ) one can compare directly @xmath7-values in different energy bands , at least as long as the bands are not too different from each other . because of this , in the rest of the paper i will compare @xmath7 values even if they come from different bands ; ( ii ) changes of the low - end of the energy band affect the rms amplitudes by @xmath49 of its value or less ( in the case of instrument gain changes and ageing , since changes of the low - energy boundary are smaller than those in the case of 4u 0614 + 09 previously discussed , the effect is probably much less important ) . in the next section , when i plot the qpo parameters against luminosity and against each other , i divide the rms amplitudes in the band going from @xmath50 to @xmath51 kev by @xmath52 to make them compatible with the rms amplitude values obtained in the full pca band . the values quoted in tables [ table_l ] and [ table_u ] , however , are the ones actually measured and presented in the original papers , in the energy band indicated there . once i have collected all @xmath1 and @xmath0 values for each qpo for each source , and the frequencies @xmath27 and @xmath53 at which those maximum values occur , i use figure 1 in @xcite to get the corresponding source luminosity : using the qpo frequency as input , i read off the luminosity of the source at that frequency from that figure . for three sources , 4u 160852 , 4u 163653 , and 4u 182030 ( see below ) , the maximum rms amplitudes of @xmath19 occur in states of the source in which the qpo frequencies are outside the range of frequencies in figure 1 of @xcite . in those three cases i either search the literature for flux measurements of the source in that state , or i extract spectra from the corresponding _ rxte _ observations , and calculate the luminosities myself , or both . below i explain those three cases in more detail . the luminosities that i report here are uncertain for a number of reasons : first , there is a statistical error in the fluxes derived from model fitting . given that these sources are quite bright when khz qpos are detected , and that _ rxte _ has a large effective area , these errors are small , usually less than @xmath54 . the second source of error is the accuracy with which one can determine the luminosity using the results of @xcite for a given qpo frequency . this is not so much how accurately one can read @xmath55 from their plot , but the fact that for each source there is not a single x - ray intensity or x - ray flux value that corresponds to a given qpo frequency . this is the so - called parallel - track phenomenon ( e.g. , * ? ? ? * ) . in the extreme cases , the x - ray intensity of the source at a given qpo frequency may by as much as a factor of 3 different ( e.g. 4u 160852 , * ? ? * ) . on average , however , the range of intensities at fixed qpo frequency is smaller , of the order of @xmath56 @xcite . the third source of error is the use of the @xmath57 kev flux as a measure of the bolometric flux of the source . this is discussed in detail by @xcite , and i therefore refer the reader to that paper . this effect contributes uncertainties of the order of @xmath58 of the reported luminosity values . the fourth source of error , if @xmath1 and @xmath0 depended not on the luminosity of the source , @xmath55 , but on the luminosity of the source normalized to its own eddington luminosity , @xmath59 , would be the use of a single eddington luminosity to normalise the observed luminosities of all the sources in the sample , as done by @xcite . of course , if @xmath1 and @xmath0 depended on @xmath55 , and not on @xmath59 , this would be of no concern . the value used by @xcite is @xmath60 , which is the eddington luminosity for a @xmath61 neutron star accreting matter with cosmic abundance . since all these systems are thought to have had a common evolutionary path , and to have accreted between @xmath62 and @xmath63 @xmath64 , the use of a single eddington luminosity introduces an uncertainty that is of the order of @xmath65 or less . ( notice that changes of the chemical composition of the accreting material may introduce a difference of up to a factor of 1.7 in the value of the eddington luminosity ; in the unlikely case that the neutron stars in this sample had different equations of state , the different redshifts at the surface of the neutron star may add an extra factor 1.5 . ) finally , the largest uncertainty in the luminosity comes from the error in the estimate of the distance to these sources ( see * ? ? ? * for a discussion in the context of these sources ) ; this uncertainty can introduce errors of up to 60% in the luminosity ( see , e.g. , * ? ? ? it should be mentioned that for gx340 + 0 and gx 5 - 1 , @xcite only give upper limits to their distances , hence for those two sources the luminosities in @xcite should be taken as upper limits . in tables [ table_l ] and [ table_u ] and in figure [ low ] i use a fixed error of @xmath65 in the values of @xmath59 as indicative of the error in the luminosity . it is clear from the previous discussion that this is a lower limit to the real error in this quantity . i now add a few remarks about special situations of some of the data for sources i include in this paper : 1 . : to measure @xmath4 , i use the same data in @xcite . i accumulate power density spectra every 64 seconds of data up to a nyquist frequency of 2048 hz ; i average a variable number of consecutive power spectra , in all cases less than 15 , in order to obtain significant qpo detections , and i search for high - frequency ( @xmath66 250 hz ) peaks . if i detect two qpo peaks , the one with the lowest frequency is @xmath18 ; if i only detect one peak , i decide whether it is @xmath18 or @xmath19 on the basis of the the rms - frequency relation ( e.g. , * ? ? ? if it is @xmath67 i discard it , if it is @xmath18 i measure @xmath26 . + individual @xmath26 measurements have relatively large errors . since @xmath26 is a function @xmath27 , i then sort the results on the basis of @xmath27 , and i calculate average @xmath26 values over a narrow range of @xmath27 . as usual , this procedure reduces the errors in @xmath26 at the expense of slightly underestimating its maximum value . + i do not use the results of @xcite because in that case they calculate @xmath7 using whole observations ; although they shift the individual power spectra of each observation ( * ? ? ? * see below ) , still in many of those observations the frequency of the qpo drifts by several tens of hz , which has a negative effect in the determination of @xmath7 . nevertheless , i find a @xmath4 value that is larger than , but still consistent with , the value in @xcite . + for this source , @xmath68 comes from @xcite ; they find this rms amplitude at a qpo frequency @xmath29 ( see table [ table_u ] ) that is outside the frequency range in @xcite , and hence no simultaneous luminosity measurement is available in @xcite . van straaten ( 2003 ) measured this rms amplitude during their interval c ; this corresponds to regions 1 and 2 in @xcite , who find that the bolometric flux of the source in those regions is @xmath69 . interval c of @xcite corresponds to the _ rxte _ obsid 30062 - 01 - 01 - 05 . i therefore downloaded those data from the _ rxte _ archive , i produce a light curve of this observation and find that there are no x - ray bursts or any anomaly in the data ; i then extract an x - ray spectrum using the standard2 data following the procedures described on the _ rxte _ web pages . i fit this spectrum with a model consisting of a blackbody and a power law plus a gaussian emission line at around 6.5 kev , all affected by interstellar absorption . the model fits the data well , with a @xmath70 per degree of freedom of about 1 . the @xmath57 kev unabsorbed flux of the source in this observation is @xmath71 , consistent with the flux of @xcite . using the same distance to this source as in @xcite , @xmath72 kpc , this flux corresponds to a luminosity @xmath73 , or @xmath74 using the eddington luminosity used by @xcite , @xmath75 . to measure @xmath4 , i use the same data presented in @xcite . the procedure to produce power spectra , identify the qpos , and measure @xmath26 is the same as the one described for 4u 160852 . the value of @xmath4 that i find is slightly higher than , but still consistent with , the one found by @xcite . + for this source , @xmath68 comes from @xcite ; they find this rms amplitude at a qpo frequency @xmath29 ( see table [ table_u ] ) that , as in the case of 4u 160852 , is outside the frequency range in @xcite , and hence no simultaneous luminosity measurement is available in @xcite . @xcite find this rms amplitude value during their interval c , which corresponds to the _ rxte _ obsid 60032 - 05 - 10 - 000 and 90409 - 01 - 01 - 02 . i produce a light curve of these observations and find that there is an x - ray burst in the first one , and no x - ray bursts or any anomaly in the data of the second one ; i then follow the procedures described on the _ rxte _ web pages to extract an x - ray spectrum from the standard2 data of the second observation . i fit this spectrum with a model consisting of a blackbody and a power law plus a gaussian emission line at around 6.5 kev , all affected by interstellar absorption . the model fits the data well , with a @xmath70 per degree of freedom of about 1 . the @xmath57 kev unabsorbed flux of the source in this observation is @xmath76 . using the same distance to this source as in @xcite , @xmath77 kpc , this flux corresponds to a luminosity @xmath78 , or @xmath79 using the eddington luminosity used by @xcite , @xmath60 . 3 . to measure @xmath4 and @xmath80 , i use the same data as in @xcite and @xcite . the procedure to produce power spectra , isolate @xmath18 , and measure @xmath26 is the same as the one described for 4u 160852 and 4u 163653 . to detect @xmath19 and measure @xmath30 , i proceed as follows : i group the power spectra on the basis of the frequency of @xmath18 , @xmath27 , such that within each group @xmath27 does not change by more than 50 hz . i then shift the power spectra such that in each group the frequency of @xmath18 is the same in all power spectra , and then i calculate an average power spectrum per group . this procedure eliminates the effect of the drift in @xmath18 , and since the frequency difference in this source is more or less constant when @xmath27 changes @xcite , it also corrects the effect of the drift in @xmath19 . i then fit the power spectra with two lorentzians that represent the qpos , and a constant that represents the poisson counting noise . the fit yields , among other parameters , @xmath30 . + for this source , the @xmath68 value is from @xcite ; they find this rms amplitude at a qpo frequency @xmath29 ( see table [ table_u ] ) that , as in the case of 4u 160852 and 4u 163653 , is outside the frequency range in @xcite , and hence no simultaneous luminosity measurement is available in @xcite . @xcite report this rms amplitude at a position in the colour - colour diagram of 4u 182030 that is consistent with the interval @xmath81 of @xcite . ( @xmath82 is a parameter that measures the position along the c - shaped track traced out by the source in a colour - colour diagram . ) during their observation , @xcite find that the intensity of the source is 297 counts s@xmath83 pcu@xmath83 , while in their observation , @xcite find the source at 312 counts s@xmath83 pcu@xmath83 . for their observation , @xcite find a @xmath57 kev luminosity of @xmath84 for an assumed distance @xmath85 kpc . @xcite use a distance to ngc 6624 , the globular cluster that contains 4u 182030 , @xmath86 kpc . to be consistent with the other luminosities for this source , which were taken from @xcite , i convert the luminosity in @xcite to the one corresponding to the distance to 4u 182030 in @xcite . this corresponds to @xmath87 , or @xmath88 using the eddington luminosity used by @xcite , @xmath89 . + 4 . to measure @xmath4 , i use the same data presented in @xcite and @xcite . the procedure to produce power spectra , identify the qpos , and measure @xmath26 is the same as the one described for 4u 160852 . 5 . for this source , @xcite recently reported a systematic study of @xmath26 and @xmath25 as a function of @xmath27 ; i therefore use @xmath90 and @xmath91 that are reported there . there are , however , very few reports of the properties of @xmath19 in the literature @xcite ; no reliable value of @xmath1 or @xmath0 are available ( see also fig . i therefore do not discuss @xmath19 of this source in this paper . 6 . for this source only one khz qpo has been detected @xcite , which appears to be @xmath18 @xcite , and hence there are no qpo parameters available for @xmath19 . 7 . since there are only three measurements of @xmath92 for cyg x-2 ( fig . [ q ] ) , i do not consider @xmath92 for this source in the rest of this paper . also , there are only two measurements of @xmath93 for this source , but the upper limits @xcite seem to indicate that the maximum @xmath93-value has been measured , and hence here i use @xmath93 in the analysis . 8 . to avoid detector safety triggers , telemetry saturation , and to reduce the dead - time effects produced by the high count rate of sco x-1 , some observations of this source were carried out with the source slightly off - axis , with some of the five proportional counter units of the pca switched off , recording only photons detected by the upper anode chain of the pca , recording only photons from a limited energy range , or using a combination of these constraints @xcite . despite all these efforts , the source count rate during these observations remained high , at a level in which the dead - time of the pca is still unknown . this means that the rms values ( taken from * ? ? ? * ) are very uncertain , and in fact could be larger than reported . for that reason , for sco x-1 in this paper i report the maximum rms amplitude as a lower limit . figure [ low ] shows the dependence of the maximum coherence and maximum rms amplitude of both khz qpos as a function of source luminosity in units of the eddington luminosity for a @xmath61 neutron star . from this figure it is apparent that @xmath3 and @xmath2 both decrease more or less exponentially with @xmath59 . for @xmath2 the e - folding scale is @xmath94 , while @xmath3 decreases significantly faster with @xmath59 than @xmath2 , with an e - folding scale @xmath95 . using roughly the same sample of sources , @xcite had already noticed that the rms amplitude of the upper khz qpo decreases as the luminosity of the source increases . from the same figure , it is also apparent that at low luminosity , @xmath4 first increases with @xmath55 up to @xmath96 , and then it decreases exponentially , at a faster rate than @xmath2 , with an e - folding scale @xmath97 . on the other hand , @xmath6 does not show any significant trend with luminosity ; although there is a hint that it stays constant up to @xmath98 , then it increases up to @xmath99 , and after that it decreases for increasing luminosity , statistically a fit with a function that represents that behavior is not significantly better than a fit of just a constant . the gap in figure [ low ] at @xmath100 separates the atoll sources at low @xmath55 , and the z sources at high @xmath55 ( see * ? ? ? * for a definition of atoll and z sources ) . that gap would be occupied by the four intermediate - type sources , gx 9 + 1 , gx 9 + 9 , gx 3 + 1 , and gx 13 + 1 , which so far have not shown any khz qpos @xcite . the upper limit to the rms amplitude of the qpo in these sources ranges from 1.6% to 2.6% . since the range of luminosities spanned by these sources @xcite is @xmath101 , these upper limits are much lower than would be expected from the interpolation of the trends of @xmath5 and @xmath3 with @xmath59 in figure [ low ] . from figure [ low ] it is apparent that the maximum rms amplitude of both khz qpos and the maximum coherence of the lower khz qpo are consistently lower in the z sources than in the atoll sources . a valid question would then be whether this difference could be due to a bias in the way in which the qpos of z and atoll sources are measured , combined with the frequency drift of the qpos ( see [ selection ] ) . a comparison , however , of the width of the lower khz qpo over a short time interval in a z and an atoll source shows that this is not the case ( as i describe in 2 , the rms amplitudes are less affected by the frequency drift ) . to see this , i compare an _ rxte _ observation of the z source sco x-1 from may 25 1996 ( see * ? ? ? * ; * ? ? ? * ) , and another _ rxte _ observation of the atoll source 4u 160852 from march 26 1998 ( see * ? ? ? * ) . in the sco x-1 observation the power spectrum shows two simultaneous khz qpos that move in frequency ; @xmath18 moves from @xmath102 hz to @xmath103 hz , and @xmath19 moves from @xmath104 hz to @xmath105 hz . in the observation of 4u 160852 the power spectrum also shows two simultaneous khz qpos that move in frequency . in this case @xmath18 moves from @xmath102 hz to @xmath106 hz , and @xmath19 moves from @xmath107 hz to @xmath108 hz . to compare the qpos in both observations over the same frequency range , i calculate an average power spectrum of the sco x-1 data using only intervals in which @xmath109 hz ; this corresponds to 24 individual power spectra each of them 16-s long . in the case of 4u 160852 i calculate an average power spectrum of the whole observation , corresponding to 41 individual power spectra each of them 64-s long . to correct for any residual drift in the qpo frequency , i apply the shift - and - add procedure described in @xcite . the results , however , do not change significantly if i average the individual power spectra without first applying this method , most likely because the frequency drift is small in both cases . from the above power spectra i find that for sco x-1 @xmath110 , whereas for 4u 160852 @xmath111 , which shows that in the z source sco x-1 , even over very short time intervals ( in this case 384 s ) , the lower khz qpo is significantly broader than in the atoll source 4u 160852 . this in turn shows that the drop of the maximum qpo coherence of the lower khz qpos at high luminosities in figure [ low ] is real . table [ table_l ] shows that in all sources the maximum coherence factor of the lower khz qpo , @xmath4 , occurs more or less at the same frequency ( the spread is @xmath112 ) ; table [ table_u ] shows that the same is true for the maximum coherence factor of the upper khz qpo , @xmath6 ( the spread in this case is @xmath113 ) . since the lifetime of an oscillation of frequency @xmath11 and coherence factor @xmath7 is roughly @xmath114 , the results described above also indicate that the maximum lifetime of @xmath18 , @xmath115 , first increases and then decreases with @xmath59 , whereas the maximum lifetime of @xmath19 , @xmath116 , is independent of @xmath59 . @xcite have shown that there is a general correlation between the average source luminosity and the average source hardness , h1 , defined as the ratio of the count rate in the @xmath117 kev band to the count rate in the @xmath118 kev band , measured with _ heao-1 _ ( see * ? ? ? * ) . in that respect , as expected , the plots of @xmath1 and @xmath0 vs. hardness in figure [ h1 ] show that @xmath3 and @xmath2 both increase with the hardness ratio h1 , @xmath4 increases with h1 and then it decreases for 4u 0614 + 09 , the hardest source in this sample , and as in the plots as a function of luminosity , @xmath6 is consistent with being constant with h1 . at the risk of pointing out something obvious , the opposite behaviour of the coherence and amplitude of the qpos to that in figure [ low ] is due to the anticorrelation between h1 and @xmath59 . i caution the reader that contrary to the @xmath59 values that i present in tables [ table_l ] and [ table_u ] and i plot in figure [ low ] , for this figure i use average values of the spectral hardness measured several years before the khz qpos were discovered . notice that gx 340 + 0 , for which the distance , and hence the luminosity , is uncertain ( see [ luminosity ] ) , appears in figure [ h1 ] close to the other z sources ( open symbols ; gx 340 + 0 is the point at h1 @xmath119 , just to the left of gx 17 + 2 at h1 @xmath120 ) ; since h1 is a distance - independent parameter , this suggests that the distance to gx 340 + 0 is not too much in error . it is interesting to notice in this figure that there is no gap in the distribution of sources as a function of hardness between the z and atoll sources . this is opposite to what is apparent in the plot of qpo parameters vs. luminosity , where there is a gap corresponding to the intermediate - type , the gx - sources ( see above ) . the h1 values of the gx - sources range from 0.03 to 0.14 . however , both in figure [ low ] and [ h1 ] there appears to be a gap in the values of @xmath121 , and perhaps also @xmath2 and @xmath3 between the two types of sources ( see also below ) . figure [ rmsq ] shows the relation between @xmath1 and @xmath122 for the same qpo . to produce this figure ( and the next one ) , i plot the @xmath1 vs. the @xmath0 values for the same source , even if they occur at slightly different luminosities . i also produced plots of @xmath1 vs. @xmath0 combining values for sources that are closest in luminosity , even if this means combining values from different sources , but the trends described below remain the same as with the previous choice . except for the case of 4u 0614 + 09 , @xmath4 and @xmath2 are positively correlated with each other . concerning 4u 0614 + 09 , the hardest source in the sample and the one at the lowest luminosity , it is as if in this case @xmath4 were too low for @xmath2 , or conversely , as if @xmath2 were too high for @xmath4 . this is already apparent in figure [ low ] , where @xmath18 of 4u 0614 + 09 ( low luminosity part of the plots ) looses coherence without a similar decrease of its rms amplitude . figure [ rmsq ] also shows that @xmath80 is independent of @xmath3 . let us examine the case of the low @xmath4 value in 4u 0614 + 09 to check whether it is real : the measurements of @xmath26 in this source come from @xcite , and are made over a different energy band than those of the other sources . for 4u 0614 + 09 @xcite use a band that starts at @xmath123 kev , whereas for other sources at slightly higher luminosity the @xmath26 values are measured using the whole pca band , nominally starting at @xmath124 kev ( see table [ table_l ] ) . in [ energy ] i compare measurements of @xmath26 as a function of @xmath27 ( and also @xmath30 as a function of @xmath29 ) for this source over the full pca band ( @xmath46 kev ; * ? ? ? * ) and over the @xmath125 kev band @xcite , and i find that both sets of values are compatible with each other . this indicates that most likely the difference in @xmath4 between 4u 0614 + 09 and other sources at slightly higher luminosity is real . recently , barret ( private communication ) did a similar analysis on the data of 4u 0614 + 09 as he did for instance for 4u 163653 @xcite , and he finds that @xmath26 in 4u 0614 + 09 is @xmath126 times smaller than in 4u 163653 , consistent with the results in table [ table_l ] . this confirms that @xmath4 in 4u 0614 + 09 is in fact smaller than in the other sources at similar but slightly higher luminosity , and hence that the decrease of @xmath4 at low luminosity is real . as i already mentioned , there appears to be a gap between the @xmath4 and @xmath2 values of the z and atoll sources , with the coherence showing the largest gap ( figure 3 , lower panel ) . 4u 0614 + 09 appears to be the only source to ( partially ) break this rule , since it has a @xmath4 value that is intermediate between those of z and atoll sources . similarly , the upper panel of figure 3 shows a gap between the @xmath3 values , but not for the @xmath6 values , of the z and atoll sources . while this can indicate a dependence on luminosity ( see fig . [ low ] ) or on spectral hardness ( see fig . [ h1 ] ) , this could also point to a difference between z and atoll sources . it is worth noting , however , that there is still a trend of @xmath127 and both rms amplitudes _ within _ the atoll sources in figure [ low ] . furthermore , there is a significant trend in the relations between @xmath1 and @xmath0 within the atoll sources ; e.g. , the relation between @xmath127 and @xmath128 ( lower panel of figure [ rmsq ] ) is @xmath129 different from a constant . all this suggests that the distinction between z and atoll sources can not be the ( only ) explanation for this difference . the plots of @xmath1 and @xmath0 of one qpo vs. the same parameters of the other qpo are shown in figure [ cross - plots ] . from this figure there are three things apparent : ( i ) @xmath6 is independent both of @xmath4 and @xmath2 ( upper panels ) ; ( ii ) as in the case of the plot of @xmath2 vs. @xmath4 , except for 4u 0614 + 09 , @xmath3 is positively correlated with @xmath4 . as in the case of @xmath18 , it appears as in 4u 0614 + 09 , the source with the lowest luminosity in the sample , the amplitude of @xmath19 , @xmath130 , were too high for the coherence of @xmath18 , @xmath121 , or alternatively , the coherence of @xmath18 , @xmath121 , were too low for the amplitude of @xmath19 , @xmath130 ( lower left panel ) . finally , ( iii ) the amplitudes of both khz qpos are positively correlated with each other , including in this case the low - luminosity source 4u 0614 + 09 ( lower right panel ) . from the lower panels of this figure , it is also apparent that there is a gap between the @xmath3 values of the z and atoll sources , similar to the ones described above to summarize the results from figures [ rmsq ] and [ cross - plots ] , the maximum amplitude and coherence of @xmath18 , @xmath2 and @xmath4 , and the maximum rms amplitude of @xmath19 , @xmath3 , are all correlated with each other ( in the case of @xmath4 , at least above @xmath131 ) , whereas the coherence of the upper khz qpo , @xmath6 , is independent of all the other parameters . i study the maximum amplitude , @xmath0 , and maximum coherence , @xmath1 , of the khz qpos as a function of luminosity and hardness for a large sample of low - mass x - ray binaries . i show , for the first time , that the maximum coherence of the lower khz qpo , @xmath121 , first increases up to @xmath131 and then decreases with luminosity , whereas the maximum coherence of the upper khz qpo , @xmath6 , is independent of luminosity . i also find that the maximum rms amplitudes of both the lower and the upper khz qpos , @xmath2 and @xmath3 , respectively , decrease monotonically with luminosity and increase monotonically with the hardness of the source . ( the dependence of @xmath3 on luminosity and hardness was first reported by * ? ? ? * ) . from the above results it follows that for all sources , @xmath5 and @xmath3 are positively correlated with each other . also , for all sources with @xmath132 , that is all sources in this paper except 4u 0614 + 09 , the hardest source in the sample and the one at the lowest luminosity , @xmath4 is positively correlated both with @xmath2 and @xmath130 . @xmath6 is independent of @xmath4 or the maximum rms amplitude of the khz qpos . since the frequencies at which the @xmath4 values occur are more or less the same in all sources , and the same is true for @xmath80 , although the frequencies in this case are higher , the dependence of @xmath4 and @xmath6 on luminosity also reflects the dependence of the maximum lifetime of the qpos on luminosity ( and spectral hardness ) . therefore , for the lower khz qpo , the maximum qpo lifetime first increases and then decreases with luminosity . for the upper khz qpo , the maximum qpo lifetime is independent of luminosity . in individual sources , both @xmath25 and @xmath26 increase with @xmath27 and then drop rather abruptly at the high end of the @xmath27 range ; @xmath28 also increases and then drops at high @xmath29 values , and @xmath30 is more or less constant or increases slightly with @xmath29 ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in the case of 4u 163653 , @xcite interpret the sudden drop of the coherence and rms amplitude of @xmath18 , together with the existence of a frequency above which @xmath18 is not detected , as evidence of the innermost stable circular orbit , isco , around the neutron star in this system . in general relativity , different from the newtonian theory of gravitation , the effective potential as a function of radial distance to the central source has a maximum . a particle in a circular orbit at that radius would be in unstable equilibrium ; if perturbed , the particle would fall onto the central object . no stable orbit around the central object is possible inside the radius of the isco , which in the schwarzschild case ( non - rotating central object ) is @xmath133 @xcite . from the results of individual sources ( see references above ) and those of the sample of sources that i present in this paper , it is apparent that the behaviour of the coherence and rms amplitude of the khz qpos as a function of the _ qpo frequency _ in _ individual sources _ is similar to the behaviour of the maximum coherence and maximum rms amplitude of the khz qpos as a function of _ luminosity _ in the _ sample of sources_. since in individual sources there is a general relation between qpo frequencies and source intensity , in the sense that at higher intensity the qpos generally appear at higher frequencies ( but remember the parallel - track effect ) , this raises the question of whether the same mechanism may be behind both behaviours . at first sight , this may appear problematic because it is known that the relation between qpo frequency and intensity in individual sources is more complex than the general frequency - luminosity trend described above , and detailed qpo frequency vs. intensity plots show the so - called parallel - track phenomenon ( see e.g. , * ? ? ? however , the link between these two behaviours need not be luminosity , but could be the high - energy emission ( or hardness ) in these systems . on one hand , in the sample of sources the maximum coherence and maximum rms amplitude of the khz qpo , except the maximum rms amplitude of @xmath19 , appear to correlate fairly well with spectral hardness ( see figure [ h1 ] ) , while on the other hand in individual sources the qpo frequencies are well correlated with the spectral hardness of the source @xcite , the index of the power law that fits the high - energy part of the x - ray spectrum @xcite , or @xmath16 , a parameter that measures the position along the track traced out by the source in a colour - colour or colour - intensity diagram ( called @xmath134 and @xmath135 in the z and atoll sources , respectively ; see e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . from this , it follows that in individual sources there should be a relation between qpo rms amplitude and coherence on one hand and spectral hardness on the other . to my knowledge , a plot like that has never been published , but can be easily constructed using published results . for instance , in the case of 4u 160852 , this can be done by combining the plots of qpo frequency vs. hardness of figure 3 in @xcite and the plots of qpo coherence and qpo rms amplitude vs. qpo frequency of figure 2 in @xcite and figure 3 in @xcite , respectively . in figure [ 1608 ] i show the rms amplitude ( upper panel ) and coherence ( lower panel ) of the lower khz qpo ( left panel ) and the upper khz qpo ( right panel ) in 4u 160852 as a function of hardness . in this case the hardness is defined ( see * ? ? ? * ) as the logarithm of the ratio of the pca count rate in the @xmath136 kev band to the pca count rates in the @xmath137 band . ( notice that the hardness in this plot is defined over a different energy band than in the case shown in figure [ h1 ] . ) the lines connecting the data points indicate the evolution of qpo frequency , which generally increases from right to left . this figure shows the drop of the rms amplitude and coherence of the lower khz qpo and of the rms amplitude of the upper khz qpo at the very high end of the qpo frequency range ( cf * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) the trend seen in figure [ 1608 ] for the coherence of the lower khz qpo in 4u 160852 is rather similar to the behaviour of the maximum coherence of the lower khz qpo for the sample of sources in figure [ h1 ] ( left panel ; notice again that the hardness is defined over a different energy band in each figure ) : when 4u 160852 is hard , to the right of figure [ 1608 ] , the coherence is low , while for the hardest source in the sample , to the right of the left panel of figure [ h1 ] , the maximum coherence is also low . when 4u 160852 becomes softer , to the left of figure [ 1608 ] , the coherence first increases , and then drops abruptly ( see black points in figure [ 1608 ] , lower panel ) , while in the sample of sources the maximum coherence behaves in a similar way . the behaviour of the rms amplitude in 4u 160852 and the maximum rms amplitude in the sample of sources is more or less similar , except that in the case of 4u 160852 when the source is hard the rms amplitude is low , increases as the source gets softer and then drops abruptly ( see black points in figure [ 1608 ] , upper panel ) , whereas in the sample of sources moving from hard to soft sources the maximum rms amplitude remains more or less constant and at the end drops . to summarize , the relevant conclusion from this comparison is this : both the rms amplitude and coherence of the lower khz qpo in 4u 160852 and the maximum rms amplitude and maximum coherence of the lower khz qpo in the sample of sources drop abruptly when the hardness decreases . the comparison between individual sources and the sample of sources suggests that the same mechanism is responsible for the drop of coherence and rms amplitude of the lower khz qpo with qpo frequency in individual sources as well as for the drop of maximum qpo coherence and maximum qpo rms amplitude with luminosity in the sample of sources . the comparison between 4u 160852 and the sample of sources in the previous paragraph suggests that most likely the mechanism is related to the high - energy emission in these systems . this does not necessarily mean that the fractional emission at high energies ( represented by the hardness or x - ray colors ) is the root mechanism that drives all qpo parameters ( qpo frequency , coherence , and rms amplitude ) . for instance , one possibility ( there could be many others ) is that the ( instantaneous ) mass accretion rate sets the size of the inner radius of the disc @xcite , which in turn determines the qpo frequency , as well as the relative contribution of the high - energy part of the spectrum to the total luminosity . if the efficiency of the modulation mechanism and the lifetime of the oscillations that produce the qpo depended upon the emission from the high - energy part of the source spectrum ( see [ modulation ] for a discussion of possible ways in which this could happen ) , observationally it would appear as if the coherence and rms amplitude of the qpo depended upon the qpo frequency , and hence upon the radius in the disc at which the qpo is produced . the sudden drop of the coherence and rms amplitude of the qpo at some qpo frequency would then be associated to a dynamical peculiarity in the accretion disc , for instance the isco . observing the same source repeatedly would not allow to distinguish the above scenario from one in which qpo coherence and rms amplitude were actually set by qpo frequency or the dynamics in the accretion disc . to distinguish between these two options , one would need to observe a sample of sources of khz qpos for which the mass - accretion rate , and hence the relative contribution of the high - energy part of the spectrum to the total emission , was different from source to source . in that case , qpo coherence and rms amplitude would drop for sources accreting mass at higher rates , even if the frequency of the qpo was more or less the same from source to source . since , as i have shown in this paper , this is the general behaviour observed in sources of khz qpos , it is reasonable to infer that a mechanism similar to the one i described in the previous paragraph is effective in setting the coherence and rms amplitude of the khz qpos . if this is correct , this also implies that the drop of qpo coherence and rms amplitude as a function of qpo frequency in individual sources can not be due to effects of the isco . note also that in individual sources not just the rms amplitude of the khz qpos , but also the rms amplitude of other lower - frequency qpos decrease with increasing qpo frequency . for instance , in four atoll sources , 4u 172834 , 4u 160852 , 4u 0614 + 09 , and 4u 163653 , the rms amplitudes of the `` bump '' , a qpo at @xmath138 hz , the `` hump '' , a qpo at @xmath139 , and the hectohertz qpo at @xmath140 hz , all drop as the frequencies of the khz qpos increase , in a similar fashion as the amplitude of the upper and lower khz qpos ( e.g. , * ? ? ? * and references therein ; see also there a description of these other qpos ) . this also argues against the interpretation of the isco as the cause of the drop of the rms of the khz qpos , and indicates that the amplitudes of _ all _ variability components are set by the same mechanism which , as i suggested , could be the same one that governs the high - energy spectral component . the idea that the rms amplitude and coherence of the qpos depend on the high - energy emission , and not on qpo frequency , can in principle be tested ; for instance , if one of the khz qpo sources showed a sudden change in the high - energy part of the spectrum , but the change of qpo frequency was less sudden . this could for example occur if the component that sets the high - energy emission in the spectrum of these sources and the one that sets the qpo frequency had different responses to changes of the mass accretion rate . this scenario is similar to the one proposed by @xcite to explain the parallel tracks in the frequency vs. intensity plots of these sources . if this was the case , one would observe a khz qpo that had a coherence and rms amplitude that would not match those expected on the basis of the @xmath141 and @xmath142 relations in , e.g. , @xcite . i would like to point out that figure 1 in @xcite appears to show an effect like this : individual measurements of the coherence of the lower khz qpo at the same qpo frequency differ significantly from one another . e.g. , in 4u 163653 at @xmath143 hz , @xmath92 ranges from @xmath144 to @xmath145 . this difference , however , may still be due to uncorrected drifts of the qpo frequency during the intervals in which @xcite measured the coherence of the qpo , although according to their description they had taken this into account in producing their figure . how can the mechanism that produces the high - energy emission in these systems change the coherence and rms amplitude of the khz qpos ? there is general agreement that the oscillation mechanism that produces the quasi - periodic variability must be in the disk ( see the references to theoretical models in [ intro ] ) ; it is very easy to come up with characteristic dynamical frequencies in the disk that match the observed frequencies of the khz qpos , whereas it appears more difficult to have a `` clock '' somewhere else in the accretion flow . nevertheless , from the early observations it became clear that emission from the disk alone can not explain the rms amplitudes of the khz qpos , since in some cases the modulated luminosity in the qpos is @xmath146 of the total emission , whereas at the same time the emission of the disk is @xmath147 or less . furthermore , the steep increase of the rms amplitude with energy imply a large modulation of the emitted flux , up to @xmath148 at @xmath149 kev ( see e.g. , * ? ? ? * ) , at energies where the contribution of the disk is negligible . hence , while the oscillation mechanism probably takes place in the disk , most likely the modulation mechanism is associated to the high - energy spectral component in these sources . the high - energy part of the x - ray spectrum of low - mass x - ray binaries can be described in terms of thermal comptonization @xcite . in low - luminosity atoll sources @xcite , the hard spectral changes are explained as due to changes of the properties of this component as the average mass - accretion rate , @xmath150 , changes . for instance , for 4u 160852 , @xcite find that at low inferred @xmath150 , the comptonizing component has a relatively high temperature , @xmath151 kev , and is optically thin , @xmath152 . the high - energy part of the spectrum resembles a rather flat power law ( power - law index @xmath153 ) , with a high - energy cut - off above @xmath154 kev @xcite . at high inferred @xmath150 , @xcite find that the comptonizing component becomes cooler , with @xmath155 kev and the optical depth increases , @xmath156 . the high - energy part of the spectrum resembles a power law that is steeper ( power - law index @xmath157 ) with a high - energy cut - off that is at lower energies than in the low @xmath150 case . in the case of high - luminosity z - sources @xcite , the comptonizing component is relatively cool and optically thick as in the case of the low - luminosity atoll sources at high @xmath150 @xcite . using a simple time - dependent comptonization model , @xcite calculated the spectrum of variability that would be produced by oscillations in the ( i ) injection rate of seed photons , ( ii ) density , and ( iii ) temperature of the comptonizing medium ( see also * ? ? ? they find that to reproduce the rms spectrum of the lower khz qpo in 4u 160852 , the variability must be mostly driven by an oscillation of the density of the comptonizing medium . they also find that in the case of variations of the density of the comptonizing medium , the largest rms variability in the @xmath158 kev range occurs as the optical depth of the medium is smallest ( see their figure 4 , panel b ) . if the high - energy emission is due to comptonization , the results of @xcite , together with the global correlation between luminosity and spectral hardness @xcite , provide a possible explanation , at least qualitative , of the dependence of the rms amplitude of the khz qpos on luminosity and hardness . at low luminosity , corresponding to low mass accretion rate and hard spectra , the comptonizing plasma is optically thin , and hence the amplitude of the variability is high . when mass accretion rate increases , the luminosity increases and the source becomes softer , and since the optical depth of the comptonizing plasma increases , the amplitude of the variability decreases . this would explain both the drop in rms amplitude of the khz qpos at high qpo frequency in individual sources , and that of the maximum rms amplitude in the sample of sources at high luminosity ( low hardness ) . in individual sources , the rms amplitude of the khz qpos also decreases at low frequencies , which is more difficult to explain in this simplified scenario . also , in this scenario there is no straightforward explanation of the drop of the coherence of the lower khz qpo at high frequencies in individual sources , and at high luminosities in the sample of sources . @xcite have analyzed the rms energy spectra as a function of frequency of two low - mass x - ray binaries , the z source gx 340 + 0 and the atoll source 4u 160852 . they find that the variability , including that in the frequency range of the khz qpos , is primarily due to variations of the luminosity of the boundary layer , with the emission of the accretion disk being much less variable . they also find ( see also * ? ? ? * ) that in gx 340 + 0 , the contribution of the boundary layer to the observed emission decreases as mass accretion rate increases , as the source moves along the z - shaped track in the color - color diagram from the horizontal branch to the normal branch ( see * ? ? ? * for an explanation of the branches in the color - color diagram of z sources ) . @xcite speculate that this decrease could be due either to obscuration of the boundary layer by a thickened accretion disk , a quantitative change in the structure of the boundary layer , or a complete disappearance of the boundary layer , when @xmath159 . in this scenario , the decrease of the rms variability of the khz qpos at high luminosities ( high mass accretion rate ) would be related to the lower relative contribution of the boundary layer , which produces the bulk of the qpo variability , to the total emission . the obscuration mechanism can not explain , however , the decrease of the coherence of the qpo at high luminosity . one way to explain the loss of coherence would be if the modulation mechanism in the boundary layer would damp out the oscillations . for instance , the flow of mass onto the boundary layer could be modulated more or less periodically by the disc , and emission would proceed as a series of shots when mass reaches the boundary layer . changes in the configuration of the boundary layer , as those proposed by @xcite , could increase the damping and reduce the lifetime of the qpo . this is of course just speculation ; it remains to be seen whether it is possible to explain the decrease of qpo coherence by significant changes in the properties of the boundary layer . @xcite suggested that there is a critical luminosity for low - mass x - ray binaries with a neutron - star primary , @xmath160 erg s@xmath83 ; sources below that level show a hard power - law component , whereas sources above that that level do not . figure [ low ] shows a sort of dichotomy between the maximum rms amplitude and maximum coherence of the khz qpos in the z sources , for which @xmath161 and the atoll sources , for which @xmath162 . the separation , however , occurs at a luminosity that is @xmath163 times larger than suggested by @xcite . nevertheless , at least for the lower khz qpo , the maximum coherence and maximum rms amplitude change significantly and rather smoothly within the atoll class ; this may indicate that , if the qpo properties depend upon the emission of the power - law component , the relative importance of this component may change in a less abrupt manner than suggested by @xcite . the @xmath164 correlation ( fig . [ cross - plots ] ) suggests that the same mechanism sets the amplitude of both qpos . but in each source @xmath2 and @xmath3 occur at different luminosities ( see tables [ table_l ] and [ table_u ] ) , and hence at different times , therefore the mechanism can not be acting simultaneously on both khz qpos . ( in fact , in each source @xmath2 occurs at a higher frequency of the upper khz qpo , @xmath29 than @xmath165 ; see , e.g. , * ? ? ? * ; * ? ? ? * ) . if one ignores the case of 4u 0614 + 09 , the source with the lowest luminosity in the sample , also @xmath4 is correlated with @xmath2 as well as with @xmath3 . from this it is tempting to infer that the mechanism that sets the rms amplitude of the khz qpos sets the coherence of the lower khz qpo as well , at least for @xmath132 . at low luminosities either this mechanism is not effective in setting @xmath4 , or another mechanism comes into play that counteracts the effect . the lack of a correlation between @xmath6 and @xmath4 , @xmath2 , or @xmath3 indicates that the mechanism that sets the latter three quantities does not have a significant effect on the coherence of the upper khz qpo . as i showed in [ results ] , for 4u 0614 + 09 either @xmath121 is too low for @xmath2 , or @xmath2 is too high for @xmath4 . but since the maximum coherence for @xmath18 in 4u 0614 + 09 lies away from the @xmath166 as well as from the @xmath167 correlations , it appears more likely that in this case @xmath4 is too low both for the @xmath2 and @xmath3 , and not the other way around . this hypothesis is reinforced by the fact that a larger @xmath4 would bring 4u 0614 + 09 closer to the other atoll sources in figures [ rmsq ] and [ cross - plots ] , whereas lower @xmath2 and @xmath165 values would put 4u 0614 + 09 in the gap of the correlations in those two figures . if this is the case , it is unclear what mechanism could be responsible for the drop of coherence without a simultaneous drop of rms amplitude . comparison of the upper right panels of figures [ h1 ] and [ 1608 ] suggests that the behaviors of @xmath30 in individual sources ( in this case 4u 160852 ) and @xmath6 in the sample of sources as a function of hardness are also similar , in that both the coherence of the upper khz qpo in individual sources , and the maximum coherence of the upper khz qpo in the sample of sources are more or less independent of the hardness of the source . however , the lack of correlation between @xmath30 and qpo frequency in 4u 160852 appears to be exceptional , and in other sources , both atoll ( see , e.g. , * ? ? ? * ; * ? ? ? * ) as well as z - sources ( see , e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , @xmath30 increases with qpo frequency , and hence decreases with hardness . from the data in figure [ h1 ] it is not possible to discard a similar slow decrease of @xmath6 vs. hardness . in black - hole systems , high - frequency qpos have been observed in the range of tens up to @xmath168 hz with typical rms amplitude and coherence values @xmath169% and @xmath170 , respectively @xcite . the luminosities of these systems when high - frequency qpos are observed are @xmath171 erg s@xmath83 ( e.g. , in xte j1550564 ; * ? ? ? * ; * ? ? ? * ) or , for the typical mass of the black holes , @xmath172 , @xmath173 . compared to the neutron - star systems in figure [ low ] , in black - hole systems the rms amplitude and coherence of the high - frequency qpos are too low for @xmath59 of the source , but they would be consistent with the values for the z sources at the same luminosity , i.e. , the luminosities not being normalized to the source own eddington luminosity ( of course , to plot the black - hole sources in figure [ low ] one would have to divide their luminosities by @xmath174 , the eddington luminosity for a @xmath175 neutron star , as was done for the other sources in that plot ) . it remains to be seen whether in black - hole systems the modulation mechanism of the high - frequency qpos would be similar to that of the khz qpos in neutron - star systems . i study the maximum amplitude , @xmath0 , and maximum coherence , @xmath1 , of the khz qpos as a function of luminosity and hardness for a dozen low - mass x - ray binaries . i find that : 1 . the maximum coherence of the lower khz qpo , @xmath4 , first increases up to @xmath131 and then decreases with luminosity . the maximum coherence of the upper khz qpo , @xmath6 , is independent of luminosity . 3 . the maximum rms amplitudes of both the lower and the upper khz qpos , @xmath2 and @xmath3 , respectively , decrease monotonically with luminosity 4 . both @xmath3 and @xmath2 increase with the source hardness , @xmath4 first increases with hardness and then drops for the hardest source in the sample , and @xmath176 is independent of hardness . 5 . the relation between @xmath1 and @xmath0 with luminosity in the sample of sources is similar to the relation between @xmath7 and @xmath12 with qpo frequency in individual sources . the similarity extends also to hardness in the sample of sources and in at least one individual source , 4u 160852 . the above argues against the interpretation that the drop of qpo coherence and qpo rms amplitude at high qpo frequency in individual sources is due to effects related to the innermost stable orbit around the neutron star in these systems . the drop of coherence and rms amplitude of the khz qpos , both in individual sources and in the sample could be produce by changes in the properties of the region in the accretion flow where the flux that produces the qpo is modulated . i thank didier barret and cole miller for valuable discussions on the ideas that i present in this paper . i thank diego altamirano for comments on an earlier version of this manuscript . i also thank michiel van der klis for insightful comments that helped improve the paper . this research has made use of data obtained through the high energy astrophysics science archive research center online service , provided by the nasa / goddard space flight center . the netherlands institute for space research ( sron ) is supported financially by nwo , the netherlands organisation for scientific research .	i study the behaviour of the maximum rms fractional amplitude , @xmath0 and the maximum coherence , @xmath1 , of the kilohertz quasi - periodic oscillations ( khz qpos ) in a dozen low - mass x - ray binaries . i find that : ( i ) the maximum rms amplitudes of the lower and the upper khz qpo , @xmath2 and @xmath3 , respectively , decrease more or less exponentially with increasing luminosity of the source ; ( ii ) the maximum coherence of the lower khz qpo , @xmath4 , first increases and then decreases exponentially with luminosity , at a faster rate than both @xmath5 and @xmath3 ; ( iii ) the maximum coherence of the upper khz qpo , @xmath6 , is more or less independent of luminosity ; and ( iv ) @xmath0 and @xmath1 show the opposite behaviour with hardness of the source , consistent with the fact that there is a general anticorrelation between luminosity and spectral hardness in these sources . both @xmath0 and @xmath1 in the sample of sources , and the rms amplitude and coherence of the khz qpos in individual sources show a similar behaviour with hardness . this similarity argues against the interpretation that the drop of coherence and rms amplitude of the lower khz qpo at high qpo frequencies in individual sources is a signature of the innermost stable circular orbit around a neutron star . i discuss possible interpretations of these results in terms of the modulation mechanisms that may be responsible for the observed variability . [ firstpage ] stars : neutron x - rays : binaries x - rays : individual : 4u 0614 + 09 x - rays : individual : 4u 160852 x - rays : individual : 4u 163653 x - rays : individual : 4u 172834
it has recently been noted that the matter produced in heavy ion collisions at rhic can not be weakly coupled but instead behaves as a good liquid @xcite . the evidence for the sqgp is large but mainly consists of the following two points : 1 . the observed collective flows at rhic can be explained by hydrodynamics showing that the dissipative lengths are very short and 2 . binary bound states are seen to exist in lattice simulations above @xmath1 and are also predicted @xcite using lattice interparticle potentials . since a complete quantum description tends to rely on perturbative methods other approaches have to be adopted in order to perform calculations at strong coupling . one approach as was discussed in @xcite , is to model the strongly interacting quark and gluon quasiparticles as a classical non - relativistic colored coulomb gas . this model is analyzed using molecular dynamics ( md ) simulations in which real time correlators can be extracted . the interaction between quarks was taken as a coulomb potential with a strong repulsive core in order to _ mock up _ quantum - mechanical effects . a second example is a model treating charmonium in the sqgp @xcite , where initial ensembles of quark pairs from pqcd events are evolved classically according to a langevin model with an additional interaction potential included between quark - antiquark pairs . the interaction potential was taken from lattice @xmath2 internal energy data with quantum - mechanical effects mimicked by simply turning off the potential below the approximate bohr radius . if these models are to be refined it is necessary that we consider quantum - mechanical effects more carefully . the construction of an effective interaction which mimics the effects of quantum dynamics was first performed by kelbg @xcite . the goal of this work is to generate classical effective @xmath3 potentials which take into account quantum - mechanical effects , first the dynamics and then the symmetry or anti - symmetry which must exist in the wavefunction for the @xmath4 pair . we want to find an effective potential @xmath5 whose @xmath6 boltzmann factor yields the diagonal term of the @xmath7 density matrix : @xmath8 for example , if this corrected interaction were used in the modeling of charmonium discussed in the previous paragraph , the final equilibrium distribution obtained by the fokker - planck evolution would be the correct quantum - mechanical distribution . this would obviously improve estimates for the intermediate states as well . these effective interactions would also have use in the md simulations discussed earlier . in this work we calculate effective quark - antiquark potentials from the internal energy of static @xmath9 pairs calculated in @xcite based on lattice calculations of the @xmath9 @xmath10 energy in @xcite . section [ veff ] of this paper summarizes a number of previous works @xcite , including the definition of the slater sum and its use in creating effective potentials . the original results of this paper start in section [ veff_lat ] , where the methods presented in section [ veff ] are applied to potentials extracted from lattice data . anticipating applications of this work to many - particle systems , we define the @xmath11-particle slater sum : @xmath12 where @xmath13 are the n - particle energy eigenfunctions and @xmath14 is the thermal wavelength . integrating the slater sum in coordinate space yields the partition function : @xmath15 in the case of ideal particles using boltzmann statistics the partition function is given as @xmath16 . the corresponding free energy can then be separated in the following way : @xmath17 substituting eq . [ pf ] for @xmath18 in the above equation yields : @xmath19 this expression should now be compared to the classical counterpart of the free energy : @xmath20 in the classical limit one obtains : @xmath21 and near the classical limit one can include multi - body interactions due to the quantum correlations between particles : @xmath22 where @xmath23 and @xmath24 are effective two- and three - body interactions respectively and can be expressed in terms of two- and three - particle slater sums as : @xmath25 keeping only the first term , @xmath23 , in the series for @xmath26 accounts for quantum effects up to first order in the coupling parameter , @xmath27 where @xmath28 are respectively the ion charge , the wigner - seitz radius @xmath29 and the temperature . the second term @xmath24 introduces corrections to the effective potential of order @xmath30 . the previous section motivates us to take the binary approximation : @xmath31 where @xmath23 is defined as in eq . [ eq : effpot ] . this approximation is exact for a 2-body systems ( for example the model discussed in section [ intro ] where a heavy quark only interacts with its diagonal partner ) . for the remainder of this paper , we now focus on an explicit calculation of @xmath23 . our method involves examining the two - particle density matrix : @xmath32 where the diagonal term is proportional to the needed slater sum . the two - particle density matrix can then be factored into center - of - mass and relative components , @xmath33 . the two - body effective potential can then be found in the limit as @xmath34 and @xmath35 . since @xmath36 one is left with solving for @xmath37 . acting on @xmath38 with the one - particle hamiltonian ] , @xmath39 where @xmath40 is the reduced mass of the two particle system and @xmath41 is the relative coordinate it can be shown that @xmath42 satisfies : @xmath43 the boundary condition on the density matrix is given from the fact that \{@xmath13 } is a complete basis : @xmath44 we use the same method of solution on eq . [ eq : bloch ] which is given in @xcite . the solution to the free - particle density matrix ( v=0 in eq . [ eq : bloch ] ) is given as @xmath45 where @xmath46 and @xmath47 is the reduced mass of the two - particle system . we then define the effective potential as in the introduction : @xmath48 the effective two body potential @xmath23 of interest is simply the diagonal part of @xmath49 when @xmath50 . substituting the above form for the density matrix [ eq : general ] into the bloch equation [ eq : bloch ] one finds that @xmath49 satisfies : @xmath51 in the limit that @xmath52 , @xmath53 , or @xmath54 is small , eq . [ eq : exact ] may be linearized : @xmath55 the solution of which we call @xmath56 . as has been emphasized in @xcite , this condition is far more permissive than the typical condition from perturbation theory that @xmath52 be small . this equation can be solved exactly : @xmath57 where the green function @xmath58 is given by : @xmath59 \exp(-\frac{m(\|\vec r-\vec r_1\|+\|\vec r\p-\vec r_1\|)^2}{2\beta } ) \label{eq : green}\ ] ] for a spherically symmetric potential @xmath60 , the angular integration in the above equation can be performed and the final result for the diagonal term is given as @xmath61 is : @xmath62dr_1 \label{eq : u1}\ ] ] equation [ eq : u1 ] will be the starting point for our future analysis of lattice potentials . it can be shown that the same result can be found by solving for the slater sum @xmath63 explicitly using plane waves for the wavefunction and solving perturbatively in the coupling constant . it has already been shown in @xcite that the linearized approximation which yielded eq . [ eq : approx ] is satisfactory assuming the contribution from bound states is small . the term dropped from the linearized equation , @xmath64 , becomes significant when the gradient of the effective potential is on the order of the temperature . this is exactly the condition leading to large contributions from the lowest bound states to the density matrix . in order to take into account the effect of bound states explicitly we follow @xcite , where the bound and free states are considered as two separate contributions to the slater sum . both the partition function and the two - particle slater sum can be separated as @xmath65 where the operator @xmath66 projects out or removes the free component . the terms on the right - hand side correspond to the free and bound state contribution to the slater sum , respectively . there are various ways of defining the projection operator @xmath66 , so that it separates the slater sum into a part dominated by the free contribution ( call it @xmath67 ) and a part dominated by the bound state contribution . in principle , these are merely different conventions , however since we will identify @xmath67 with the slater sum we already calculated , some conventions are better than others . explicitly , one can use the riewe - rompe convention which acts as a sharp cutoff between the bound and continuum states . more specifically , the projection operators acting on the two particle slater sum yield @xmath68 where @xmath69 . using this convention , we would be adding a term to the slater sum even where @xmath54 is small , where our calculation needs no correction . in order to circumvent this , we use the brillouin - planck - larkin ( bpl ) convention since it has a continuous transition at the continuum edge , @xmath70 . \label{eq : bound}\ ] ] the advantage of this convention can be seen by examining small and large @xmath54 . when @xmath54 is small , the coefficient @xmath71 , and no significant contribution is added to the free slater sum . when @xmath54 is large , @xmath72 ( @xmath73 is negative ) , making the bound state part the dominant contribution to the slater sum , as it should be at low temperatures . in the above expressions @xmath74 is the free - particle slater sum and can be found by substituting the result from eq . [ eq : u1 ] into eq . [ eq : effpot ] . the second term includes a sum over bound states with wavefunctions given as solutions to the schrdinger equation . we make one final consideration in creating effective potentials : symmetry conditions for identical particles . in the case of the identical fermions , the two - particle density matrix , written as an imaginary time propagator , becomes @xmath75 for our situation where two particles interact according to a potential only dependent on @xmath76 , their relative separation , this formula dramatically simplifies : @xmath77 \label{eq : antisym}\ ] ] again , we will be interested in the limit @xmath50 . we may calculate explicitly the off - diagonal term of the density matrix needed in eq . [ eq : antisym ] , or we may follow the approach in @xcite and approximate the off - diagonal term using only the off diagonal free particle density matrix @xmath78 the final form for the correctly symmetrized density matrix @xmath79 is given as : @xmath80e^{-\beta u_1(\vec{r})}\ ] ] where a minus sign is used for fermions and a plus sign is for bosons . we now apply the methods outlined in the previous section to the case of the coulomb potential . we first note that equation [ eq : u1 ] can be integrated exactly in the case of a coulomb potential , @xmath81 with the result : @xmath82\biglb)\ ] ] in figure [ fig : cbound ] we show the results of including bound states for the case of the coulomb potential . the dotted green and blue curves show the results for @xmath83 for values of @xmath84 and @xmath85 respectively . the corresponding solid curves show the results of including the first lowest bound state in the sum in eqn [ eq : bound ] . the solid red curve shows the result at @xmath86 where the result @xmath87 vanishes but the effective potential sits at the energy of the lowest bound state @xmath88 in atomic units . also shown on this plot ( black triangles ) are numerical results of a monte - carlo simulation of the density matrix . paths in imaginary time were sampled from the free - particle distribution by way of the levy construction , and the actions of these paths were calculated and averaged according to their weights determined from the sampled , free - particle distribution of paths . see ceperley @xcite for an extremely useful and pedagogical introduction into monte - carlo methods for calculating @xmath11-particle ( bose ) density matrices . . dotted lines show the kelbg potential @xmath83 and corresponding solid line shows potential including correction from lowest bound state . the triangles are the results of pimc ( see text ) . ] in fig [ fig : symf ] we show the result for two identical electrons after the correct symmetrization is preformed . again the dotted lines show the result of @xmath87 and the corresponding solids lines are the result after anti - symmetrization of the density matrix [ eq : symf ] . now we focus on the problem at hand , obtaining an effective potential appropriate for simulations of heavy flavor quarks or plasma quasiparticles . internal and free energies for static quark - antiquark pairs have been calculated at temperatures above deconfinement @xcite , and either of these energies could be used to construct an effective interaction . however , it is not clear which potential would best be used in the simulation of dynamical heavy quarks . shuryak and zahed @xcite have argued that the rapid rotation of the quarks in bound states causes the timescales of interest to be relatively short , and therefore the appropriate energy for the interaction would be the internal energy . however , mocsy and petreczky @xcite , have found that potential models which use the free energy instead of the internal energy as the interaction potential best describe the spectral functions which were extracted from the euclidean correlators on the lattice using the maximal entropy method . while this does not yet rule out the argument of shuryak and zahed ( there is still significant uncertainty in the spectral function obtained using the maximal entropy method ) , it clearly establishes that at the time of this publication , there is no clear answer concerning which potential best describes the interaction between heavy quarks . with no specific bias towards the use of either energy for describing the interaction , we will consider the internal energy . we parameterize the internal energy in the temperature range @xmath89 as follows : @xmath90 this potential is similar in form to the screened cornell potential as first proposed in @xcite . the first part of the potential is the usual screened coulomb potential with temperature dependent coupling ( @xmath91 ) and debye mass ( @xmath92 ) . these values were extracted from the lattice data and parameterized as follows . for the coupling constant we use a form similar to that from the leading log order renormalization group equation , @xmath93^{-1}$ ] . the fit to the debye mass goes as @xmath94 which is consistent with the lowest - order perturbative calculations @xcite with additional corrections to account for the data better : @xmath95 . the second part of the potential takes on the same form as a screened linear confining potential . we choose a different functional form than the usual @xmath96 in order that the potential remains flat at intermediate and large distances instead of having a maximum and then becoming repulsive at larger @xmath97 as in the case of the cornell potential . we find that a parameterization of the _ string tension _ as @xmath98 reproduces the lattice data reasonably well across a large range of temperatures . the final parameters are @xmath99 and @xmath100 . the potential we use is then given as : @xmath101 subtracting away the internal energy at @xmath102 is convenient when we use the bpl convention for the bound state projection operator @xmath103 . the results of the fit to the internal energy compared to the lattice data as well as the resulting potentials is shown in fig [ fig : potlat ] . any parametrization of the lattice results for the heavy quark potentials will be strongly dependent on temperature . so far , we have only considered creating effective potentials from temperature - independent potentials ( for example , the coulomb potential ) . starting from temperature - dependent potentials is straightforward . however , because there seems to be little discussion of this in the literature , we show how our results may be generalized to the temperature - dependent case . consider the following partial differential equation , @xmath104 this equation is identical to the bloch equation ( eq . [ eq : bloch ] ) but with the hamiltonian , @xmath105 , taken at the fixed temperature @xmath54 . because the potential in @xmath105 is at a fixed temperature , the solution to [ eq : blochtemp ] is the same as before , @xmath106 by setting @xmath107 we find that the right - hand side is exactly the density matrix . therefore the prescription for a temperature - dependent hamiltonian is simple : fix the potential at the temperature of interest for all @xmath108 , then solve for the density matrix exactly as we have done in the previous sections ( _ i.e. _ either using the bloch equation or numerically with pimc methods ) . finally set @xmath109 . this example shows how the analogy between the matrix elements of a density matrix , and imaginary- time propagators , breaks down when the potential has temperature dependence . using the lattice potential from the previous section the methods of section [ veff ] can be applied . do the previous results justify using the kelbg potential to approximate the effective @xmath9 potential ? well , since the divergent term in eq . [ eq : param ] is the screened coulomb term at small @xmath97 , we may work in units where @xmath110 . the question then is whether or not @xmath54 is sufficiently small in units of the bohr energy @xmath111 . the answer is yes , since in the extreme case where the bohr energy is the highest and the temperature is at the lowest ( near @xmath1 ) , we actually have @xmath112 , where our results for the coulomb case clearly show the kelbg potential working well down to @xmath113 . so we are confident that our methods will work well for the temperature range of interest . in fig . [ fig : bc ] we show the results for @xmath83 for charmonium and bottomonium ( dotted blue and green lines respectively ) at a temperature @xmath114 . as in the coulomb case the effective potential takes on a finite value at @xmath115 . . the dotted ( upper ) curves show @xmath87 and the solid ( lower ) curves show the result after the orthogonalization to the lowest bound state . ] it is also possible to include the effect of the lowest bound state as was done in the coulomb case . the resultant energy for the lowest bound state of charmonium , bottomonium and also a light quark bound state having reduced mass @xmath116 gev is shown in fig . [ fig : bound ] . also shown in fig . [ fig : bc ] as solid lines is the effective potential when the lowest bound state is included using the bpl formulation . at @xmath114 the effect of the bound state is to lower the potential by about 5% for charmonium and by about 15% for bottomonium . the effect is even smaller at higher temperatures . in order for the results given above to be useful in future simulations we now quote some useful results for kelbg - like screened coulomb and linear potentials . we note that equation [ eq : u1 ] can be integrated exactly in the case of a screened coulomb potential , @xmath117 with the result : @xmath118\ ] ] and for a linear potential @xmath119 : @xmath120\ ] ] which can be used as an approximation to the _ confinement _ term in the lattice potential at small distances . however , the inclusion of bound states which are most prominent at low temperature and higher masses must be calculated numerically from eq . [ eq : bound ] . we have used methods developed for treating effective potentials in electron - ion plasmas in order to generate potentials for quarkonium above deconfinement that properly take into account quantum effects and symmetry considerations . the nature of the potential changes at small distances ( @xmath121 ) and reaches a finite value at zero as is also seen in the coulomb case . these effective potentials should be used in classical transport simulations as well as in molecular dynamic simulations of the quark - gluon plasma when @xmath122 is sufficiently small . e. v. shuryak and i. zahed , phys . d * 70 * , 054507 ( 2004 ) [ arxiv : hep - ph/0403127 ] . a. mocsy and p. petreczky , phys . d * 73 * , 074007 ( 2006 ) [ arxiv : hep - ph/0512156 ] . f. karsch , m. t. mehr , and h. satz , _ z. phys . c. _ * 37 * , 617(1988 )	using methods previously developed by kelbg and others for creating effective potentials for electron - ion plasmas , we investigate quarkonium potentials above deconfinement . using results for the internal energy of a static quark - antiquark ( @xmath0 ) pair , a set of effective potentials are constructed that take into account quantum effects and symmetry conditions . bound state effects are explicitly included in order to account for the strongly coupled nature of the plasma . it is hoped that these effective potentials will be useful in simulations of heavy quarks or plasma quasiparticles when the dynamics is treated classically . = 1
plasma - based accelerators @xcite can produce extremely large accelerating gradients , enabling compact sources of high - energy beams . rapid experimental progress has occurred in the field of laser - driven plasma accelerators , and electron beams accelerated to multi - gev energies have been demonstrated@xcite using an intense laser driving a plasma wave in cm - scale plasmas . generation of these electron beams relied on self - injection from the background plasma in highly - nonlinear plasma waves @xcite . in the regime where background plasma electrons are self - trapped , experiments show that beams with sub - micron normalized transverse emittance can be produced @xcite . to reduce the electron beam emittance further it has been proposed to use a laser to ionize electrons at a trapped phase in a plasma wake that is independently excited by a particle beam @xcite or an intense laser @xcite . the laser - based method @xcite relies on two laser pulses of different colors : a long wavelength pulse , with large ponderomotive force and small peak electric field , excites a plasma wake without fully ionizing a high - z gas ; a short - wavelength injection pulse , with small ponderomotive force and large peak electric field , co - propagating and delayed with respect to the wake drive laser , ionizes a fraction of the remaining bound electrons at a trapping phase of the wake , generating an electron beam . figure [ fig:1d ] illustrates the basic principle of the two - color ionization injection method . this two - color , two - pulse ionization injection concept was first proposed in ref . . an additional numerical study was performed in ref . subsequent to refs . . in this paper we discuss the trapping condition , the distribution of ionized electrons , and the dynamics of the trapped electron beam . we present expressions for the transverse emittance , parallel and orthogonal to the laser polarization , of the trapped electron beam , valid for beam or laser wakefield drivers.@xcite for the all - optical , two - color ionization injection method , it is natural to consider a 10-@xmath0 m co@xmath1 laser as the long - wavelength drive pulse , generating the plasma wake , followed by a ti : al@xmath1o@xmath2 laser ( frequency - doubled , with 0.4 @xmath0 m wavelength ) for ionization injection . progress in co@xmath1 laser technology has opened the possibility of sub - ps pulse durations , that would enable efficient ( i.e. , resonant , with duration of order the plasma period ) plasma wakefield excitation at plasma densities @xmath3 @xmath4 , and such laser systems are expected to become available in the next several years @xcite . we present an example of two - color ionization injection using a short - pulse , co@xmath1 drive laser pulse and a frequency - doubled , ti : al@xmath1o@xmath2 injection laser pulse . [ cols="^ " , ] the two - color ionization injection concept relies on a large plasma wave ( or wakefield ) excited by a long - wavelength drive laser pulse , with wavelength @xmath5 and normalized vector potential amplitude @xmath6 , where @xmath7 and @xmath8 are the electron charge and mass , respectively , and @xmath9 is the speed of light . for efficient wakefield generation the drive pulse should be approximately resonant with the plasma , with @xmath10 , where @xmath11 is the drive laser length and @xmath12 is the plasma frequency , with @xmath13 the electron number density . co - propagating and delayed with respect to the drive pulse is an ionization laser pulse , with wavelength @xmath14 and normalized vector potential amplitude @xmath15 . the ionization pulse electric field @xmath16 is sufficiently large , @xmath17 , to ionize remaining bound electrons in the gas , that may be trapped in the wakefield . trapping requires that the electrons are ionized at the proper wake phase and that the wake amplitude is sufficiently large . behind the drive laser pulse , and assuming @xmath18 , this condition may be expressed as @xcite @xmath19 where @xmath20 is the phase position in the plasma wakefield of the ionized electron ( initially at rest ) , with @xmath21 is the lorentz factor of the plasma wave phase velocity , and @xmath22 is the potential of the wakefield ( normalized to @xmath23 ) , with @xmath24 in the minimum amplitude of the potential . here we have assumed the wakefield near the axis is approximately described by the 1d nonlinear wake potential @xcite , whose extrema satisfy , @xmath25 ^ 2 -1 \right\}^{1/2 } , \label{eq : phim}\ ] ] where @xmath26 is the peak of the accelerating wakefield normalized to @xmath27 . the optimal ionization phase @xmath28 for trapping is at the peak of the wake potential , @xmath29 and @xmath30 . at this phase location , the trapping condition is @xmath31 . in the limit @xmath32 , the minimum wakefield amplitude required for trapping is @xmath33 the longitudinal wakefield amplitude excited by a resonant , circularly - polarized , gaussian laser pulse is @xmath34 , such that the required laser amplitude for trapping is @xmath35 . figure [ fig:1d ] illustrates an example of the concept of two - color ionization injection . figure [ fig:1d](a ) shows the drive laser vector potential @xmath36 ( pulse centered at the origin ) , the field of the excited plasma wake @xmath26 ( red dashed curve ) , the ionization laser vector potential [ pulse centered at @xmath37 @xmath38 , and the trapping condition eq . , where @xmath39 indicates an electron ionized at rest will be trapped ( blue curve ) . figure [ fig:1d](b ) shows the electric field of the drive and ionization lasers ( normalized to the peak drive laser field ) , and the fraction of the ionized gas state @xmath40 ( red dashed curve ) . the normalized parameters shown in fig . [ fig:1d ] correspond to the case of a 10-@xmath0 m wavelength ( co@xmath1 ) drive laser , with @xmath41 ( linear polarization ) and 0.47 ps duration ( fwhm laser intensity ) , propagating in a uniform gas , producing a plasma with electron density @xmath42 @xmath4 , and resonantly exciting a wakefield . the ionization pulse , with @xmath43 ( linear polarization ) , @xmath44 @xmath0 m ( frequency - doubled ti : al@xmath1o@xmath2 laser ) , and 120 fs duration ( fwhm intensity ) , ionizes kr@xmath45 to kr@xmath46 ( ionization potential @xmath47 ev ) at a trapped wake phase . tunneling ionization will determine the distribution of electrons ionized , and , hence the trapped charge and emittance . the transverse phase space distribution of the ionized electrons was derived in ref . . the rms radius of the transverse distribution of ionized electrons is @xcite @xmath48 where @xmath49 is the ionization laser spot size [ i.e. , @xmath50 and @xmath51 where @xmath52 is the classical electron radius , @xmath53 is the fine structure constant , and @xmath54 is the potential of the state of the gas used for ionization injection , normalized to @xmath55 ev . the parameter @xmath56 is the normalized laser field amplitude @xmath57 , and @xmath58 is satisfied at the ionization threshold . ( here we assume the keldysh parameter @xmath59^{1/2}$ ] satisfies @xmath60 , such that tunneling ionization is the dominant ionization mechanism . ) for an ionization laser pulse located at the proper phase @xmath28 of a plasma wave of sufficient amplitude @xmath26 , given by eq . , electrons ionized at rest will be on trapped orbits , and the amount of trapped charge will be determined by the number of ionized electrons : @xmath61 , where @xmath40 is the ionization fraction , @xmath62 is the number of ions with electrons in the proper charge state , and @xmath63 is the interaction length . for interaction with a single species gas , @xmath64 , where @xmath13 is the electron density and @xmath65 is the number of electrons ionized by the drive laser ( or pre - ionized ) . if the length of the high - z gas region is longer than the ionization laser rayleigh range , then the total charge will be limited by the diffraction of the ionizing laser , @xmath66 , and , hence , we expect the charge to scale as @xmath67 with the interaction length limited by diffraction . as shown below , the emittance scales as @xmath68 , and , hence , there is a trade - off between increasing the trapped charge ( @xmath69 , limited by diffraction ) and reducing the emittance . the achievable beam emittance using ionization injection into a wakefield , driven by a charged particle beam or an intense laser , is determined by the initial phase space distribution of the trapped electrons and the plasma focusing forces.@xcite . the rms of the momentum distribution of the ionized electrons , in the plane of laser polarization , is approximately @xcite @xmath70 where @xmath71 is given by eq . . for a small ionization injection laser amplitude @xmath72 , the momentum gain from the ponderomotive force of the ionizing laser may be neglected . using eqs . and , the initial thermal emittance of the beam generated by ionization injection is @xcite @xmath73 equation is independent of the driver of the plasma accelerator ( e.g. , laser or particle beam ) and is determined from the ionization physics , i.e. , the distribution of electrons produced via tunneling ionization . unless injected matched , the emittance of the trapped beam will grow due to phase mixing in the plasma wakefield . an electron trapped in the plasma wakefield , i.e. , ionized at the proper phase to satisfy the trapping condition eq . , will begin to accelerate and rotate in transverse phase space . for an electron with initial transverse position @xmath74 and normalized momentum @xmath75 , the transverse position and momentum will evolve as @xmath76 = \gamma^{-1/4 } r_{\beta0 } \cos ( k_\beta z+ \varphi ) , \\ u_x & = p_x / m_ec \simeq \gamma^{1/4 } \left [ u_0 \cos ( k_\beta z ) - x_0 k_{\beta 0 } \sin ( k_\beta z)\right ] = - \gamma^{1/4 } k_{\beta 0 } r_{\beta0 } \sin ( k_\beta z + \varphi ) , \end{aligned}\ ] ] where @xmath77^{1/2}$ ] is the initial betatron amplitude , the energy @xmath78 is increasing as the electron is accelerated in the wakefield , @xmath79 is the betatron wavenumber in the wakefield , and @xmath80 is the initial betatron wavenumber . for laser drivers with parameters satisfying @xmath81 ( assuming circular polarization ) , or particle beam drivers , the driver will produce a co - moving ion cavity with @xmath82 . averaging the betatron amplitude over the initial beam distribution yields @xcite @xmath83 { w_i^2}\delta^2/2 \label{eq : r_beta}\ ] ] in the plane of ionization laser polarization , and @xmath84 orthogonal to the plane of laser polarization . if the ionization injection region is short ( due to a finite high - z gas region or short ionization injection pulse rayleigh range ) , followed by a post acceleration region that is sufficiently long to allow betatron phase mixing , then the normalized emittance will saturate to @xmath85^{1/2 } = k_{\beta 0 } \langle r_{\beta 0}^2 \rangle /2 . \label{eq : sat}\ ] ] the initial beam phase - space distribution eq . will determine the final saturated emittance eq . . matched injection in the plane of laser polarization , where the thermal emittance eq . equals the saturated emittance eq . , occurs for laser - plasma parameters such that @xmath86 . if continuous ionization injection occurs over the length of the plasma accelerator , the saturated emittance will asymptote to @xmath87 . c inferno-2-color-3.eps ( 81,43)@xmath88[nm ] ( 81,22 ) in this section we present an example of beam generation using two - color ionization injection with a short - pulse co@xmath1 laser driving the wakefield and a frequency - doubled ti : al@xmath1o@xmath2 ionizing laser . modeling was done using the inf&rno @xcite and warp @xcite pic codes of blast ( berkeley lab accelerator simulation toolkit ) @xcite . figure [ fig : co2 ] shows the evolution of the transverse normalized emittance of the trapped electron beam versus propagation distance @xmath89 . in fig . [ fig : co2 ] , a 10 @xmath0m - wavelength , 10 j , co@xmath1 laser with @xmath90 ( linear polarization ) , 0.47 ps duration ( fwhm laser intensity ) , and 155 @xmath0 m spot size ( @xmath91 mm ) propagates in krypton gas , ionizing the gas up to kr@xmath45 , producing a plasma with electron density 10@xmath92 @xmath93 ( gas density of @xmath94 @xmath93 ) and driving a wake . a frequency - doubled ( 0.4 @xmath0 m wavelength ) ti : al@xmath1o@xmath2 laser with @xmath95 ( linear polarization ) , 120 fs duration ( fwhm laser intensity ) , @xmath96 m spot size ( @xmath97 mm ) , delayed with respect to the co@xmath1 laser , ionizes kr@xmath45 to kr@xmath46 ( @xmath98 ev ) and generates a trapped electron beam . the solid black and red solid curves are the normalized transverse emittance of the trapped electron beam , parallel and orthogonal to the laser polarization plane , respectively , calculated using the inf&rno module of blast . for a small ionization injection laser amplitude @xmath72 , propagating in a uniform gas jet , the final normalized transverse emittances @xcite are , using eqs . , , and , @xmath99 \frac{a_i}{\lambda_i } \left ( \frac{3 \pi r_e}{4 \alpha^4}\right ) \left ( \frac{u_h}{u_i}\right)^{3/2 } , \label{eq : e - para } \\ \epsilon_\perp & = k_{\beta 0 } w_i^2 \frac{a_i}{\lambda_i } \left ( \frac{3 \pi r_e}{4 \alpha^4}\right ) \left ( \frac{u_h}{u_i}\right)^{3/2 } , \label{eq : e - perp}\end{aligned}\ ] ] where @xmath88 ( @xmath100 ) is the transverse normalized emittance parallel ( orthogonal ) to the plane of ionization laser polarization . shown in fig . [ fig : co2 ] are the theoretical predictions : ( dashed black line ) @xmath88 emittance in the laser polarization plane eq . ( [ eq : e - para ] ) and ( dashed red line ) @xmath100 emittance orthogonal to the laser polarization plane eq . ( [ eq : e - perp ] ) . for these drive laser - plasma parameters , @xmath101 . good agreement is achieved between the pic simulation and the theoretical estimates eqs . ( [ eq : e - para ] ) and ( [ eq : e - perp ] ) . phase mixing is not complete after 5 mm , and the emittance orthogonal to the plane of laser polarization is still varying . the charge contained in the beam modeled with the pic code is 27 pc . this approximately agrees with the order - of - magnitude estimate of the trapped charge over a rayleigh length , assuming @xmath102 , @xmath103 which , for the parameters of fig . [ fig : co2 ] , yields @xmath104 pc . in general , a trade - off exists between the reducing the emittance ( @xmath68 ) and increasing the trapped charge ( @xmath105 , with interaction length limited by diffraction ) . figure [ fig : rs ] shows the theoretical estimates for the beam emittance [ parallel , eq . ( dashed black curve ) , and orthogonal , eq . ( dashed red curve ) , to the ionization laser polarization ] and the charge , eq . ( dashed blue curve ) versus ionization laser spot size ( all other parameters are the same as fig . [ fig : co2 ] ) . the points in fig . [ fig : rs ] are pic simulation results ( calculated using the inf&rno module of blast ) for the emittance in the laser polarization plane @xmath106 $ ] ( black ) and orthogonal to the laser polarization plane @xmath100 [ ] ( red ) , and the charge @xmath107 $ ] ( blue ) . c rs.eps ( 20,30)@xmath88[nm ] ( 20,18 ) ( 70,23 ) ultra - low emittance electron beams , on the order of tens of nm , can be generated in a plasma accelerator using ionization injection in a wakefield @xcite . in this paper we have reviewed the trapping physics , distribution of ionized electrons , and the trapped beam dynamics . expressions for the achievable beam emittance using ionization injection into a plasma wakefield were presented.@xcite note that these results are independent of the wakefield driver ( laser @xcite or particle beam @xcite ) . an experimentally - relevant example was presented of ultra - low emittance beam generation by two - color ionization injection using a short - pulse , co@xmath1 laser driving the wakefield and a frequency - doubled , ti : al@xmath1o@xmath2 laser ionizing a krypton gas . in this example the interaction region was limited by the rayleigh range of the ionizing laser . in general , a trade - off exists between the reducing the saturated emittance ( @xmath68 ) and increasing the trapped charge ( @xmath69 , with interaction length limited by diffraction ) . self - consistent numerical modeling of the beam injection for this example was performed and compared to analytic estimates , showing good agreement . it should be noted that this example is not fully optimized , and , depending on the laser - plasma parameters , other gases may be considered ( e.g. , ne ) for improved performance . this work was supported by the director , office of science , office of high energy physics , of the u.s . department of energy under contract nos . de - ac02 - 05ch11231 . this research used computational resources of the national energy research scientific computing center , which is supported by the office of science of the u.s . department of energy under contract no . de - ac02 - 05ch11231 . this work was supported in part by the national basic research program of china ( grant no . 2013cba01504 ) and the national natural science foundation of china ( grant no . 11405107 ) . x. wang , r. zgadzaj , n. fazel , z. li , s. a. yi , x. zhang , w. henderson , y .- y . chang , r. korzekwa , h .- e . tsai , c .- h . pai , h. quevedo , g. dyer , e. gaul , m. martinez , a. c. bernstein , t. borger , m. spinks , m. donovan , v. khudik , g. shvets , t. ditmire , and m. c. downer , `` quasi - monoenergetic laser - plasma acceleration of electrons to 2 gev , '' _ nature communications _ * 4 * , p. 1988 , june 2013 . w. p. leemans , a. j. gonsalves , h .- s . mao , k. nakamura , c. benedetti , c. b. schroeder , c. tth , j. daniels , d. e. mittelberger , s. s. bulanov , j .- l . vay , c. g. r. geddes , and e. esarey , `` multi - gev electron beams from capillary - discharge - guided subpetawatt laser pulses in the self - trapping regime , '' _ phys . _ * 113 * , p. 245002 , december 2014 . c. benedetti , c. b. schroeder , e. esarey , f. rossi , and w. p. leemans , `` numerical investigation of electron self - injection in the nonlinear bubble regime , '' _ phys . plasmas _ * 20 * , p. 103108 , october 2013 . g. r. plateau , c. g. r. geddes , d. b. thorn , m. chen , c. benedetti , e. esarey , a. j. gonsalves , n. h. matlis , k. nakamura , c. b. schroeder , s. shiraishi , t. sokollik , j. van tilborg , cs . tth , s. trotsenko , t. s. kim , m. battaglia , t. sthlker , and w. p. leemans , `` ultra - low - emittance electron bunches from a laser - plasma accelerator measured using single - shot x - ray spectroscopy , '' _ phys . lett . _ * 109 * , p. 064802 , august 2012 . r. weingartner , s. raith , a. popp , s. chou , j. wenz , k. khrennikov , m. heigoldt , a. r. maier , n. kajumba , m. fuchs , b. zeitler , f. krausz , s. karsch , and f. grner , `` ultralow emittance electron beams from a laser - wakefield accelerator , '' _ phys . rev . st accel . beams _ * 15 * , p. 111302 , november 2012 . b. hidding , g. pretzler , j. b. rosenzweig , t. knigstein , d. schiller , and d. l. bruhwiler , `` ultracold electron bunch generation via plasma photocathode emission and acceleration in a beam - driven plasma blowout , '' _ phys . _ * 108 * , p. 035001 , january 2012 . yu , e. esarey , j .- l . vay , c. b. schroeder , c. benedetti , c. g. r. geddes , s. g. rykovanov , s. s. bulanov , m. chen , and w. p. leemans , `` low transverse emittance electron bunches from two - color laser - ionization injection , '' in _ proc . spie _ , * 8779 * , p. 877908 , spie digital library , may 2013 . x. l. xu , y. p. wu , c. j. zhang , f. li , y. wan , j. f. hua , c .- h . pai , w. lu , p. yu , c. joshi , and w. b. mori , `` low emittance electron beam generation from a laser wakefield accelerator using two laser pulses with different wavelengths , '' _ phys . st accel . beams _ * 17 * , p. 061301 , june 2014 . c. b. schroeder , j .- l . vay , e. esarey , s. s. bulanov , c. benedetti , l .- l . yu , m. chen , c. g. r. geddes , and w. p. leemans , `` thermal emittance from ionization - induced trapping in plasma accelerators , '' _ phys . rev . beams _ * 17 * , p. 101301 , october 2014 . c. benedetti , c. b. schroeder , e. esarey , c. g. r. geddes , and w. p. leemans , `` efficient modeling of laser - plasma accelerators with inf&rno , '' in _ proc . of 2010 advanced accelerator concepts workshop _ , g. nusinovich and s. gold , eds . , * 1299 * , pp . 250255 , aip , ( ny ) , 2010 .	ultra - low emittance ( tens of nm ) beams can be generated in a plasma accelerator using ionization injection of electrons into a wakefield . an all - optical method of beam generation uses two laser pulses of different colors . a long - wavelength drive laser pulse ( with a large ponderomotive force and small peak electric field ) is used to excite a large wakefield without fully ionizing a gas , and a short - wavelength injection laser pulse ( with a small ponderomotive force and large peak electric field ) , co - propagating and delayed with respect to the pump laser , to ionize a fraction of the remaining bound electrons at a trapped wake phase , generating an electron beam that is accelerated in the wake . the trapping condition , the ionized electron distribution , and the trapped bunch dynamics are discussed . expressions for the beam transverse emittance , parallel and orthogonal to the ionization laser polarization , are presented . an example is shown using a 10-@xmath0 m co@xmath1 laser to drive the wake and a frequency - doubled ti : al@xmath1o@xmath2 laser for ionization injection .
we investigate the shrinkage properties of the partial least squares ( pls ) regression estimator . it is known ( e.g. @xcite ) that we can express the pls estimator obtained after @xmath2 steps in the following way : @xmath3 where @xmath4 is the component of the ordinary least squares ( ols ) estimator along the @xmath5th principal component of the covariance matrix @xmath6 and @xmath7 is the corresponding eigenvalue . the quantities @xmath8 are called shrinkage factors . we show that these factors are determined by a tridiagonal matrix ( which depends on the input output matrix @xmath9 ) and can be calculated in a recursive way . combining the results of @xcite and @xcite , we give a simpler and clearer proof of the shape of the shrinkage factors of pls and derive some of their properties . in particular , we show that some of the values @xmath8 are greater than @xmath1 ( this was first proved in @xcite ) . + we argue that these `` peculiar shrinkage properties '' @xcite do not necessarily imply that the mean squared error ( mse ) of the pls estimator is worse compared to the mse of the ols estimator : in the case of deterministic shrinkage factors , i.e. factors that do not depend on the output @xmath10 , any value @xmath11 is of course undesirable . but in the case of pls , the shrinkage factors are stochastic they also depend on @xmath12 . even if latexmath:[$p\left ( \left that the mse is worse than the mse of the ols estimator . in particular , bounding the absolute value of the shrinkage factor by @xmath1 does not automatically yield a lower mse , in disagreement to what was conjectured in e.g. @xcite . + having issued this warning , we explore whether bounding the shrinkage factors leads to a lower mse or not . it is very difficult to derive theoretical results , as the quantities of interest - @xmath14 and @xmath8 respectively - depend on @xmath10 in a complicated , nonlinear way . as a substitute , we study the problem on several artificial data sets and one real world example . it turns out that in most cases the mse of the bounded version of pls is indeed smaller than the one of pls , although the improvement is tiny . + the paper is organized as follows : in section [ pre ] we introduce the notation and in section [ kspaces ] we recall some propertie of krylov spaces . in section [ pls ] we define the pls estimator and in section [ sectiontri ] we provide mathematical results that are needed in the rest of the paper . after explaining the notion of shrinkage in section [ shrinkage ] we derive the formulas for the pls shrinkage factors in section [ shrinkpls ] and derive some of their properties . in section [ simulation ] , we report on the results of the experiments . the paper ends with a conclusion . we consider the multivariate linear regression model @xmath15 with @xmath16 the numbers of variables is @xmath17 , the number of examples is @xmath18 . for simplicity , we assume that @xmath19 and @xmath10 are scaled to have zero mean , so we do not have to worry about intercepts . we have @xmath20 we set @xmath21 . the singular value decomposition of @xmath19 is of the form @xmath22 with @xmath23 we have @xmath24 and @xmath25 . + set @xmath26 . the eigendecomposition of @xmath27 is @xmath28 the eigenvalues @xmath7 of @xmath27 ( and any other matrix ) are ordered in the following way : @xmath29 the moore - penrose inverse of a matrix @xmath30 is denoted by @xmath31 . + the ordinary least squares ( ols ) estimator @xmath32 is the solution of the optimization problem @xmath33 set @xmath34 the ols estimator is given by the formula @xmath35 set @xmath36 finally , we need a result on the shape of the moore - penrose inverse of a symmetric matrix . [ penrose ] let @xmath37 be a symmetric matrix with eigendecomposition @xmath38 with eigenvalues @xmath7 . set @xmath39 as @xmath40 we can write @xmath41 then @xmath42 the four properties that we have to check are 1 . @xmath43 , 2 . @xmath44 , 3 . @xmath45 , 4 . @xmath46 . as @xmath47 is symmetric , the polynomial @xmath48 is symmetric as well , which proves the first two conditions . next note that it suffices to prove the and properties @xmath49 and @xmath50 for the diagonal matrix @xmath51 with @xmath52 . this is true as @xmath53 we have @xmath54 the third property of the moore - penrose inverse is @xmath55 which is equivalent to @xmath56 which is obviously true . the fourth property follows as easily . the degree of the polynomial @xmath57 is @xmath58 . the proposition is valid no matter if we count the non - zero eigenvalues with or without multiplicities . we count the eigenvalues with multiplicities in order to connect the polynomial to the characteristical polynomial in the regular case : if @xmath47 is a regular matrix , @xmath57 is linked to the characterictical polynomial @xmath59 in the following way : @xmath60 set @xmath61 the columns of @xmath62 are called the krylov sequence of @xmath27 and @xmath63 . + the space spanned by the columns of @xmath62 is called the krylov space of @xmath27 and @xmath63 and denoted by @xmath64 . we recall some basic facts on the dimension of the krylov space that are needed in the rest of the paper . set @xmath65 ( the vector @xmath66 is defined in ( [ te ] ) ) and @xmath67 we have @xmath68 suppose that @xmath69 for some @xmath70 . using the eigendecompostion of @xmath27 this equation is equivalent to @xmath71 as @xmath72 is an invertible matrix , this is equivalent to @xmath73 for @xmath74 . hence , each element @xmath75 is a zero of the polynomial @xmath76 this is a polynomial of degree @xmath77 . as it has @xmath78 different zeroes , it must be trivial , i.e. @xmath79 . if @xmath80 we have @xmath81 . it is clear that @xmath82 as @xmath83 . assume that there is a set @xmath84 of @xmath85 linear independent vectors in the krylov sequence @xmath62 . set @xmath86 hence @xmath87 . the condition that @xmath84 is linear independent is equivalent to the following : there is no nontrivial polynomial @xmath88 such that @xmath89 for @xmath75 . as the polynomial @xmath90 is of degree @xmath91 and @xmath92 , there is always a nontrivial solution of equation ( [ dep ] ) . we sum up the two results : [ dimkm ] we have @xmath93 in particular @xmath94 it is not our aim to give an introduction to the partial least squares ( pls ) method and refer to @xcite . we take a purely algebraic point of view as in @xcite . the @xmath95 estimator @xmath96 is the solution of the constrained minimization problem @xmath97 we call @xmath2 the number of steps of pls . it follows that any solution of this problem is of the form @xmath98 where @xmath99 is the solution of the unconstrained problem @xmath100 plugging this into the formula for the ols estimator ( cf . section [ pre ] ) we get the pls estimator obtained after @xmath2 steps can be expressed in the following way : @xmath101^{- } \left(k^{(m)}\right)^t b\,.\end{aligned}\ ] ] it should be clear that we can replace the matrix @xmath62 in equation ( [ bhat ] ) by any matrix @xmath102 , as long as its columns span the space @xmath64 . in fact , in the nipals algorithm ( see @xcite ) , an orthogonal basis of @xmath64 is calculated with the help of the gram - schmidt procedure . denote by @xmath103 this orthogonal basis of @xmath64 . of course , this basis only exists if @xmath104 , which might not be true for all @xmath105 . the maximal number for which this holds is @xmath106 ( see proposition [ dimkm ] ) . note however that @xmath107 ( see ( [ chainkm ] ) ) and the solution of the optimization problem does not change anymore . hence for the rest of the paper , we make the assumption that @xmath108 we have @xmath109 we show that @xmath110 . by definition @xmath111 with @xmath112 ( recall that @xmath113 is the rank of @xmath27 ) . on the other hand , any vector @xmath114 lies in @xmath115 if and only if there is a polynomial @xmath90 of degree @xmath116 such that @xmath117 it follows that @xmath118 . as @xmath119 we have @xmath120 . set @xmath121 where @xmath102 is as defined in equation ( [ gram ] ) . [ tri ] the matrix @xmath122 is symmetric and positive semidefinite . furthermore @xmath122 is tridiagonal , i.e @xmath123 for @xmath124 . the first two statements are obvious . let @xmath125 . as @xmath126 , the vector @xmath127 lies in the subspace @xmath128 . as @xmath129 , the vector @xmath130 is orthogonal on @xmath128 , in other words @xmath131 as @xmath122 is symmetric , we also have @xmath123 which proves the assertion . we will see in section [ shrinkpls ] that the matrices @xmath122 and their eigenvalues determine the shrinkage factors of the pls estimator . to prove this , we list some properties of @xmath122 in teh following sections . a symmetric tridiagonal matrix @xmath132 is called unreduced if all subdiagonal entries are non - zero , i.e @xmath133 for all @xmath5 . all eigenvalues of an unreduced matrix are distinct . set @xmath134 [ unreduced ] if @xmath135 , the matrix @xmath122 is unreduced . more precisely @xmath136 for all @xmath137 . set @xmath138 and denote by @xmath139 the basis obtained by gram - schmidt . its existence is guaranteed as we assume that @xmath135 . for simplicity of notation , we assume that the vectors @xmath140 are not normalized to have length 1 . by definition @xmath141 as the vectors @xmath140 are pairwisse orthogonal , it follows that @xmath142 we conclude that @xmath143 note that the matrix @xmath144 is obtained from @xmath122 by deleting the last column and row of @xmath122 . it follows that we can give a recursive formula for the characteristical polynomials @xmath145 of @xmath122 . we have @xmath146 and @xmath147 . + we want to deduce properties of the eigenvalues of @xmath122 and @xmath27 and explore their relationship . denote the eigenvalues of @xmath122 by @xmath148 all eigenvalues of @xmath149 are eigenvalues of @xmath27 . first note that @xmath150 as the columns of the matrix @xmath151 form an orthonormal basis of @xmath152 , @xmath153 is the matrix that represents @xmath154 with repect to this basis . as any eigenvalue of @xmath155 is obviously an eigenvalue of @xmath27 , the proof is complete the following theorem is a special form of the cauchy interlace theorem . in this version , we use a general result from @xcite and exploit the tridiagonal structure of @xmath122 . [ ci2 ] each interval @xmath156\end{aligned}\ ] ] @xmath157 contains a different eigenvalue of @xmath158 ( @xmath159 ) . in addition , there is a different eigenvalue of @xmath160 outside the open interval @xmath161 . this theorems ensures in particular that there is a different eigenvalue of @xmath27 in the interval @xmath162 $ ] . theorem [ ci2 ] holds independently of assumption ( [ ass ] ) . by definition , for @xmath163 @xmath164 here @xmath165 , so @xmath166 an application of theorem 10.4.1 in @xcite gives the desired result . [ distinct ] if @xmath122 is unreduced , the eigenvalues of @xmath122 and the eigenvalues of @xmath144 are distinct . suppose the two matrices have a common eigenvalue @xmath167 . it follows from ( [ charpol ] ) and the fact that @xmath122 is unreduced that @xmath167 is an eigenvalue of @xmath168 . repeating this , we deduce that @xmath169 is an eigenvalue of @xmath170 , a contradiction , as @xmath171 [ not1 ] in general it is not true that @xmath122 and a submatrix @xmath172 have distinct eigenvalues . consider the case where @xmath173 for all @xmath5 . using equation ( [ charpol ] ) we conclude that @xmath174 is an eigenvalue for all submatrices with @xmath2 odd . if @xmath135 , we have @xmath175 . @xmath122 is positive semidefinite , hence all eigenvalues of @xmath122 are @xmath176 . in other words , @xmath177 if and only if its smallest eigenvalue @xmath178 is @xmath179 . using theorem [ ci2 ] we have @xmath180 as @xmath135 , the matrix @xmath122 is unreduced , which implies that @xmath122 and @xmath144 have no common eigenvalues ( see [ distinct ] ) . we can therefore replace the first @xmath181 by @xmath182 , i.e. the smallest eigenvalue of @xmath144 is @xmath179 . in general , it is not true that @xmath183 . an easy example is @xmath184 we have @xmath185 i.e. @xmath186 . on the other hand @xmath187 it is well known that the matrices @xmath122 are closely related to the so - called rayleigh - ritz procedure , a method that is used to approximate eigenvalues . for details consult e.g. @xcite . we have presented two estimators for the regression parameter @xmath188 ols and pls which also define estimators for @xmath189 via @xmath190 one possibility to evaluate the quality of an estimator is to determine its mean squared error ( mse ) . in general , the mse of an estimator @xmath191 for a vector - valued parameter @xmath192 is defined as @xmath193 \\ & = & e\left [ \left(\hat\theta -\theta\right)^t\left(\hat \theta -\theta\right ) \right]\\ & = & \left(e\left [ \hat \theta \right ] -\theta \right)^t\left(e\left [ \hat \theta \right ] -\theta \right)+ e\left [ \left(\hat \theta^t - e\left[\hat\theta\right]\right)^t \left(\hat \theta^t - e\left[\hat\theta\right]\right ) \right]\,.\end{aligned}\ ] ] this is the well - known bias - variance decomposition of the mse . the first part is the squared bias and the second part is the variance term . we start by investigating the class of linear estimators , i.e. estimators that are of the form @xmath194 for some matrix @xmath84 that does not depend on @xmath10 . the ols estimators are linear : @xmath195 @xmath196 is the projection @xmath197 onto the space that is spanned by the columns of @xmath19 . + recall the regression model ( [ linreg ] ) . let @xmath198 be a linear estimator . we have @xmath199&= & sx\beta \\ \text{var}\left[\hat \theta \right]&=&\sigma^2 \text{tr}\left(ss^t\right)\,.\end{aligned}\ ] ] the estimator @xmath200 is unbiased as @xmath201 & = & s_2 x\beta \\ & = & p_{l(x ) } x\beta\\ & = & x\beta\,.\end{aligned}\ ] ] the estimator @xmath32 is only unbiased if @xmath202 : @xmath203&=&e\left[\left(x^tx\right)^- x^t y\right]\\ & = & \left(x^tx\right)^- x^t e\left[y\right]\\ & = & \left(x^tx\right)^- x^t x\beta \\ & = & \beta\,.\end{aligned}\ ] ] let us now have a closer look at the variance term . for @xmath32 we have @xmath204 hence @xmath205 next note that @xmath196 is the operator that projects on the space spanned by the columns of @xmath19 . it follows that @xmath206 and that @xmath207 we conclude that the mse of the estimator @xmath32 depends on the eigenvalues @xmath208 of @xmath209 . small eigenvalues of @xmath27 correspond to directions in @xmath19 that have very low variance . equation ( [ varbeta ] ) shows that if some eigenvalues are small , the variance of @xmath32 is very high , which leads to a high mse . + one possibility to ( hopefully ) decrease the mse is to modify the ols estimator by shrinking the directions of the ols estimator that are responsible for a high variance . this of course introduces bias . we shrink the ols estimator in the hope that the increase in bias is small compared to the decrease in variance . + in general , a shrinkage estimator for @xmath188 is of the form @xmath210 where @xmath211 is some real - valued function . the values @xmath212 are called shrinkage factors . examples are * principal component regression @xmath213 and * ridge regression @xmath214 where @xmath215 is the ridge parameter . we will see in section [ shrinkpls ] that pls is a shrinkage estimator as well . it will turn out that the shrinkage behavior of pls regression is rather complicated . + let us investigate in which way the mse of the estimator is influenced by the shrinkage factors . if the shrinkage estimators are linear , i.e. the shrinkage factors do not depend on @xmath10 , this is an easy task . let us first write the shrinkage estimator in matrix notation . we have @xmath216 the diagonal matrix @xmath217 has entries @xmath212 . the shrinkage estimator for @xmath10 is @xmath218 we calculate the variance of these estimators . @xmath219 and @xmath220 next , we calculate the bias of the two shrinkage estimators . we have @xmath221&= & s_{shr,1 } x\beta\\ & = & u\sigma d_{shr } \sigma^-u^t \beta \,.\end{aligned}\ ] ] it follows that @xmath222 -\beta \right)^t\left(e\left[s_{shr,1 } y\right ] -\beta \right)\\ & = & \left(u^t \beta\right)^t \left(\sigma d_f \sigma^- -\text{id}\right)^t \left(\sigma d_f \sigma^- -\text{id}\right)\left(u^t \beta\right)\\ & = & \sum_{i=1 } ^{p^ * } \left(f(\lambda_i ) -1\right)^2 \left(u_i ^t \beta \right)^2\,.\end{aligned}\ ] ] replacing @xmath223 by @xmath224 it is as easy to show that @xmath225 for the shrinkge estimator @xmath226 and @xmath227 defined above we have @xmath228 if the shrinkage factors are deterministic , i.e. they do not depend on @xmath10 , any value @xmath229 increases the bias . values @xmath230 decrease the variance , whereas values @xmath231 increase the variance . hence an absolute value @xmath0 is always undesirable . the situation is completely different for stochastic shrinkage factors . we will discuss this in the following section . + note that there is a different notion of shrinkage , namely that the @xmath232- norm of an estimator is smaller than the @xmath232-norm of the ols estimator . why is this a desirable property ? let us again consider the case of linear estimators . set @xmath233 for @xmath234 . we have @xmath235 the property that for all @xmath236 @xmath237 is equivalent to the condition that @xmath238 is negative semidefinite . the trace of negative semidefinite matrices is @xmath239 . furthermore @xmath240 , so we conclude that @xmath241 it is known ( see @xcite ) that @xmath242 in this section , we give a simpler and clearer proof of the shape of the shrinkage factors of pls . basically , we combine the results of @xcite and @xcite . it turns out that some of the factors @xmath8 are greater than 1 . we try to explain why these `` peculiar shrinkage properties '' do not necessarily imply that the mse of the pls estimator is increased . + denote by @xmath243 the polynomial associated to @xmath122 that was defined in proposition [ penrose ] , i.e. @xmath244 recall that the eigenvalues of @xmath122 are denoted by @xmath245 . it follows that @xmath246 by definition of pls , @xmath247 hence there is a polynomial @xmath248 of degree @xmath249 with @xmath250 . suppose that @xmath135 . we have @xmath251 by proposition [ penrose ] , @xmath252 we plug this into equation ( [ bhat ] ) and obtain @xmath253 recall that the columns of @xmath102 form an orthonormal basis of @xmath254 . it follows that @xmath255 is the operator that projects on the space @xmath254 . in particular @xmath256 for @xmath257 . this implies that @xmath258 suppose that @xmath135 . if we denote by @xmath4 the component of @xmath259 along the @xmath5th eigenvector of @xmath27 then @xmath260 where @xmath261 is the polynomial defined in ( [ tm ] ) . ( @xcite ) this follows immediately from the proposition above . we have @xmath262 we now show that some of the shrinkage factors of pls are @xmath263 . for each @xmath264 , we can decompose the interval @xmath265 $ ] into @xmath266 disjoint intervals if @xmath267 . ] @xmath268 such that @xmath269 set @xmath270 . it follows from equation ( [ tm ] ) that the zero s of @xmath271 are @xmath272 . as @xmath122 is unreduced , all eigenvalues are distinct . set @xmath273 and @xmath274 . define @xmath275\mu_{i } ^{(m)},\mu_{i+1 } ^{(m)}[$ ] for @xmath276 . by definition , hence @xmath271 is non - negative on the intervals @xmath278 if @xmath279 is odd and @xmath271 is non - positive on the intervals @xmath278 if @xmath279 is even . it follows from theorem [ ci2 ] that all interval @xmath278 contain at least one eigenvalue @xmath7 of @xmath280 . in general it is not true that @xmath281 for all @xmath7 and @xmath282 . using the example in remark [ not1 ] and the fact that @xmath283 is equivalent to the condition that @xmath7 is an eigenvalue of @xmath122 , it is easy to construct a counterexample . using some of the results of section [ sectiontri ] , we can however deduce that some factors are indeed @xmath284 . as all eigenvalues of @xmath285 and @xmath286 are distinct ( c.f . proposition [ distinct ] ) , we see that @xmath287 for all @xmath5 . in particular @xmath288 more generally , using proposition [ distinct ] , we conclude that @xmath289 and @xmath290 is not possible . in practice i.e. calculated on a data set the factors seem to be @xmath284 all of the time . + furthermore @xmath291 to proove this , we set @xmath270 . we have by definition @xmath277 . furthermore , the smallest positive zero of @xmath271 is @xmath245 and it follows from theorem [ ci2 ] and proposition [ distinct ] that @xmath292 . hence @xmath2930,1]$ ] . + using theorem [ ci2 ] , more precisely @xmath294 it is possible to bound the terms @xmath295 from this we can derive bounds on the shrinkage factors . we will not pursue this further , readers who are interested in the bounds should consult @xcite . instead , we have a closer look at the mse of the pls estimator . + in section [ shrinkage ] we showed that a value @xmath296 is not desirable , as the variance of the estimator increases . note however , that in the case of pls , the factors @xmath8 are stochastic ; they depend on @xmath10 - in a nonlinear way . for @xmath297 we have the following situation : if we set @xmath298 and @xmath299 , we have to compare @xmath300 note that the rhs is not necessarily smaller than the lhs , even if @xmath301 . an easy counterexample is @xmath302 the lhs is @xmath303 . + among others , @xcite proposed to bound the shrinkage factors of the pls estimator in the following way . set @xmath304 and define a new estimator : @xmath305 if the shrinkage factors are numbers , this will improve the mse ( cf . section [ shrinkage ] ) . but in the case of stochastic shrinkage factors , the situation is completely unclear . consider again the example @xmath306 . set @xmath307 in this case @xmath308 so it is not clear whether the modified estimator bound leads to a lower mse , which was conjectured in e.g. @xcite . + the above example ( involving @xmath309 and @xmath310 ) is of course purely artificial . it is not clear whether the shrinkage factors behave this way . it is hard if not infeasable to derive statistical properties of the pls estimator or its shrinkage factors , as they depend on @xmath10 in a complicated , nonlinear way . as an alternative , we compare the two different estimators on different data . in this section , we explore the difference between the methods pls and bound . we investigate three artificial datasets and one real world example . in all examples , we rescale @xmath19 and @xmath10 to have zero mean and unit variance . + let us start with the artificial datasets . of course , artificial datasets do not reflect many real world situations , but we have the advantage that we know the true regression coefficient @xmath188 and that we have an unlimited amount of examples at hand . we can estimate the mse of any of the four estimators : for @xmath311 we generate a sample @xmath10 and calculate the estimator @xmath312 . we define @xmath313 for all examples , we choose @xmath314 . in our first example we generate @xmath315 examples in the following way : the input data is the realistion of a @xmath316 dimensional normally distributed variable with expectation @xmath317 and covariance matrix @xmath318 defined as @xmath319 the regression coefficient @xmath188 is the random permutation of @xmath320 with @xmath321 . + next we determine the variance of the error term . we do this by considering several signal - to - noise - ratios ( stnr ) . this quantity is defined as @xmath322 we set @xmath323 and determine the corresponding value of @xmath324 . we generate @xmath325 samples @xmath10 and calculate the four estimators . + the following figures show the estimated mse for @xmath188 and @xmath189 respectively . the solid lines with the @xmath326 s correspond to pls . the lines with the @xmath327 s correspond to bound . + we see that bound is better in all cases , although the improvement is not dramatic . we should remark that both method pick the same ( optimal ) number of steps most of the times . the difference between the two methods is especially tiny ( but non - zero ) in the first step . we do not have an explanation for this phenomenon . the mse is the same for the last step @xmath328 as in this case @xmath329 in this example , we generate @xmath330 examples . the input data is the realisation of a @xmath331 dimensional random variable with distribution @xmath332 . the covariance matrix is defined as in the first example ( with @xmath316 replaced by @xmath331 ) . again , the coefficients of @xmath188 are a random permutation @xmath333 with @xmath334 . we consider the signal - to - noise - ratios @xmath335 . + the results are qualitatively the same as those from the first example . bound is better all of the times , the optimal number of steps are the same for both methods . the input data is generated as in the second example , in particular , we have @xmath336 . this time , we only generate @xmath337 examples . the coefficients of the regression vector @xmath188 are realizations of a @xmath338 distibuted random variable . we investigate the signal - to - noise - ratios @xmath335 . as we have more variables than examples , we do not investigate estimators for @xmath339 : different vectors @xmath340 can lead to @xmath341 , so it does not make sense to determine the bias of an estimator for @xmath339 . instead , we only show the figures for @xmath342 and @xmath343 . again , the estimated mse of bound is lower than the estimated mse of pls . this example is taken from @xcite . a survey investigated the degree of job satisfaction of the employees of a company . the employees filled in a questionnaire that consisted of @xmath344 questions regarding their work environment and one question ( the response variable ) regarding the degree to which they are satisfied with their job . the answers of the employees were summerized for each of the @xmath345 departments of the company . + we compare the two methods pls and bound on this data set . for each @xmath346 we determine the 10fold crossvalidation error . [ cv ] the method bound is slightly better than pls on this data set : the cv error for the optimal number of components ( which is @xmath347 ) is 0.2698 for bound and 0.2747 for pls . it is remarkable that in this example the cv error of bound exceeds the cv error of pls in some cases . it is not clear if this is due to the small number of examples ( which makes the estimation unprecise ) or if this can also happen `` in theory '' . this paper consists of two parts . in the first part , we gave alternative and hopefully clearer proofs of the shrinkage factors of pls . in particular , we derived the fact that some of the shrinakge factors are @xmath0 . we explained in detail that this would lead to an unnecessarily high mse if pls was a linear estimator . this is however not the case and we emphasized that bounding the absolute value of the shrinkage factors by @xmath1 does not automatically lead to a lower mse . + in the second part , we investigated the problem numerically . experiments on simulated and real world data showed that it might be better to adjust the shrinkage factors so that their absolute value is @xmath348 - a method that we called bound . the difference between bound and pls was not dramatic however . besides , the scale of the experiments was of course way too small , so it would be light - headed if we concluded that we should always use bound instead of pls . + nevertheless , the experiments show that it is worth exploring the method bound in more detail . one drawback of this method is that we have to adjust the shrinkage factors `` by hand '' . if bounding the shrinkage factors tends to lead to better results , we might modify the original optimization problem of pls such that the shrinkage factors of the solution are bounded . we might modfify @xmath27 and @xmath63 to obtain a different krylov space or replace @xmath64 by a different set of feasible solutions . i would like to thank ulrich kockelkorn who eliminated innumerable errors from earlier versions of this paper and who gave a lot of helpful remarks . i would also like to thank jrg betzin for our extensive discussions on pls .	we present a formula for the shrinkage factors of the partial least squares regression estimator and deduce some of their properties , in particular the known fact that some of the factors are @xmath0 . we investigate the effect of shrinkage factors for the mean squared error of linear estimators and illustrate that we can not extend the results to nonlinear estimators . in particular , shrinkage factors @xmath0 do not automatically lead to a poorer mean squared error . we investigate empirically the effect of bounding the the absolute value of the partial least squares shrinkage factors by @xmath1 .
let @xmath0 be a free group of rank @xmath33 , with generators @xmath34 . it is convenient to take @xmath35 to be finite , although our results apply just as well to the infinite - rank case . let @xmath1 be a subgroup of finite index . the action of @xmath0 by left multiplication on @xmath29 can be thought of as a homomorphism from @xmath0 to the symmetric group @xmath36 . of course , @xmath0 acts transitively , and the subgroup @xmath1 can be recovered as the stabiliser of the trivial coset . we summarise this paragraph as follows . [ r : permutations ] subgroups of @xmath0 of index @xmath37 correspond canonically to transitive actions of @xmath0 on pointed sets of cardinality @xmath37 . we now switch to a topological point of view . let @xmath38 be a rose with @xmath35 petals , that is , a graph with precisely one vertex @xmath39 and @xmath35 edges . ( all our results can be generalised to the situation in which @xmath38 is an arbitrary finite graph . however , the case in which @xmath38 is a rose is sufficient for our purposes . ) we fix once and for all an isomorphism @xmath40 by orienting each edge of @xmath38 and labelling it with a generator of @xmath0 . by standard covering space theory , a subgroup @xmath1 of @xmath0 corresponds to a connected , pointed covering space @xmath41 with the covering map inducing an isomorphism @xmath42 . the orientation and labelling of the edges of @xmath38 pulls back to an orientation and labelling of the edges of @xmath43 . conversely , any orientation and labelling of the edges of @xmath43 determines a combinatorial map @xmath44 that sends vertices to vertices and edges to edges . we will only consider such maps , and we will usually think of them in terms of the corresponding labelled and oriented graph @xmath43 . covering maps to @xmath38 are easily characterised in terms of the labelling and orientation on @xmath43 . a labelling and orientation on a graph @xmath43 are said to satisfy the _ covering condition _ if , for each vertex @xmath45 of @xmath43 and each label @xmath46 , there is exactly one incoming and one outgoing edge labelled @xmath46 at @xmath45 . the proof of the following lemma is a simple exercise . a combinatorial map of non - empty , connected graphs @xmath44 is a covering map if and only if it is bijective on the links of vertices . equivalently , @xmath44 is a covering map if and only if the corresponding labelling and orientation on the graph @xmath43 satisfy the covering condition . when a covering map is restricted to a subgraph , we obtain an immersion . a combinatorial map @xmath44 is an _ immersion _ if it is injective on links of vertices . again , this can be characterised in terms of the labelling and orientation on @xmath43 . a labelling and orientation on a graph @xmath43 are said to satisfy the _ immersion condition _ if , for each vertex @xmath45 of @xmath43 and each label @xmath46 , there is at most one incoming and one outgoing edge labelled @xmath46 at @xmath45 . just as before , the proof of the following lemma is an easy exercise . a combinatorial map of non - empty , connected graphs @xmath44 is an immersion if and only if the corresponding labelling and orientation on the graph @xmath43 satisfy the immersion condition . the immersion condition and the covering condition are illustrated in figure [ fig : conditions ] . it follows from standard covering space theory that @xmath0 acts on the vertex set of @xmath43 by path lifting . to be precise , any element @xmath47 can be thought of as a loop in @xmath38 , represented by a continuous map @xmath48\to x$ ] with @xmath49 . covering space theory asserts that , for each vertex @xmath50 , there is a unique lift @xmath51\to y$ ] with @xmath52 . the action of @xmath53 on the vertices of @xmath43 is defined by @xmath54 furthermore , the map that sends @xmath55 induces an @xmath0-equivariant bijection between @xmath29 and the vertices of @xmath43 . we can summarise this discussion in the following remark . [ r : coverings ] subgroups of @xmath0 correspond to connected , based , labelled , oriented graphs that satisfy the covering condition . the index of the subgroup is equal to the number of vertices of the graph . it is apparent from remarks [ r : permutations ] and [ r : coverings ] that connected , based , labelled , oriented graphs satisfying the covering condition are in bijection with transitive permutation actions of @xmath0 on based sets . indeed , given such a graph @xmath43 , as observed above the group @xmath0 acts by path lifting on the vertices . the action of the generator @xmath46 can be seen by restricting attention to the edges of @xmath43 labelled by @xmath46 : by the covering condition , these edges form a union of topological circles ; each circle corresponds to a cycle under the action of @xmath46 ; taken together , these circles give the cycle type of @xmath46 . conversely , given an action of @xmath0 by permutations on a set @xmath56 , we can build a covering graph @xmath43 . take @xmath56 to be the vertex set of @xmath43 . there is an oriented edge labelled @xmath46 from @xmath57 to @xmath58 if @xmath59 . by construction , this graph @xmath43 satisfies the covering condition . in this section , we recall stallings s proof of marshall hall s theorem @xcite . the key observation is that one can complete any immersion to a covering map without increasing the number of vertices . [ l : completion ] let @xmath60 be a finite graph , let @xmath38 be a rose and let @xmath61 be an immersion . the immersion factors as @xmath62 where @xmath60 is a subgraph of @xmath43 , every vertex of @xmath43 is a vertex of @xmath60 , and @xmath44 is a covering map . as above , we can think of @xmath60 as an oriented , labelled graph that satisfies the immersion condition . for each @xmath5 , let @xmath63 be the set of vertices of @xmath60 that adjoin an outgoing edge labelled @xmath46 , and let @xmath64 be the set of vertices of @xmath60 that adjoin an incoming edge labelled @xmath46 . counting the edges labelled @xmath46 , we see that @xmath65 whence @xmath66 choose any bijection between @xmath67 and @xmath68 . we can use this additional bijection to add new oriented edges labelled @xmath46 to @xmath60 . let @xmath43 be the result of carrying this out for each @xmath5 . by construction , @xmath43 satisfies the covering condition . the proof is illustrated in figure [ fig : hall ] . , satisfies the immersion condition . after adding the grey edges , the resulting graph @xmath43 satisfies the covering condition.,width=192 ] [ t : marshall hall ] let @xmath0 be a finitely generated free group , let @xmath69 be a finitely generated subgroup , and let @xmath70 . there is a homomorphism @xmath3 from @xmath0 to a finite group such that @xmath4 for all @xmath5 . we identify @xmath0 with @xmath71 for @xmath38 a suitable rose . let @xmath72 be the covering map corresponding to @xmath1 . each @xmath24 lifts at @xmath73 to a path @xmath74\to x'$ ] ; by assumption , @xmath75 . we now consider the topology of the graph @xmath76 . let @xmath77 be a maximal tree . because @xmath1 is finitely generated , it follows from the seifert van kampen theorem that @xmath78 is a finite union of edges @xmath79 . for each @xmath80 , let @xmath81 be an edge path from @xmath73 to the initial vertex of @xmath82 and let @xmath83 be an edge path from @xmath73 to the terminal vertex of @xmath82 . let @xmath84 that is , @xmath60 is a finite , connected subgraph of @xmath76 that contains the paths @xmath85 and that also carries the fundamental group of @xmath76 . the labelling of @xmath76 restricted to @xmath60 satisfies the immersion condition , so we may apply lemma [ l : completion ] to construct a finite - sheeted covering space @xmath43 . consider the action of @xmath0 on the vertices of @xmath43 . every element of @xmath1 is represented by a loop in @xmath60 . therefore , @xmath1 is contained in the stabiliser of @xmath73 . on the other hand , @xmath75 for each @xmath5 , and so @xmath24 does not stabilise @xmath73 . in this section , we show how to modify stallings s construction to force the action of @xmath0 on the vertices of @xmath43 to be alternating or symmetric . recall some basic terminology from the theory of symmetric group actions . consider a transitive action by a group @xmath6 on a finite set of order @xmath86 , in other words a homomorphism @xmath87 with transitive image ; the integer @xmath86 is called the _ degree _ of the action . the smallest degree of a non - trivial cyclic subgroup of the image of @xmath6 is called the _ minimal degree _ of the action . the action is _ primitive _ if it does not preserve any proper partition of the finite set . we will use a classical theorem of jordan , which gives a criterion for an action to be symmetric or alternating ( in other words , for the homomorphism @xmath87 to have image of index at most two ) @xcite . see theorem 3.3d of @xcite for a modern treatment . [ l : alternating completion ] assume that @xmath93 . let @xmath60 be a finite graph and let @xmath38 be a rose with @xmath35 petals . any immersion @xmath61 that is not a covering map factors as @xmath62 where @xmath44 is a finite - sheeted covering map and the action of @xmath94 on the vertex set of @xmath43 is alternating . as before , we think of @xmath60 as a labelled , oriented graph . because the immersion @xmath61 is not a covering map , the labelling and orientation on the graph @xmath60 do not satisfy the covering condition , so for some @xmath5 there are ( not necessarily distinct ) vertices @xmath95 of @xmath60 such that @xmath96 does not adjoin an outgoing edge labelled @xmath46 and @xmath97 does not adjoin an incoming edge labelled @xmath46 . without loss of generality , we may take @xmath98 . by adding edges to @xmath60 as in the proof of lemma [ l : completion ] , we may assume that @xmath96 and @xmath97 are the only vertices that adjoin fewer than @xmath99 half - edges . let @xmath37 be the number of vertices of @xmath60 . before we proceed with the details of the proof , we will give an outline . the labelled , oriented graph @xmath43 is constructed from @xmath60 by attaching a large labelled , oriented graph @xmath100 on which @xmath101 acts trivially . we need to ensure that @xmath43 has the following properties : 1 . the labelling and orientation on @xmath43 satisfy the covering condition ; 2 . the number of vertices of @xmath43 is prime ( which ensures that the action of @xmath0 on the vertices of @xmath43 is primitive ) ; 3 . the generator @xmath101 fixes a large number of the vertices of @xmath43 ( this ensures that the action of @xmath0 on the vertices of @xmath43 has small minimal degree ) . at this point it will follow from theorem [ t : jordan s theorem ] that the action of @xmath0 on the vertices of @xmath43 is either alternating or symmetric . in order to force the action to be alternating , we will also attach a carefully chosen labelled , oriented graph @xmath102 . a schematic diagram of the construction of @xmath43 is given in figure [ fig : schematic ] . we will now give the details of the construction of @xmath100 . take a graph homeomorphic to an interval with @xmath103 vertices , denoted @xmath104 . label each edge @xmath105 and orient the edges consistently , so that the immersion condition is satisfied . attach an oriented edge labelled @xmath46 from @xmath106 to itself for each @xmath107 and for each @xmath80 . the result is a labelled , oriented graph @xmath100 , with @xmath86 vertices , that satisfies the immersion condition . only two vertices of @xmath100 adjoin fewer than @xmath99 half edges : one does not adjoin an incoming edge labelled @xmath105 and the other does not adjoin an outgoing edge labelled @xmath105 . an example is shown in figure [ fig : w_n ] . next , consider any sequence @xmath108 ( for @xmath109 . we will define a labelled , oriented graph @xmath102 , with four vertices , denoted @xmath110 . by changing the sequence @xmath111 , we will be able to change the signs of the permutations defined by the generators of @xmath0 . the construction is as follows . there is an oriented edge labelled @xmath105 from @xmath112 to @xmath113 ; for each @xmath114 , there are oriented loops of length one attached to @xmath112 and @xmath113 . for each @xmath115 , if @xmath116 then there are oriented loops of length one labelled @xmath46 attached to @xmath117 and @xmath118 ; if @xmath119 then @xmath117 and @xmath118 are the two vertices of an oriented loop of length two labelled @xmath46 . if @xmath120 then @xmath112 and @xmath117 are the two vertices of an oriented loop of length two labelled @xmath101 , and likewise @xmath113 and @xmath118 are the two vertices of such a loop ; if @xmath121 then @xmath110 are the vertices of an oriented loop of length four labelled @xmath101 . two examples are shown in figure [ fig : v_s ] . we are nearly ready to construct @xmath43 from the pieces @xmath60 , @xmath100 and @xmath102 . however , we first construct a similar labelled , oriented graph @xmath123 , in which we will not try to control the parity of the action of @xmath0 . let @xmath124 be the smallest prime such that @xmath125 and let @xmath126 . let @xmath127 for @xmath128 . construct @xmath123 from @xmath129 by attaching four oriented edges labelled @xmath105 in any way that makes the result connected and that satisfies the covering condition . so @xmath0 acts on the vertex set of @xmath123 , and we can read off the cycle types by looking at the edges . because @xmath123 is connected , the action is transitive , and because the number of vertices @xmath124 is prime , it follows that the action is primitive . by construction , @xmath130 for any @xmath80 , and so we see that the minimal degree of the action is at most @xmath131 . it follows from jordan s theorem that the action is alternating or symmetric . finally , we will modify @xmath123 to force the action to be alternating . for each @xmath5 , consider the sign @xmath132 of the action of @xmath46 on the vertex set of @xmath123 . now construct @xmath43 from @xmath123 by replacing @xmath133 by @xmath102 . as before , we see that the action of @xmath0 on the vertex set of @xmath43 is symmetric or alternating , but the modification ensures that each generator @xmath46 acts as an even permutation . therefore , the action of @xmath0 on the vertex set of @xmath43 is alternating . the construction of @xmath43 is illustrated in figure [ fig : schematic ] . note that , in order to ensure that the resulting action is symmetric rather than alternating , one can simply change the sequence @xmath134 in one place . we therefore also have a symmetric version of the previous lemma . [ l : symmetric completion ] assume that @xmath93 . let @xmath60 be a finite graph . any immersion @xmath61 that is not a covering map factors as @xmath62 where @xmath44 is a finite - sheeted covering map and the action of @xmath94 on the vertex set of @xmath43 is symmetric . theorem [ t : alternating quotients ] follows from lemma [ l : alternating completion ] in exactly the same way that theorem [ t : marshall hall ] follows from lemma [ l : completion ] . likewise , theorem [ t : symmetric quotients ] follows from lemma [ l : symmetric completion ] .	we strengthen marshall hall s theorem to show that free groups are locally extended residually alternating . let @xmath0 be any free group of rank at least two , let @xmath1 be a finitely generated subgroup of infinite index in @xmath0 and let @xmath2 be a finite subset . then there is a surjection @xmath3 from @xmath0 to a finite alternating group such that @xmath4 for any @xmath5 . the techniques of this paper can also provide symmetric quotients . [ d : residually finite ] a group @xmath6 is _ residually @xmath7 _ ( for some class of groups @xmath7 ) if , for any @xmath8 , there is a surjection @xmath9 with @xmath10 . many groups are known to be residually finite , and it is natural to ask whether one can restrict attention to smaller classes of finite groups . katz and magnus proved that free groups are residually alternating , hence residually simple @xcite ( see also @xcite , @xcite and @xcite _ et seq . _ ) ; in the topological context , long and reid showed that many hyperbolic 3-manifold groups are residually @xmath11 @xcite . one of the most obvious generalisations of definition [ d : residually finite ] replaces the trivial subgroup with an arbitrary finitely generated subgroup @xmath1 . ( in many cases , it is too much to expect infinitely generated subgroups to satisfy this property see remark [ r : erf ] . ) this consideration leads to the following definition . a group @xmath6 is said to be _ locally extended residually finite _ ( _ lerf _ , also often called _ subgroup separable _ ) if , for any finitely generated subgroup @xmath12 and any @xmath13 , there is a surjection @xmath3 from @xmath6 onto a finite group such that @xmath14 . [ r : erf ] without the requirement that the subgroup @xmath1 be finitely generated , such a group @xmath6 is called _ extended residually finite _ or _ erf_. it is well known that the _ @xmath15-bausmlag solitar group _ @xmath16 is not residually finite . by considering the kernel of the natural map @xmath17 , it follows easily that non - abelian free groups are not erf . marshall hall jr proved that free groups are locally extended residually finite @xcite . his theorem has been reinterpreted topologically and generalised to much larger classes of groups @xcite . however , to the best of the author s knowledge , no results have been proved that restrict the finite quotients that arise to a more specific class . the aim of this note is to begin to fill this gap by proving that free groups are what one might call locally extended residually alternating . [ t : alternating quotients ] let @xmath0 be a free group of rank greater than one . let @xmath1 be a finitely generated subgroup of infinite index in @xmath0 and let @xmath18 be a finite subset of @xmath19 . there is a surjection @xmath3 from @xmath0 onto some finite alternating group @xmath20 such that @xmath4 for all @xmath5 . a small modification of the argument gives symmetric , rather than alternating , quotients . [ t : symmetric quotients ] let @xmath0 be a free group of rank greater than one . let @xmath1 be a finitely generated subgroup of infinite index in @xmath0 and let @xmath18 be a finite subset of @xmath19 . there is a surjection @xmath3 from @xmath0 onto some finite symmetric group @xmath21 such that @xmath4 for all @xmath5 . in the case when @xmath7 is the class of all finite groups , it is equivalent to state the theorem for a single element @xmath22 instead of for a finite subset @xmath23 : to deduce the latter from the former , simply take the product of the quotients across all @xmath24 . this works because a product of finite groups is finite . as a product of alternating groups is not alternating and a product of symmetric groups is not symmetric , we give the stronger statements . the hypothesis that @xmath1 is of infinite index in @xmath0 is necessary . for instance , suppose that @xmath1 is a normal subgroup of finite index in @xmath0 . for any surjection @xmath25 with @xmath26 , @xmath27 is a normal subgroup of @xmath28 and is therefore either the whole of @xmath28 or trivial . in the latter case , it follows that @xmath29 maps onto @xmath28 . but @xmath0 has many finite quotients that do not map onto @xmath28 . stallings interpreted hall s original proof that free groups are lerf using the topology of graphs , reducing it to the topological fact that any immersion of finite graphs can be completed to a covering map @xcite . wiegold s proof that free groups are residually even alternating @xcite uses a classical theorem of jordan , which asserts that if the minimal degree of a primitive permutation group is small enough , then that group must be symmetric or alternating @xcite . the proofs of theorems [ t : alternating quotients ] and [ t : symmetric quotients ] combine jordan s theorem with the covering theory of graphs . the key technical result is lemma [ l : alternating completion ] . in the proof , we show how to complete an immersion of a finite graph into the rose to a covering map in such a way that the resulting permutation action of @xmath0 on the vertices is primitive and satisfies jordan s condition . this proves that free groups are , so to speak , locally extended residually symmetric - or - alternating . a small modification of this construction then forces the action to be alternating ; a slightly different modification forces the action to be symmetric . in order to apply jordan s theorem , we must ensure that the action of @xmath0 on the vertices of the cover we construct is primitive . in this paper we do so in the simplest possible way , by requiring the number of vertices of the cover to be prime . thus , we actually prove that free groups are locally extended residually alternating - of - prime - degree . alternatively , one could ensure that the action is primitive by forcing it to be 2-transitive . this is possible , using a more complicated construction , of which we do not give the details here . via this more complicated construction , one can obtain different restrictions on the degrees of the alternating quotients . after free groups , the fundamental groups of surfaces are the next examples to consider . let @xmath30 be a closed , orientable , hyperbolic surface and let @xmath1 be a finitely generated subgroup of infinite index in @xmath31 . for any @xmath32 there is a surjection @xmath3 from @xmath6 onto a finite alternating group with @xmath14 . in section [ s : permutations and coverings ] , we recall the very well known facts that subgroups of free groups can be viewed in terms of either permutation representations or coverings of graphs , and observe that it is easy to pass from one point of view to the other . in section [ s : hall s theorem ] we revisit stallings s proof of marshall hall s theorem . finally , in section [ s : alternating quotients ] we bring in jordan s theorem and prove theorems [ t : alternating quotients ] and [ t : symmetric quotients ] .
supernova remnants ( snrs ) in the galaxy are believed to be most likely accelerators of cosmic - rays ( crs ) below knee , in an energy range less than @xmath510@xmath6 ev . the highest energy @xmath1-rays in the galaxy are emitted from three young and very - high - energy ( vhe ; @xmath7 @xmath8 100 gev ) @xmath1-ray snrs , rx j1713.7@xmath23946 @xcite , rx j0852.0@xmath24622 @xcite , and hess j1731@xmath2347 @xcite . in these snrs , it is suggested that cr protons or electrons are accelerated efficiently up to 10800 tev or 140 tev , respectively ( e.g. , * ? ? ? all the three show pure synchrotron x - rays with no thermal features , and acceleration of tev cr electrons is well established . the origin of the @xmath1-rays in these snrs , either hadronic or leptonic , has been under debate ; the gev @xmath1-ray observations show hard @xmath1-ray spectrum similar to that of the leptonic @xmath1-rays @xcite , whereas a recent comparative study between @xmath1-rays and the interstellar protons is consistent with the hadronic origin of @xmath1-rays @xcite . it is notable that the numerical simulations on the shock - cloud interaction @xcite lend support for the hadronic @xmath1-rays , since the low energy cr protons producing the gev @xmath1-rays can not penetrate into the densest part of the cloud cores , leading to the hard spectrum just presented by @xcite . a similar conclusion is drawn by @xcite and @xcite . even though the leptonic scenarios are still debated from a different perspective ( e.g. , * ? ? ? * ) . the young snr rx j1713.7@xmath23946 ( also known as [email protected] ) is one of the best targets for testing the action of the cr electrons , because the snr emits strong synchrotron x - rays and vhe @xmath1-rays @xcite . rx j1713.7@xmath23946 has a large apparent diameter of @xmath31 degree at a small distance of @xmath31 kpc @xcite . subsequently , the x - ray properties of rx j1713.7@xmath23946 were studied by using @xmath9-@xmath10 , @xmath0 , and @xmath11 . @xmath11 observations revealed filamentary structures of x - rays and short time variability ( @xmath31 yr ) of one of them @xcite , which is interpreted as due to rapid particle acceleration of cr electrons and cooling in @xmath31 yr scale with mg magnetic field . if this is correct , the cr electrons should be accelerated efficiently in the bohm diffusion regime @xcite . additionally , when the cosmic - rays are accelerated , the magnetic fields amplified by cosmic - ray streaming instability @xcite . moreover , different processes of amplifying magnetic fields may be working in during shock - cloud interaction ( e.g. , * ? ? ? * ; * ? ? ? the amplified magnetic field may intensify the non - thermal x - rays , whereas the synchrotron loss due to the amplified field may decrease the number of cr electrons via synchrotron loss ( e.g. , * ? ? ? in order to test such actions , we need a detailed comparison between the electron spectrum and the interstellar matte . @xcite presented spatial distribution of absorbing column density @xmath12(x - ray ) and photon index @xmath13 obtained by @xmath9-@xmath10 at angular resolutions smoothed to 230 arcmin . the @xmath12(x - ray ) distribution shows significant variations over the whole snr ( 0.4 @xmath14 10@xmath15 @xmath16 @xmath17 @xmath12(x - ray ) @xmath17 1.1 @xmath14 10@xmath15 @xmath16 ) . the authors carried out an analysis of x - ray absorption and compared the x - rays with co , hi , and visual extinction and showed that the interstellar medium ( ism ) is correlated with the @xmath12(x - ray ) . additionally , the value of photon index @xmath13 has also shown the strong variation ( 1.8 @xmath18 2.6 ) . this trend was confirmed by a more detailed analysis of @xmath9-@xmath10 data @xcite . @xcite and @xcite detected the hard synchrotron x - rays up to 40 kev from the snr by using @xmath0 . the spectra are fitted by interstellar absorbed power - law model with the exponential rolloff , suggesting cr electrons close to the bohm diffusion limit @xcite . @xcite presented multi - transition co data and comparing with the @xmath0 x - rays toward the northwest of rx j1713.7@xmath23946 . these authors found that the x - rays and molecular clumps are well - correlated at 1 pc scale , but anti - correlated at 0.1 pc scale . the latter shows a trend as rim - brightened x - rays around the dense molecular clumps . subsequently , @xcite compared 18 co clumps and a hi clump with their surrounding x - rays and concluded that the trend is commonly found over the whole snr . additionally , the authors found the strong correlation between the x - ray intensity and the molecular mass interacting with the snr . these results indicate that the synchrotron x - rays are closely related to the surrounding co gas . to explore this correlation , we should estimate the physical parameters such as the absorbing column density @xmath12(x - ray ) , photon index @xmath13 and x - ray flux , in angular resolution high enough to compare with the ism distributions . such quantitative comparisons with the ism mark an important step toward understanding the spectral variation of x - rays and efficient cosmic - ray acceleration . our aim is to compare the spatial distribution of x - ray properties with that of the ism clumps . for this aim , we carried out the spectral extraction and model fitting in the regions gridded at 28 arcmin scales over the entire snr . in the present paper , we show a detailed comparison of the spatial distribution among @xmath12(x - ray ) , @xmath13 , x - ray flux and ism both molecular and atomic gas at angular resolutions raging from 2 arcmin to 8 arcmin depending on the photon statistics . section [ section : observations ] gives observations and data reduction of @xmath0 x - rays , nanten co and atca @xmath19 parkes hi . section [ section : results ] consists of three subsections : section [ typical ] presents the typical spectra of x - rays ; section [ characterization ] spatial and spectral characterization of the x - rays ; and [ comparisonism ] a detailed comparison with the ism . section [ section : discussion ] consists of three subsection : in section [ absorbingcol ] discusses the spatial variation of absorbing column density is discussed ; section [ relationship ] the relationship between x - ray flux and ism ; and section [ efficient ] the efficient acceleration of cosmic - ray electrons . the conclusions are given in section [ section : conclusion ] . we analyzed the x - ray dataset archive obtained by @xmath0 ( data archives and transmission system ; darts at isas / jaxa ) . this dataset consists of 17 pointings taken at 2005 september ( swg ; 3 pointings ) , 2006 september and october ( ao1 ; 10 pointings ) , 2010 february ( ao4 : 4 pointings ) and we mainly used 15 pointing of on sources . these data were already analyzed and published elsewhere @xcite . we used software heasoft version 6.11 with pipeline processing version 2.0 or 2.4 and with standard event selection criteria ( cleaned event files ) . xis mosaic image of rx j1713.7@xmath23946 in the energy band 15 kev @xcite . the color scale indicates count rate on a square root scale and is unit of 10@xmath20 counts s@xmath21 pixel@xmath21 . the solid boxes are those used for spectral analysis . we also show the spectra enclosed by the black solid lines ( see also figure [ fig2 ] ) . the dashed boxes correspond to the regions used to extract the background spectrum . the two point sources , 1wga j1714.4@xmath23945 and 1wga j1713.4@xmath23949 , are circled with magenta and those regions were removed from spectral analysis.,width=325 ] the x - ray images are those used in figure [ fig1]a of @xcite . figure [ fig1 ] shows xis mosaic image ( 15 kev ) in rx j1713.7@xmath23946 @xcite . color scheme is in a square - root scale in 10@xmath20 counts s@xmath21 pixel@xmath21 ( pixel size is @xmath316.7@xmath22 ) smoothed with a gaussian kernel with fwhm @xmath345@xmath22 . this image is subtracted for non x - ray background ( nxb ) and corrected for the vignetting effect by xrt ( see more details in @xcite @xcite section 2.3 ) . the x - ray spectra of rx j1713.7@xmath23946 is represented by the absorbed power - law model ( e.g. , * ? ? ? this model is for synchrotron x - rays with photoelectric absorption and the model fitting gives three parameters , absorbing column density @xmath12(x - ray ) , and the photon index @xmath13 , in addition to the absorption - corrected x - ray flux . we show typical x - ray spectra in figure [ fig2 ] . we shall explain the method to derive the three parameters above . first , the snr was divide into @xmath3600 regions of 2@xmath23 grids , where the number of x - ray counts is more than 160 counts per grid in the energy band 15 kev . for the @xmath3600 regions , only the data taken with fi ccd ( xis 0 , 2 , and 3 ) were used to derive x - ray spectra . in order to compare with the data of the interstellar medium , co and hi , the angular resolution was set to the highest value ( @xmath0 hpd @xmath32@xmath24 ) . in this case , the signal leaking into / out of neighboring grid cells is less than 50 @xmath25 , because we generated arfs ( ancillary response files ) with 4 arcmin grid images for each region . this signal leaking is not critical in obtaining the large - scale trend of the photon index , absorbing column density , and fluxes . in addition , the background spectrum was estimated in the four regions shown in figure [ fig1 ] and the western background spectrum ( @xmath2617@xmath2710@xmath2831@xmath29 , @xmath30 = @xmath239@[email protected]@xmath22 ) was used for every region . the two prominent point sources ( 1wga j1714.4@xmath23945 and 1wga j1713.4@xmath23949 ) in the field are excluded in the analysis . subsequently , the spectra of each region taken with xis 0 , 2 , and 3 are summed up and was fit by the absorbed power - law model to generate rmf ( redistribution matrix file ) by xisrmfgen and arf by xissimarfgen @xcite . ) . the solid lines represent the best - fit absorbed power - law model . the lower part shows the residuals from the best - fit models . the north and center spectra can be fitted by the same absorbing column density ( @xmath32(x - ray)@xmath30.5 @xmath14 10@xmath15 @xmath16 ) but with different photon index ( north : @[email protected] , center : @[email protected] ) . the center and south spectra are the same photon index ( @xmath342.7 ) but with different absorbing column densities ( center : @xmath32(x - ray ) = 0.45@xmath35 @xmath14 10@xmath15 @xmath16 , south : @xmath32(x - ray ) = 0.89@xmath36 @xmath14 10@xmath15 @xmath16).,width=325 ] next , in order to reduce the statistical relative errors , some neighboring regions on source were summed up and the spectra were again fit by the absorbed power - law model . specifically , we combined two or more 2 arcmin grids so that the relative error of @xmath12(x - ray ) becomes 30@xmath25 or less . the combined grids are selected in the neighboring regions , which have roughly the same values in @xmath12(x - ray ) . figure [ fig1 ] shows the final division with 305 regions as indicated by solid boxes . about one third of the whole snr is 2@xmath24 grid ( 4 arcmin@xmath4 ) and 80@xmath25 of that is better than 4@xmath24 grid ( 16 arcmin@xmath4 ) . these values are much better than the previous studies ( e.g. , * ? ? ? * of that is worse than 6@xmath24 grid ( 36 arcmin@xmath4 ) ) . each spectrum is binned to include at least 100 counts per individual energy bin . after the binning the energy ranges a below 0.4 kev and above 12 kev bin are excluded in the fitting . the best fit parameters in the fitting were used to derive absorbing column density @xmath12(x - ray ) , photon index @xmath13 , and absorption - corrected flux in 310 kev , @xmath37 , as shown in figure [ fig3 ] . if the background spectrum is changed to that for other three regions , the values of @xmath12(x - ray ) only changed systematically with 1020@xmath25 and it has no effect on the global trend in figure [ fig3 ] . @xmath38co(@xmath39=10 ) and hi datasets are nanten galactic plane survey ( ngps ) and southern galactic plane survey ( sgps ) taken with nanten , and atca @xmath19 parkes telescopes @xcite , respectively . the angular resolutions are hpbw @xmath32@xmath406 for co and @xmath32@xmath402 for hi , similar to @xmath0 xis hpd @xmath32@xmath24 . the velocity resolutions and typical rms noise fluctuations are ( co ) 0.65 km s@xmath21 and 0.3 k ch@xmath21 and ( hi ) 0.82 km s@xmath21 and 1.9 k ch@xmath21 , respectively . the velocity integrated intensities of co and hi , @xmath41(co ) and @xmath41(hi ) , are converted into molecular column density @xmath42(h@xmath43 ) and atomic column density @xmath12(hi ) and the total proton column density is obtained as @xmath12(h@xmath43+hi ) = 2@xmath14@xmath42(h@xmath43 ) + @xmath12(hi ) . a relationship @xmath42(h@xmath43 ) ( @xmath16 ) = @xmath44 ( @xmath16 ( k km s@xmath21)@xmath21)@xmath14@xmath41(co ) ( k km s@xmath21 ) is used where @xmath44 = 2.0@xmath1410@xmath45 ( @xmath16 ( k km s@xmath21)@xmath21 ) @xcite . the hi line is generally assumed to be optically thin and a relationship @xmath12(hi ) = 1.823@xmath1410@xmath46 @xmath14 @xmath41(hi ) is used @xcite . the regions where the hi is optically thick and hence need to apply the correction for self - absorption . details of the method is given by @xcite in section 3.3.3 . figure [ fig2 ] shows @xmath0 xis 0 + 2 + 3 spectra in the three typical regions , the north ( @xmath47 , @xmath30 ) = ( 17@[email protected]@xmath29 , @xmath239@[email protected]@xmath22 ) , the center ( @xmath47 , @xmath30 ) = ( 17@[email protected]@xmath29 , @xmath239@[email protected]@xmath22 ) , and the south ( @xmath47 , @xmath30 ) = ( 17@[email protected]@xmath29 , @xmath240@[email protected]@xmath22 ) , and the results of the model fitting . the lower parts show residuals in the fitting indicating that the fitting is reasonably good . the absorbing column density @xmath12(x - ray ) is @xmath30.5@xmath1410@xmath15 @xmath16 both in the north and the center , but the photon index is different as @xmath13 = [email protected] and @xmath13 = [email protected] for the north and the center , respectively . on the other hand , the south has large absorbing column density @xmath12(x - ray ) = 0.89@xmath4810@xmath15 @xmath16 and photon index @xmath13 @xmath32.7 . these differences are seen as significantly different spectral shapes in figure [ fig2 ] . figure [ fig3]a shows absorbing column density @xmath12(x - ray ) in the individual regions , where the @xmath0 xis 15 kev intensity is overlayed as contours . we aimed at achieving that the relative error is 30@xmath25 at maximum at the 90@xmath25 confidence level by tuning the pixel size . consequently , the average relative error and its maximum value are confined to be 14@xmath25 and 30@xmath25 , respectively , at the 90@xmath25 confidence level . this accuracy is high enough to assess the spatial variation of the absorbing column density and photon index as discussed later . we also defined the region name in figure [ fig3]d . we find that the global distribution of @xmath12(x - ray ) shows a shell - like structure in figure [ fig3]a . in the southeast and the center , @xmath12(x - ray ) is as small as 0.40.5 @xmath1410@xmath15 @xmath16 ( region in blue ) . on the other hand , the northeast and the northwest show medium values of @xmath12(x - ray ) = 0.71.0 @xmath14 10@xmath15 @xmath16 ( region in green ) , and the southwest and south show the largest @xmath12(x - ray ) with 1.11.4 @xmath1410@xmath15 @xmath16 ( region in red ) . therefore , the absorbing column density varies from 0.4@xmath1410@xmath15 @xmath16 to 1.4@xmath1410@xmath15 @xmath16 within the snr . these values are mostly consistent with the previous studies with @xmath9-@xmath10 ( * ? ? ? * typical angular resolution @xmath49 ) , while the angular resolution and the source coverage are better in the present study . figure [ fig4]a shows absorbing column density @xmath12(x - ray ) and visual extinction ( @xmath50 in unit of mag ) derived from the digitized sky survey i ( dss ; * ? ? ? the visual extinction is smoothed to the same binning with the x - rays . the values of @xmath50 are roughly classified into the three levels ; low values ( @xmath50 @xmath51 2 mag in blue ) in the center of the snr , medium values ( 2 mag @xmath51 @xmath50 @xmath51 3 mag in green ) in the northeast and northwest regions , and high values ( @xmath50 @xmath8 3 mag in red ) in the southwest and the south . the general trend of @xmath50 is similar to that of @xmath12(x - ray ) . figure [ fig4]b shows a correlation plot between absorbing column density @xmath12(x - ray ) and visual extinction , showing a good correlation with a correlation coefficient of @xmath30.83 . figures [ fig3]b and [ fig3]c show the distributions of photon index @xmath13 and absorption - corrected flux @xmath52 , respectively . their relative errors and maximum errors are @xmath36@xmath25 and 13@xmath25 in figure [ fig3]b , and @xmath37@xmath25 and 23@xmath25 in figure [ fig3]c at the 90@xmath25 confidence level , respectively . like absorbing column density , these quantities show significant spatial variation . in particular , photon index is largest with @xmath13 @xmath33 ( in orange color ) toward the center and the south . on the other hand , regions with a small photon index ( @xmath53 ) are distributed as islands inside the snr from the north to the southeast and from the southwest to the northwest . both regions have areas of @xmath54 arcmin@xmath4 and @xmath55 arcmin@xmath4 , respectively . the absorption - corrected flux @xmath52 is similar to the 15 kev x - rays contours , especially at the brightest peaks in the northwest , west , and southwest . we also find another strong peak in the north . these flux excesses are not due to a systematic error and are possibly connected with the ism distribution as discussed in section [ comparisonism ] . vs. the proton column density @xmath32(h@xmath43+hi ) , ( b ) the photon index @xmath13 vs. the proton column density @xmath32(h@xmath43+hi ) , and ( c ) the photon index @xmath13 vs. the proton column density @xmath32(h@xmath43+hi ) , respectively . the regions are defined as follow figure [ fig5 ] : @xmath56 for east , @xmath57 for west . the histograms are also shown at lower or right side of each plot.,width=298 ] overlays of the interstellar gas at @xmath58 = @xmath220 to 2 km s@xmath21 with absorption - corrected flux @xmath52 and photon index @xmath13 are shown in figure [ fig5]a and figure [ fig5]b , respectively , where the ism is total proton column density @xmath12(h@xmath43+hi ) including derived from the co and hi . the velocity range of the ism associated is derived by @xcite and is supported by @xcite . based on a good correlation between the ism and x - rays enhanced by the interaction with the snr . we also annotated the positions ( center of gravity ) of the molecular clumps ( crosses ) and the hi clump ( circle ) defined by @xcite . additionally , we will hereinafter refer to the regions between the clumps as inter - clump . figure [ fig5]a shows a trend that the x - rays are enhanced toward the regions with enhanced ism in a pc scale . this trend is most significant in the west of the snr , where the ism is rich . on the other hand , in a sub - pc scale , we find that each bright spot of x - rays is lying around the co and hi clumps except for the molecular clump in southwest ( @xmath47 = 17@[email protected]@xmath29 , @xmath30 = @xmath239@[email protected]@xmath22 ; named as clump c ) . these results are consistent with the previous study @xcite and the present work has made it possible to evaluate the trend more quantitatively . figure [ fig6]a shows a correlation plot between @xmath52 and the ism density ; the linear correlation coefficients ( hereafter lcc ) are @xmath30.55 ( for the whole ) , @xmath30.19 ( the eastern half , on the left side of @xmath47 @xmath317@xmath2713@xmath2838@xmath29 ) and @xmath30.54 ( the western half , on the right side of @xmath47 @xmath317@xmath2713@xmath2838@xmath29 ) , respectively . the lcc@xmath3 0.55 does not show strong correlation but the number of sample in the plot @xmath3300 indicates a positive correlation at a 5@xmath25 confidence level according to a t - test , because the value of test statistic ( @xmath311.7 ) is greater than that of student s t distribution ( @xmath31.97 ) ( e.g. , * ? ? ? * ) . in figure [ fig5]b , we note that the regions having the hard spectrum are seen toward the enhanced ism in the west , whereas the regions of the hard spectrum are also found toward the diffuse ism in the eastern half . most outstanding are the ism peaks toward / around the two x - ray peaks in the northwest and in the southwest . in addition , we find four regions of the hard spectrum in the southeast where the ism is diffuse . figure [ fig6]b shows a correlation plot showing that the relationship between the ism and photon index is different between the west and east . in the western half , the spectrum is hard when the ism is dense and the spectrum is soft where the ism is diffuse ( lcc @xmath59 ) . in the eastern half , the ism is diffuse and the photon index shows variation ( lcc @xmath60 ) , where the ism density is not apparently related to the variation of the photon index for the whole region lcc is estimated to be @xmath61 . additionally , we show correlation plots between the photon index @xmath13 and the absorption - corrected flux @xmath52 in figure [ fig6]c . lccs are @xmath62 ( the whole ) , @xmath63 ( the eastern half ) and@xmath64 ( the western half ) , respectively . the histograms toward the west and east regions show a similar trend to the photon index distribution with respected the ism density , while the west region has much higher x - ray fluxes than the east . we shall discuss about these results later in section [ efficient ] . we summarize the main aspects of the present analysis as follows ( figures [ fig5 ] and [ fig6 ] ) ; 1 . in the western half of the snr , the ism density is high with significant h@xmath43 and the x - rays are enhanced . in the eastern half of the snr , the ism density is low , being dominated by hi , and the x - rays are weak . the overall correlation between the ism density and the x - rays is however not significantly high with a correlation coefficient of @xmath30.55 ( figure [ fig6]a ) . 2 . the smallest x - ray photon index @xmath65 is seen around / toward the two co peaks in the the western half , and toward the region of low ism density in the eastern half ( figure [ fig5]b ) . the largest x - ray photon index around 3 is found toward the central region of the snr as well as in the southern edge ( figure [ fig5]b ) . the photon index shows a good correlation with the x - rays and the low photon index is seen toward regions of intense x - rays and vice versa ( figure [ fig6]c ) . there is an offset by 0.3 in this correlation between the eastern and western halves of the snr in the sense that the x - rays are more intense in the western half , where the ism is denser than in the eastern half . the observations with @xmath0 xis have enabled us to estimate detailed distributions of the absorbing column density @xmath12(x - ray ) and photon index @xmath13 in rx j1713.7@xmath23946 at a low background level and high photon statics . here in the discussion , we first discuss the absorbing column density @xmath12(x - ray ) . the spatial distribution of @xmath12(x - ray ) delineates the snr shell as shown in figure [ fig3]a . this distribution shows a good correspondence with visual extinction shown in figure [ fig4]a . the regression in a straight line calculated from a least - squares fitting gives @xmath12(x - ray ) ( @xmath16 ) = 3.0@xmath1410@xmath66@xmath67 ( magnitude ) for the scatter plot in figure [ fig4]b . the numerical factor is slightly larger than that of the conventional relation @xmath12 ( @xmath16 ) = 2.5@xmath1410@xmath66@xmath67 ( magnitude ) @xcite . this is understandable if we consider the distance of rx j1713.7@xmath23946 is 1 kpc , since the visual extinction tends to be under - estimated for a distance larger than a few 100 pc by the foreground stars . it is not necessarily true that all @xmath12(x - ray ) is physically associated with the snr . here we need to take into account the contribution of the local gas between the snr and the sun ( figure [ fig7]a ) . @xcite already estimated the foreground component @xmath68(h@xmath43+hi ) ( figure [ fig7]a ) . figure [ fig7]b shows the distribution of absorbing column density [ @xmath32(x - ray)@xmath2@xmath68(h@xmath43+hi ) in the snr , which gives a shell - like distribution of @xmath12(x - ray ) more clearly than figure [ fig3]a . the absorption toward the center of the snr is @xmath30.2@xmath1410@xmath15 @xmath16 , whereas that toward the outer boundary is shell - like with absorbing column density of @xmath30.50.8@xmath1410@xmath15 @xmath16 . this shell represents a cavity wall of the ism created by the stellar wind of the snr progenitor . the inside of the cavity is highly evacuated with density lower than @xmath69 @xmath70 and the dense clumps in the cavity wall have higher density like @xmath310@xmath410@xmath71 @xmath70 ( e.g. , * ? ? ? we shall discuss the absorption toward the southeast - rim . in this region @xcite identified cold hi gas that corresponds to the vhe @xmath1-ray shell . the cold hi with low spin temperature of @xmath340 k has density around 100 @xmath70 , less than @xmath31000 @xmath70 threshold density for the collisional excitation of the co emission . the proton column density of the cold hi without co emission is estimated to be @xmath30.5@xmath1410@xmath15 @xmath16 from the hi self - absorption in the southeast rim of rx j1713.7@xmath23946 @xcite . the present absorption in the snr calculated from x - rays has also column density of 0.40.5@xmath1410@xmath15 @xmath16 ( figure [ fig7]b ) which is consistent with the cold hi . ccccccc @xmath32(x - ray ) & @xmath72 & & 0.83 & & & figure [ fig5]b + @xmath52 & @xmath32(h@xmath43+hi ) & & 0.55 & 0.16 & 0.53 & figure [ fig6]a + @xmath13 & @xmath32(h@xmath43+hi ) & & @xmath73 & @xmath74 & @xmath75 & figure [ fig6]b + @xmath13 & @xmath52 & & @xmath76 & @xmath77 & @xmath78 & figure [ fig6]c + @xmath32(x - ray)@xmath2@xmath68(h@xmath43+hi ) & blue - shifted ism & & 0.25 & 0.18 & 0.25 & figure [ fig9]a + @xmath32(x - ray)@xmath2@xmath68(h@xmath43+hi ) & red - shifted ism & & 0.39 & 0.45 & 0.39 & figure [ fig9]b [ tab1 ] we compared the absorbing column density [ @xmath12(x - ray ) @xmath2 @xmath79(h@xmath43+hi ) ] ( figure [ fig7]b ) with @xmath32(h@xmath43+hi ) , the proton column density physically associated with the x - ray emitting shell , in two velocity ranges of the interacting gas in figures [ fig8 ] and [ fig9 ] . figures [ fig8]a and [ fig8]b show [ @xmath12(x - ray ) @xmath2 @xmath79(h@xmath43+hi ) ] overlayed on the ism proton column density @xmath32(h@xmath43+hi ) at @xmath58 @xmath3 @xmath80 to @xmath81 km s@xmath21 and @xmath3 @xmath29 to 2 km s@xmath21 . this comparison shows that the ism distribution has generally good correspondence with the x - ray absorbing column density . it is however to be noted that the blue - shifted ism shows a poorer correlation than that of the red - shifted ism with the x - ray absorbing column density as shown in figure [ fig9 ] , scatter plots between the ism and the x - ray absorption column density . in figure [ fig9 ] we divide the plots into the red- and blue - shifted ism . the blue - shifted ism in the west shows a poor correlation with the x - ray absorbing column density of [email protected] , while the red - shifted ism in the west has a higher correlation of [email protected] than the blue - shifted ism ( see also table [ tab1 ] ) . this indicates that the red - shifted ism is located on the far - side of the snr , and it is in the opposite sense to what is expected in an expanding motion of the shell - like ism . the relative position of the ism is explicable if the pre - existent cloud motion is dominant for the dense co gas instead of expansion , as already suggested to explain the velocity distribution of the hi self - absorption @xcite . @xcite showed by magnetohydrodynamic ( mhd ) numerical simulations that the x - ray intensity is closely correlated with the ism by the enhanced magnetic field around dense clumps due to turbulence in the shock - cloud interaction . the interaction creates correlation between the ism and x - rays at a pc scale via the magnetic field . observations by @xcite showed that pc - scale correlation is seen between the x - ray intensity and the clump mass interacting with the snr blast waves , as well as anti - correlation between them in a sub - pc scale due to exclusion of the cr electrons in the dense clumps . the distributions of the x - ray intensity in figure [ fig5]a at grid sizes of 28 arcmin ( 0.62.4 pc ) show their interrelation at a pc scale ; we see trend that the x - rays are enhanced in the west where the ism is rich , and that the x - rays are depressed in the east where the ism is poor . this is consistent with what is expected in the shock - cloud interaction scheme . the higher dispersion in figure [ fig6]a may be in part ascribed to anti - correlation at a sub - pc scale , consistent with the suggestion by @xcite . based on the present results on the photon index ( figure [ fig3]b ) , we discuss the efficient cosmic - ray acceleration in rx j1713.7@xmath23946 . as shown by the previous works @xcite , the photon index in the 110kev range reflects rolloff energy @xmath82 of the synchrotron x - rays . the small ( large ) photon index corresponds to large ( small ) rolloff energy . according to the standard dsa scheme , the rolloff energy of synchrotron photons @xmath82 is given as follows when the synchrotron cooling is effective @xcite , @xmath83 where @xmath84 is the shock speed and @xmath85 ( @xmath81 ) a gyro - factor , the degree of magnetic field fluctuations . in particular the limit of @xmath86 = 1 is called as bohm limit corresponding to highly turbulent conditions and , accordingly , the rolloff energy is likely determined by the shock speed and turbulence . the present results suggest that in the western half of the snr , where the ism is rich , the shock - cloud interaction is effective and that turbulence is enhanced around dense ism clumps . on the other hand , in the eastern half of the snr where the ism is poor the dsa alone is mainly working without shock - cloud interaction . @xcite presented that the innermost cavity of 34 pc radius is surrounded by the ism shell consisting of the dense clumps and the inter - clump gas . the dense clumps have density of 10@xmath4 to 10@xmath71 @xmath70 @xcite and are mainly distributed in the western half toward the galactic plane . the inter - clump gas has density @xmath87 @xmath70 , as given by the upper limit from no thermal x - rays @xcite , and we shall adopt density of the inter - clump gas to be 1 @xmath70 for discussion . we assume that density in the cavity is @xmath30.25 @xmath70 , the same with that estimated toward gum nebula where the gas is swept up by the strong stellar winds from @xmath88 pup @xcite . in this scheme , dsa is efficiently taking place in the cavity and the accelerated crs are injected into the ism shell . in the western half of the snr outside the center ( figure [ fig3]d ) , where the shock - cloud interaction is working ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , the magnetic field may be amplified up to @xmath31 mg as indicated by the short time variation of @xmath31 yr in the synchrotron x - rays by @xmath11 , where the bohm limit ( @xmath86 = 1 ) is a good approximation @xcite . on the other hand , in the eastern half the shock speed is not much decelerated in the lower density . because the shock speed @xmath84 is proportional to 1/@xmath89 and decreases in the dense ism , where @xmath90 is the average number density of the ism . the average ism density in the eastern half is about 1/4 of that in the western half from the co / hi observations and the shock speed @xmath84 becomes larger by a factor of 2 . then , @xmath86 in the southeast becomes larger by a factor of 4 , but the term ( the shock speed @xmath84)@xmath4 is more effective . the trend in the photon index is thus largely explained in the dsa scheme by considering the effect of the ism gas , and in the both regions the rolloff energy becomes larger and the photon index smaller as @xmath53 . in figure [ fig5]b , the center of the snr ( figure [ fig3]b ) having the large photon index @xmath91 is shown by orange color . we suggest that the electrons there have lower rolloff energy due to synchrotron cooling over the last 1000 yrs ; for magnetic field of 10 @xmath92 g the cr electron cooling time at 10 kev is small as @xmath3500 yrs , whereas that at 1 kev is large as @xmath31500 yrs , leading to low rolloff energy in the central 34 pc . we also note that such spectral softening can be ascribed to the energy loss due to adiabatic expansion ( e.g. , * ? ? ? one would expect that the photon index becomes even larger toward the denser regions , if only @xmath84 determines the photon index . interestingly , the photon index becomes small toward the clump c and the region between the two dense clumps , d and l in figure [ fig5]b . it is also notable that the x - rays are enhanced toward these regions having small photon indexes . in the shock - cloud interaction scheme , the magnetic field is amplified in the interacting region , leading to higher synchrotron loss and smaller rolloff energy . this can lead to lower x - ray intensity and lower rolloff energy toward the dense clumps , while obviously we need a more elaborate work to quantitatively affirm this . it remains to be explored if the dsa scheme can accommodate the high x - ray intensity , the high ism density , and the small photon index consistently . also we here suggest that some additional acceleration mechanism may possibly be working to accelerate cr particles in the west where the ism is rich . such mechanisms possibly include the 2nd order fermi acceleration @xcite , magnetic reconnection in the turbulent medium @xcite , reverse shock acceleration ( e.g. , * ? ? ? * ) , non - liner effect of dsa ( e.g. , * ? ? ? * ) and so on . to summarize , we suggest that the ism and its distribution have a significant impact on the scheme of the particle acceleration via the shock - cloud interaction . this suggests the tight relationship between the snr and the ism . the present study explored the cr electron behavior into significant details not achieved in the previous works , and presented that the ism density affects significantly the cr electron rolloff energy . in particular , dense molecular clumps excites turbulent in the shock waves , amplifying the magnetic field . the cr electron rolloff energy is increased in spite of such amplified field , and we suggest that additional electron acceleration may be taking place in such turbulent regimes around the dense clumps . in near future , cherenkov telescope array ( cta ) will provide information on the spectral distribution of @xmath1-rays over the snr at 1 arcmin resolution and will allow us to investigate the spectra and acceleration of both the cr protons and electrons . in addition , the soft x - ray spectrometer ( sxs ) of astro - h will probe yet undetected thermal x - rays by sensitive high - energy resolution spectroscopy in rx j1713.7@xmath23946 . we can then probe the fraction of the energy of the snr blast waves used for cosmic - ray acceleration . and the hard x - ray imager ( hxi ) of astro - h will resolve photon index distribution at 10 kev or higher , possibly allowing us to constrain electron energy spectrum models by comparing that below 10 kev ( e.g. , * ? ? ? these future instruments will enable more accurate measurements of the efficient cosmic - ray acceleration . we summarize the present work as follows . 1 . we have estimated the spatial distribution of absorbing column density , photon index , and absorption - corrected flux ( 310 kev ) comparable to the scale of the ism distribution , a few arcmin , by using @xmath0 archival data with low background . 2 . the x - ray flux shows enhancement toward the dense ism . this is consistent with the shock - cloud interaction model by @xcite that x - rays become bright around the dense cloud cores due to turbulence amplification of magnetic field . 3 . photon index shows a large variation within the snr from @xmath13 = 2.12.9 . the photon index shows smallest values around the dense regions of cloud cores as well as toward diffuse regions with no molecular gas . this trend can be described as a rolloff energy variation of cr electrons @xcite . we present possible parameters to explain the variation of the rolloff energy under the dsa scheme . the enhanced intensity and harder spectra of the x - rays toward the dense clumps may require additional electron acceleration toward the dense clumps possibly via magnetic reconnection or other mechanism incorporating magnetic turbulence . 4 . the absorbing column density shows a good correlation with the visual extinction . we found that the southeast - rim identified by hi self - absorption , which shows enhanced vhe @xmath1-rays has a clear counterpart in the x - ray absorption too , lending new support to the hi self - absorption interpretation . nanten2 is an international collaboration of 10 universities ; nagoya university , osaka prefecture university , university of cologne , university of bonn , seoul national university , university of chile , university of new south wales , macquarie university , university of sydney , and university of eth zurich . this work was financially supported by a grant - in - aid for scientific research ( kakenhi , no . 24@xmath9310082 , no . 21253003 , no . 23403001 , no . 22540250 , no . 22244014 , no . 23740149 , no . 23740154 ( t.i . ) , no . 22740119 , and no . 24224005 ) from the ministry of education , culture , sports , science and technology of japan ( mext ) . this work was also financially supported by the young research overseas visits program for vitalizing brain circulation ( r2211 ) and the institutional program for young researcher overseas visits ( r29 ) by japan society for the promotion of science ( jsps ) and by the mitsubishi foundation and by the grant - in - aid for nagoya university global coe program , quest for fundamental principles in the universe : from particles to the solar system and the cosmos , from mext . this research made use of data obtained from data archives and transmission system ( darts ) , provided by center for science - satellite operation , and data archive ( c - soda ) at isas / jaxa . 99 abdo , a. a. , ackermann , m. , ajello , m. , et al . 2011 , , 734 , 28 acero , f. , ballet , j. , decourchelle , a. , et al . 2009 , , 505 , 157 aharonian , f. a. , akhperjanian , a. g. , aye , k .- , et al . 2004 , , 432 , 75 aharonian , f. , akhperjanian , a. g. , bazer - bachi , a. r. , et al . 2005 , , 437 , l7 aharonian , f. , akhperjanian , a. g. , bazer - bachi , a. r. , et al . 2006 , , 636 , 777 aharonian , f. , akhperjanian , a. g. , bazer - bachi , a. r. , et al . 2006 , , 449 , 223 aharonian , f. , akhperjanian , a. g. , bazer - bachi , a. r. , et al . 2007 , , 464 , 235 aharonian , f. , akhperjanian , a. g. , bazer - bachi , a. r. , et al . 2007 , , 661 , 236 bertsch , d. l. , dame , t. m. , fichtel , c. e. , et al . 1993 , , 416 , 587 berezhko , e. g. , & vlk , h. j. 2006 , , 451 , 981 berezhko , e. g. , & vlk , h. j. 2008 , , 492 , 695 berezhko , e. g. , & vlk , h. j. 2010 , , 511 , a34 butt , y. m. , torres , d. f. , combi , j. a. , dame , t. , & romero , g. e. 2001 , , 562 , l167 cassam - chena , g. , decourchelle , a. , ballet , j. , et al . 2004 , , 427 , 199 dickey , j. m. , & lockman , f. j. 1990 , , 28 , 215 dobashi , k. , uehara , h. , kandori , r. , et al . 2005 , , 57 , 1 ellison , d. c. 2001 , , 99 , 305 ellison , d. c. , & vladimirov , a. 2008 , , 673 , l47 ellison , d. c. , patnaude , d. j. , slane , p. , & raymond , j. 2010 , , 712 , 287 ellison , d. c. , slane , p. , patnaude , d. j. , & bykov , a. m. 2012 , , 744 , 39 enomoto , r. , tanimori , t. , naito , t. , et al . 2002 , , 416 , 823 fang , j. , zhang , l. , zhang , j. f. , tang , y. y. , & yu , h. 2009 , , 392 , 925 fermi , e. 1949 , physical review , 75 , 1169 fukui , y. , moriguchi , y. , tamura , k. , et al . 2003 , , 55 , l61 fukui , y. , sano , h. , sato , j. , et al . 2012 , , 746 , 82 fukui , y. 2013 , in astrophysics and space science proc . 34 , 2nd session of the sant cugat forum on astrophysics , ed . diego f. torres @xmath19 o. reimer ( berlin : springer ) , 249 gabici , s. , & aharonian , f. a. 2014 , , 445 , l70 giacalone , j. , & jokipii , j. r. 2007 , , 663 , l41 gorenstein , p. , harnden , f. r. , jr . , & tucker , w. h. 1974 , , 192 , 661 hiraga , j. s. , uchiyama , y. , takahashi , t. , & aharonian , f. a. 2005 , , 431 , 953 huang , c .- y . , park , s .- e . , pohl , m. , & daniels , c. d. 2007 , astroparticle physics , 27 , 429 hoshino , m. 2012 , physical review letters , 108 , 135003 h.e.s.s . collaboration , abramowski , a. , acero , f. , et al . 2011 , , 531 , a81 inoue , t. , yamazaki , r. , & inutsuka , s .- 2009 , , 695 , 825 inoue , t. , yamazaki , r. , inutsuka , s .- , & fukui , y. 2012 , , 744 , 71 ishisaki , y. , maeda , y. , fujimoto , r. , et al . 2007 , , 59 , 113 jenkins , e. b. , & savage , b. d. 1974 , , 187 , 243 taylor j. r. 1982 an introduction to error analysis ( 2nd ed . ; california : usb ) kishishita , t. , hiraga , j. , & uchiyama , y. 2013 , , 551 , a132 koyama , k. , kinugasa , k. , matsuzaki , k. , et al . 1997 , , 49 , l7 li , h. , liu , s. , & chen , y. 2011 , , 742 , l10 lucek , s. g. , & bell , a. r. 2000 , , 314 , 65 malkov , m. a. , & oc drury , l. 2001 , reports on progress in physics , 64 , 429 malkov , m. a. , diamond , p. h. , & sagdeev , r. z. 2005 , , 624 , l37 maxted , n. i. , rowell , g. p. , dawson , b. r. , et al . 2012 , , 422 , 2230 mcclure - griffiths , n. m. , dickey , j. m. , gaensler , b. m. , green , a. j. , haverkorn , m. , & strasser , s. 2005 , , 158 , 178 . moriguchi , y. , tamura , k. , tawara , y. , et al . 2005 , , 631 , 947 moraitis , k. , & mastichiadis , a. 2007 , , 462 , 173 morlino , g. , amato , e. , & blasi , p. 2009 , , 392 , 240 muraishi , h. , tanimori , t. , yanagita , s. , et al . 2000 , , 354 , l57 pannuti , t. g. , allen , g. e. , houck , j. c. , & sturner , s. j. 2003 , , 593 , 377 pfeffermann , e. , & aschenbach , b. 1996 , in roentgenstrahlung from the universe , mpe report 263 , ed . h. u. zimmermann , j. h. trmper , & h. yorke , 267 reimer , o. , & pohl , m. 2002 , , 390 , l43 sano , h. , sato , j. , horachi , h. , et al . 2010 , , 724 , 59 sano , h. , tanaka , t. , torii , k. , et al . 2013 , , 778 , 59 slane , p. , gaensler , b. m. , dame , t. m. , et al . 1999 , , 525 , 357 takahashi , t. , tanaka , t. , uchiyama , y. , et al . 2008 , , 60 , 131 tanaka , t. , uchiyama , y. , aharonian , f. a. , et al . 2008 , , 685 , 988 uchiyama , y. , aharonian , f. a. , & takahashi , t. 2003 , , 400 , 567 uchiyama , y. , aharonian , f. a. , takahashi , t. , et al . 2005 , high energy gamma - ray astronomy , 745 , 305 uchiyama , y. , aharonian , f. a. , tanaka , t. , takahashi , t. , & maeda , y. 2007 , , 449 , 576 wallerstein , g. , & silk , j. 1971 , , 170 , 289 wang , z. r. , qu , q .- y . , & chen , y. 1997 , , 318 , l59 weaver , r. , mccray , r. , castor , j. , shapiro , p. , & moore , r. 1977 , , 218 , 377 yamazaki , r. , ohira , y. , sawada , m. , & bamba , a. 2014 , research in astronomy and astrophysics , 14 , 165 yuan , q. , liu , s. , fan , z. , bi , x. , & fryer , c. l. 2011 , , 735 , 120 zhang , l. , & yang , c. y. 2011 , , 63 , 89 zirakashvili , v. n. , & aharonian , f. 2007 , , 465 , 695 zirakashvili , v. n. , & aharonian , f. a. 2010 , , 708 , 965	we have carried out a spectral analysis of the @xmath0 x - ray data in the 0.412 kev range toward the shell - type very - high - energy @xmath1-ray supernova remnant rx j1713.7@xmath23946 . the aims of this analysis are to estimate detailed x - rays spectral properties at a high angular resolution up to 2 arcmin , and to compare them with the interstellar gas . the x - ray spectrum is non - thermal and used to calculate absorbing column density , photon index , and absorption - corrected x - ray flux . the photon index varies significantly from 2.1 to 2.9 . it is shown that the x - ray intensity is well correlated with the photon index , especially in the west region , with a correlation coefficient of 0.81 . the x - ray intensity tends to increase with the averaged interstellar gas density while the dispersion is relatively large . the hardest spectra having the photon index less than 2.4 are found outside of the central 10 arcmin of the snr , from the north to the southeast ( @xmath3430 arcmin@xmath4 ) and from the southwest to the northwest ( @xmath3150 arcmin@xmath4 ) . the former region shows low interstellar gas density , while the latter high interstellar gas density . we present discussion for possible scenarios which explain the distribution of the photon index and its relationship with the interstellar gas .
let @xmath1 be the symmetric group of @xmath9:=\{1,2,\dots , n\}$ ] . for @xmath10 , define the _ inversion _ statistic @xmath11 by @xmath12 and the _ cycle _ statistic @xmath13 by @xmath14 we say a permutation statistic over @xmath1 is _ mahonian _ if it is equidistributed with @xmath11 , and is _ stirling _ if with @xmath15 . for example , it is well known that the _ right to left minimum _ statistic @xmath16 , defined by @xmath17 is stirling . recently petersen found a new mahonian statistic @xmath0 , called the _ sorting index _ ( see section 2 for definition ) , and proved the following : [ thm : petersen ] the pairs of statistics @xmath2 and @xmath3 have the same joint distribution over @xmath1 . also , we have @xmath18_q-1),\end{aligned}\ ] ] where @xmath19_q:=1+q+q^2+\dots + q^{r-1}$ ] . a combinatorial proof is found by chen et al . @xcite via a bijection @xmath20 such that @xmath21 where @xmath22 means @xmath23 . the bijection @xmath24 turns out to be of the form @xmath25 a composition of @xmath26-code and @xmath27-code introduced by foata and han @xcite . by defining the set - values statistics _ right - to - left minimum letters _ @xmath28 , _ left - to - right maximum places _ @xmath29 and the _ cycle set _ @xmath30 respectively by @xmath31 @xmath32 and @xmath33 foata and han derived the following set - valued joint equidistribution results . [ thm : foata_han ] the followings hold . 1 . for @xmath34 , we have @xmath35 2 . the set - valued statistics @xmath36 , @xmath37 , @xmath38 are symmetric and joint equidistributed over @xmath1 . the motivation of this work is to generalize above two theorems , to see if there is a set - valued version of petersen s result or other pairs of set - valued statistics having the same distribution la foata and han . it turns out that we can have them both . throughout the paper a statistic is set - valued if and only if the first letter is in capital . by introducing new set - valued statistics @xmath5 , @xmath6 , @xmath7 and @xmath8 ( see section 2 for definitions ) and corresponding ordinary statistics @xmath39 , @xmath40 , our first main theorem extends simultaneously both petersen and foata - han s results . [ thm : main ] we have : 1 . for @xmath41 , the following holds : @xmath42 2 . the quadruple statistics @xmath43 and @xmath44 have the same joint distribution over @xmath1 , and @xmath45_q+yq^{r-1}-1-q^{r-1}).\end{aligned}\ ] ] theorem [ thm : main ] generalizes theorem [ thm : petersen ] and theorem [ thm : foata_han ] ( 1 ) . we will see that theorem [ thm : foata_han ] ( 2 ) will be generalized in a later theorem . by abuse of terminology , a set - valued statistic is called mahonian ( or stirling ) if the corresponding ordinary statistic is so . our second result is to find triples of set - valued stirling statistics having the same joint distribution as @xmath46 and @xmath47 . by switching between certain set - valued statistics by applying invese , reverse , or complement operations on permutations , we obtain @xmath48 more ( and @xmath49 partial ) triples of set - valued statistics having the same joint distribution . see theorem [ thm : set_all ] for the complete list . from these we obtain two sets of symmetric and equidistributed pairs of set - valued statistics , the first of which includes those three pairs in theorem [ thm : foata_han ] ( 2 ) . the third part of the work is to derive quadruples of statistics which are joint equidistributed with @xmath50 and @xmath51 . note that the first statistic is mahonian and the others are stirling . again by switching among statistics , in theorem [ thm : sta_all ] we derive @xmath52 more ( and @xmath53 partial ) quadruples of statistics having the same joint distribution . the rest of the paper is organized as follows . definitions and preliminary results will be put in section [ sec : preli ] . in section [ sec : main_proof ] we prove theorem [ thm : main ] . section [ sec : triples ] is devoted to triples of set - valued statistics , and section [ sec : quadruples ] to quadruples of ordinary statistics . we first introduce the @xmath27- and @xmath26-code of foata and han @xcite , which are the key tools of this paper . given @xmath34 , define its _ lehmer code _ @xcite by @xmath54 where @xmath55 let @xmath56 . it is clear that @xmath57 is a bijection . @xmath27-code _ of a permutation @xmath58 is defined by @xmath59 for example , let @xmath60 . then @xmath61 and @xmath62 . @xmath26-code _ of @xmath58 is defined in the following way . for each @xmath63 , let @xmath64 be the smallest integer such that @xmath65 . define @xmath66 equivalently , @xmath67 can be determined from the cycle decomposition of @xmath58 . assume that @xmath68 appears in a cycle @xmath69 . if @xmath68 is the smallest element of @xmath69 , then set @xmath70 ; otherwise , choose @xmath71 to be the first element @xmath72 in @xmath69 with respect to the reverse direction such that @xmath73 . for example , let @xmath74 . then @xmath75 . by definition it is easy to see that @xmath26-code is a bijection from @xmath1 to @xmath76 . define the _ left - to - right minimum letters _ statistic @xmath6 , the set - valued _ left - to - right minimum places _ statistic @xmath77 and the _ left to right minimun _ statistic @xmath39 respectively by @xmath78 @xmath79 and @xmath80 it is easy to see that @xmath81 . for an integer sequence @xmath82 , define @xmath83 , the set of indices with values @xmath84 . it turns out that @xmath77 and @xmath6 correspond to lehmer code and @xmath27-code respectively . the proof of the following lemma is directly by definition . we have @xmath85 hence it is natural to consider the statistic corresponding to @xmath26-code . define the set - valued statistic @xmath8 by @xmath86 for example , @xmath87 . we have the following combinatorial interpretation of @xmath8 , which explains the somewhat awkward notation , standing for the _ left - to - right minimum of the shifted cycle containing @xmath84_. for @xmath34 , write the the cycle containing @xmath84 in the way that @xmath84 is at the end of the cycle and denote the resulting cycle @xmath88 . then @xmath89 by regarding @xmath88 as a word . by the definition of @xmath26-code , @xmath90 if and only if @xmath91 and all letters on the left of @xmath68 in @xmath88 are larger than @xmath68 . in other words , @xmath68 is a left - to - right minimum letter in @xmath88 . then we have @xmath92 . for the running example , @xmath93 and the shifted cycle is @xmath94 , hence @xmath95 . also , @xmath6 and @xmath8 are related via @xmath24 . [ lem : afterphi1 ] we have @xmath96 from , and the definition of @xmath24 , we have @xmath97 the goal of this subsection is to define and investigate the set - valued statistics @xmath7 and @xmath5 . first we need the concept of the _ induced set_. [ defi : induce_set ] given @xmath98 , define its _ induced set _ @xmath99 according to the following algorithm : 1 . set @xmath100 with the initial values @xmath101 and @xmath102 . 2 . for @xmath68 from @xmath103 down to @xmath84 do the followings : 1 . let @xmath104 be the @xmath105-th smallest element among @xmath106 , 2 . add ordered pairs @xmath107 into @xmath108 for those @xmath109 with @xmath110 . if there is no such @xmath72 then skip this step . 3 . delete @xmath104 from @xmath106 . 3 . define @xmath111 for example , for @xmath112 we have [ cols="^,^,^,<",options="header " , ] and the generating function with respect to the first three statistics in each quadruple is @xmath113_q+yq^{r-1}-1-q^{r-1}).\ ] ] observe that @xmath114 and @xmath115 for @xmath116 . the proof is done via the following mappings . @xmath117 from the list we can read off many pairs of symmetric and joint equidistributed statistics . in each of the following items , the pairs of statistics are symmetric and joint equidistributed over @xmath1 . 1 . @xmath118 , @xmath119 , @xmath120 , @xmath121 , @xmath122 , and @xmath123 . the generating function with respect to each pair is @xmath124 . 2 . @xmath125 , @xmath126 , @xmath127 , and @xmath128 . in this short paper we generalize simultaneously petersen and foata - han s results to more than two statistics and find many triples or quadruples of statistics having the same joint distribution over @xmath1 . in all quadruples , the first statistic is mahonian while the others are stirling , and we then read off many symmetric equidistributed pairs of stirling statistics . however , it is well known that there are pairs of mahonian statistics with a symmetric joint distribution as well @xcite , for examples , @xmath129 and @xmath130 are two of them . hence it would be interesting to generalize our results further to include more mahonian statistics . on the other hand , the generating function obtained in theorem [ thm : main ] only involves three of the four statistics , hence a natural question is to find a four - variable generating function including @xmath131 as well . we leave them to the interested reader .	recently petersen defined a new mahonian index @xmath0 over the symmetric group @xmath1 and proved that @xmath2 and @xmath3 have the same joint distribution . foata and han proved that the pairs of set - valued statistics @xmath4 have the same joint distribution over @xmath1 . in this paper we introduce the set - valued statistics @xmath5 , @xmath6 , @xmath7 and @xmath8 and generalize simultaneously results of petersen and foata - han and find many equidistributed triples of set - valued statistics and quadruples of statistics .
the general form of the cabbibo - kobayashi - maskawa ( ckm ) @xcite matrix can be described using the generalized wolfenstein parametrization ] @xcite as @xmath2 product of the first and the last column of the ckm matrix under the unitarity requirement yields the standard form of the unitarity triangle ( ut ) , presented in the complex @xmath3 plane in fig . [ fig : ut ] ( left ) are frequently denoted as @xmath4 and @xmath5 , respectively . ] . measurements of various observables , from which individual ckm elements can be determined , impose constraints on the sides and angles of the ut , shown in fig . [ fig : ut ] ( right ) . the overlap area determines an allowed region for the position of the ut apex . comparison of individual measurements , described below , enable a consistency check of the ckm description of quark mixing and can give hints of yet unobserved phenomena . the @xmath6 element of the ckm matrix has traditionally been obtained from the measured branching fractions of @xmath8 decays , the semileptonic decays of charged and neutral kaons @xmath9 . recently , a new measurement of the @xmath10 branching fraction in the electron decay mode was presented by the brookhaven e865 experiment @xcite . the result @xmath11 is about 2.3 standard deviations higher than the world average of previous measurements of this quantity @xcite . the measured branching ratio , where the last error comes from the uncertainty in the normalization decay modes used in the experiment , can be converted into the value of @xmath12 . the error on this value receives a significant contribution from the theoretical estimate of the @xmath13 decay form factor @xcite . using the value of @xmath14 , evaluated most precisely from the super - allowed nuclear fermi beta decays @xcite , and the above result for @xmath6 , the unitarity requirement in the first row of the ckm matrix is perfectly satisfied can be safely neglected . ] . a possible disagreement between the world averages of @xmath15 and @xmath16 will be addressed by the forthcoming precision results from kloe and na48 experiments . due to a reasonably good theoretical understanding and a satisfactory statistical power of the sample , the determination of @xmath17 and @xmath18 elements relies on the measurements of semileptonic @xmath19 meson decays , using either exclusive or inclusive methods . within the exclusive methods , the @xmath20 decay channel is the most prominent one for the @xmath21 transitions . decays into @xmath22 and @xmath23 mesons and a lepton pair were used to probe the @xmath24 transitions . inclusive methods comprise a determination of the total b meson semileptonic width combined with measurements of various moments of distributions that reduce the theoretical uncertainties entering the ckm element extraction . the differential decay rate for the @xmath20 can be expressed in terms of a four - velocity transfer @xmath25 using a single @xmath19 meson decay form factor @xmath26 , which coincides with the isgur - wise function up to the heavy quark symmetry breaking terms @xcite . experimentally , the efficiency corrected differential decay rate @xmath27 is measured and extrapolated to the maximum momentum transfer ( @xmath28 ) . measurement uncertainties are dominated by the systematic errors as seen in fig . [ fig : vcb_ex_sum ] ( left ) @xcite . at the @xmath29 energies , the @xmath30 resolution is influenced by a spread of @xmath19 meson momenta . another important source of the systematic error is the modeling of the @xmath31 background . at the @xmath32 energies the kinematic constraints enable a better @xmath30 resolution and an effective @xmath31 rejection , but the efficiency for the reconstruction of low momentum pions from @xmath33 falls rapidly at low values of @xmath30 . using the interval of values @xmath34 @xcite and the average value of @xmath35 one obtains @xmath36 where the first error represents the experimental and the second the theoretical uncertainty . the relation between the semileptonic decay width of @xmath19 mesons and the @xmath17 can be written as @xmath37 . while the average of measurements of the left hand side of equation contributes about 1% to the relative error on the @xmath17 , the theoretical factor @xmath38 introduces an error of about 5% @xcite . the solution is provided in the framework of operator product expansion , where the semileptonic width is expressed to order @xmath39 using two parameters and heavy quark masses . the same parameters enter the expressions for differential distributions of several observables like the energy of the lepton in semileptonic decays , the invariant mass of the produced hadronic system or the energy of the photon in radiative b meson decays @xcite . measurements of moments of those distributions enable the relevant parameters determination and hence help reducing the theoretical uncertainty in the extraction of @xmath17 from the measured semileptonic width . a summary of @xmath40 measurements at @xmath32 energies is shown in fig . [ fig : vcb_ex_sum ] ( right ) @xcite . the precision is almost identical to the measurements performed by the lep experiments , which benefit from the fact that the total lepton energy spectrum can be used in analyzes . combination of results together with the measured @xmath19 meson lifetime yields a value of @xmath41 mev @xcite . cleo and delphi experiments measured the moments ( up to the third ) of distributions mentioned above @xcite . taking as an input the value of the semileptonic width , @xmath17 evaluations @xcite can be summarized as @xcite @xmath42 the experimental error includes contributions from moments and semileptonic width measurements , while the theoretical error reflects the uncertainties due to the perturbative qcd and @xmath43 series truncation . the exclusive methods of @xmath18 determination consist of @xmath44 measurements over a limited interval of lepton pair invariant mass ( @xmath45 ) , lepton momentum or lepton energy , in order to suppress the dominant @xmath21 background . one or more dimensional fits to distributions of kinematical variables , separating the @xmath46 transitions from the background , make use of isospin relations between the charged and neutral @xmath47 and @xmath48 decay modes . in order to obtain the full @xmath49 , where @xmath50 represents a light meson , an extrapolation to the full range of fitted observables is necessary , introducing a model dependence . the results are shown in fig . [ fig : vub_ex_sum ] ( left ) . error bars mark the statistical uncertainty , while the total line represents the quadratic sum of statistical and experimental systematic errors ; for the latter the largest contributions arise from detector simulation used in neutrino reconstruction and @xmath51 background modeling . lines below the results show the theoretical uncertainties . they are obtained using various form factors , calculated from lattice qcd , light cone sum rules or quark models , with different methods and are thus not directly comparable . the belle experiment made a first measurement of @xmath52 @xcite , however the value of @xmath18 is not quoted . in order to reduce the theoretical uncertainty , the cleo experiment has performed a measurement of @xmath53 ( last line of fig . [ fig : vub_ex_sum ] ( left ) ) in distinct @xmath45 bins @xcite . the advantage of such an approach is a lower systematic error arising from the @xmath45 dependence of the efficiency , and the discriminating power of the @xmath45 distribution for different form factor models . for the exclusive @xmath18 determination no official average is provided by the heavy flavor averaging group @xcite up to date . using the first four results of fig . [ fig : vub_ex_sum ] ( left ) , and treating half of the systematic error as completely correlated ( due to @xmath51 background ) , one obtains the value @xmath54 illustrating the uncertainty dominated by the theory . the @xmath51 semileptonic width is sensitive to @xmath18 in an analog way as the @xmath55 is sensitive to @xmath17 . however , in isolating the @xmath24 transitions , one is restricted to a limited interval of kinematic variables , separating the signal from the major @xmath21 background . while stringent requirements on the lepton energy , hadronic or lepton pair invariant mass in semileptonic decays lead to a better purity of the selected sample , the selection of events in a vicinity of the phase space boundaries introduces large non - perturbative corrections to theoretical predictions and hence a larger theoretical error @xcite . a number of measurements using individual or several of the above mentioned observables for @xmath51 separation were reported . for the reconstruction of the invariant mass of the hadronic system ( @xmath56 ) , produced in a semileptonic @xmath19 meson decay at @xmath19 factories , one needs to separate the decay products of individual @xmath19 mesons . a large number of recorded @xmath19 decays enables the use of a sample , where one of the @xmath19 mesons is fully reconstructed through its @xmath21 decay modes . using this method , babar collaboration determines the yield of @xmath57 events at low values of @xmath56 @xcite . the corresponding @xmath18 value is shown in the last line of fig . [ fig : vub_ex_sum ] ( right ) . to separate decay products , belle collaboration uses a `` simulated annealing '' technique , based on maximization of a likelihood function for a correct and wrong assignment of particles , by iteratively exchanging particles between the two @xmath19 mesons @xcite . selection based on the reconstructed value of @xmath56 and @xmath45 yields the result marked in fig . [ fig : vub_ex_sum ] ( right ) as `` belle ( @xmath58 ) '' . similarly as in the case of the exclusive @xmath18 determination , uncertainties of all results shown are dominated by the theoretical error introduced by the extrapolation of @xmath51 yield to the full interval of kinematic observables . as an illustration , the average of babar , belle and cleo measurements @xcite using the lepton energy for @xmath24/@xmath21 separation , yields @xmath59 when the theoretical error is taken as completely correlated , while experimental systematic error is treated as uncorrelated . the @xmath62 value is determined from the ratio of @xmath63 and @xmath64 meson oscillation frequencies , @xmath65 and @xmath66 . the use of the ratio instead of the individual measurements is motivated by a partial cancellation of the theoretical uncertainties . while the world average of @xmath65 has a relative error of 1.2% @xcite and is completely dominated by the measurements performed at b factories @xcite , the @xmath67 limits steam from lep experiments and will until the lhc era remain in the domain of the tevatron collider . the amplitude method for the @xmath64 oscillations measurements consists of fitting the observed decay time distribution with the amplitude of oscillations as a free parameter . the fitted amplitudes at given values of @xmath66 are then converted into the lower limit on the oscillation frequency . the world average of these measurements is shown in fig . [ fig : delta_ms_sum ] ( left ) , yielding @xmath68 at 95% c.l . @xcite . taking the @xmath69 ratio and the lattice qcd results for @xmath70 @xcite , one arrives at the limit @xmath71 a detailed overview of @xmath72 measurements , most prominently in the @xmath73 decays , is given in @xcite . averaging results of babar and belle collaborations in all charmonium decay modes yields @xmath74 @xcite , a 7% relative precission result , where the error is still limited by statistics . an interesting feature evolves in the measurements of the cp violation using the @xmath75 decay mode with only penguin processes contributing to the amplitude . these theoretically clean decays are expected to yield a value of @xmath72 consistent with the above quoted result . however , the present situation reveals a discrepancy of the two values , the latter being @xmath76 @xcite . with an increased statistics one might expect these measurements to reveal new phenomena beyond the sm . the angle @xmath77 of the ut can be determined examining the @xmath78 decays @xcite . the amplitude for the process involves a tree digram as well as a significant contribution from the penguin processes , and hence the interpretation of the cp asymmetry measurements is complicated . while the belle result @xcite points to a direct cp violation in these processes ( fig . [ fig : delta_ms_sum ] ( right ) ) , the result of the babar collaboration @xcite is consistent with no direct cp violation . limits obtained on the angle @xmath77 from the belle measurement , taking into account a range of predictions for the ratio of penguin and tree amplitude contributions , are @xmath79 at 95.5% c.l . the ckm fitter group @xcite provides a fit to the measured observables and range of theoretical parameters in order to constrain the region in the @xmath80 plane . in the so called rfit approach one maximizes the @xmath81 , where measurements @xmath82 and theoretical predictions @xmath83 , depending on parameters @xmath84 , enter the experimental part @xmath85 . theoretical part @xmath86 equals unity if the set of parameters is within an allowed range of predictions and vanishes otherwise . using the set of inputs described at @xcite one obtains a 90% c.l . region in @xmath80 plane shown in fig . [ fig : ckm_fit ] ( left ) . at the present , measurements of @xmath72 , @xmath18 and @xmath67 all impose severe constraints on the position of the ut apex . [ fig : ckm_fit ] ( right ) includes results on @xmath72 and @xmath87 by the belle experiment @xcite , showing a consistency of measurements . * @xmath7 is measured with a relative precision of @xmath88 in the nuclear super - allowed fermi transitions . the measurement uncertainty is completely theoretically dominated . * @xmath6 : a new measurement of @xmath89 branching fraction by e865 experiment resolves the question of the unitarity in the first row of ckm matrix . a possible disagreement between the charged and neutral @xmath89 decay modes will be addressed in the future by kloe and na48 experiments . * @xmath17 , determined from @xmath90 decays , is limited by the uncertainty in the decay form factor normalization . inclusive determinations will be improved with new measurements of moments of @xmath91 , @xmath45 and @xmath56 distributions . * @xmath18 : the @xmath45 dependent measurements have started in exclusive decay channels . new inclusive measurements of @xmath92 using @xmath56 and @xmath45 variables will be performed by babar and belle experiments . theoretical ambiguities are expected to be resolved through tests of different models ( in exclusive channels ) and determination of moments of differential distributions ( @xmath93 transitions ) . * @xmath60 , @xmath61 : while results on @xmath65 are already very precise , further important constraints on ut are hoped for from @xmath67 measurements by d0 and cdf experiments . * @xmath72 has become a precision measurement , new phenomena might arise in the cp asymmetry measurements using the @xmath94 decay channel . * @xmath87 determinations have just started and although complicated , these measurements will in the future provide an interesting constraint on the ut . 99 m. kobayashi , t. maskawa , prog . . phys . * 49 * , 652 ( 1973 ) . l. wolfenstein , phys . . lett . * 51 * , 1945 ( 1983 ) ; a.j . buras _ et al . rev . d50 , 3433 ( 1994 ) ; m. schmidtler , k.r . schubert , z. phys . c53 , 347 ( 1992 ) . a. sher _ et al . _ , hep - ex/0305042 ( 2003 ) . k. hagiwara _ et al . rev . d66 , 010001 ( 2002 ) . v. cirigliano _ et al . j. c23 , 121 ( 2002 ) . hardy , i.s . towner , eur . j. a15 , 223 ( 2002 ) . i. caprini _ et al . _ , b530 , 153 ( 1998 ) . heavy flavor averaging group , http://www.slac.stanford.edu/xorg/hfag/ a.f . rev . d53 , 2491 ( 1996 ) . lett . * 87 * , 251807 ( 2001 ) ; d. cronin - hennessy _ et al . * 87 * 251808 ( 2001 ) ; a.h . rev . d67 072001 ( 2003 ) ; m. calvi _ et al . _ , hep - ex/0210046 ( 2002 ) . m. battaglia _ et al . lett . b556 , 41 ( 2003 ) ; c.w . rev . d67 , 054012 ( 2003 ) . a. stocchi , workshop on the ckm unitarity triangle , durham , uk , 2003 . et al . _ , hep - ex/0307075 ( 2003 ) . s.b . athar _ et al . _ , hep - ex/0304019 ( 2003 ) . m. luke , workshop on the ckm unitarity triangle , durham , uk , 2003 . d. del re , electroweak interactions and unified theories , moriond , france , 2003 . a. sugiyama , electroweak interactions and unified theories , moriond , france , 2003 . b. aubert _ et al . _ , hep - ex/0207081 ( 2002 ) ; k. abe _ et al . _ , contributed paper to lepton photon 2003 , batavia , usa , 2003 ; a. bornheim _ et al . _ , 88 * , 231803 ( 2002 ) . k. abe , xxiii physics in collision , zeuthen , germany , 2003 ; y. pan , ibido . l. lellouch , nucl . * 117 * , 127 ( 2003 ) . t. browder , lepton photon 2003 , batavia , usa , 2003 . k. abe _ et al . _ , * , 012001 ( 2003 ) . b. aubert _ lett . * 89 * , 281802 ( 2002 ) . h. h@xmath95cker _ et al . j. c21 , 225 ( 2001 ) ; http://ckmfitter.in2p3.fr/ h. sagawa , flavor physics and cp violation , paris , france ( 2003 ) .	an overview of experimental determinations of different ckm elements is given with an emphasis on @xmath0 and @xmath1 extraction . measurements are compared to the standard model predictions and constraints on the unitarity triangle , arising from combination of measurements , are presented . = 14.5pt
in the last two decades , developments in infrared ( ir ) technologies , such as @xmath2 arrays and the mid - ir _ spitzer space telescope _ , have enabled large - scale studies of star formation processes in molecular clouds ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? one major discovery from such surveys of nearby clouds is that a large fraction ( @xmath370%-90% ) of stars form in embedded clusters @xcite . the spatial distribution and hierarchical structure of embedded clusters in giant molecular clouds ( gmcs ) should reveal physical processes in cluster formation ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? it is particularly important to understand the influence of initial clusters on later star formation in adjacent cloud material . the expansion of the hii region around massive stars in young stellar clusters can , in principle , sequentially trigger formation of second - generation ob clusters , as described by @xcite ( see also * ? ? ? * ; * ? ? ? * ; * ? ? ? however , the operation of triggering in specific cases of ob associations in gmcs is often unclear . as one of the most massive galactic gmcs , the rosette molecular cloud ( rmc ) is ideal for studying the formation and possible triggering of embedded star clusters . with a total extent of @xmath3150 pc @xcite and @xmath4 of molecular gas @xcite , it lies only 1.4 kpc from the sun @xcite . * hereafter tfm03 ) and ( * ? ? ? * hereafter rl08 ) review past and present research on this well - studied and important star formation region . its orientation is perpendicular to the line - of - sight with a large , annular hii region at its tip known as the rosette nebula . this hii region is powered by the @xmath5 myr old ob cluster ngc 2244 . the molecular gas exhibits the strongly clumped structure characteristic of turbulent gmcs @xcite . ( * ? ? ? * henceforth , pl97 ) identified seven embedded clusters in the dense molecular cloud cores from stellar concentrations in near - ir ( nir ) @xmath2 images which are associated with iras mid - ir sources ( see figure 1a in paper i ) . they suggested that both triggered and spontaneous star formation may be contributing to the forming clusters in this cloud . the recent deeper flamingos nir imaging survey of rosette ( * ? ? ? * hereafter refl08 ) reveal four more clusters across the complex . they find an increasing fraction of younger stars ( identified from @xmath6-band excesses associated with hot circumstellar disks ) with increasing distance from ngc 2244 , implying a temporal sequence of star formation across the complex . four new clusters and over 700 stars with dusty disks are located in mid - ir maps obtained with the irac and mips detectors on the _ spitzer space telescope _ both the flamingos and @xmath7 studies argue that the embedded star formation is not principally produced by sequential triggering from the ngc 2244 ob cluster . lying 2@xmath8 from the galactic plane , ir imaging surveys in the rosette region are unavoidably contaminated by the overwhelming field star populations . therefore , identification of young embedded populations has relied on selecting stars with ir color excesses produced by dusty circumstellar disks . because the many diskfree pre - main sequence stars ( pms ) ( weak - lined t tauri stars , wtts ) that are cluster members can not be identified in this way , statistical subtraction of the contaminating field population is needed to estimate the full embedded population from ir images . even in very young clusters , the wtts may outnumber stars with prominent disks ( e.g. , * ? ? ? * ; * ? ? ? pms stars both with and without disks are well - known to emit x - rays that can penetrate the heavy extinction of gmcs @xcite . the level of x - ray emission from young stellar populations is elevated @xmath9 times above the levels of x - ray emission from the old galactic disk population @xcite . as a consequence , x - ray observations , particularly with the high spatial resolution _ chandra x - ray observatory _ , are highly efficient in selecting pms members in stellar clusters . @xmath10 images provide remarkably complete and unbiased pms populations with subarcsecond positions . x - ray surveys thus complement ir surveys in establishing the census of embedded pms populations in gmcs which in turn can address issues relating to triggered star formation . @xmath10 has made recent advances in the study of star formation triggering in the vicinity of hii regions produced by the cep ob3b , trumpler 37 ( ic 1396 ) , ngc 6618 ( m 17 ) , and trumpler 16 ( carina ) ob associations and a high latitude cluster @xcite . a previous x - ray study of the rmc was presented by @xcite based on an image from the _ rosat _ satellite . they report faint x - ray sources associated with embedded ir - emitting low mass star clusters . however the poor spatial resolution and soft energy band of _ rosat _ prevents identification of individual embedded stars . the diffuse and stellar emission from the central rosette nebula cluster ngc 2244 were studied in detail in tfm03 and paper i. we now examine a mosaic of four _ chandra _ images in the rmc , presented without detailed analysis in tfm03 , to identify the pms population and study the cluster formation process in molecular clouds , complementing the nir studies ( pl97 , rl08 , refl08 ) . figure 1 outlines the location of our images with respect to the molecular gas and dust emission and the nir clusters . section [ sec : obs.sec ] describes the observations with [ sec : counterpart.sec ] outlining the stellar counterparts of the x - ray sources . important individual stars are discussed in [ sec : previous ] . characterization of the embedded clusters , and non - clustered ( distributed ) population , follows in [ sec : cluster ] . in particular we discover a previously unidentified older cluster ( [ sec : xa ] ) and investigate clustered star formation in the particular environment of the rosette complex ( [ sec : distribsf ] ) . in [ sec : discuss ] we summarize our findings and discuss the implications for star formation modes in the rmc . the rmc was observed with the imaging array of _ chandra _ s advanced ccd imaging spectrometer ( acis - i ) in a sequence of 20 ks exposures obtained in january 2001 ( obsids 1875 - 77 ) . the observatory and instrument are described by @xcite . details on the data filtering , point source detection and extraction , variability analysis , and spectral fitting procedures using the _ acis extract _ ( ae ) software package @xcite are given in paper i. this paper also defines the quantities presented in the source tables below . the resulting rmc source list contains 395 x - ray detections which we designate rmcx 1 through 395 . we divide these into a primary list of 347 highly reliable sources ( table [ tbl : ch5_primary ] ) and a secondary list of 48 tentative sources ( table [ tbl : ch5_tentative ] ) with @xmath11 likelihood of being spurious background fluctuations based on poisson statistics . this approach has been adopted in our previous papers on _ chandra _ observations of star forming regions ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) ; it allows the reader to evaluate the validity of faint sources individually . table [ tbl : ch5_primary ] and table [ tbl : ch5_tentative ] have a format that is identical to tables 1 and 2 in paper i. a _ chandra _ mosaic of the rmc region ( 0.5 - 8 kev ) overlaid with source extraction regions ( calculated with ae ) is shown in figure [ fig : mosaic ] . the image here is shown at reduced resolution ( 2 per pixel ) . during the @xmath12 ks observations , eleven sources display significant variability ( @xmath13 , shown in column 15 of tables [ tbl : ch5_primary ] and [ tbl : ch5_tentative ] , where @xmath14 is the significance of the statistic for 1-sided kolmogorov - smirnov test comparing a uniform count rate model to the distribution of source event time stamps ) . these are associated with magnetic reconnection flares which are responsible for most of pms x - ray emission @xcite . only one of the variable sources ( source # 119 ) has more than 100 net counts due to the short exposure ; its light curve shows a slow 3-fold rise in emission over 2 hours followed by a decay over the subsequent 3 hours . such slow - rise x - ray flares are not infrequently seen from pms stars @xcite . spectral analysis results for 164 sources with photometric significance _ signif _ @xmath15 ( _ signif _ defined as the net counts value divided by the error on that value ; see table [ tbl : ch5_primary ] footnote ) are presented in table [ tbl : ch5_thermal_spectroscopy ] . the table entries are described in the table footnotes . acis spectral distributions were fit to thermal plasma models ( apec ; , * ? ? ? * ) with interstellar soft x - ray absorption ( wabs ; , * ? ? ? * ) convolved with the telescope mirror and detector spectral response . best - fit absorbing column densities range from negligible to @xmath16 @xmath17 , equivalent to a visual absorption of @xmath18 200 mag @xcite . temperatures range from @xmath19 kev to the hardest measurable by acis which we truncate at @xmath20 kev . the range of total band ( @xmath21 kev ) absorption corrected luminosities ( @xmath22 ) derived from spectral modeling is @xmath23 ergs s@xmath24 . assuming a 2 kev plasma temperature and an average @xmath25 mag visual extinction ( @xmath26 @xmath17 ) , the faintest on - axis detection in table [ tbl : ch5_tentative ] corresponds to a limiting luminosity of @xmath27 ergs s@xmath24 . note however that the patchy extinction throughout our fields will change the limiting sensitivity of the observation somewhat ( [ sec : sensitivity ] ) . the brightest source in the field is source # 119 with 369 acis counts in the 20 ks exposure . this count rate is not high enough to cause photon pile - up in the detector . the x - ray sources , which are generally located with precision better than 0.5 in the 2mass / hipparcos reference frame , are associated with counterparts at other wavelengths to establish their stellar properties . optical coverage of the rmc in the literature is poor because of the high visual extinction towards the molecular cloud and the contamination by foreground stars . few spectral types are measured . we therefore use optical and ir photometric catalogs including : whole - sky usno - b1.0 catalog ( * ? ? ? * hereafter usno ) ; 2mass all - sky catalog of point sources @xcite ; and the university of florida flamingos survey of giant molecular clouds ( pi : e. lada ) . the flamingos observations of the rosette complex fields and ir data reduction are described by romn - ziga ( 2006 ) and refl08 . we have also associated our sources with the coordinates of @xmath2850,000 stars seen in @xmath7 irac 3.6@xmath29 m images ( poulton et al . 2008 , private communication ) . positional coincidence criteria are used to associate acis x - ray sources with optical and nir ( onir ) sources as described in the appendix of @xcite . associations between acis sources and onir sources are reported in table [ tbl : ch5_counterparts ] . of the 395 acis sources in the rmc , 299 ( 76% ) have an onir counterpart identified . @xmath2 magnitudes from flamingos photometry are reported if available ; we adopt 2mass photometry for six bright stars ( rmcx # 25 , 65 , 164 , 185 , 209 , 342 ) that are saturated in flamingos ( @xmath30 mag , romn - ziga 2006 ) . nir counterparts of 279 acis sources have measurements in all three @xmath2 bands . the simbad and vizier catalog services were examined for complementary information ; these results are reported in the table footnotes . useful sources include the tass mark iv photometric survey of the northern sky @xcite and two large optical surveys of the rosette @xcite that partly covered our _ chandra _ fields . source # 65 is too bright ( @xmath31 ) for its spectral type ( g0 ) at the rmc s distance , leading to its classification as a foreground star . the rest of the stars with known spectral types show colors and magnitudes consistent with being intermediate- to high - mass cloud members . we have examined the spatial distribution of the 96 x - ray sources that do not have associated onir counterparts in the context of digital sky survey optical plates and co emission maps @xcite . over 20 of these sources are located in the brightest nebulosity of the photodissociation region ( pdr ) of the rosette nebula , where source detection becomes extremely difficult at onir wavelengths . we suspect that most of these are cloud members . about 50 others are concentrated in regions where the dense molecular cores are located and show x - rays with harder median energy ( @xmath32 ) . most of these are probably heavily obscured new members of the embedded clusters , although a few may be extragalactic sources coincidentally lying behind the embedded clusters . only a small fraction of the x - ray sources , perhaps 10 of the 96 that do not have onir counterparts , seem to be randomly distributed across the fields with a surface density consistent with the extragalactic background of active galactic nuclei . extragalactic contaminants are seen in all @xmath10 fields of young stellar clusters . considering the contamination analysis of these earlier studies , the absorption by the rmc gas , and the short 20 ks exposures considered here , we estimate a contamination of @xmath33% from extragalactic sources and @xmath33% from older galactic field star populations among the 395 acis sources , and even fewer among the sources with onir counterparts listed in table [ tbl : ch5_counterparts ] . the sensitivity limits of our x - ray observations are spatially uneven due to the patchy absorption in the rmc and to degraded point spread function ( psf ) of the _ chandra _ mirrors at large off - axis angles . based on previous experience with wavelet source detection and the acis extract software @xcite , we estimate that the far - off - axis sensitivity is approximately 2 times worse than on - axis ( see [ sec : new_sensitivity ] ) . sources with 8 or more net counts would be detected anywhere in the three fields ; over 60% of the sources in table [ tbl : ch5_primary ] lie in this complete regime . the typical extinction is @xmath34 in the rmc regions with molecular concentrations and @xmath35 in less obscured regions ( refl08 ) . assuming a @xmath36 kev thermal plasma as the spectral model of a typical pms star @xcite , the absorption - corrected luminosity limit of 8 acis counts in 20 ks in the hard band ( 28 kev ) is @xmath37 erg s@xmath24 in obscured regions and @xmath38 erg s@xmath24 in less obscured regions of the rmc ( the observed luminosity in the 0.58 kev band @xmath39 erg s@xmath24 and @xmath40 erg s@xmath24 , respectively ) . due to the strong empirical correlation between @xmath41 and pms stellar mass @xcite , these luminosity limits correspond to a mass limit around @xmath42 . adopting the x - ray luminosity function ( xlf ) of the _ chandra _ orion ultradeep project ( coup ) observation of the orion nebula cluster as a calibrator for other pms stellar populations , we can estimate the completeness of our rmc observations with respect to the full underlying pms population . using the xlf of @xcite , we find that the rmc observations probe @xmath12% ( @xmath43% ) of the pms population down to the stellar limit of @xmath44 m@xmath45 in the obscured ( lightly obscured ) regions of the rmc . these fractions are adopted to obtain estimates of the total stellar populations in various rmc structures below ( [ sec : cluster ] and table [ tbl : cls_counts ] ) . the only previously known o star in our fields , hd 46485 ( = bd+04@xmath461318 , o7v ; * ? ? ? * ; * ? ? ? * ) is detected with 310 counts in the 20 ks exposure ( rmcx # 164 ) . a single - temperature thermal plasma model gives an acceptable fit to its x - ray spectrum with @xmath47 kev and a low absorbing column @xmath48 @xmath17 . the fit can be improved with a two - temperature thermal plasma model ( figure [ fig : spectra ] ) , with a soft component ( @xmath49 kev ) , a second harder component ( @xmath50 kev ) , and an absorbing column of @xmath51 @xmath17 ( @xmath52 ) . the soft plasma and the derived x - ray luminosity , @xmath53 erg s@xmath24 , are similar to the ngc 2244 o stars ( e.g. , hd 46223 , hd 46202 ; paper i ) , consistent with being associated with instabilities in the radiatively accelerated wind . note that some o stars exhibit a much harder x - ray component ; for example , @xmath54 ori c ( o7v ) ionizing the orion nebula has a component with @xmath55 kev ( e.g. * ? ? ? this hard component is attributed to a magnetically channeled wind shock ( mcws ) . these differences in the x - ray properties of o stars of similar spectral types may be attributed to the various magnetic field strength in different stars ; in this case , hd 46485 likely does not have a strong field . measurements of magnetic properties of massive stars ( e.g. , * ? ? ? * ; * ? ? ? * ) will be able to further evaluate the link between x - ray emission and the presence of magnetic fields in ob stars . hd 46485 does not appear to have any relationship to the embedded clusters of the rmc . it is much less absorbed and lies outside all of the clusters discussed here or in the infrared literature . the reported proper motion of hd 46485 is small ( @xmath56 = -0.26 mas / yr , @xmath57 = -1.55 mas / yr , the hipparcos catalogue ; * ? ? ? * ) , and the radial velocity is @xmath58 km s@xmath24 @xcite . it might be a runaway o star from ngc 2244 , or a somewhat older member of the open clusters in the larger ob association @xcite . this star in pl97 s cluster # 2 ( pl2 ) exhibits unusually red colors in the nir , with @xmath59 and @xmath60 . these colors imply both unusually strong reddening ( @xmath61 ) and a strong @xmath6-band excess . it is also a bright msx point source ( msx6c g206.7804 - 01.9395 ) . the @xmath10 source has only 21 net counts but the x - ray properties are all consistent with a highly absorbed source : the median photon energy is very hard ( 4.6 kev ) which , based on the orion nebula calibration between @xmath32 and @xmath62 @xcite , corresponds to absorption @xmath63 @xmath17 or @xmath64 . the observed x - ray luminosity in the hard band ( @xmath65 kev ) is @xmath66 erg s@xmath24 . while not impossibly high for a lower mass pms star , this is at the top @xmath33% of the pms x - ray luminosity function @xcite . among protostars , only one of several dozen class i in the orion nebula region flares up to a comparable x - ray level ( e.g. , coup # 554 in omc 1s , * ? ? ? * ; * ? ? ? although the evolutionary stage of pms stars can not be directly established from hard - band x - ray properties because magnetic reconnection flares are present throughout the class i - iii phases , our past experience suggests that sources with absorption exceeding @xmath67 @xmath17 are protostars where most of the absorption arises in local envelopes rather than the larger - scale molecular cloud @xcite . a protostellar classification is also indicated by the rapidly rising flux ( high @xmath68 @xmath29 m flux ratio ) and roughly flat @xmath69 @xmath29 m spectral energy distribution measured with the @xmath7 and @xmath70 satellites @xcite . rmcx # 89 = iras 06306 + 0437 is also a massive system . with a dereddened @xmath6 magnitude around @xmath71 , it has a bolometric luminosity associated with an early b or late o star ( see the nir color magnitude diagram in paper i ) . as o stars generally have softer x - ray spectra than pms stars , the correction from hard band to the total @xmath21 kev band is larger ; for the o6-o7 star @xmath54 c ori in the orion nebula cluster , this correction is a factor of @xmath72 . if we assume the same correction , the estimated absorption - corrected total band x - ray luminosity is @xmath73 erg s@xmath24 for rmcx # 89 . this source is similar to a candidate massive yso found in the @xmath10 observation of the pismis 24 cluster in the ngc 6357 star forming complex @xcite . in fact , iras 06306 + 0437 is a known water maser source @xcite . it was included in a cs(2 - 1 ) survey of iras sources with far - ir color characteristics of ultra compact hii regions , but not detected @xcite . @xcite observed iras 06306 + 0437 among other water masers not associated with known hii regions or low mass ysos and detected nh@xmath74(1,1 ) emission coincident with the iras / msx infrared peak . they suggested the ammonia core may be a high mass protostellar candidate . the @xmath70 catalog notes that the mir source is extended . we examine the dusty environment of iras 06306 + 0437 in figure [ fig : source89 ] which we extracted from the archived irac and mips data from the _ spitzer _ satellite . the star lies within a bright region of heated dust surrounded by an elongated complex of infrared filaments extending over several arcminutes which is probably associated with the larger photodissociation region of the rosette nebula . a smaller - scale view of the dust emission around iras 06306 + 0437 is shown by @xcite . they note that this star does not have an accompanying concentration of young class i or ii stellar companions ; i.e. that pl 2 is not a true stellar cluster . figure [ fig : source89 ] shows 25 x - ray selected stars within 1 pc of rmcx # 89 . we have classified these as class i , ii or iii based on a preliminary examination of their nir colors , including two other class i protostars , three class ii systems , and @xmath12 class iii systems . this is consistent with the concentration of nir @xmath6-band excess stars found in pl 2 ( figure 7 in refl08 ) . another candidate protostar , rmcx # 72 ( @xmath75 ) , is close to rmcx # 89 , lying @xmath76 to the east in a secondary dust condensation . this further indicates the youth of this region . we thus conclude that rmcx # 89 is a massive proto - o star lying in a sparse cluster of pms stars , most of which are older class iii systems . the implied age offset @xmath77 that the central o star is younger than the cluster of lower - mass pms stars @xmath77 is similar to that recently inferred from @xmath10 study of the embedded w3 main cluster @xcite . @xcite list the five brightest @xmath6-band sources in the pl4 region . the two bright stars bg / irs1 and bg / irs2 ( acronyms used in block et al . 1993 = bg ) form a close pair whose spectra show br@xmath78 and are likely herbig ae / be ( haebe ) stars @xcite . rmcx # 209 and # 207 are matched to bg / irs2 and bg / irs4 , respectively . both have @xmath350 counts and the spectral fits yield @xmath55 kev with absorption @xmath79 @xmath17 for rmcx # 207 and @xmath80 @xmath17 for rmcx # 209 . this absorption column ( @xmath81 ) agrees with the @xmath82 mag derived by @xcite . the absorption corrected luminosities are @xmath83 erg s@xmath24 . their hard x - ray spectra are consistent with the high x - ray temperatures ( @xmath320 mk ) seen in x - ray surveys of haebe stars @xcite . rmcx # 210 coincides with bg / irs 5 but has too few counts for spectral fitting . the brightest x - ray source in our @xmath10 fields is rmcx # 119 with 369 acis counts . the x - ray spectrum can be roughly fit with a standard two - temperature thermal plasma model but there seems to be some abundance anomaly around 1 kev . the derived x - ray luminosity is @xmath84 erg s@xmath24 , too high for most lower - mass pms stars but consistent with the range of haebe systems . the stellar counterpart is bright , with @xmath85 , and was found not to have strong h@xmath86 emission in a follow - up study of @xmath87 sources in the rosette region @xcite . its location in the nir color - magnitude diagram indicates an intermediate mass star , although possibly with weak or absent accretion as suggested by the weak h@xmath86 emission @xcite . five features of excited gas seen in the optical band [ sii ] survey by @xcite lie in our @xmath10 field . these are probably herbig - haro outflows from protostars . we have carefully examined the x - ray fields in the vicinity of these sources . in a region 6 southwest of pl 2 , the x - ray emitting star rmcx # 51 is found near the extended [ sii ] feature rmc - f . its ir counterpart is a heavily reddened star with @xmath88 , which has no significant @xmath6-band excess . this star may be responsible for the [ sii ] emission feature . rmc - h is an arc - shaped feature of excited gas outlining a globule of gas northwest of pl 2 with a possible hh outflow originating in the globule . three x - ray sources are seen coincidently tracing the edge of the globule . one of them is the class i protostar rmcx # 58 which may be powering the jet - like outflow . @xcite suggests that rmc - l is an hh outflow driven by a star in pl 4 associated with iras 06314 + 0427 . in this context , the x - ray detected class i protostar rmcx # 315 may be the exciting source as it lies eastwards on the axis of symmetry of the bow - shock feature . the 3.5 pc separation between the star and shock is rather large , but comparable large - scale collimated flows have been detected in other star forming regions ( e.g. , hh80/81 , @xmath89 pc long ; marti et al.1993 ) . other [ sii ] features such as rmc - e , rmc - g , and rmc - m do not have an obvious correlation with our x - ray sources . visual examination of the spatial distribution of x - ray sources in figure [ fig : mosaic ] shows a non - random clustering . this is also evident in the earlier presentation of the field shown in figure 6a of tfm03 . we identify the x - ray source concentrations , which can be considered to be ` star clusters , ' using two methods in [ sec : smooth]-[sec : neighbor ] . the newly identified xa cluster is describe in [ sec : xa ] and the previously established embedded xb and xc clusters are discussed in [ sec : xbxc ] . tables [ tbl : cls_comparison]-[tbl : cls_counts ] provide quantitative results . @xcite and @xcite found that smoothed maps of the stellar distribution derived from @xmath10 source lists are effective in revealing the structure of young stellar clusters . for example , in m 17 , the x - ray stellar density maps highlighted the region s complex structure , including the compact ngc 6618 cluster , a triggered population along the edge of the hii region , several small embedded clusters , and a previously unknown subcluster . a similar approach applied to 2mass star distributions revealed the ngc 2237 subcluster on the western side of the rosette nebula @xcite . compared to infrared maps , x - ray maps are less affected by contamination from galactic field stars and show richer populations than disky ir - excess samples . smoothed maps of the @xmath10 sources in the rmc are shown in figure [ fig : ch5_ssd ] . the three acis fields have closely matched 20 ks exposures , only a small sensitivity bias is present between the on - axis and off - axis areas of each field , and contamination by extraneous sources is small ( [ sec : counterpart.sec ] ) . the three rows of panels in figure [ fig : ch5_ssd ] show the total source sample , lightly obscured sources with @xmath90 kev , and heavily obscured sources with @xmath91 kev . the left and right panels show maps with 3 and 2 radius ( 1.2 and 0.8 pc ) smoothing kernels , respectively . three large - scale structures are seen as overdensities in the total sample , irrespective of the adopted smoothing kernel size . we define these regions as xa , xb , and xc and trace their outline from a contour of the 3 kernel . we note seven small - scale overdensities within the three major structures in the 2 map of the total population . we label these xa1-xa3 , xb1-xb3 , and xc1 . in addition , a distinct substructure of heavily obscured stars extending southward from xc1 is seen in panel figure [ fig : ch5_ssd]@xmath92 which we call xc2 . note that the substructures are defined from smoothed contours on different maps and are not contiguous ; the star totals in the main structures are thus larger than the sum of the stars in the substructures . we label stars lying outside all overdensities `` distributed '' stars in the rmc . while the regions are defined based on subjectively chosen surface density contours , we are confident that these structures are not artifacts caused by the slightly degraded x - ray sensitivities at large off - axis angles on the acis - i ccds . the overdensities are not located at the centers of the fields ( where the telescope response is maximized ) and represent a factor of @xmath93 difference in stellar density which can not be solely attributed to the instrumental sensitivity gradients . a second technique for identifying stellar clusters is analysis of the distribution of nearest neighbors @xcite . this was used by refl08 to establish the content of the seven pl97 clusters in the @xmath2 catalog of the rmc . we apply the same nearest neighbor method used by refl08 based on the local density derived from the distances to the 10 nearest neighbors to each star . contours of the resulting nearest neighbor maps for our _ chandra _ rmc observations are shown in figure [ fig : chandra_contour ] for the total , unobscured , and obscured populations . the locations and morphologies of the resulting concentrations closely resemble the structures that we defined using the smoothed surface density maps for x - ray sources . this indicates the robustness of the inferred clustering irrespective of the methodology used to smooth the star distribution . the maps shown in figure [ fig : ch5_ssd ] used all detected x - ray sources . however , the sensitivity across the field is not uniform ; there is a bias towards detecting faint sources in the area close to the individual field aim points and in the overlapping regions with deeper exposure . it is possible that this sensitivity variations can create spurious structure in an intrinsically smooth distribution of sources . we evaluate the importance of this effect by creating a smoothed source map similar to figure [ fig : ch5_ssd ] which removes the sensitivity degradation due to telescope vignetting , effective exposure and background across the fields . following @xcite and @xcite , we estimate the sensitivity across the x - ray imaging mosaic assuming a poisson model . the resulting relation can be approximately represented by @xmath94 , where @xmath95 is the required number of counts for a source required to be detected , and @xmath96 is the background counts measured in a local region . the coefficients are obtained by constructing a lower envelope in the plot of _ net full _ counts against _ bkgd full _ counts tabulated in table [ tbl : ch5_primary ] . we then construct a sensitivity map across the mosaic by dividing a smoothed map of @xmath95 by the telescope exposure maps in units of @xmath97 s ( shown in figure [ fig : expmap ] , see also appendix of tfm03 ) . the resulting sensitivity map is typically @xmath98 counts @xmath17 s@xmath24 near the aimpoint and @xmath99 near the edge of a typical 20 ks field . it reaches @xmath100 in a small region around subcluster xa1 where the field overlaps the much deeper @xmath3100 ks exposure of ngc 2244 discussed in paper i. this corresponds to a @xmath38 counts limit across the 20 ks exposures and @xmath340 counts in the overlapping region around xa1 . applying this spatially varying counts threshold removes the sensitivity bias from telescope degradation and overlapping fields , and provides a complete sub - sample of our x - ray detections . this procedure removes 30% of the weaker sources in table 1 ( the remaining sources with number of counts above the detection threshold are marked in table 1 , see table footnote ) . the smoothed stellar surface density map of this reduced , spatially uniform sample is shown in figure [ fig : uniform_contour ] . the three major x - ray clusters , xa , xb , and xc , are present as prominent overdensities . some substructures are less significant than in the full sample map ; xa2 and xb1 are weakest with densities @xmath101 times above the background level . additional possible substructures emerge : an enhancement at ( @xmath102 , @xmath103 ) east of xa3 and at ( @xmath104 , @xmath105 ) northwest of xc1 . the reality of these features must be confirmed with deeper x - ray observations . the three main stellar enhancements and seven secondary enhancements are listed in table [ tbl : cls_comparison ] with approximate sizes . most are around 3 or 1.2 pc in extent . four of the seven substructures @xmath77 xb2 , xb3 , xc1 , and xc2 @xmath77 are associated with clusters identified from @xmath2 and @xmath7 images ( pl97 , rl08 , poulton et al . figure [ fig : co_ir ] shows the outline of the three main structures on the molecular co map of @xcite . it shows that these same four substructures xb2 , xb3 , xc1 , and xc2are associated with dense molecular regions . x - ray cluster xa , on the other hand , lies in the evacuated region on the eastern side of the h@xmath86-bright rosette pdr . located in the north - west side of the north - westernmost acis field where molecular material has been removed by the expanding hii region , its stars are less obscured compared to the xb and xc concentrations embedded in the molecular cloud . this is shown in columns @xmath106 of table [ tbl : cls_counts ] : 30% of xa stars have @xmath107 kev , corresponding to absorption @xmath108 @xmath17 or @xmath109 , compared to @xmath110% in concentrations xb and xc . most xa stars are concentrated in subcluster xa3 centered at ( @xmath86 , @xmath111 ) = ( 06:32:38,+04:46:18 ) . xa1 and xa2 represent small extensions to the northwest and southwest , respectively . sixty - one stars are in the xa x - ray sample . scaling by the incompleteness factor for lightly obscured regions given in [ sec : sensitivity ] , the estimated total stellar population in xa is 300 stars . this is about 15% of the total population estimated for the central ngc 2244 cluster using the same xlf - based method ( @xmath32000 ; paper i ) . the lightly absorbed xa cluster was not clearly identified in previous ir surveys of the cloud . the location of substructure xa1 is separated by @xmath12 from iras 06297 + 0453 . however a close inspection of iras , dss , and 2mass images led us to conclude that this is a spatial coincidence : iras 06297 + 0453 is associated with the very bright foreground k0v star bd + 04 1304 ( @xmath112 , @xmath113 ) and does not belong to the rmc . the xa cluster was not covered in the earlier pl97 survey . six stars with @xmath6-band excesses are present in the flamingos survey ( table [ tbl : cls_counts ] ) , insufficient to highlight a cluster using the 10th - nearest neighbor method . however , it does appear as an enhancement in the smoothed surface density map of all 2mass sources shown by ( * ? ? ? * figures 1a and 2a ) , appearing as a stellar enhancement between xb and the main ngc 2244 cluster . a group of 32 stars with disky mir spectral energy distributions is found within our xa contour in the @xmath7 irac / mips , but again these were not identified as a distinct cluster by ( * ? ? ? * see their figures 6b ) . rather they appear as an eastward extension of the central ngc 2244 cluster . the cluster is more spatially distinct from the main cluster in the acis x - ray mosaic of @xmath320ks pointings towards ngc 2244 and rmc ( see figure 6a in tfm03 ) , which is not sensitive to obscuration . the reason that the xa cluster emerges more clearly in our x - ray survey than in infrared surveys is probably due to presence of nebular contamination . it lies on a bright ridge of enhanced h brackett-@xmath86 and spatially variable pah emission ( figures 1 and 4 of * ? ? ? these contaminants limit the detection or photometric accuracy of stars in certain infrared bands which then inhibits identification of disk emission from multiband spectral energy distributions . as noted above and in table [ tbl : cls_comparison ] , the xb and xc concentrations of x - ray stars shown in figures [ fig : ch5_ssd ] and [ fig : chandra_contour ] correspond to embedded clusters established from @xmath70 , 2mass , flamingos , and @xmath7 studies . their stellar populations inferred from nir studies have been most thoroughly described by rl08 . the xb and xc clusters have higher fractions of obscured stars than the xa cluster ( 46% and 41% @xmath114 30% ; table [ tbl : cls_counts ] ) . this , combined with their spatial relationship to dense molecular structures ( figure [ fig : co_ir ] ) confirms that they are at least partly embedded in cloud material . the main advantage of our x - ray survey is that disk - free ( pms class iii ) members are readily detected , subject only to the sensitivity limitations given in [ sec : sensitivity ] . in most infrared surveys , class iii systems can not be readily distinguished from the strong contamination of galactic field stars . infrared surveys are thus often limited to counting disky ( pms class i and ii ) systems . the x - ray samples are thus usually larger than the ir - excess samples , and provide a less biased way to estimate disk fractions which is relevant to stellar ages . the xb cluster is the sparsest of the three x - ray concentrations with 52 detected stars and an estimated total population of 200 stars ( table [ tbl : cls_counts ] ) . it has an `` l''-shaped structure with three subcomponents of similar ( @xmath33 pc ) sizes . subcluster xb2 associated with @xmath7 cluster c @xcite appears more embedded than the other subclusters . subcluster xb3 is associated with nir cluster pl 2 and mir source iras 06306 + 0437 . as discussed in [ sec : rmcx89 ] above , this structure is dominated by the embedded massive star , rmcx # 89 . both xb1 and xb2 contain bright msx mid - ir sources ( see footnotes to table [ tbl : ch5_counterparts ] ) . the xc cluster is the richest of the three main x - ray structures in the cloud with 160 detected stars . in the principal concentration of stars , xc1 , roughly 2/3 are lightly obscured and 1/3 are heavily obscured . this rich cluster was first noted in @xmath115-band imaging by @xcite , studied spectroscopically in @xcite , and noted as a nir cluster by pl97 , @xcite , and @xcite . the sparser xc2 component is only evident in the stellar surface density maps for the obscured population ( figure [ fig : ch5_ssd]@xmath116 and @xmath92 ) indicating it is deeply embedded . this region also appears much darker in the dss optical image . it appears as a curved southern extension of the rich xc1 cluster in the x - ray maps , and is associated with the nir clusters pl 5 , refl 8 , and the @xmath7 cluster e ( table [ tbl : cls_comparison ] ) . the stellar structure may extend beyond the southeastern edge of our acis field . in contrast to xc1 , only 1/4 of the stars in xc2 are lightly obscured as measured by the @xmath32 indicator of soft x - ray absorption . xc2 is thus the most deeply embedded stellar cluster in the rmc regions we examine here . x - ray properties can not discriminate very young clusters , dominated by accreting class i - ii systems , from older clusters , dominated by non - accreting class iii systems . the reason for this is well - established : the x - ray luminosity function and flaring properties do not greatly change as low mass stars evolve from their class i through iii phases ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? x - ray selected samples are thus relatively unbiased with respect to age , and their ages can then be estimated from their rapidly evolving infrared properties . here we focus on the high - quality @xmath2 photometry available from the flamingos survey ( refl08 ) for x - ray sources within the boundaries of clusters xa , xb , and xc presented above . figure [ fig : ch5_ccd ] shows the @xmath117 vs. @xmath118 color - color diagrams for the three clusters . the region between the left two dashed lines is associated with class iii objects ( diskless wtts ) reddened by interstellar extinction . to the right of this band are sources exhibiting significant @xmath6-band excess ; here we require @xmath119 based on flamingos photometric errors for the sources to be k - band excess stars . the region between the middle and right - most dashed lines is occupied by class ii objects ( pms stars with circumstellar accretion disks ) , and protostars still possessing thick envelopes ( class i objects ) lie beyond the right - most dashed line . star counts in various regions are summarized in the middle columns of table [ tbl : cls_counts ] . the colors of most sources in cluster xa are consistent with pms stars with small reddening , @xmath120 , typical of stars in the central ngc 2244 cluster illuminating the rosette nebula ( paper i ) . a few class ii sources are present showing higher @xmath121 ; these may lie within xa or behind xa in the molecular cloud . sources in cluster xb show a wide range of absorptions ; some sources are reddened with @xmath122 and the @xmath6-excess sources are among the most reddened . this may suggest the xb subclusters are distributed along the line of sight , at different cloud depths . three class i sources are found in this region . despite the smaller number of sources in region xb , it has more ir - excess sources than the more populous xa region . subcluster xb1 has 11 detected x - ray stars with an inferred total population around 40 stars , but has no ir - excess systems . xb2 has more mir - excess than nir - excess stars , while xb3 has more nir - excess than mir - excess stars . we thus tentatively infer that the xb region is not homogeneous and coeval . subcluster xb3 , which includes rmcx # 89 = iras 06306 + 0437 discussed in [ sec : rmcx89 ] , is clearly the youngest structure in the region as the presence of nir excess in the stars indicates inner disks have not been cleared yet . x - ray sources in cluster xc are mostly lightly obscured pms stars with @xmath123 and a dozen heavily obscured sources with @xmath124 . the xc cluster is very populous in the x - ray map , and only 5 - 10% of these stars have infrared excesses in the nir or mir bands . this , together with the presence of only one very young system in figure [ fig : co_ir ] , indicates that the region is not dominated by active star formation today despite the presence of two @xmath70 sources resolved into several ir clusters ( pl 4 , pl 5 , refl 8 and @xmath7 s e ) . the very faint xc source rmcx # 315 is associated with a faint nir source with unusual colors , @xmath125 ; it is either a class i system or a very heavily reddened class ii system . nir @xmath126 vs. @xmath117 color - magnitude diagrams shown in figure [ fig : ch5_cmd ] provide an approximate range of the mass distribution and local absorptions for the same stars shown in figure [ fig : ch5_ccd ] . no candidate new ob stars appear in the rmc nir color - magnitude diagram . this stands in marked contrast to @xmath10 studies of other prominent star forming regions where dozens of new obscured ob stars were identified by virtue of their x - ray emission . these include westerlund 1 @xcite , and rcw 38 @xcite , ngc 6357 @xcite , rcw 108 @xcite . there is no reason to believe that massive star formation is suppressed in the rmc ; more likely the population of the individual embedded cluster is simply too small to contain massive stars as expected from the initial mass function . we conclude that cluster xa , showing less absorption and very few ir - excess stars , appears older than clusters xb and xc . the latter structures are not unified , homogeneous clusters but rather collections of smaller stellar concentrations . similar to the results of refl08 , we find no clear spatial pattern of cluster ages in this region . xc is the most populous and , with only @xmath127% fraction of ir - excess sources and no more than one protostar , is not very young . active star formation is strongest in subcluster xb2 with several protostars including the massive system rmcx # 89 = iras 06306 + 0437 . figure [ fig : co_ir ] shows the x - ray selected rmc stars , classified by evolutionary stage from the nir color - color diagram ( [ sec : clusevol ] ) , superposed on contours representing @xmath128co emission @xcite . the younger class i and ii sources are mostly seen around the co molecular ridge near cluster xb and the molecular material near cluster xc . class iii stars appear more dispersed throughout the cloud and dominate the xa region where molecular material is no longer present . we can interpret this trend as reflecting the drifting of older class iii stars away from their birth sites @xcite . for example , with a characteristic transverse speed of @xmath31 km s@xmath24 , a star would travel 1 pc ( @xmath129 ) in 1 myr . this is sufficient to disperse such stars originally formed in the clusters throughout the @xmath10 fields in a few myrs , resulting in a `` dispersed '' population . the relationship of cluster xa to molecular gas is clearly different than for clusters xb and xc . xa lies within the bright nebulosity of the rosette hii region where most of the molecular materials are gone . this cluster of less embedded and older pms stars ( see [ sec : clusevol ] ) is likely an early population triggered by the expansion of the hii region . its similarity to ngc 2237 , a similar lightly obscured cluster on the western side of the rosette nebula , will be discussed in our paper iii ( wang et al . , in preparation ) . the overdensity of stars in xb , including the nir pl 2 cluster , is situated on the ` rim ' of molecular material adjacent to the expanding rosette pdr . it includes xb2 , the youngest subcluster in our examined portion of the rmc . this cluster , and possibly the similar pl 1 cluster located south - west of our @xmath10 fields in a molecular clump , are reasonably interpreted as triggered by the expanding shell of the ngc 2244 hii region . the molecular clumps are probably undergoing photoevaporation and will eventually become unobscured clusters like region xa . region xc contains the large and most extended population of embedded stars , including several infrared clusters . its relationship to a dense molecular clump has been noted by @xcite and refl08 . our analysis above ( [ sec : clusevol ] ) suggests that subcluster xc1 ( pl 4 ) is partially emerging from the molecular core while xc2 is younger and still embedded . as discussed by refl08 and @xcite , it is not clear that star formation here has been triggered by the expanding hii region . since the x - ray sample provides an estimate of the total stellar population of each cluster , we can roughly estimate the star formation efficiency ( sfe ) of molecular material in the rmc . @xcite estimate the molecular masses of 70 clumps in the rmc using an empirical co - to - h@xmath130 ratio @xmath131 @xmath132 molecules @xmath17 ( k km s@xmath24)@xmath24 . mass estimates are made assuming either local thermodynamic equilibrium or gravitational binding of gas motions within clumps . if we assume an average stellar mass of 0.4 m@xmath45 @xcite , the sfe @xmath133 is approximately 15% in region xb and 11% in region xc , adopting the lte clump masses . if the gravitational clump masses in @xcite are used , the sfe becomes 24% and 32% in xb and xc , respectively . these values are similar to the sfes inferred for nearby embedded rich clusters : ngc 2071 ( sfe@xmath13412% ) and ngc 2024 ( 33% ) in the orion molecular cloud @xcite , @xmath135 oph ( 9% ; * ? ? ? * ) , and mon r2 ( 25% ; * ? ? ? as noted by rl08 , the xc ( @xmath3pl 4 ) cluster lies at the high end of sfes ; this may indicate that on - going star formation has significantly depleted gas in the region . infrared observations have established that @xmath136% of stars in gmcs form in embedded clusters @xcite . unclustered pms stars have been established mainly in the orion molecular cloud complex , but even there it has been difficult to quantify their population and star formation processes @xcite . @xcite present new survey results from _ maps of nearby star forming clouds , and many show a significant distributed population . distributed star formation in more distant clouds like the rmc is largely unstudied due to the heavy contamination from galactic field stars outside of dense clusters . modern astrophysical models of star formation in gmcs suggests that individual stars may commonly form asynchronously in supersonically turbulent cloud clumps @xcite . because our x - ray identified young stellar population suffers little contamination , we are able to examine the fraction of stars that are formed both in clusters and at low stellar densities in the rmc . the total number of sources in the three fields is 553 , of which 303 lie within the contours we use to define clusters xa , xb , and xc ( table [ tbl : cls_counts ] ) . among the 250 stars distributed outside the clusters , at most 10% of the total detections could be galactic and extragalactic contaminants given the short exposure time and the presence of the molecular cloud ( [ sec : counterpart.sec ] ) . this implies that @xmath335% of stars could be formed in a distributed fashion throughout the rmc region , consistent with estimates from infrared studies ( e.g. * ? ? ? * ; * ? ? ? * ) . after correcting the sensitivity variation across the fields ( [ sec : new_sensitivity ] ) , the fraction of distributed stars is @xmath345% , which implies that clustered star formation would then be the governing mechanism in the rmc but distributed star formation is rather important . however , further interpretation of this fraction should proceed with caution . there are at least three sources of complexity here . first , our definition of cluster boundaries based on smoothed stellar density maps ( figure [ fig : ch5_ssd ] ) is arbitrary , which are about 0.8 stars arcmin@xmath137 for the x - ray - sampled stellar density . in comparison , in refl08 the cluster boundaries are defined by 0.2 stars arcmin@xmath137 contour levels in the nearest neighbor density map of ir - excess sources . they find that 60% stars in the complex are in clusters . the clustering fraction reaches 86% if only the stars within the molecular cloud are considered . besides the differences in the field coverage of the x - ray and ir surveys ( see also [ sec : discuss_xir ] ) , we could get different distributed fractions with different , reasonable choices of cluster boundaries . second , a significant number of lower mass pms stars ( @xmath138 m@xmath45 ) is missing from our survey due to short x - ray exposures ( [ sec : sensitivity ] ) . while there may be no preference for lower - mass stars to be clustered or distributed , it is possible that higher - mass stars preferentially form in denser clusters from primordial mass segregation ( e.g. , * ? ? ? third , the distributed stars may have been born in a clustered environment but have moved into a wider region . numerical simulations show that the formation process of a cluster of stars is highly dynamic and chaotic ; stellar encounters violently eject some of the young stars from multiple systems @xcite . sparse clusters in particular can lose @xmath139 of their stars over a few myrs . thus the observed distributed populations may not have formed _ in situ_. an underlying theme to this and similar studies concerns the relationships between x - ray and infrared surveys of star forming molecular clouds to uncover young stellar populations . each as advantages and disadvantages . with instruments like flamingos , nir surveys can cover large cloud areas to faint magnitudes . these catalogs can include virtually the entire cloud population at low to moderate extinctions ( @xmath140 ) , but often suffer bad ( often factors @xmath141 ) contamination by foreground and background field stars . used by themselves , pms samples are restricted to stars with nir - bright inner disks , usually class ii and lightly obscured class i systems . the resulting nir - excess samples can constitute only a small fraction ( @xmath142% ) of the total stellar populations which are often dominated by systems with cool or absent disks . with space - based instruments like @xmath7 s irac , mir surveys also cover large cloud areas with very high sensitivity and penetration to @xmath64 . as with nir studies , mir observations are dominated by field star populations and samples are restricted to mir - excess systems . the nir - excess stars are recovered and new subpopulations emerge : embedded class i and some class 0 systems , low luminosity pms brown dwarf systems , and a few older systems with transitional mir - excess disks ( e.g. , * ? ? ? * ; * ? ? ? diskfree class iii members are catalogued but can not be distinguished from contaminants . x - ray surveys have very different characteristics because they trace magnetic reconnection flaring rather than stellar and disk bolometric emission . older galactic field stars with much lower flaring levels are nearly absent , and most extragalactic contaminants are readily removed by the absence of stellar counterparts . x - rays can penetrate @xmath143 but with reduced sensitivity . x - ray studies also give estimates of the total stellar populations , and measurements of absorption for each star individually . while some class i and ii systems are recovered , the main advantage of x - ray observations is that disk - free class iii pms members are readily detected , subject to the sensitivity limitations which correlate with stellar mass . since many clusters are dominated by class iii stars , the x - ray populations can considerably exceed the nir- or mir - excess populations . a significant drawback is areal coverage : the @xmath144 telescope has a @xmath43 field but suffers confusion in crowded fields , while the _ chandra x - ray observatory _ has a 17 field of view with subarcsecond resolution . this limitation can be somewhat mitigated with mosaics as described here . the combination of x - ray and infrared studies are more effective at uncovering pms populations in molecular clouds than any one method . we find here that the x - ray samples are quite free from contamination and can be 10 times richer than the nir- or mir - excess samples ( table [ tbl : cls_counts ] ) . this greatly assists in tracing the spatial structure of star formation in the cloud . subsequent infrared photometry then gives the evolutionary classification , absorption and approximate mass of each x - ray - selected star without bias towards luminous disks ( figures [ fig : ch5_ccd]-[fig : ch5_cmd ] ) . there is good agreement between our x - ray survey and refl08 nir survey ( and other surveys ) . taking this combined x - ray / infrared approach , we emerge with a number of results on the pms population in the portion of the rmc examined with @xmath10 . first , we confirm past findings that star formation is not spatially uniform in the cloud , and we estimate that about 1/3 of the stars lie in unclustered environments ( [ sec : distribsf ] ) . this supports a large body of infrared - based survey work giving similar results in other clouds @xcite . second , we estimate the total population in the examined portion of the cloud to be about 1700 stars down to the substellar limit ( table [ tbl : cls_counts ] ) . this estimate is based on an extrapolation of the empirical pms @xmath41-mass correlation @xcite and knowledge of our x - ray sensitivity limits ( [ sec : sensitivity ] ) . this is similar to the total population of @xmath145 stars found for the central ngc 2244 cluster which illuminates the rosette nebula hii region ( paper i ) . third , in contrast to the single concentrated ob cluster at the center of the rosette nebula , the rmc population is divided into three smaller clusters which themselves can be divided into sparse subclusters . portions of x - ray clusters xb and xc are associated with well - established infrared clusters pl 2 and pl 4 ( [ sec : xbxc ] ) . the x - ray stellar distributions and absorptions provide more detail on these structures . xc is richest with a total population around 800 stars ; subcluster xc2 is more deeply embedded with younger stars than xc1 . similarly in cluster xb , group xb2 has more protostars than xb1 . these structures are thus inhomogeneous and not coeval . we caution that , our current identifications of structural components are from figure [ fig : ch5_ssd ] ( and figure [ fig : chandra_contour ] ) rather than figure [ fig : uniform_contour ] , as the addition of more sources is critically needed for substructure identification . while some substructures were cross - identified in the infrared ( see table [ tbl : cls_comparison ] ) , the weak substructures certainly require further confirmation regarding the details of their locations and sizes . subcluster xb2 = pl 2 is sparse but contains the highest - mass star in our examined region of the rmc , the heavily obscured , infrared - excess o9/b0 star rmcx # 89 = iras 06306 + 0437 ( [ sec : rmcx89 ] ) . there is only one lower - mass protostar , rmcx # 72 , lies at a projected distance of 0.3 pc to rmcx # 89 . the absence of a substantial cluster surrounding this massive protostar has interesting implications for star formation at the edge of hii regions where triggering may be active . the system can not realistically be dynamically ejected from a more crowded region , as it is still embedded in its dense dusty envelope . one possibility is that the massive star formed in isolation . we recall that hd 46223 , one of the two o4 stars in the unobscured ngc 2244 cluster lies in a puzzling location near the southeast boundary of the nebula and , based on our @xmath10 study , has no lower mass companions ( paper i ) . iras 06306 + 0437 may be like hd 46223 seen at an earlier phase . another possibility is that the massive star has formed before ( except for rmcx # 72 ) an accompanying cluster of lower mass stars . if this scenario applies , the star formation sequence is opposite to the one @xcite inferred for the embedded w3 main clusters , where the ob stars appeared to have formed after , not before , an extended cluster of lower - mass pms stars . while it is reasonable that the density of obscured x - ray sources decreases significantly towards the hii region ionized by the ngc 2244 cluster , it was surprising that the new lightly obscured cluster xa appeared in the x - ray survey in a region with no molecular material ( [ sec : xa ] ) . infrared surveys likely missed this structure due to the paucity of infrared - excess stars and confusion by the bright pah and atomic nebular emission , a stellar enhancement associated with xa can be discerned in 2mass , flamingos , and @xmath7 stellar distributions . its location and low fraction of ir - excess sources suggest that it may have formed through an early epoch of triggered star formation produced by the expanding rosette nebula hii region . it appears similar to the small ngc 2237 cluster on the western side of the rosette nebula ( * ? ? ? * refl08 ) . a detailed @xmath10 study of ngc 2237 is presented in paper iii . while paper iii will present a more comprehensive discussion of star formation in the rosette complex , some immediate inferences can be made concerning the rmc structures presented here . ngc 2244 is the youngest ob cluster among the sub - associations @xcite . @xcite argue that sequential star formation extends into the rmc following the compression and triggering process described by @xcite . however , neither refl08 , @xcite nor the present study support this model in the xb and xc region of the rmc . no clear spatial sequence of stellar ages is seen . it is plausible , however , that cluster xa was triggered by the expanding rosette hii region . at present it lies in ( or , at least , projected upon ) the ionized region rather than in the molecular cloud . it has the smallest fraction of infrared - excess stars of the clusters examined here , and hence was likely the earliest formed . we thank bruce elmegreen and zhiyun li for helpful discussions on cluster structure and formation . we thank travis rector and mark heyer for kindly providing the kpno mosaic images of the rosette nebula and the co emission maps of the rosette complex , respectively . thanks c. poulton for providing the full source catalog of spitzer / irac 3.6@xmath29 m detections . this work was supported by _ chandra x - ray observatory _ grants go1 - 2008x and go3 - 4010x awarded to l.k.t . flamingos was designed and constructed by the ir instrumentation group ( pi : r. elston ) at the university of florida , department of astronomy with support from nsf grant ast97 - 31180 and kitt peak national observatory . the data were collected under the noao survey program , `` towards a complete near - infrared spectroscopic survey of giant molecular clouds '' ( pi : e. lada ) and supported by nsf grants ast97 - 3367 and ast02 - 02976 to the university of florida . e.a.l . also acknowledges support from nasa ltsa nng05d66 g . this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by nasa and the national science foundation . this research has made use of the simbad database and the vizier catalogue access tool , operated at cds , strasbourg , france . , j. , cesaroni , r. , caselli , p. , catarzi , m. , codella , c. , comoretto , g. , curioni , g. p. , curioni , p. , di franco , s. , felli , m. , giovanardi , c. , olmi , l. , palagi , f. , palla , f. , panella , d. , pareschi , g. , rossi , e. , speroni , n. , & tofani , g. 1994 , , 103 , 541 , p. , townsley , l. k. , getman , k. v. , & bauer , f. 2002 , acis extract , an acis point source extraction package , pennsylvania state university , http://www.astro.psu.edu/xray/docs/tara/ae_users_guide.html , e. , whitney , b. a. , babler , b. l. , indebetouw , r. , meade , m. r. , watson , c. , wolff , m. j. , wolfire , m. g. , bania , t. m. , benjamin , r. a. , clemens , d. p. , cohen , m. , devine , k. e. , dickey , j. m. , heitsch , f. , jackson , j. m. , kobulnicky , h. a. , marston , a. p. , mathis , j. s. , mercer , e. p. , stauffer , j. r. , & stolovy , s. r. 2004 , , 154 , 322 , r. m. , skrutskie , m. f. , van dyk , s. , beichman , c. a. , carpenter , j. m. , chester , t. , cambresy , l. , evans , t. , fowler , j. , gizis , j. , howard , e. , huchra , j. , jarrett , t. , kopan , e. l. , kirkpatrick , j. d. , light , r. m. , marsh , k. a. , mccallon , h. , schneider , s. , stiening , r. , sykes , m. , weinberg , m. , wheaton , w. a. , wheelock , s. , & zacarias , n. 2003 , 2mass all sky catalog of point sources . ( the irsa 2mass all - sky point source catalog , nasa / ipac infrared science archive . http://irsa.ipac.caltech.edu/applications/gator/ ) , ii , n. j. , allen , l. e. , blake , g. a. , boogert , a. c. a. , bourke , t. , harvey , p. m. , kessler , j. e. , koerner , d. w. , lee , c. w. , mundy , l. g. , myers , p. c. , padgett , d. l. , pontoppidan , k. , sargent , a. i. , stapelfeldt , k. r. , van dishoeck , e. f. , young , c. h. , & young , k. e. 2003 , , 115 , 965 , n. , feigelson , e. d. , getman , k. v. , townsley , l. , broos , p. , flaccomio , e. , mccaughrean , m. j. , micela , g. , sciortino , s. , bally , j. , smith , n. , muench , a. a. , garmire , g. p. , & palla , f. 2005 , , 160 , 530 , p. m. , chapman , n. , lai , s .- p . , evans , ii , n. j. , allen , l. e. , jrgensen , j. k. , mundy , l. g. , huard , t. l. , porras , a. , cieza , l. , myers , p. c. , mern , b. , van dishoeck , e. f. , young , k. e. , spiesman , w. , blake , g. a. , koerner , d. w. , padgett , d. l. , sargent , a. i. , & stapelfeldt , k. r. 2006 , , 644 , 307 , d. g. , levine , s. e. , canzian , b. , ables , h. d. , bird , a. r. , dahn , c. c. , guetter , h. h. , harris , h. c. , henden , a. a. , leggett , s. k. , levison , h. f. , luginbuhl , c. b. , martini , j. , monet , a. k. b. , munn , j. a. , pier , j. r. , rhodes , a. r. , riepe , b. , sell , s. , stone , r. c. , vrba , f. j. , walker , r. l. , westerhout , g. , brucato , r. j. , reid , i. n. , schoening , w. , hartley , m. , read , m. a. , & tritton , s. b. 2003 , , 125 , 984 , m. a. c. , lindegren , l. , kovalevsky , j. , hoeg , e. , bastian , u. , bernacca , p. l. , crz , m. , donati , f. , grenon , m. , van leeuwen , f. , van der marel , h. , mignard , f. , murray , c. a. , le poole , r. s. , schrijver , h. , turon , c. , arenou , f. , froeschl , m. , & petersen , c. s. 1997 , , 323 , l49 , k. e. , harvey , p. m. , brooke , t. y. , chapman , n. , kauffmann , j. , bertoldi , f. , lai , s .- p . , alcal , j. , bourke , t. l. , spiesman , w. , allen , l. e. , blake , g. a. , evans , ii , n. j. , koerner , d. w. , mundy , l. g. , myers , p. c. , padgett , d. l. , salinas , a. , sargent , a. i. , stapelfeldt , k. r. , teuben , p. , van dishoeck , e. f. , & wahhaj , z. 2005 , , 628 , 283 1@xmath146 & [email protected] & & 98.135158 & 4.618251 & 0.7 & 6.0 & & 7.5 & 3.4 & 0.5 & 6.7 & 0.91 & & 1.9 & @xmath148 - 5 & g ... & & 9.8 & 3.9 + 3@xmath146 & [email protected] & & 98.152431 & 4.603172 & 0.9 & 6.6 & & 6.2 & 3.2 & 0.8 & 4.4 & 0.89 & & 1.6 & -4.6 & .... & a & 17.9 & 2.6 + 4@xmath146 & [email protected] & & 98.158706 & 4.547265 & 1.2 & 9.9 & & 16.8 & 4.9 & 2.2 & 0.6 & 0.91 & & 3.1 & @xmath148 - 5 & g ... & & 10.6 & 1.3 + 5@xmath146 & [email protected] & & 98.164603 & 4.608042 & 0.6 & 6.3 & & 12.3 & 4.1 & 0.7 & 3.5 & 0.90 & & 2.6 & @xmath148 - 5 & .... & a & 18.0 & 1.8 + 6@xmath146 & [email protected] & & 98.164639 & 4.523587 & 1.2 & 9.9 & & 11.7 & 4.4 & 3.3 & 3.9 & 0.91 & & 2.3 & @xmath148 - 5 & .... & a & 16.2 & 1.9 + 7@xmath146 & [email protected] & & 98.169734 & 4.614833 & 0.5 & 5.9 & & 14.4 & 4.4 & 0.6 & 9.6 & 0.90 & & 2.9 & @xmath148 - 5 & .... & a & 18.1 & 3.1 + 8@xmath146 & [email protected] & & 98.176413 & 4.538329 & 1.3 & 9.0 & & 7.0 & 3.5 & 2.0 & 6.7 & 0.90 & & 1.7 & -3.6 & .... & a & 17.3 & 3.6 + 9@xmath146 & [email protected] & & 98.188575 & 4.535152 & 0.7 & 8.4 & & 21.4 & 5.3 & 1.6 & 1.9 & 0.90 & & 3.6 & @xmath148 - 5 & .... & c & 17.5 & 0.8 + 10@xmath146 & [email protected] & & 98.192141 & 4.603596 & 0.8 & 7.3 & & 10.8 & 4.1 & 2.2 & 7.7 & 0.90 & & 2.3 & @xmath148 - 5 & .... & a & 35.3 & 2.4 + 11@xmath146 & [email protected] & & 98.194699 & 4.642082 & 0.5 & 4.5 & & 5.8 & 3.0 & 0.2 & 0.8 & 0.89 & & 1.6 & @xmath148 - 5 & .... & a & 18.5 & 1.4 + 2@xmath146 & [email protected] & & 98.137138 & 4.577017 & 1.4 & 8.4 & & 5.9 & 3.4 & 2.1 & 0.6 & 0.90 & & 1.5 & -2.8 & g ... & & 13.0 & 0.7 + 12 & [email protected] & & 98.197650 & 4.530632 & 1.3 & 7.9 & & 4.5 & 3.0 & 1.5 & 4.0 & 0.90 & & 1.2 & -2.3 & .... & a & 17.6 & 2.1 + 16 & [email protected] & & 98.205124 & 4.573882 & 1.1 & 7.8 & & 4.9 & 3.4 & 3.1 & 3.7 & 0.90 & & 1.2 & -1.8 & .... & a & 33.1 & 7.8 + 26 & [email protected] & & 98.226879 & 4.623523 & 0.9 & 6.3 & & 4.6 & 3.0 & 1.4 & 3.1 & 0.90 & & 1.3 & -2.6 & .... & & 35.6 & 3.4 + 29 & [email protected] & & 98.232618 & 4.571979 & 1.1 & 7.2 & & 5.2 & 3.4 & 2.8 & 0.0 & 0.89 & & 1.3 & -2.1 & .... & a & 34.6 & 1.3 + 33 & [email protected] & & 98.242303 & 4.501051 & 1.1 & 6.6 & & 3.3 & 2.5 & 0.7 & 3.5 & 0.89 & & 1.1 & -2.3 & .... & a & 15.9 & 5.2 + 36 & [email protected] & & 98.252556 & 4.508536 & 0.9 & 5.9 & & 3.5 & 2.5 & 0.5 & 0.0 & 0.90 & & 1.1 & -2.7 & .... & a & 17.6 & 1.3 + 53 & [email protected] & & 98.284762 & 4.700962 & 1.0 & 7.3 & & 4.6 & 3.2 & 2.4 & 2.4 & 0.89 & & 1.2 & -1.9 & .... & a & 33.9 & 1.6 + 57 & [email protected] & & 98.285532 & 4.609538 & 0.3 & 4.0 & & 4.7 & 3.0 & 1.3 & 0.0 & 0.90 & & 1.3 & -2.7 & g ... & & 23.3 & 0.9 + 58 & [email protected] & & 98.288891 & 4.634532 & 1.1 & 6.1 & & 6.9 & 3.7 & 3.1 & 5.9 & 0.89 & & 1.6 & -2.8 & .... & a & 35.4 & 3.3 + 5 & [email protected] & 12.3 & 2.6 & & @xmath149 21.8 @xmath150 & 2.2 & @xmath151 53.4 @xmath150 & & 29.67 & 29.98 & 30.02 & 30.15 & 30.42 & & + 6 & [email protected] & 11.7 & 2.3 & & 22.1 & _ 2.0 _ & 53.7 @xmath152 & & 29.72 & 30.14 & 30.21 & 30.28 & 30.65 & & + 7 & [email protected] & 14.4 & 2.9 & & @xmath153 22.2 @xmath154 & 7.0 & @xmath155 53.6 @xmath156 & & 29.48 & 30.48 & 30.55 & 30.52 & 30.75 & & + 9 & [email protected] & 21.4 & 3.6 & & _ 20.0 _ & @xmath157 0.2 @xmath158 & @xmath155 53.8 @xmath152 & & 30.45 & 27.47 & 27.47 & 30.45 & 30.48 & & + 10 & [email protected] & 10.8 & 2.3 & & 22.1 & 3.5 & 53.3 & & 29.28 & 29.98 & 30.05 & 30.06 & 30.34 & & + 14 & [email protected] & 14.6 & 2.7 & & _ 20.0 _ & @xmath155 1.2 @xmath159 & @xmath160 52.9 @xmath152 & & 29.77 & 29.10 & 29.10 & 29.85 & 29.86 & & + 17 & [email protected] & 10.2 & 2.2 & & @xmath161 22.2 @xmath154 & 3.4 & 53.2 & & 29.07 & 29.91 & 30.00 & 29.97 & 30.29 & & + 18 & [email protected] & 29.9 & 4.4 & & @xmath162 21.9 @xmath154 & @xmath149 1.7 @xmath163 & @xmath151 53.7 @xmath164 & & 29.86 & 30.02 & 30.07 & 30.24 & 30.60 & & + 21 & [email protected] & 10.4 & 2.3 & & @xmath155 22.7 @xmath152 & _ 2.0 _ & @xmath151 53.8 @xmath150 & & & 30.06 & 30.32 & 30.08 & & & + 22 & [email protected] & 15.1 & 2.9 & & @xmath161 22.1 @xmath154 & @xmath165 4.0 & @xmath151 53.3 @xmath164 & & 29.36 & 30.10 & 30.16 & 30.17 & 30.43 & & + 1 & [email protected] & & & & & & & & & & & & @xmath166 + 2 & [email protected] & & & & & & & & & & & & @xmath167 + 3 & [email protected] & & & & & & & & & & & & @xmath167 + 4 & [email protected] & & 0945@xmath770095483 & & & 14.54 & & & & & & & @xmath167 + 5 & [email protected] & & & & & & & 063239@xmath147043629 & 14.69 & 13.45 & 12.54 & & @xmath166 + 6 & [email protected] & & & & & & & & & & & & @xmath167 + 7 & [email protected] & & & & & & & & & & & & @xmath167 + 8 & [email protected] & & & & & & & & & & & & @xmath167 + 9 & [email protected] & & 0945@xmath770095525 & 18.41 & 16.59 & 14.86 & 06324526@xmath1470432074 & 063245@xmath147043207 & 13.68 & 12.99 & 12.79 & aaa000 & @xmath166 + 10 & [email protected] & & & & & & & 063246@xmath147043610 & 14.69 & 14.29 & 14.06 & & @xmath166 + xa & 06:32:38 & @xmath14704:46:18 & @xmath168 & & & & + xa1 & 06:32:24 & + 04:50:30 & @xmath169 & & & & 06297@xmath1470453 + xa2 & 06:32:25 & + 04:39:00 & @xmath170 & & & & + xa3 & 06:32:38 & + 04:46:18 & @xmath171 & & & & + & & & & & & & + xb & 06:33:15 & + 04:35:11 & @xmath172 & & & & + xb1 & 06:32:54 & + 04:37:00 & @xmath173 & & & & + xb2 & 06:33:14 & + 04:31:30 & @xmath174 & & & c & + xb3 & 06:33:15 & + 04:35:11 & @xmath175 & & pl 2 & & 06306@xmath1470437 + & & & & & & & + xc & 06:34:12 & + 04:28:15 & @xmath176 & & & & + xc1 & 06:34:12 & + 04:28:15 & @xmath177 & & pl 4 & e & 06314@xmath1470427 + xc2 & 06:34:24 & + 04:22:00 & @xmath178 & & pl 5 , refl 8 & e & 06317@xmath1470426 + xa & 91 & 27 & 30% & 300 & & 61 & 6 & 10% & 2% & & 32 & 11% + xa1 & 6 & 2 & 33% & 20 & & 3 & 0 & 0% & 0% & & 2 & 6% + xa2 & 4 & 1 & 25% & 15 & & 3 & 0 & 0% & 0% & & 0 & 0% + xa3 & 29 & 5 & 17% & 100 & & 23 & 4 & 17% & 4% & & 11 & 11% + & & & & & & & & & & & & + xb & 52 & 24 & 46% & 200 & & 37 & 8 & 22% & 4% & & 16 & 8% + xb1 & 11 & 6 & 55% & 40 & & 5 & 0 & 0% & 0% & & 0 & 0% + xb2 & 5 & 4 & 80% & 20 & & 3 & 1 & 33% & 5% & & 8 & 40% + xb3 & 14 & 7 & 50% & 20 & & 11 & 4 & 36% & 20% & & 1 & 5% + & & & & & & & & & & & & + xc & 160 & 66 & 41% & 800 & & 131 & 9 & 7% & 1% & & 62 & 8% + xc1 & 67 & 22 & 33% & 340 & & 57 & 5 & 9% & 1% & & 17 & 5% + xc2 & 17 & 13 & 76% & 90 & & 14 & 1 & 7% & 1% & & 16 & 18% + & & & & & & & & & & & & + distributed & 250 & 105 & 42% & 1100 & & & & & & & & + # stars arcmin@xmath137 ) map for all rmc sources smoothed with a 3 radius kernel . ( b ) the same as ( a ) but using a 2 radius kernel . ( c ) the same as ( a ) but for the unobscured population . ( d ) the same as ( b ) but for the unobscured population . ( e ) the same as ( a ) but for the obscured population . ( f ) the same as ( b ) but for the obscured population . the density scaling is the same between maps of the unobscured population and obscured population for a fair comparison.[fig : ch5_ssd],title="fig:",scaledwidth=40.0% ] # stars arcmin@xmath137 ) map for all rmc sources smoothed with a 3 radius kernel . ( b ) the same as ( a ) but using a 2 radius kernel . ( c ) the same as ( a ) but for the unobscured population . ( d ) the same as ( b ) but for the unobscured population . ( e ) the same as ( a ) but for the obscured population . ( f ) the same as ( b ) but for the obscured population . the density scaling is the same between maps of the unobscured population and obscured population for a fair comparison.[fig : ch5_ssd],title="fig:",scaledwidth=40.0% ] # stars arcmin@xmath137 ) map for all rmc sources smoothed with a 3 radius kernel . ( b ) the same as ( a ) but using a 2 radius kernel . ( c ) the same as ( a ) but for the unobscured population . ( d ) the same as ( b ) but for the unobscured population . ( e ) the same as ( a ) but for the obscured population . ( f ) the same as ( b ) but for the obscured population . the density scaling is the same between maps of the unobscured population and obscured population for a fair comparison.[fig : ch5_ssd],title="fig:",scaledwidth=40.0% ] # stars arcmin@xmath137 ) map for all rmc sources smoothed with a 3 radius kernel . ( b ) the same as ( a ) but using a 2 radius kernel . ( c ) the same as ( a ) but for the unobscured population . ( d ) the same as ( b ) but for the unobscured population . ( e ) the same as ( a ) but for the obscured population . ( f ) the same as ( b ) but for the obscured population . the density scaling is the same between maps of the unobscured population and obscured population for a fair comparison.[fig : ch5_ssd],title="fig:",scaledwidth=40.0% ] # stars arcmin@xmath137 ) map for all rmc sources smoothed with a 3 radius kernel . ( b ) the same as ( a ) but using a 2 radius kernel . ( c ) the same as ( a ) but for the unobscured population . ( d ) the same as ( b ) but for the unobscured population . ( e ) the same as ( a ) but for the obscured population . ( f ) the same as ( b ) but for the obscured population . the density scaling is the same between maps of the unobscured population and obscured population for a fair comparison.[fig : ch5_ssd],title="fig:",scaledwidth=40.0% ] # stars arcmin@xmath137 ) map for all rmc sources smoothed with a 3 radius kernel . ( b ) the same as ( a ) but using a 2 radius kernel . ( c ) the same as ( a ) but for the unobscured population . ( d ) the same as ( b ) but for the unobscured population . ( e ) the same as ( a ) but for the obscured population . ( f ) the same as ( b ) but for the obscured population . the density scaling is the same between maps of the unobscured population and obscured population for a fair comparison.[fig : ch5_ssd],title="fig:",scaledwidth=40.0% ] # stars arcmin@xmath137 ) map for a sub - sample of rmc sources ( photon flux @xmath179 count @xmath17 s@xmath24 ) smoothed with a 3 radius kernel ( top panel ) and a 2 radius kernel ( lower panel ) . the cluster regions and substructure regions from figure [ fig : ch5_ssd ] are shown.[fig : uniform_contour],title="fig:",scaledwidth=54.0% ] # stars arcmin@xmath137 ) map for a sub - sample of rmc sources ( photon flux @xmath179 count @xmath17 s@xmath24 ) smoothed with a 3 radius kernel ( top panel ) and a 2 radius kernel ( lower panel ) . the cluster regions and substructure regions from figure [ fig : ch5_ssd ] are shown.[fig : uniform_contour],title="fig:",scaledwidth=54.0% ] vs. @xmath118 color - color diagram for _ chandra _ stars with high - quality @xmath2 photometry ( error in both @xmath117 and @xmath118 colors @xmath180 mag ) in regions a , b , and c as defined in figure [ fig : chandra_contour ] . the green circles and red triangles represent sources with and without significant @xmath6-band excess ( @xmath119 ) respectively . the black solid and long - dash lines denote the loci of ms stars and giants , respectively , from bessell & brett ( 1988 ) . the purple dash dotted line is the locus for classical t tauri stars from meyer et al . ( 1997 ) , and the cyan solid line is the locus for haebe stars from lada & adams ( 1992 ) . the blue dashed lines represent the standard reddening vector with crosses marking every @xmath25 mag.[fig : ch5_ccd],title="fig:",width=249 ] vs. @xmath118 color - color diagram for _ chandra _ stars with high - quality @xmath2 photometry ( error in both @xmath117 and @xmath118 colors @xmath180 mag ) in regions a , b , and c as defined in figure [ fig : chandra_contour ] . the green circles and red triangles represent sources with and without significant @xmath6-band excess ( @xmath119 ) respectively . the black solid and long - dash lines denote the loci of ms stars and giants , respectively , from bessell & brett ( 1988 ) . the purple dash dotted line is the locus for classical t tauri stars from meyer et al . ( 1997 ) , and the cyan solid line is the locus for haebe stars from lada & adams ( 1992 ) . the blue dashed lines represent the standard reddening vector with crosses marking every @xmath25 mag.[fig : ch5_ccd],title="fig:",width=230 ] vs. @xmath118 color - color diagram for _ chandra _ stars with high - quality @xmath2 photometry ( error in both @xmath117 and @xmath118 colors @xmath180 mag ) in regions a , b , and c as defined in figure [ fig : chandra_contour ] . the green circles and red triangles represent sources with and without significant @xmath6-band excess ( @xmath119 ) respectively . the black solid and long - dash lines denote the loci of ms stars and giants , respectively , from bessell & brett ( 1988 ) . the purple dash dotted line is the locus for classical t tauri stars from meyer et al . ( 1997 ) , and the cyan solid line is the locus for haebe stars from lada & adams ( 1992 ) . the blue dashed lines represent the standard reddening vector with crosses marking every @xmath25 mag.[fig : ch5_ccd],width=230 ] vs. @xmath117 color - magnitude diagram using the same sample and symbols as figure [ fig : ch5_ccd ] . the purple solid line and dashed line is the 2 myr isochrone and the 1 myr isochrone for pms stars from @xcite , respectively . both isochrones are shown here because of the uncertainty in stellar ages . the dash dotted line marks the location of zero age main sequence ( zams ) stars . the blue dashed lines represent the standard reddening vector with asterisks marking every @xmath25 mag and the corresponding stellar masses are marked.[fig : ch5_cmd],title="fig:",width=259 ] vs. @xmath117 color - magnitude diagram using the same sample and symbols as figure [ fig : ch5_ccd ] . the purple solid line and dashed line is the 2 myr isochrone and the 1 myr isochrone for pms stars from @xcite , respectively . both isochrones are shown here because of the uncertainty in stellar ages . the dash dotted line marks the location of zero age main sequence ( zams ) stars . the blue dashed lines represent the standard reddening vector with asterisks marking every @xmath25 mag and the corresponding stellar masses are marked.[fig : ch5_cmd],title="fig:",width=259 ] vs. @xmath117 color - magnitude diagram using the same sample and symbols as figure [ fig : ch5_ccd ] . the purple solid line and dashed line is the 2 myr isochrone and the 1 myr isochrone for pms stars from @xcite , respectively . both isochrones are shown here because of the uncertainty in stellar ages . the dash dotted line marks the location of zero age main sequence ( zams ) stars . the blue dashed lines represent the standard reddening vector with asterisks marking every @xmath25 mag and the corresponding stellar masses are marked.[fig : ch5_cmd],width=259 ]	we explore here the young stellar populations in the rosette molecular cloud ( rmc ) region with high spatial resolution x - ray images from the _ chandra x - ray observatory _ , which are effective in locating weak - lined t tauri stars as well as disk - bearing young stars . a total of 395 x - ray point sources are detected , 299 of which ( 76% ) have an optical or near - infrared ( nir ) counterpart identified from deep flamingos images . from x - ray and mass sensitivity limits , we infer a total population of @xmath0 young stars in the survey region . based on smoothed stellar surface density maps , we investigate the spatial distribution of the x - ray sources and define three distinctive structures and substructures within them . structures b and c are associated with previously known embedded ir clusters , while structure a is a new x - ray - identified unobscured cluster . a high mass protostar rmcx # 89 = iras 06306 + 0437 and its associated sparse cluster is studied . the different subregions are not coeval but do not show a simple spatial - age pattern . disk fractions vary between subregions and are generally @xmath1% of the total stellar population inferred from the x - ray survey . the data are consistent with speculations that triggered star formation around the hii region is present in the rmc , but do not support a simple sequential triggering process through the cloud interior . while a significant fraction of young stars are located in a distributed population throughout the rmc region , it is not clear they originated in clustered environments .
in this appendix , we calculate the loop diagrams for the m1 radiative decays . we present the calculation of @xmath0 , the calculations of the decays @xmath86 ( @xmath87 ) are obtained by replacing @xmath152 with @xmath153 and @xmath154 with @xmath6 ( @xmath155 ) . the triangle loop diagrams in fig . [ triangle ] have a similar form as the triangle loop diagrams in the hadronic decays , therefore the notation used here will be the almost the same as that of ref . we refer the reader to that paper for explicit expressions for the integrals . the amplitude from fig . [ triangle](a ) is @xmath156 where @xmath157 , @xmath158 is relevant for loops with neutral @xmath26 mesons and @xmath159 is relevant for loops with charged and strange @xmath26 mesons . here @xmath160 only differs from the function defined in ref . @xcite by omitting a factor of @xmath161 from the denominator . [ triangle](b ) contributes @xmath162 where the functions @xmath163 and @xmath164 are again the same as functions in ref . @xcite up to a factor of @xmath161 . fig . [ triangle](c ) contributes @xmath165 fig . [ triangle](d ) contributes @xmath166 where @xmath167 is relevant for loops with neutral @xmath26 mesons and @xmath168 is relevant for loops with charged and strange @xmath26 mesons . finally , fig . [ triangle](e ) gives @xmath169 in addition to the triangle diagrams with the couplings of @xmath26 and @xmath170 mesons to the magnetic field from eq . ( [ mag ] ) , there are also two triangle diagrams with the coupling of the photon to charged @xmath26 and @xmath170 mesons that arises due to gauging their kinetic terms . these are shown in fig . [ gauge ] . the sum of these two diagrams yields @xmath171 so far we have only included the interactions coupling the photon to @xmath26 and @xmath170 mesons . there are additional diagrams where the photons couple to @xmath60 and @xmath172 mesons that give an equal contribution . [ contact ] shows the loop diagrams with the contact interaction that arises from gauging the coupling @xmath4 . [ contact](a ) yields @xmath173 where @xmath174 for @xmath175 and @xmath176 . [ contact](b ) is related to fig . [ contact](a ) by charge conjugation so @xmath177 the graphs in fig . [ contact](c ) and fig . [ contact](d ) both vanish , so the total contribution to the amplitude from loops with contact interactions is @xmath178 . only diagrams with charged and strange @xmath26 mesons in the loop will contribute . the total amplitude from loop diagrams in fig . [ triangle ] with neutral @xmath26 mesons is given by @xmath179\ } \ , .\end{aligned}\ ] ] adding the five diagrams with the photon coupling to a @xmath60 or @xmath172 doubles this contribution . the contribution from diagrams of fig . [ triangle ] with charged and strange charmed mesons in the loops is obtained by substituting @xmath180 with @xmath181 and @xmath182 with @xmath183 . in addition , the contributions from the diagrams in fig . [ gauge ] and fig . [ contact ] with charged and strange charmed mesons in the loops need to be included . the full decay amplitude is @xmath184 where the superscript @xmath185 indicates a contribution from loops with charged ( strange ) @xmath26 mesons . the decay rate for @xmath186 is given by @xmath187=\frac{1}{8\pi}(\sqrt{m_{j/\psi}m_{\eta_c}}m_{full})^2\frac{\|\vec{q}\,\|}{m_{j/\psi}^2}.\end{aligned}\ ] ]	we investigate the effect of intermediate charmed meson loops on the m1 radiative decays @xmath0 and @xmath1 as well as the isospin violating hadronic decays @xmath2 using heavy hadron chiral perturbation theory ( hh@xmath3pt ) . the calculations include tree level as well as one loop diagrams and are compared to the latest data from cleo and bes - iii . our fit constrains the couplings of 1s and 2s charmonium multiplets to charmed mesons , denoted @xmath4 and @xmath5 , respectively . we find that there are two sets of solutions for @xmath4 and @xmath5 . one set , which agrees with previous values of the product @xmath6 extracted from analyses that consider only loop contributions to @xmath2 , can only fit data on radiative decays with fine - tuned cancellations between tree level diagrams and loops in that process . the other solution for @xmath4 and @xmath5 leads to couplings that are smaller by a factor of 2.3 . in this case tree level and loop contributions are of comparable size and the numerical values of the tree level contributions to radiative decays are consistent with estimates based on the quark model as well as non - relativistic qcd ( nrqcd ) . this result shows that tree level hh@xmath3pt couplings are as important as the one loop graphs with charmed mesons in these charmonium decays . the couplings @xmath4 and @xmath5 are also important for the calculations of the decays of charmed meson bound states , such as the @xmath7 , to conventional charmonia . many of the static properties and decays of charmonium states can be understood within a framework in which these states are viewed as non - relativistic bound states of charm and anticharm quarks . this includes the quark model @xcite as well as the modern qcd - based approach of non - relativistic qcd ( nrqcd ) @xcite , which allows for systematic treatment of charmonium properties as an expansion in @xmath8 and @xmath9 , where @xmath9 is the relative velocity of the charm - anticharm quarks . despite many successes there remain specific transitions that are not well understood quantitatively . examples of decays that are not completely understood are the hadronic decays @xmath10 , and the radiative decays to @xmath0 and @xmath11 . the hadronic decays violate isospin , in the case of a final state with @xmath12 , or @xmath13 , when the final state is @xmath14 . as a consequence the ratio of this decay is sensitive to light quark masses @xcite . the value of the light quark mass ratio extracted from the measured decay rates @xcite , @xmath15 , differs significantly from the result extracted from meson masses in chiral perturbation theory , @xmath16 @xcite . for the radiative decays the experimentally measured rates differ from quark model expectations . for example , a non - relativistic quark model calculation of @xmath17 yields a prediction of @xmath18 ( @xmath19 ) kev , vanishes in the non - relativistic quark model due to the vanishing overlap of the orbital wavefunctions of the @xmath20 and the @xmath21 , and is no longer zero once relativistic corrections are taken into account . however , quark models that include relativistic corrections still have trouble reproducing the correct rate for @xmath22 @xcite . ] whereas the experimental results are 1.57 @xmath23 0.38 ( 0.97 @xmath23 0.14 ) kev @xcite . in ref . @xcite , nrqcd is used to analyze the decay @xmath24 , and the authors show that @xmath25 corrections can lower the rate so that the theoretical prediction is consistent with data . however , no attempt has been made to understand radiative decays of @xmath20 in this framework . for reviews of these puzzles and others in charmonium physics , see refs . @xcite . recently , ref . @xcite proposed that the hadronic decays mentioned above are dominated by loop diagrams with virtual @xmath26 mesons . the decays are calculated using heavy hadron chiral perturbation theory ( hh@xmath3pt ) @xcite , in which the charmonia are treated non - relativistically and coupled to the @xmath26 mesons and goldstone bosons in a manner consistent with heavy quark and chiral symmetries . in a non - relativistic theory the @xmath26 meson kinetic energy scales as @xmath27 and momentum scales as @xmath28 , where @xmath29 is a @xmath26 meson mass and @xmath30 is the typical velocity of the @xmath26 mesons in the loops . with this scaling , ref . @xcite showed that the loop diagrams with @xmath26 mesons should be enhanced over tree level couplings by a factor of @xmath31 . the rates for @xmath32 and @xmath33 are sensitive to the product @xmath6 , where the @xmath34 coupling to @xmath26 mesons is @xmath4 and the @xmath20 coupling to @xmath26 mesons is @xmath5 . ref . @xcite found a value of @xmath6 consistent within errors with the two experimentally measured rates . this resolves the disagreement between the value of @xmath35 extracted from these decays and other extractions , since the prediction for the ratio of rates in terms of @xmath35 relied on the rates being dominated by the tree level hh@xmath3pt coupling . the value of @xmath6 extracted by ref . @xcite is consistent with power counting estimates of @xmath4 and @xmath5 , which are both expected to be @xmath36 up to constants of order unity . other hadronic and radiative charmonium decays are also analyzed within the same formalism in refs . @xcite . the goal of this paper is to apply the same theory to the radiative decays mentioned above . one of our aims is to check whether the theory can also successfully resolve puzzles in radiative decays as one would hope . it is also important to check that couplings extracted from the hadronic decays are consistent with data on radiative decays . an important aspect of our analysis is that unlike refs . @xcite , tree level counterterms are included in our calculations of both hadronic and radiative decays . ref . @xcite argued for an additional factor in the loop graphs of @xmath37 , for @xmath38 , which would compensate the @xmath31 enhancement of the loops . this factor , and the fact that @xmath39 is not very small , support including both the loops and tree level interactions in the calculation , which we will do in this paper . this can have an important impact of the extracted values of the couplings @xmath4 and @xmath5 . finally , an additional motivation for our analysis is that the extracted couplings are important for the physics of the @xmath7 and other recently discovered charmonium bound states that have been interpreted as charmed meson molecules . if the @xmath7 is a charmed meson bound state , then the coupling @xmath40 is an important theoretical input for calculations of @xmath41 , so extraction of @xmath4 and @xmath5 is relevant to unconventional as well as conventional charmonia . for theoretical calculations of @xmath7 to conventional charmonia using effective field theory , see refs . @xcite . our main result is that in order to obtain a consistent fit to both radiative decays as well as the hadronic decays considered in refs . @xcite , counterterm contributions must be included and the values of @xmath4 and @xmath5 will then be smaller than estimated in an analysis containing only the loop diagrams by a factor of 2.3 . this decreases the overall size of the loop amplitude by a factor of 5 . it is not possible to get reasonable agreement with radiative decay data without including counterterms . since nrqcd is the microscopic theory of charmonia , and does not include loop effects from charmonia , one is tempted to identify the result of a calculation of the @xmath0 amplitude in nrqcd with the tree level coupling in hh@xmath3pt this is somewhat tenuous as the bare coupling in our theory has an infinite piece that must cancel the linear divergence in the meson loop integrals . nevertheless , we regard it as satisfying that the size of the counterterms we extract in our fit with the smaller values of @xmath4 and @xmath5 are consistent within a factor of 2 with the quark model and nrqcd calculations of the radiative transitions . for other extractions of the couplings @xmath4 and @xmath5 in different theoretical frameworks , see , e.g. , refs . @xcite . in refs . @xcite , the charmed meson loop corrections to radiative @xmath34 and @xmath20 decays are studied in a version of hh@xmath3pt with relativistic propagators and couplings , as well as form factors at the vertices that regulate ultraviolet divergences . the form factors introduce an additional parameter into the calculations . these authors did not attempt to simultaneously fit the hadronic decays but used values of @xmath4 and @xmath5 consistent with those obtained in refs . @xcite . their results are also consistent with the experimental data on the radiative decays . the effective hh@xmath3pt lagrangian relevant to the hadronic decays is @xcite @xmath42+\frac{\delta}{4}tr[h^{\dagger}_{a}\vec{\sigma}h_{a}\vec{\sigma } ] -\frac{g}{2}tr[h^{\dagger}_ah_b\vec{\sigma}\cdot\vec{u}_{ab } ] \\ \nonumber & + & i\frac{a}{4}\left(tr[j'\sigma^ij^{\dagger}]-tr[j^{\dagger}\sigma^ij']\right)\partial^i(\chi_-)_{aa } + i\frac{g_{2}}{2}tr[j^{\dagger}h_{a}\vec{\sigma}\cdot\overleftrightarrow{\partial}\bar{h}_a]+ h.c . \ , .\end{aligned}\ ] ] here @xmath43 and @xmath44 are the charmed and anti - charmed meson multiplets with @xmath45 and @xmath46 denoting the vector and pseudoscalar charmed mesons , respectively , and @xmath47 denotes the charmonium multiplets with @xmath48 and @xmath49 . the @xmath50 are the pauli matrices , @xmath51 and @xmath52 denote flavor indices , and @xmath53 . the first two terms in eq . ( [ lag ] ) are kinetic terms for the charmed mesons , @xmath54 is the hyperfine splitting , and @xmath55 is the mass of pseudoscalar ( vector ) charmed meson . the third term contains the interactions of @xmath26 mesons with the goldstone boson fields which are contained in @xmath56 where @xmath57 is a @xmath58 matrix of goldstone boson fields and @xmath59 mev is the pion decay constant . there are identical terms for the @xmath60 mesons which are not explicitly shown . the tree level couplings for @xmath61 come from the term with coupling constant @xmath62 . the factor @xmath63 is defined by @xmath64 , where @xmath65 , @xmath66 @xmath67 and @xmath68 are the light quark masses and @xmath69 . finally , the 1s charmonia couple to the @xmath26 mesons via the last term with coupling @xmath4 . the same term , with @xmath70 and @xmath4 replaced with @xmath71 and @xmath72 , couples the @xmath73 charmonia to charmed mesons . the tree level decay amplitudes are @xcite , @xmath74 where @xmath75 and @xmath76 . to leading order in the chiral expansion , these factors may be expressed in terms of light meson masses : @xmath77 and @xmath78 . the @xmath79 mixing must also be included , and the mixing angle is @xmath80 when this mixing is included the first matrix element in eq ( [ amps ] ) is multiplied by 3/2 . the loop diagrams contributing to the decay have been evaluated in refs . @xcite . since the decay to @xmath81 vanishes in the isospin ( @xmath13 ) limit , the diagrams cancel in the sum over @xmath82 , and @xmath83 appearing in the loop in the limit that all these mesons are degenerate . mass differences between the mesons render the cancellation incomplete and are responsible for the finite contribution . for electromagnetic decays , we need to add couplings to the magnetic field and gauge the interactions in eq . ( [ lag ] ) . the tree level coupling of the charmonia to the magnetic fields is given by @xcite @xmath84 + \frac{\rho^\prime}{2}(tr[j^\prime\vec{b}\cdot\vec{\sigma}j^{\dagger } ] + h.c . ) + \frac{\rho^{\prime \prime}}{2}tr[j^\prime\vec{b}\cdot\vec{\sigma}j^{\prime \,\dagger } ] \ , , \end{aligned}\ ] ] where @xmath85 is the magnetic field . due to the presence of pauli matrices these terms break heavy quark spin symmetry . the first term is responsible for the decay @xmath0 , the second for @xmath86 , and the third for @xmath87 . for the loop corrections to the radiative decays , we must also include the coupling of the charmed mesons to the magnetic field , which is given by @xcite @xmath88 + \frac{e}{2m_{c}}q'tr[h_{a}^{\dagger}\vec{\sigma}\cdot\vec{b}h_{a}],\end{aligned}\ ] ] where @xmath89 , @xmath90 , and @xmath91 is the mass of charm quark . these terms are responsible for the decays @xmath92 . including leading as well as @xmath93 suppressed terms is crucial for reproducing observed @xmath94 rates @xcite . ref . @xcite finds that a good fit to the experimental rates is obtained for the values @xmath95 gev and @xmath96 . these couplings enter the radiative decays of charmonia through the triangle diagrams shown in fig . [ triangle ] . ) .,title="fig:",scaledwidth=100.0% ] there are also interactions that arise from gauging the derivatives in eq . ( [ lag ] ) . gauging the derivatives in the kinetic term for the @xmath26 mesons leads to couplings to the photon which contribute to the radiative decays via triangle loop diagrams shown in fig . [ gauge ] . ) .,title="fig:",scaledwidth=80.0% ] gauging the coupling @xmath97 leads to a contact interaction that directly couples charmonia , heavy mesons and the photon field , which is given by @xmath98+h.c.\end{aligned}\ ] ] where @xmath99 or @xmath100 only , i.e. , only charged and strange @xmath26 mesons appear in the interaction term of eq . ( [ contactint ] ) . the loop diagrams with contact interactions are shown in fig . [ contact ] . the tree level amplitude for the @xmath101 decay , for example , is @xmath102 where @xmath103 denotes the momentum of photon , and @xmath104 and @xmath105 are the polarization vectors of the photon and @xmath34 , respectively . the corresponding decay rate is @xmath106&=&\frac{1}{8\pi}\|\sqrt{m_{j/\psi}m_{\eta_c } } \mathcal{m}_0\|^2\frac{\|\vec{q}\,\|}{m_{j/\psi}^2 } , \nonumber \\ & = & \frac{\rho^2}{12 \pi } \frac{m_{\eta_c}}{m_{j/\psi } } \|\vec{q}\,\|^3 \ , .\end{aligned}\ ] ] here the factor @xmath107 comes from the normalization of nonrelatvistic fields in hh@xmath3pt . in the non - relativistic quark model , @xmath108 , and the predicted decay rate is about a factor of 2 too large . the contribution to the amplitude from meson loops is also proportional @xmath109 so the loops give an additive shift to the @xmath110 term for each decay . we evaluate the loops in pure dimensional regularization so linear divergences do not appear and the corrections from all loops are finite . the full rate is eq . ( [ decay ] ) with @xmath111 replaced by @xmath112 , where the @xmath112 includes both the tree level interaction and the contributions from neutral , charged , and strange meson loops . the explicit expression for @xmath112 can be found in the appendix . before proceeding to our fits to the data , we will briefly discuss the power counting for the diagrams we have shown . as stated earlier , for non - relativistic @xmath26 mesons one takes @xmath113 , @xmath114 , so the propagators scale as @xmath115 and the loop integration measure is @xmath116 . the vertices coupling the charmonium to @xmath26 mesons carry a factor of @xmath114 . to estimate @xmath117 one may take the difference between an external and two internal mesons , so @xmath118 where @xmath119 is the mass of one of the external charmonia and @xmath120 is the mass of two @xmath26 mesons in the loop . this leads to an estimate ranging from @xmath121 ( @xmath122 and @xmath123 to @xmath124 ( @xmath125 and @xmath126 ) . naively with this counting the triangle diagrams scale as @xmath127 . the first factor comes from the loop integration factor , the second from the propagators , the third factor from the derivative couplings of charmonium to @xmath26 mesons , and the factor of @xmath103 is the photon or pion momentum which comes from the coupling of these particles to @xmath26 mesons . the diagrams with the contact interaction scale as @xmath128 which is the same as the triangle graph . this is because there is no derivative in the contact interaction with the photon , the derivative in the charmonium @xmath26 meson coupling must turn into the factor of @xmath103 required by gauge invariance or chiral symmetry and there are only two propagators . the factor @xmath103 is common to all diagrams including the tree level diagrams . factors of @xmath29 are compensated by other dimensionful couplings so we will focus only on counting powers of @xmath39 from here on . so the triangle graphs and graphs with the contact interaction are @xmath39 suppressed relative to the tree level interactions . however , these hadronic decays violate either isospin or @xmath13 and the radiative decays violate heavy quark symmetry so there are cancellations between graphs due to heavy meson mass differences that are missed by this power counting . one can formally modify the power counting counting in the following way . the inverse propagator for a non - relativistic meson can be written as @xmath129 , where the residual mass term in the propagator has been split into a term @xmath52 which is common to all @xmath26 meson states and a term @xmath130 contains @xmath13 breaking and hyperfine splittings that are different for different @xmath26 mesons . expand the @xmath26 meson propagators as @xmath131 the graph in which all propagators contribute only the first term is zero by symmetry . in order to get a non - vanishing result at least one propagator in the graph must give a contribution from the second term , then the power counting says the graph is enhanced by a factor of @xmath132 , which makes the graph @xmath31 enhanced rather than @xmath39 suppressed relative to the tree level diagrams @xcite . since @xmath39 is not very small this could be compensated by other numerical factors . in practice it is easier to simply calculate the graphs with the unexpanded propagators but expanding the propagator makes it clear that after summing over all graphs one gets a @xmath31 enhancement . in this paper , we will take the viewpoint that the leading one loop diagrams are of roughly the same size as the tree level contributions and include both in the decays , then try to simultaneously fit the radiative and hadronic decays mentioned above . a separate question is whether higher order chiral corrections are under control . certainly some chiral corrections are suppressed as argued for different charmonium radiative decays in ref . @xcite . but in a subgraph with a ladders of single pion exchanges between a pair of @xmath26 mesons , non - relativistic power counting shows that the ladder with @xmath133 single pion exchanges is suppressed relative to one with @xmath134 pion exchanges by a factor @xmath135 @xcite where @xmath136 is the relative momentum of the @xmath26 mesons . this would require @xmath137 with @xmath138 to be less than @xmath139 . in some channels a resummation of single pion exchanges may be needed to do accurate calculations . such a resummation is beyond the scope of this paper . here we are simply interested in the impact that including the tree level interactions and simultaneously fitting the radiative and hadronic decays has on the values of @xmath4 and @xmath5 and therefore the size of @xmath26 meson loop contributions to charmonium decays . to constrain the parameters @xmath4 and @xmath5 , we determine the parameter @xmath62 and the product @xmath6 from the measured rates for @xmath140 $ ] and @xmath141 $ ] . because the predictions for the decay rates are quadratic in @xmath6 , this does not completely determine @xmath6 , but yields two possible solutions . then we fix the relative size of the two couplings using the relation @xmath142 , which follows if the dimensionless coupling of the @xmath34 and @xmath20 to @xmath26 mesons is the same @xcite . once @xmath4 and @xmath5 are determined this way from the hadronic decays , the only parameters remaining in the radiative decays are @xmath110 , @xmath143 , and @xmath144 , which can be determined from the three decay rates @xmath145 $ ] , @xmath146 $ ] , and @xmath147 $ ] . .the numerical results for fitting parameters to hadronic and radiative charmonium decays . the fit is explained in the text . the quark model @xcite predictions for the parameters @xmath110 , @xmath143 and @xmath148 are shown in the bottom line . errors are due to experimental uncertainties only . [ cols="<,^,^,^,^,^",options="header " , ] in summary , we have computed the decay rates for @xmath0 and @xmath149 including tree level and one loop diagrams with charmed mesons in hh@xmath3pt . we combined our results with the decay rates for @xmath61 found in refs . @xcite , used the relationship @xmath150 , and fit the five remaining coupling constants simultaneously . including tree level couplings is essential for simultaneously reproducing all the decay rates . a smaller value of @xmath151 is required to avoid large cancellations between tree level and charmed meson loop contributions to the radiative decay . the tree level couplings @xmath110 and @xmath148 in this fit are consistent ( to within a factor of 2 ) with expectations based on the quark model and nrqcd . this work was supported in part by the director , office of science , office of nuclear physics , of the u.s . department of energy under grant numbers de - fg02 - 05er41368 ( t.m . ) and de - fg02 - 05er41367 ( d.y . ) . we thank the department of energy s institute for nuclear theory at the university of washington for its hospitality during the completion of this work .
active galaxies , galaxies accreting significant material onto a central supermassive blackhole , are some of the most intriguing objects in the universe . the first active galaxies were identified based on their extreme luminosities , emission line properties and ability to vary in luminosity over short time periods . since then , several selection techniques have been employed to identify and study the active galaxy population . however , these techniques also imposed redshift limits on the selected samples . quasars , the most luminous class of active galactic nuclei ( agn ) , can be identified out to high redshifts using most selection techniques , while the intrinsically fainter agns ( e.g. seyfert galaxies ) are mostly limited to local samples . this has led to a prominent gap in our current understanding of the agn phenomenon ; the connection between bright qsos which exist in large numbers at redshifts of z@xmath323 and low - luminosity seyferts which have historically been studied at much lower redshifts . understanding how the population of faint agns evolves with redshift is of particular importance for determining the frequency and total space density of agns at earlier epochs . this has obvious implications for determining their total contribution to the x - ray , ir and uv backgrounds . large numbers of low - luminosity agns have been proposed to explain the ionization of the intergalactic medium at high redshift ( steidel & sargent 1989 ) , although recent observational and theoretical results differ on the amount of their true contribution ( barger @xmath7 2003a ; schirber & bullock 2003 ) . additionally , the faint end of the agn luminosity function ( lf ) is an important constraint on evolutionary models such as pure luminosity and luminosity - dependent density evolution ( e.g. boyle @xmath7 2000 ) . the agn luminosity function has been well studied out to high redshifts for qsos ( e.g. hartwick & schade 1990 ) and several works measure the local luminosity functions for seyfert galaxies ( cheng @xmath7 1985 ; huchra & burg 1992 ; koehler @xmath7 1997 ; ulvestad & ho 2001 ; londish @xmath7 2000 ; maia @xmath7 2003 ; hao @xmath7 2005 ) . most seyfert lfs have been based on optical spectroscopically selected nuclei . however , spectroscopically selected samples of agns at high redshift are harder to obtain due to the inevitably lower signal - to - noise spectra , combined with the lack of suitable diagnostic emission lines in the optical portion of the spectrum that allow identifying agns . another identifying characteristic of active galaxies is their variability . qsos and seyfert galaxies have long been known as variable objects , with significant optical flux changes occurring on timescales of months to several years . while variability has been primarily used to identify qsos ( hawkins 1986 ; koo @xmath7 1986 ; hook @xmath7 1994 ) , seyfert 1 galaxies have also been shown to vary in broad - band photometric surveys ( e.g. peterson @xmath7 1998 ) . a survey for variable sources in sa57 revealed many optically extended , seyfert - like galaxies ( bershady @xmath7 1998 ) which generally had higher variability amplitudes than the more luminous qsos . this result suggests that variability is a good technique for selecting intrinsically faint qsos and seyfert nuclei . the purpose of this study is to employ two independent and complementary selection techniques to identify seyfert - luminosity agns to redshifts of z@xmath01 . we analyze the deep extragalactic evolutionary probe ( deep ) spectroscopic survey data of the groth - westphal survey strip ( gss ; groth @xmath7 1994 ) to detect and measure emission lines to determine the ionizing source within the galaxy and thus probe the population of active galaxies to z@xmath01 . we further select candidate agns based on nuclear variability . three - quarters of the gss was re - imaged with hst @xmath07 years after the original wfpc2 images were obtained . we identify galaxies whose nuclei have significantly varied over this time interval in order to select potential active galaxies . thus , we compare / contrast the types of agn identified using these two established optical selection techniques . in addition , we investigate how the agn identified via optical selection manifest themselves in deep x - ray surveys of the gss . this work constitutes the first comparison of these three agn selection techniques for z@xmath01 galaxies and provides an important benchmark for future studies of agn in deep surveys . we describe the spectroscopic observations and line measurements in 2 . in 3 , we discuss the hst photometric survey and the identified variable galaxies . the selected agn / agn candidates are discussed and compared to typical gss galaxies in 4 and 5 . in 6 , the x - ray sources in the gss are discussed and compared to the results of the spectroscopic and variability surveys . we compute the relative fraction of type 1 and 2 agns in 7 and summarize the results in 8 . the deep extragalactic evolutionary probe ( deep ; koo @xmath7 1996 ; simard @xmath7 2002 ; vogt @xmath7 2005 ; weiner @xmath7 2005 ) is a galaxy redshift survey designed to study the formation and evolution of distant field galaxies and large scale structure to z@xmath01 . this paper uses the first phase of deep ( deep1 ) which was carried out using the low - resolution imaging spectrometer ( lris ; oke @xmath7 1994 ) on the keck telescopes . the second phase of deep ( deep2 ; davis @xmath7 2003 ) is currently under way and uses the deimos spectrograph ( faber @xmath7 2005 ) on the keck 2 telescope . the groth survey strip is one of the fields targeted by deep for spectroscopic observations and comprises 28 contiguous wfpc2 fields located at 14h17m+52 , imaged in the v(f606w ) and i(f814w ) filters . the fields are numbered 4 through 31 diagonally across the sky . the gss object ids used throughout this paper indicate the gss field ( first 2 digits ) and the wide field ccd number ( third digit ) in which each object was identified . optical spectra were obtained through multi - slit masks for 813 objects in the gss between 1995 and 2001 . the selection of objects is extensively discussed by vogt @xmath7 ( 2005 ) , while the description of the reduction of spectroscopic data and redshift determination are found in weiner @xmath7 ( 2005 ) . in summary , each mask was observed with a blue and red grating for a combined wavelength coverage of @xmath05000 8200 @xmath8 depending on the slit placement on the mask . the spectral resolution is fwhm=2.9 @xmath8 in the blue and 4.2 @xmath8 in the red corresponding to 2 3 pixel sampling per resolution element . typical exposure times were one hour per grating setup with fainter targets exposed for a few hours . since most galaxies have sizes comparable to the seeing disk ( @xmath01@xmath9 ) , spectral information along the spatial direction is not available in most cases and a one - dimensional spectrum has been produced for each object by summing several pixels along the spatial axis . six - hundred and twenty galaxy spectra were obtained in the survey and 576 have high enough signal - to - noise to yield redshifts with confidence levels of 90% and higher and allow measuring spectral line parameters ( fluxes , widths and equivalent widths ) . for the purpose of identifying active galaxies , we measure several emission and absorption lines indicative of agn activity . we considered only emission / absorption lines with signal - to - noise measurements greater than 5:1 in this paper . the line measurements were performed using two independent fitting techniques . the primary fitting algorithm was developed by one of the authors ( a. c. p. ) and employs a maximum likelihood estimation to perform an automated gaussian fit to emission / absorption lines . the extracted sky spectrum was used to determine the noise at each position along the spectrum . for about 10% of the galaxies , however , the sky spectrum necessary for the error determination was not available in the deep database . for these objects we manually fit the line profiles using splot in iraf as described in kobulnicky @xmath7 ( 2003 ) . each line was fit individually with a gaussian profile except for the [ oii ] doublet , where two gaussian profiles were fit simultaneously and the sum recorded . we empirically estimate the error in the line flux as @xmath10@xmath11=@xmath12@xmath13rms , where 12 is the number of pixels summed in a given emission line for this resolution , and rms is the root - mean - squared variations in an adjacent offline region of the spectrum . the equivalent width error is determined as @xmath14 where l , c , @xmath10@xmath11 , and @xmath10@xmath15 are the line and continuum levels in photons and their associated 1@xmath10 uncertainties . the error in the continuum is determined by fitting the baseline regions around each line . we used both fitting techniques to measure a broad sample of emission and absorption lines from several galaxy spectra in order to compare the measured values and errors . the line flux , width , continuum values and errors were in general agreement to within 5% . for equivalent width measurements , the determination of the continuum level is particularly important . in almost all cases , the continuum has been averaged over a large number of pixels in the wavelength direction , which allows this value to be well determined . the typical s / n in the continuum is well above 5 around the majority of emission / absorption lines for most sources . those with the weakest continuum measurements ( just under 1 count per 1500s of exposure time ) have s / n of @xmath01.5 to 2 . active galactic nuclei can be detected in galaxy spectra through the presence of broad , permitted lines covering a wide range of ionization . the galaxy spectra in the gss span a wide redshift range ( 0.05@[email protected] ) and thus we do not have complete coverage of any one particular line for the entire set of spectra . therefore , we have identified and measured several emission lines known to be broadened in seyfert 1 nuclei or qsos ( i.e. broad - line agns or blagns ) . these lines include the strongest hydrogen balmer lines ( h@xmath16 and h@xmath17 ) , mgii ( @xmath182800@xmath8 ) and cii ( @xmath182326@xmath8 ) . we also measured ciii ] ( @xmath181909@xmath8 ) , civ ( @xmath181549@xmath8 ) and ly@xmath16 ( @xmath181215@xmath8 ) for the higher redshift sources . the definition of broad " for blagns is not well defined in the literature . the criterion in terms of fwhm ranges between 1000 and 2000 km / s ( e.g. weedman 1977 , steidel @xmath7 2002 ) . recently , hao @xmath7 ( 2005 ) found a bimodal distribution of h@xmath16 fwhms in a study of over 80,000 galaxy spectra from the sloan digital sky survey ( sdss ) . these results suggest that blagns can be classified as those galaxies with fwhm(h@xmath16 ) @xmath2 1200 km / s . for the sake of uniformity , we have applied this criterion to all possibly broadened emission lines in the spectral range of the gss galaxies to identify blagns . in fact , the measurements discussed here show little ambiguity in the identification of blagns which display marked differences from their narrow - line counterparts . of the 576 gss spectra , 563 have at least one of the necessary emission lines within the spectral range . h@xmath16 is in the wavelength range for 129 galaxies and found in emission for 55 sources though none displays widths greater than 400 km / s . the h@xmath17 line is in emission in 226 out of 410 possible galaxy spectra . one galaxy , 283_3452 , displays broadened h@xmath17 with a velocity of 4900 km / s . only seven out of 201 spectra with mgii in range display this rare emission line . of these seven galaxies , six are broad with velocities between 4300 and 8300 km / s . two of the six display considerable self - absorption of the broadened mgii emission line preventing an accurate measurement of the velocity width ( 085_5273 and 152_6235 ) . the mgii emission in 152_6235 is particularly weak compared to the broad emission lines in the other galaxies . however , there is also evidence of ciii ] at the blue edge of the spectrum which also appears slightly broad . this adds some confidence to the classification of this galaxy as a blagn , though we note that this classification is less certain than the others . one other galaxy with narrow mgii emission , 092_3172 , also displays mgii absorption lines on either side of the emission line but the overall shape of the emission does not appear to be broad . two of 48 galaxies with cii in range show the line in emission . one of these , 083_5407 , is broadened and also contains broad mgii . none of the spectra contain civ emission . only three high redshift sources contain ly@xmath16 in their wavelength range . two of these show the line in emission with widths less than 300 km / s . thus , a total of 7 galaxies , one with broad h@xmath17 , five with broad mgii alone and one with mgii and cii , display broadened emission lines meeting our criteria . these galaxies are listed in table 1 in order of increasing redshift . the spectra and images of these objects are shown in the first 7 panels of figures 1 and 2 . the presence of [ nev](@xmath183426@xmath8 ) indicates an ionizing potential of at least 97.02 ev , too high to be produced through star - formation alone but possible through accretion onto a supermassive blackhole ( e.g. hall @xmath7 2000 ) . we detect [ nev ] in only 3 of the 393 galaxies with this line within their spectral range ( see table 2 and figures 1 and 2 ) . the velocity widths of the [ nev ] line range from 390 to 650 km / s . one of these , 073_7749 , shows additional balmer absorption lines indicative of a post - starburst galaxy . the other two are also classified as blagns based on broad h@xmath17 and mgii . objects 292_3076 and 083_0815 also show weak [ nev ] emission , though it is below our signal - to - noise ratio for a significant detection . broadened [ oiii](@xmath185007@xmath8 ) is also an indicator of agn activity . the analysis of sdss spectra ( hao 2003 ; zakamska @xmath7 2003 ) reveals that agns typically have fwhm([oiii ] ) @xmath2 400 km / s whereas widths for normal star - forming galaxies are much lower . the [ oiii ] line is in the spectral range of 386 galaxies , being detected in 192 . only two ( 142_4838 and 292_3076 ) reveal broadened [ oiii ] with velocities of 450 km / s and 820 km / s respectively and are listed in table 2 ( spectra and images in figures 1 and 2 ) . object 142_4838 is also a blagn . a number of criteria have been used to identify narrow - line agns ( nlagns ) , which can be differentiated from normal star - forming galaxies using the emission - line ratios of the most prominent optical lines such as [ oii](@xmath183727@xmath8 ) , [ oiii](@xmath184959,5007@xmath8),[nii](@xmath186583@xmath8 ) , [ sii](@xmath186716,6731@xmath8 ) , h@xmath16 and h@xmath17 ( e.g. veilleux & osterbrock 1987 ) . for a large fraction of galaxies in the gss , however , many of these lines are redshifted out of the optical range . in addition , the gss spectra are not flux calibrated , making it impossible to compare flux ratios for emission lines in different regions of the spectrum . for these reasons , most traditional line ratio diagnostics are not applicable to our survey data . and [ nii]/[sii ] are close in wavelength and thus are not significantly affected by the lack of flux calibration , we did investigate this line ratio diagnostic but found no convincing agn candidates among those galaxies with these lines in their spectra . ] a novel emission line diagnostic to identify nlagns proposed by rola , terlevich & terlevich ( 1997 ; rtt ) uses the equivalent widths of the [ oii ] and h@xmath17 and allows classifying galaxies to [email protected] without the need flux calibration . two distinct zones define the agn region of the diagram , at ew(h@xmath17)@xmath110 and ew(oii)/ew(h@xmath17)@xmath23.5 . using a sample of local emission line galaxies , rtt find that 87% of the agns reside in these regions while 88% of the hii galaxies fall in the remaining region ( see their figure 3b ) . the agn and hii samples are comprised of integrated galaxy spectra while the liner spectra are extracted from the nuclear regions of the galaxies . using [email protected] emission - line galaxies from the canada - france redshift survey ( tresse @xmath7 1996 ) , rtt find good agreement between their technique and classical diagnostic diagrams . the cfrs sample also contains spatially integrated spectra as deep1 , though , with a much lower resolution . although this technique does not perfectly separate the star - forming galaxies from agns , it does a fairly good job of identifying the majority of nlagns in a sample of emission line galaxies . of the 576 galaxies in our spectroscopic survey , 318 contain [ oii ] and h@xmath17 in their spectral range , but only 115 display both lines in emission with greater than 5@xmath10 signal - to - noise . in two cases , the continuum at h@xmath17 was negative due uncertainties in the sky subtraction , making it impossible to measure the galaxies equivalent widths . with the removal of these sources , we plot the remaining 113 on the rtt diagram in figure 3a with a uniform correction of 3 @xmath8 for underlying stellar absorption ( kennicutt @xmath7 1992 , tresse @xmath7 1996 ) . this correction has the effect of moving objects toward the hii / star - forming ( sf ) galaxy region of the diagram . based on the rtt criterion , 39 out of 113 galaxies ( 35% ) fall in the agn regions of the diagram . this represents only those galaxies where the 1@xmath10 errorbars remain within the agn regions of the diagram . as an additional check on the validity of the agn criterion from rtt , we have analyzed a local sample of emission line galaxies from the kitt peak international spectroscopic survey ( kiss ) ( salzer @xmath7 2000 ; gronwall @xmath7 2004 ) using the rtt equivalent width diagram . kiss is a large - area , objective prism survey of galaxies selected on the basis of strong h@xmath16 emission . at redshifts less than z=0.1 , 318 galaxies from the kiss sample have follow - up spectroscopic observations that include the [ oii ] and h@xmath17 emission lines . twenty - six of these local galaxies are classified as agn / liners using classical line ratio diagnostic diagrams ( see gronwall @xmath7 2002 ) . figure 3b shows the kiss star - forming galaxies ( blue asterisks ) and agn / liners ( red triangles ) . we have added 3 @xmath8 for underlying stellar absorption as done for the gss galaxies . we find that a threshold of ew(oii)/ew(h@xmath17)@xmath25 ( dashed line in figures 3a and 3b ) maximizes the differentiation of the agns and normal star - forming kiss galaxy populations . with this new threshold , galaxies in the upper 2 quadrants ( agn regions ) yield a 44% probability of being an agn while the lower - right quadrant ( star - forming galaxies ) yields a 1% chance of being an agn . the lower - left quadrant is a mixture of agns and star - forming galaxies ( 13% agn ; 87% sf ) . it is unclear why there is such disagreement between the results of rtt and the kiss emission - line galaxy analysis using their diagram . however , we will use the more conservative estimate of nlagns based on the kiss galaxy sample . assuming our emission - line galaxies are like the kiss sample , the 7 gss galaxies which lie above the ew(oii)/ew(h@xmath17)@xmath25 threshold in figure 3b have a 44% chance of being nlagns . an additional 15 galaxies lie in the lower - left quandrant containing a mix of agns and star - forming galaxies . table 3 lists the 7 moderate probability " nlagns followed by the 15 lower probability nlagn candidates . the galaxy images are shown in figure 2 . a possible indicator of agn activity suggested by becker @xmath7 ( 1997 ) is the presence of strong mgii and/or feii absorption . hall @xmath7 ( 2000 ) identified 15 absorption - line selected agn candidates from 0.73@[email protected] among the cnoc2 spectra . however , they point out that mgii absorption lines of comparable strength have also been identified in starburst galaxies ( e.g. storchi - bergmann @xmath7 1995 ) and thus such sources can only be classified as agn candidates . only one of their 15 was confirmed as a blagn through ir spectroscopy . through visual inspection , we identify 14 galaxies with significant mgii and/or feii absorption among the 200 gss spectra with these absorption features in their spectral range . we list these galaxies in table 4 but do not consider them strong agn candidates since there is no further agn evidence in any of the spectra . using a variety of selection techniques , we have identified active galaxies and agn candidates among the 576 gss galaxies with high quality spectroscopic data . we find 7 blagns , three [ nev ] detected agns ( two are also blagns ) and two broadened-[oiii ] detected agn ( one is also a blagn ) . based on [ oii ] and h@xmath17 equivalent width measurements , we have also tried to identify those narrow - line galaxies that may harbor agn . we find 7 galaxies with a moderate probability of being nlagns ( 44% ) and 15 galaxies with a marginal chance of being nlagns ( 13% probability ) which we hereafter identify as agn candidates . this results in a total of 31 unique galaxies identified as agn / agn candidates . those with the highest probability of being agns are the 16 galaxies identified through the presence of broad emission lines , the moderate probability agns from the ew diagram , the [ nev ] detected galaxies and the broad [ oiii ] galaxies . to identify possible agns via variability , we have analyzed two epochs of hst images for a portion of the gss . the original gss was observed in the spring of 1994 and consisted of 27 wfpc2 fields ( numbers 4 through 6 and 8 through 31 ) imaged in v ( f606w ) and i ( f814w ) with total exposure times of 2800s in v for each field . an adjacent field , known as the westphal field ( gss field number 7 ) , was observed in the v band for 24,400s . in the spring of 2001 , 18 of the gss fields were reobserved in the v band with exposure times of 4200s . an additional 3 fields , including the westphal field , were reobserved in spring of 2002 with exposure times of 4200s ( 7200s for the westphal field ) . thus , a total of 20 of the original 27 gss fields plus the westphal field have been analyzed for variability . for the remainder of the paper , we analyze the westphal field separately due to the significant difference in exposure time . the original gss images were combined and processed using standard techniques discussed in simard @xmath7 ( 1999 ; 2002 ) . unlike the images obtained in 1994 , the second epoch of images for the 21 gss fields were obtained in dither mode to allow for the construction of higher resolution , drizzled " images . for each field , 6 exposures with sub - pixel shifts were drizzled together to produce 0.05@xmath9 resolution images for the wide field ccds using the stsdas dither package in iraf . we then block averaged these images in two - by - two pixel blocks to produce the same resolution as the first epoch images obtained in 1994 ( 0.1@xmath9/pixel ) . we adopted the object coordinate catalogs for the gss described in vogt @xmath7 ( 2005 ) . all galaxy ccd positions were visually inspected to ensure centering on the nucleus of the galaxy . in cases where the galaxy morphology was irregular , the position was centered on the brightest pixel near the approximate center of the galaxy . for each of the 21 fields reobserved in 2001/2002 , we mapped the object positions from the epoch 1 fields onto the epoch 2 fields . several of the epoch 2 fields were not well aligned with the original images , having offsets of up to @xmath080 pixels . we used geomap and geoxytran in iraf to determine the offsets and apply the transformation to the epoch 1 catalogs . a final visual inspection of the epoch 2 images overlaid with object positions ensured that identical galaxy nuclei were identified in each epoch . and 0.2@xmath9 apertures the effects of poor centering should be more enhanced for concentrated galaxies than diffuse galaxies . we found no trend in the amount of variability vs. galaxy concentration and therefore conclude that the aperture centering in both epochs is robust and not a significant source of error . ] only galaxies in the wide field ( wf ) ccds were included in our photometric survey . the planetary camera ( pc ) ccd images do not extend as deep as the wf ccds due to the different pixel scale . additionally , the small field - of - view ( fov ) offers little contribution to the total sky areal coverage . the overlapping nature of the gss fields allows for coverage of the pc fov with one of the wf ccds of an adjacent field in most cases . a total of 6604 galaxy nuclei were identified for the photometric survey in the 20 gss fields and 529 in the deeper westphal field . aperture photometry was performed using iraf phot on the galaxies in each field and epoch . we chose an aperture radius of 1.5 pixels or 0.15@xmath9 to maximize light from an unresolved nuclear component and minimize light from the underlying host galaxy . this is the approximately the measured fwhm of unresolved stars in the wfpc2 images . we used a magnitude zero - point of 22.91 as determined for v - band ( f606w ) hst images ( wfpc2 synphot update , may 1997 ) . after performing aperture photometry , nuclear variability is determined from the magnitude difference of measurements for a source at both epochs . however , charge transfer efficiency losses which have occurred over the seven year interval produce a systematic offset that correlates with position on the ccd and magnitude . this is the well known cte effect ( whitmore @xmath7 1999 ; biretta @xmath7 2001 ) which causes targets farther from the ccd readout amplifier to appear fainter than similar targets near the amplifier . sarajedini , gilliland & kasm ( 2003 ; sgk ) performed a similar analysis on images of the hubble deep field separated by five years and discuss the role of cte losses in more detail . this effect must be empirically quantified and removed from the data in order to identify varying nuclei with statistical accuracy . we found that differences in the slope of the cte relation between the ccds were significant and could easily be determined for each of the three wf ccds individually . we combined the data for the 20 gss fields , separated into individual datasets for objects identified in wf ccd # 2 , wf ccd # 3 and wf ccd # 4 . the westphal field was fitted separately . we adopted a two - step approach to determine the cte relationship with object position and magnitude . first , we fit a linear surface to the x and y object positions and the magnitude differences using surfit in iraf for each ccd . the slopes of the relationship with x and y were found to be different for each of the 3 wf ccds . the fits were used to correct the magnitude difference values for position - dependent systematic offsets . secondly , we fit the dependence of the magnitude difference with nuclear magnitude . a 3rd order fit was required to properly model the dependence in all 3 wf ccds . this correction was then applied to the data to correct for the magnitude - dependent systematic offsets . the overall corrections to magnitude difference range from 0.05 to 0.25 magnitudes with an average offset of 0.15 centering the distribution around zero . the points in figure 4 show nuclear magnitude vs. cte - corrected magnitude difference for galaxies in the 20 gss fields ( fig . 4a ) and the deeper westphal field ( fig . the magnitude cut - off at the faint end ( v@xmath20=27 for the 20 gss fields and v@xmath20=28 for the westphal field ) was identified as the point where the galaxy number counts peak . faintward of these limits , the number counts sharply decline . with these limits , our photometric survey consists of 4512 objects in the 20 gss fields and 357 in the deeper westphal field . to identify significant variables , we determine the expected photometric error for non - variable sources . ignoring obvious outliers , we binned the data along the magnitude axis in 0.5 mag bins with larger bins at the bright end to include a greater number of points , and fit the distribution of magnitude differences with a gaussian . each bin was well fit with a gaussian and the sigma width in each bin was determined . these values were then fit with a 7th order polynomial to produce a smoothly varying photometric error as a function of magnitude . the average magnitude difference ( or photometric error ) does not vary with magnitude at the bright end . we thus determined a lower limit to the photometric error at this end of the distribution . brightward of v@xmath20=23.8 , the minimum 1@xmath10 limit for the photometric error was found to be 0.058 based on the average value of all probable non - varying sources brighter than this limit . in the deeper westphal field , the minimum 1@xmath10 photometric error was found to be 0.046 for galaxies brighter than v@xmath20=24.3 . the solid line in figure 4 represents the 3.2@xmath10 threshold . we have chosen this criterion to identify galaxies which have undergone a significant nuclear flux change between the two epochs . this threshold was found to yield the most robust selection of variables by minimizing both the number of statistical outliers in the survey ( only 0.14% in a normal distribution ) and the number of objects falling just at / above the threshold . in the 20 gss fields , 24 variable nuclei are detected ( blue triangles ) from the 4512 galaxies surveyed brighter than v@xmath20=27 . one of these , 283_1832 , appears to be a supernova in the disk of the spiral galaxy 283_1831 ( sharon 2003 ) and falls beyond the y - axis limit of the figure at @xmath21mag@xmath34 . in the westphal field , 3 variable nuclei ( blue triangles ) are detected among the 357 galaxies surveyed to v@xmath20=28 . approximately 6.5 of the total number of variables are expected to be statistical outliers in a gaussian distribution ( see discussion of errors below ) . table 5 lists the 26 variable galaxies ( omitting the likely supernova ) in order of decreasing nuclear apparent magnitude with columns as follows : ( 1 ) deep i d , ( 2 ) redshift , ( 3 ) v@xmath20 internal to r=1.5 pixel aperture , ( 4 ) magnitude difference between 1994 and 2001/2002 , and ( 5 ) significance of change obtained when normalized by the expected error . images of the variable galaxies are shown in figure 2 . a comparable variability study for the hdf - n ( sgk ) provides a good comparison to the current gss variability survey . sixteen galaxies were identified in the hdf - n displaying significant nuclear flux changes over a 5 year interval . figure 4 of sgk , however , shows that the variability threshold for the hdf - n is @xmath05 times lower than that for the gss fields . the significantly lower photometric errors for non - varying sources in the hdf - n results from the first and second epoch observations being obtained with the same pointing and roll angle , such that sources were located on the same ccd pixels in each epoch . the gss fields suffered from large offsets and orientation differences between the two epochs . this resulted in less accurate removal of cte loss effects for the gss . in addition , the dithering pattern for the second epoch observations resulted in differences in the combining and processing of the individual images . the combination of these effects resulted in greater photometric noise for the gss galaxy photometry when compared to the hdf - n . as discussed in sgk , there are two distinct completeness issues that affect this type of variability survey . the first is related to the incomplete time sampling of the variable sources we wish to detect . variability surveys usually image the survey field several times over many years ( e.g. trevese @xmath7 1994 ; hawkins 2002 ) . depending on the quality of the data , these surveys have shown that virtually all known agns will be found to vary if observed periodically over several years . our study is limited by the fact that we have only two epochs with which to determine variability and therefore sample just two points on the lightcurve of a varying source . because of this , we will be incomplete in our census of agns since some varying sources will lie at magnitudes close to their original magnitude measured in the first epoch and would thus be undetected in our survey . we estimate our incompleteness due to undersampling of the lightcurve by using variability data for qsos obtained over several years . because long term variability surveys for low - luminosity agns have not yet been published , we use a sample of 42 pg quasars from giveon @xmath7 ( 1999 ) and randomly select points along the lightcurve separated by 6.5 years ( the average time interval between the epoch 1 and epoch 2 gss images ) . the giveon @xmath7 quasar lightcurves cover 7 years with a typical sampling interval of 40 days . assuming the photometric error we determined for the gss ( solid line in fig . 4 ) , we can estimate the probability that the giveon qsos would be detected in our survey . the average change in magnitude between two points on the qso lightcurve separated by 6.5 years is @xmath00.2 mag and is consistent with the intrinsic variablility amplitude of 0.14 mag reported in giveon @xmath7 ( 1999 ) . we quantify the completeness by simulating photometric measurements of the giveon qsos assuming the photometric error of our survey at a given magnitude . we then find the percentage of qsos that would be detected above the 3.2@xmath10 threshold as a function of magnitude . the completeness gets lower with increasing magnitude since the variability threshold increases at fainter magnitudes . the detection rate of qsos ranges from @xmath045% at the bright end of the distribution to about 3% at the faint end ( v@xmath20=27 ) . to find the average survey completeness , we weight the detection rate or completeness values by the number of galaxies in 0.5 mag bins . this yields a weighted detection rate of 16% for survey galaxies down to v@xmath20=27 . at a limiting magnitude of v@xmath20=26 , 29% of the qsos would be detected . at v@xmath20=24 and brighter , 45% of the qsos would be detected . thus , the incompleteness due to lightcurve sampling is a strong function of the survey s limiting magnitude . relaxing our variability threshold to 2.5@xmath10 would increase these values by @xmath067% but introduces many more spurious variables as statistical outliers . the second incompleteness issue results from the use of a fixed aperture to detect nuclear variability . we use a small , fixed aperture to minimize dilution of the agn light from the underlying host galaxy to be more sensitive to agns varying within bright hosts . the aperture size matches the fwhm of an unresolved point source in the hst images ( 0.15@xmath9 in diameter ) . the fixed aperture will include the same amount of light from an unresolved source regardless of its redshift . however , the aperture will contain a larger fraction of the underlying , resolved galaxy for higher redshift objects than for low - z sources . because of this , the dilution of the nuclear light by the galaxy increases as a function of redshift and , consequently , as a function of nuclear magnitude . this issue is discussed in more detail in sgk , where they determine that variability surveys with fixed aperture photometry become less sensitive to variable nuclei at v@[email protected] to 27.5 in hst wfpc2 images . what this essentially means is that regardless of the depth of the photometry in a wfpc2 field , only nuclei brighter than @xmath027 are being uniformly surveyed for variable agns . the fact that the photometry in the hdf - n field extended to v@xmath20=29.0 while all of the variables were found to be brighter than v@xmath20=27.0 is consistent with this effect . in the deeper westphal field we again see that although photometry extends to v@xmath20=28.0 , the variables are all brighter than v@xmath20=27.0 ( fig . thus , we consider only those galaxy nuclei brighter than v@xmath20=27.0 in all gss fields ( including the westphal field ) to be uniformly surveyed for nuclear variability . the total number of galaxies surveyed to v@xmath20=27.0 in all gss fields is 4672 ( 4512 in the 20 gss fields and 160 in the westphal field ) . since the distribution of magnitude differences between epochs is well fit with a gaussian , we expect 0.14% of galaxies to fall outside of the 3.2@xmath10 threshold as statistical outliers . with a total of 4672 galaxies , this results in @xmath06.5 galaxies . if we correct for these outliers and the expected incompleteness , the fraction of variable nuclei in our survey to v@xmath20=27.0 is @[email protected]% ( ( 26 - 6.5)/0.16/4672 ) . a recent study of the hubble ultra deep field identified @xmath01% of the galaxies to @xmath22=28 as variable over a 4 month time interval ( cohen @xmath7 2006 ) . this lower percentage may result from the fact that larger apertures encompassing the entire galaxy s flux were used and thus may be less sensitive to nuclear variations . in addition , the shorter time baseline is likely to be less complete in identifying agns than longer , multi - year surveys ( e.g. webb & malkan 2000 ) . both the spectroscopic and the variability surveys presented here have identified agn candidates . unfortunately , neither survey has 100% complete coverage of the gss and the two surveys only partially overlap each other . because the entire gss was not reobserved in 2001/2002 , and due to the large offsets between the epoch 1 and epoch 2 images for some of the repeated fields , not all galaxies with deep spectroscopy are included in the variability survey . a total of 358 galaxies are in common between the 7133 galaxies in the variability survey and the 576 galaxies with high quality spectra . figure 5 shows the 358 galaxies with both spectroscopic and variability data plotted on a diagram similar to figure 4 with nuclear magnitude along the x - axis and magnitude change on the y - axis . spectroscopically identified agn candidates from tables 1 through 3 are indicated on the diagram as follows : blagn - blue filled circles ; moderate probability nlagn identified in the upper two quadrants of figure 3b - red triangles ; lower probability nlagn identified in the lower - left quadrant of figure 3b - red open triangles ; [ nev ] selected - purple asterisks ; and broad [ oiii ] selected - green squares . most notable in figure 5 is the fact that 3 of the 4 blagns which fall within the bounds of the variability survey ( blue filled circles ) are identified as significant variables . the blagns are also the brightest variables in our survey . one possible nlagn candidate ( red open triangle in fig . 5b ) appears as a variable . furthermore , some of the high - probability nlagns ( red triangles ) are near the 3.2@xmath10 significance level in figure 5a . blagns are known to exhibit variability in their continua while nlagns primarily display flux variations within particular emission lines . since the broad , v - band images used in our photometric survey are more sensitive to continuum flux changes , it is not surprising that we find greater overlap with the spectroscopic blagns while most of the nlagns are not significant variables . figures 6 and 7 show the distribution of spectroscopically and variability detected agns , respectively , compared to all gss galaxies in redshift , absolute magnitude , apparent magnitude , morphology and color . the magnitudes shown are integrated galaxy magnitudes based on the galaxy model fitting from simard @xmath7 ( 1999 ; 2002 ) . for those galaxies without model fits , we used 0.75@xmath9 radius aperture magnitudes . since galaxies without model fits are typically fainter and smaller , these aperture magnitudes should adequately measure the total light from the galaxy . in figure 6a , galaxy redshift vs. absolute magnitude ( h@xmath23=70 , q@xmath23=0.5 ) is shown for all gss galaxies in the spectroscopic survey to z@xmath62 ( black points ) . high probability agns are indicated with red squares and agn candidates with blue triangles . the 7 blagns are further indicated with larger red squares . the agns cover the full range of redshifts for the gss survey , with the mean redshift being slightly higher for the agns ( @xmath1z@xmath2=0.73 compared to @xmath1z@xmath2=0.66 for the gss galaxies ) . the blagns have the highest average redshift ( @xmath1z@xmath2=1.1 ) and also have a mean absolute magnitude more than three magnitudes brighter ( @xmath1m@xmath24@xmath2=-22.6 ) than the general gss sample ( @xmath1m@xmath24@xmath2=-19.3 ) . the mean luminosity of the nlagns is @xmath1m@xmath24@xmath2=-19.8 . we expect galaxy nuclei hosting agns to be generally brighter than non - agn galaxies since the active nucleus will contribute significant luminosity to the galaxy . it has also been found that the host galaxies of agns are typically brighter than normal galaxies ( ho @xmath7 1997 ) . figure 7a is analogous to figure 6a for variable nuclei galaxies . here we plot all galaxies in the photometric survey for which spectroscopic data also exist ( black points ) such that redshifts are known and absolute magnitudes can be calculated . significant variables are indicated with red squares . as shown in figure 5a , 5 variables are also included in the spectroscopic survey . the three most luminous variables are the blagns detected in our variability survey . the other 2 variables have absolute magnitudes of -19.0 and -17.3 , fainter than the average gss galaxy magnitude of @xmath1m@xmath24@xmath2=-19.3 . the thick solid lines in figures 6b , c , and d are the apparent magnitude , v i color , and bulge - to - total flux histograms ( normalized ) of all gss galaxies in the spectroscopic survey . figures 7b , c , and d display these same distributions for all gss galaxies in the variability survey . the hatched / cross - hatched histograms overplotted in figures 6b 6d represent the spectroscopically detected agn candidates ( hatched region ) and higher - probability agns ( cross - hatched region ) . in figures 7b d , the hatched histograms are variable galaxies with @xmath23.2@xmath10 significance and the cross - hatched histograms represent @xmath24@xmath10 variables . the agn histograms have been arbitrarily multiplied by a factor of 5 in figure 6 and 40 in figure 7 for display purposes . the spectroscopic agns have brighter apparent magnitudes ( @xmath1v@xmath2=22.0 ) than typical gss galaxies ( @xmath1v@xmath2=23.0 ) . comparing figures 6b and 7b , the apparent magnitude distribution of the galaxies in the photometric survey extends @xmath01.5 magnitudes deeper than the spectroscopic survey . in addition , the detected variable galaxies cover the full range of magnitudes for the photometric survey with a mean value only slightly brighter ( @xmath1v@xmath2=24.0 ) than that of all gss galaxies ( @xmath1v@xmath2=24.5 ) . the situation is different in the spectroscopic survey , where the agns are significantly brighter than the typical gss galaxy . this highlights one of the strengths of the variability selection technique . such surveys can extend deeper , with considerably less telescope time , and the identified agn candidates are not biased towards the brightest galaxies but cover the full apparent magnitude range of the survey . the colors of galaxies in the spectroscopic and variability surveys are similar ( figures 6c and 7c ) . although the mean color of galaxies in the variability survey is bluer than the spectroscopic survey ( @xmath1v i@xmath2=0.63 compared to 0.79 ) , both distributions peak around v [email protected] and have a similar range . a ks - test shows that the color distributions of the variability and spectroscopically selected agns are not significantly different ( 67% significance level ) . the redder colors observed for some of the agn / agn candidates may be due to dust which has been observed to be significant in seyfert galaxies ( malkan , gorjian & tam 1998 ) . the bluest agns in both the spectroscopic and variability selected samples are the blagns which have a mean v i color of 0.45 . the bulge - to - total measurements ( b / t ) in figures 6d and 7d are from simard @xmath7 ( 1999 ; 2002 ) . these histograms represent only the subset of galaxies in the spectroscopic and variability surveys that have 2-d model information . in both histograms , we have removed the blagns that have unresolved morphologies ( three in the spectroscopic survey and two in the photometric survey ) . since the galaxy fitting algorithm did not include a point source component , the model fits for these objects are erroneous . while the other agn / agn candidates appear to have good model fits , in some cases the bulge measurement may be enhanced by an unresolved agn component in the galaxy light profile . at the current time , 3-component ( disk+bulge+point source ) modeling does not exist for the gss galaxies , so that determining the relative flux of point sources among the selected agn candidates is beyond the scope of this paper . interestingly , the morphologies of the agn / agn candidates are not significantly different from the parent population of galaxies in both the spectroscopic and variability surveys . a ks - test comparing the agn and normal galaxy distributions confirms that they are not significantly different . in both surveys , the mean b / t for agns is similar to normal gss galaxies . however , a comparison of figures 6d and 7d reveals that galaxies and agns in the variability survey are slightly more bulge - dominated than the spectroscopic survey galaxies ( @xmath1b / t@[email protected] vs. 0.25 ) . this is also evident from a visual inspection of the galaxy images in figure 2 . the galaxies and agns in the spectroscopic survey may favor more disk - dominated systems since these galaxies are likely to have the strongest emission lines and therefore have more easily identifiable redshifts . only two of the agn / agn candidates from the spectroscopic and variability selected samples , 292_3076 and 172_5049 , are among the e / s0s identified in i m @xmath7 ( 2002 ) . among these , 172_5049 , a lower - probability nlagn , is a spheroid with a particularly blue color ( i m @xmath7 2001 ) . their analysis indicates this object to be a low - luminosity spheroidal with color contours revealing a bluer central region than typical e / s0s . this evidence supports the assertion that an agn may reside in the center of this galaxy . active galactic nuclei have long been known as powerful x - ray sources . recent deep x - ray surveys of the sky have revealed that much of the x - ray background is dominated by agn activity . x - ray surveys for agns are not as hindered by some of the effects from which optical surveys suffer , such as obscuration due to dust and contamination from bright host galaxies . about 60% of the gss has been observed with the xmm - newton and chandra x - ray telescopes . gss fields 4 through 18 were observed with xmm - newton ( pi : t. miyaji ) and fields 4 through 14 with chandra acis - i ( pi : k. nandra ) . a summary of the xmm detected sources is given in miyaji @xmath7 ( 2004 ) and the chandra detections are summarized in nandra @xmath7 ( 2005 ) . a total of 31 xmm and chandra detected sources are identified within 1.5@xmath9 of an optical counterpart in the deep catalog . thirteen are detected with both chandra and xmm , 7 with xmm only and 11 with chandra only . some sources were detected in one survey and not the other due to 1 ) the slightly larger fov of the xmm survey and 2 ) the deeper flux limit of the chandra survey . table 6 lists the x - ray sources in order of decreasing full - band x - ray flux with the following columns : ( 1 ) deep i d , ( 2 ) offset between the x - ray and optical position , ( 3 ) xmm i d from miyaji @xmath7 , ( 4 ) chandra i d from nandra @xmath7 , ( 5 ) full - band ( 0.510kev ) x - ray flux in units of 10@xmath25ergs/@xmath26/s , ( 6 ) hardness ratio ( calculated as f@xmath27(2 - 10)/f@xmath27(0.5 - 2 ) ) , and ( 7 ) full - band x - ray luminosity . for sources detected with both chandra and xmm , the chandra coordinates , x - ray flux and hardness ratios are given in the table . ten of the 31 x - ray sources with optical counterparts have high quality spectroscopic data in the deep gss survey and four of these are blagns . two of the three remaining blagns in table 1 are not covered by the x - ray surveys . thus 80% of the x - ray covered blagns in the spectroscopic survey are x - ray detected . the one non - detected blagn is also the faintest optical blagn . based on the x - ray to optical flux ratio of sources detected in deep chandra surveys ( e.g barger @xmath7 2003b ) , the expected x - ray flux for an object with this optical magnitude ( approximately r@xmath324 ) ranges over 2 orders of magnitude with the faintest limit at f@xmath27(0.5 - 2)@xmath310@xmath28 ergs/@xmath26/s . since this corresponds to the soft - band detection limit of the 200ks chandra data for the gss , it is possible that this source is too faint to be detected . of the other x - ray sources with spectroscopic information , one is a [ nev ] detected agn ( 073_7749 ) . none of the 14 ew ratio detected nlagns which fell within the x - ray fov were x - ray detected . in total , 5 of the 10 x - ray sources ( 50% ) in the gss spectroscopic survey show some evidence of agn activity based on their spectroscopic signatures . if we consider all spectroscopically selected agns in the x - ray fov there are 25 galaxies , 5 of which are x - ray detected ( 20% ) . considering only the highest probability agns ( blagns , moderate probability nlagns , [ nev ] detected , and broad [ oiii ] agns ) results in 11 galaxies of which 5 are x - ray detected ( 45% ) . the blagns are the most luminous x - ray sources in the survey and are also among the softest x - ray emitters , consistent with the majority of type i , qso - like agns . the [ nev ] detected agn is among the hardest x - ray sources in the gss and is consistent with the classification of this object as highly obscured , type 2 agn . given its redshift and hardness ratio , we estimate an obsuring column density of n@[email protected]@xmath1310@xmath30 . some of the x - ray sources which were not observed in the deep1 survey or did not reveal agn signatures in their optical spectra were followed up with near - ir spectroscopic observations summarized in miyaji @xmath7 ( 2004 ) . four of the five targeted galaxies did show evidence of broad h@xmath16 emission in the near - ir ( see table 7 ) . since the signal - to - noise for these observations is much lower than the optical spectroscopic data analyzed in this paper , these results were not considered in the above statistics . however , including these 4 nir observed galaxies as spectroscopically identified agn raises the percentage of x - ray sources with agn spectroscopic signatures from 50% to 75% ( 9/12 ; the total with agn signatures becomes 5 + 4 and the total number of x - ray sources with either optical or nir spectroscopic follow - up is then 12 ) . clearly , a more complete spectroscopic survey extending into the near - ir would shed considerable light on the agn nature of faint x - ray sources . the galaxies detected in the variability survey that fall within the fov of the x - ray data are shown in figure 8 . of the 31 x - ray sources with optical counterparts , 19 are in the variability survey ( red triangles ) . sixteen are above the flux completeness limits and 4 of these are variable ( 3 in the gss fields and one in the westphal field ) . two of these 4 are also blagns while the other 2 are not covered in the spectroscopic survey . thus , we find that 25% ( 4 out of 16 ) of the x - ray sources in the variability survey are significant variables . if we consider the number of significant variables that are also x - ray sources , we find that 27% ( 4 out of 15 ) of variables in the x - ray fov are x - ray detected . all 4 of the variables are soft x - ray emitters which is consistent with the earlier assertion that the optical variables are more likely to be type 1 agns . they are also among the most luminous x - ray sources . about half of the variables that do not show x - ray emission are expected to have x - ray fluxes too faint for detection , based on typical x - ray to optical flux ratios . another two , 072_2372 and 094_6133 , are larger galaxies or clumpy objects whose optical light is probably an overestimate of the agn light and thus may also be too faint to be detected in the x - ray survey . the remaining few may be spurious variables or could have lower x - ray to optical flux ratios like those of transition objects between agn and quiescent galaxies ( barger @xmath7 2003b ) . to check for faint x - ray emission that may be just below the sensitivity limits of the x - ray data , we have stacked the chandra images at the location of spectroscopically and variability selected agns which do not individually appear as x - ray detections . within 8@xmath31 of the chandra aimpoint , there are 11 non - x - ray detected variable sources . the stacked counts , using a 1.5@xmath9 extraction radius , do not reveal a statistically significant detection ( 1.3@xmath10 ) . however , a marginally significant detection ( 2.9@xmath10 ) was found when stacking the 14 non - x - ray detected spectroscopic agn / agn candidates . these galaxies all come from the [ oii]/h@xmath17 ew selected sources listed in table 3 ( 2 from the moderate probability group and 12 from the lower probability group ) . the stacks were performed with the soft - band x - ray data ( 0.52 kev ) , where the chandra sensitivity is the greatest . there was no significant detection using the hard - band data alone or the full - band data ( 0.510 kev ) . the marginal x - ray detection of the [ oii]/h@xmath17 ew selected galaxies provides some additional supporting evidence for the agn nature of galaxies selected via this method . the lack of a detection among the stacked variable sources may be due to their likely higher redshifts and subsequent lower x - ray fluxes as discussed above . the variable sources used in the stack are generally about @xmath02 magnitudes fainter than the [ oii]/h@xmath17 ew selected galaxies . we have selected agn / agn candidates using two independent selection techniques ; spectroscopic selection based primarily on emission lines , and variability selection based on significant flux changes in galactic nuclei over a @xmath07 year interval . ideally , one would construct a luminosity function ( lf ) for both agn samples for comparison with one another as well as other similar surveys . however , lfs constructed from our datasets are likely to be highly uncertain for the following reasons : 1 ) we do not have accurate knowledge of the luminosity of the agn component in our spectroscopically or variability selected galaxies , 2 ) we have incomplete / non - uniform redshift coverage for the spectroscopic detection of agn , and 3 ) the variability survey incompleteness estimate is quite high ( 8455% ) due to poor photometric repeatability in the gss wfpc2 images ( 3.3 ) . in the following paragraphs , we make estimates to account for these issues in order to broadly place our findings in context with published agn lfs . in the spectroscopic survey , we have selected agn candidates among various subsets of galaxies in which particular emission line features could be measured . to estimate the total agn fraction among gss galaxies , we first define the sets of probable type 1 and type 2 agns . the type 1 agns are easily defined as those displaying broad emission lines ( table 1 ) . two of the three [ nev ] detected sources are also blagns . the other galaxy spectrum does not show indications of any broad emission lines even though mgii is within the spectroscopic range . therefore , we will consider the other [ nev ] detected source as a nlagn . one of the two broad [ oiii ] galaxies is also a blagn with broad mgii . the other , 292_3076 , does not show any broad lines although h@xmath17 is within the spectral range and appears as an absorption feature . we will consider this galaxy also as a nlagn , although its exact nature is uncertain . thus we include one [ nev ] galaxy and one broadened-[oiii ] galaxy in the nlagn group along with those galaxies identified via [ oii]/h@xmath17 ew in figure 3a . we compare the spectroscopic and variability survey galaxy samples with the general gss catalog presented in vogt @xmath7 ( 2005 ) containing 11,547 objects to identify any selection biases among our samples and determine a characteristic limiting magnitude for agn identification using both of our selection techniques . for comparison with vogt @xmath7 , who determined the survey completeness as a function of combined v@xmath32 and i@xmath33 magnitudes in r=0.75@xmath9 apertures , we define the magnitude @xmath34 = ( v@xmath32+i@xmath33)/2 . we investigate galaxy selection as a function of magnitude , color ( ( v - i ) using r=0.5@xmath9 apertures ) and light profile concentration ( based on the difference between r=0.15@xmath9 and 0.2@xmath9 aperture photometry ) . the variability survey is 60 to 70% complete over the full range of galaxy colors to @xmath34 = 24 . the incompleteness is largely due to the fact that only 2/3rds of the gss had second epoch data for the variability analysis . between magnitudes of 24 and 25 , the completeness drops to about 50% and becomes a function of concentration , with low - concentration sources ( i.e. low - surface brightness sources ) being less complete ( 34% ) than high - concentration sources ( 60% ) . the spectroscopic survey is generally less complete than the variability survey and the completeness drops as a function of magnitude . much of the incompleteness is due to the limited areal covereage of the deep1 spectroscopic survey . at magnitudes greater than 21 , the survey is 42% complete and drops to 30% at 23rd magnitude . between 23rd and 24th magnitude the spectroscopic survey includes just 13% of galaxies . the completeness is roughly constant as a function of color but is systematically lower for high - concentration sources ( stellar - like morphologies ) . therefore , we include a correction for the possibility of missed blagn due to this morphological bias ( described below ) . the total number of galaxies in which nlagns could be identified includes all spectra with either [ oii ] and h@xmath17 , [ oiii ] , or [ nev ] lines in the spectral range , amounting to 526 of the available 576 galaxies with spectroscopic information . using the [ oii]/h@xmath17 ew diagram probabilities detailed in 2.1.3 , and including the [ nev ] and [ oiii ] selected galaxies that do not also contain broad - lines , we estimate the total percentage of nlagns to be [email protected]% at @xmath34@xmath1924 , [email protected]% at @xmath34@xmath1923 and [email protected]% at @xmath34@xmath1922 . the nlagns lie in the redshift range 0.14@[email protected] with an average redshift of [email protected] . the nlagns have integrated galaxy absolute magnitudes between -22 to -18 ( rest - frame b - band determined using h@xmath23=70 , q@xmath23=0.5 , and k - corrections assuming a power - law spectrum with @xmath16=-1.0 ) . the fraction of blagns in the spectroscopic survey can be estimated by first determining the number of galaxies in the survey in which broad emission lines could be observed to a given limiting magnitude . to better estimate the total number of blagns , we correct for the fact that the spectroscopic survey is less complete for point - like sources . by determining the completeness of stellar - like objects vs. non - stellar objects in unity magnitude bins , and multiplying by the fraction of blagns identified among stellar - like objects at each magnitude , we estimate that @xmath03 blagns are missed to @xmath34@xmath1924 . taking this into account , the percentage of blagns among gss galaxies is [email protected]% at @xmath34@xmath1924 , [email protected]% at @xmath34@xmath1923 and [email protected]% at @xmath34@xmath1922 . the blagns in our survey lie in the redshift range 0.65@[email protected] and have integrated galaxy magnitudes of m@xmath35@xmath3 - 24 to -20 . in the variability survey , the completeness corrected fraction of galaxies containing significant variable nuclei was found to be 2.6% to v@xmath20=27.0 . this nuclear magnitude limit translates roughly to @xmath34 = 26 . in order to compare with the spectroscopic survey , we compute the fraction of galaxies containing variable nuclei to the same galaxy limiting magnitudes . a galaxy magnitude limit of @xmath34@xmath1924 imposes a v@xmath20 limit of @xmath025 . recall that the completeness in detecting nuclear variability changes as a function of magnitude . at v@xmath20@xmath1925 , the weighted completeness is 50% . correcting for this incompleteness and statistical outliers , the percentage of galaxies containing nuclear variables is [email protected]% . following the same procedure , we find the fraction of variables galaxies to be [email protected]% for @xmath34@xmath1923 and [email protected]% for @xmath34@xmath1922 . the fraction of variables in gss galaxies is systematically greater than the fraction of blagns by a few percent , though the small numbers and subsequent large errorbars on these percentages do not make them significantly so . if the variables are primarily type 1 agns , as suggested by the overlap in their samples at the bright end of the distribution , this difference is likely due to the ability of the variability technique to probe lower agn / host galaxy flux ratios . thus , about half of the detected variables may contain agns too faint to be identified on the basis of emission lines alone . this may be due to the intrinsic faintness of the agn or significant obscuration . we do not have redshift information for the majority of variability detected agns . for those with redshifts , the magnitudes lie between m@xmath35@xmath3 - 24 to -17 . making a crude estimate of the redshifts for the remaining variables ( 15 of the galaxies ) based on v i color and magnitude , the majority of variable nuclei galaxies without spectroscopic follow - up fall between m@xmath35@xmath3 - 19 to -17 . since we find no spectroscopic blagns fainter than m@xmath35@xmath3 - 20 , we can conclude that indeed the variability selection technique is more sensitive to intrinsically fainter agns and that this is the primary explanation for the difference in the numbers of detected agn . previous estimates of the fraction of type 1 and type 2 agns in the local universe reveal varying results . huchra & burg ( 1992 ; hb92 ) reported 1% of local galaxies in the cfa redshift survey as blagns ( seyfert 1s ) and another 1% as nlagns ( seyfert 2s ) . maiolino & rieke ( 1995 ) determined that at least 5% of rsa galaxies are agns with a ratio of seyfert 1s to seyfert 2s of 1:4 . this higher fraction appears to be a result of probing fainter active nuclei within galaxies of similar magnitudes . comparing the luminosity functions of seyfert galaxies determined by hb92 to that of ulvestad & ho ( 2001 ; uh01 ) helps to make this clearer . figure 2 of uh01 shows that their survey of palomar seyferts has a number density that is a factor of 510 higher than that of the cfa survey . ho & peng ( 2001 ) show that their median nuclear magnitude is almost 3 magnitudes fainter than that probed by hb92 . our spectroscopic survey is not sensitive to intrinsically faint agn within host galaxies and thus our results should be more comparable to those of hb92 , where the spectra contain light from the entire galaxy and not just the nuclear region . it is somewhat unexpected , however , that we find no blagns fainter than m@xmath35=-20 . the lf of hb92 shows a steady increase from m@xmath35=-22 extending to m@xmath35=-18 . for a proper comparison with hb92 , we must compare the fraction of galaxies containing blagns in the same absolute magnitude range . from figure 1b in hb92 , we see that about 2/3rds of the type 1 agns lie in the absolute magnitude range of our spectroscopic blagn sample ( m@xmath35@xmath3 - 22 to -20 ) which constitutes [email protected]% of the cfa redshift survey galaxies to this absolute magnitude limit . thus , we find a marginally significant increase in the fraction of galaxies containing blagns in this absolute magnitude range , rising from [email protected]% locally to [email protected]% at z@xmath31 . the nlagns in our survey cover roughly the same absolute magnitude range as the hb92 type 2 seyferts and are therefore valid for comparison . we find little evidence for any increase in the number of galaxies containing nlagns ( 1% locally to [email protected]% at @xmath1z@[email protected] ) . the increasing fraction to higher redshifts is more significant considering a brighter magnitude limit for our spectroscopic survey ( 2.5% are nlagns among galaxies brighter than 22nd magnitude ) . tresse @xmath7 ( 1996 ) found an increase in the number of nlagns to [email protected] in the cfrs , which consists of integrated galaxy spectra like the deep survey but with lower resolution . using traditional line ratio diagnostics , they find that @xmath08% of galaxies are nlagns . both the results of tresse @xmath7 and the present study are sensitive to the underlying stellar absorption from the host galaxy , particularly the h@xmath17 emission line measurement . although both studies have made estimates to account for the h@xmath17 ew measurement , uncertainties in these estimates may explain why the tresse @xmath7 fraction of nlagns among galaxies at [email protected] is greater than that found in the present study at a higher average redshift . the general result is that the fraction of nlagns among galaxies at intermediate redshifts ( [email protected] to 0.8 ) shows a moderate increase when compared with the local population . we can quantify the total agn increase by comparing both the nlagn and blagn samples to the total number identified in the cfa redshift survey . they find [email protected]% of all surveyed galaxies are agn . the total fraction of spectroscopically detected agn in our survey is [email protected]% which yields an increase of [email protected]% . we have used two independent and complementary optical selection techniques , spectroscopic and variability identification , to probe the agn population out to redshifs of z@xmath01 in the groth - westphal survey strip . agn / agn candidates have been identified among the 576 gss galaxies with high quality spectroscopic data . thirty - one galaxies have revealed agn spectroscopic signatures with 9 classified as high probability " agns , 7 as moderate probability " agns , and 15 as lower probability " agn candidates . seven of these display broad emission lines ( type 1 agns ) . we also identify 26 variable nuclei galaxies within the gss which have displayed significant flux changes ( @xmath23.2@xmath10 ) over a 6 to 7 year time interval . the primary overlap of the two surveys is the set of blagns , of which 80% appear as variable . only a few nlagns approach the 3.2@xmath10 variability significance limit , indicating that variability primarily selects type 1 blagns . the blagns have an average redshift of @xmath1z@[email protected] while the other spectroscopic and variability selected agns ( with spectroscopic data ) have redshifts closer to the mean of the general gss galaxy population ( @xmath1z@[email protected] ) . the mean luminosity of the spectroscopically selected blagns is more than three magnitudes brighter ( @xmath1m@xmath24@xmath2=-22.6 ) than the gss galaxy average ( @xmath1m@xmath24@xmath2=-19.3 ) . the nlagns are only slightly more luminous ( @xmath1m@xmath24@xmath2=-19.8 ) than the average gss galaxy . the variability survey probes galaxies @xmath01.5 magnitudes fainter than the spectroscopic survey and the agns identified via variability are more representative of the entire magnitude range of galaxies . the identified agns in both surveys have colors and morphologies similar to their parent galaxy populations . eighty percent of the blagns are detected in chandra / xmm x - ray surveys and these are among the most x - ray luminous sources in the gss . only one of the high probability nlagns is x - ray detected : a [ nev ] emission line galaxy with very hard x - ray emission . of the variable nuclei galaxies within the x - ray fov , 27% are x - ray detected . this can be compared to the hdf - n ( sgk ) where 44% of the variable nuclei galaxies were also x - ray sources . the higher percentage in the hdf - n is likely a result of better completeness of the variability survey and deeper x - ray obervations of the field ( 2ms compared to 200ks ) . the blagns and variable nuclei are among the softest x - ray sources detected . a summary of all possible agns in the gss identified via spectroscopy , variability or x - ray detection is given in table 7 . along with the gss i d and redshift if known , the table gives the spectroscopic agn classification , variability in sigma units above the detection threshold and the full - band x - ray flux from either chandra or xmm in units of 10@xmath25 ergs/@xmath26/s . we find that [email protected]% of galaxies to v@xmath5@xmath624 are nlagns while [email protected]% are blagns . the fraction of galaxies containing variable nuclei to the same limiting magnitude is [email protected]% . while it appears that variability primarily selects type 1 agns ( i.e. blagns ) , the variability survey probes intrinsically fainter galaxies ( m@xmath35@xmath19 - 17 ) than does the spectroscopic survey . this is primarily due to the ability of the high - resolution hst photometry to probe flux changes in the central , unresolved regions of the survey galaxies while the spectroscopic survey relies on integrated galaxy spectra . the fraction of spectroscopically identified blagns and nlagns at z@xmath01 reveals a marginally significant increase of [email protected]% when compared to the local population . the results of these surveys , together with recent x - ray surveys , are helping to uncover the population of agns among galaxies to z@xmath31 . it is clear that no one survey wavelength or technique is able to identify all agns in a given field and that a multiwavelength , multi - technique approach is the best way to identify and study the agn population and its evolution with redshift . as additional multiwavelength survey data becomes available for the gss and extended groth strip region ( e.g. deep vla , galex , spitzer ) , a more accurate census of the agn population to z@xmath01 will emerge . this paper is based on observations with the nasa / esa hubble space telescope , obtained at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . , under nasa contract nas5 - 26555 and observations with the keck telescope , made possible by the w. m. keck foundation and nasa . funding for deep was provided by nsf grants ast 95 - 29098 and ast 00 - 71198 . this work was also supported by the stsci grant ar-09218.04 . 283_3452 & 0.6472 & h@xmath17 & 4881 & 19.63 & -22.51 + 142_4838 & 0.8077 & mgii & & 20.75 & -21.92 + 273_4925 & 0.9853 & mgii & 4322 & 18.73 & -24.41 + 083_5273 & 1.1476 & mgii & & 22.61 & -20.90 + 083_5407 & 1.2233 & mgii , cii & 8330,8373 & 20.58 & -23.08 + 123_2458 & 1.2627 & mgii & 4816 & 20.61 & -23.13 + 152_6235 & 1.5772 & mgii & & 24.21 & -20.07 + 283_3452 & 0.6459 & 390 & yes & 19.63 & -22.51 + 142_4838 & 0.8065 & 540 & yes & 20.75 & -21.92 + 073_7749 & 0.8730 & 650 & no & 23.13 & -19.72 + 292_3076 & 0.6937 & 820 & no & 22.00 & -20.31 + 142_4838 & 0.8065 & 452 & yes & 20.75 & -21.92 + 284_4709 & 0.4498 & 57.71 & 1.25 & 3.62 & 1.08 & 19.75 & -21.54 + 193_5426 & 0.4703 & 201.98 & 16.83 & 12.06 & 2.87 & 22.32 & -19.08 + 223_6264 & 0.4715 & 87.03 & 14.07 & 9.23 & 1.44 & 21.27 & -20.13 + 154_3015 & 0.4736 & 419.72 & 8.81 & 28.07 & 1.74 & 22.34 & -19.07 + 133_2106 & 0.5328 & 124.27 & 7.91 & 16.93 & 2.80 & 23.23 & -18.46 + 084_6809 & 0.7467 & 139.08 & 2.91 & 14.61 & 4.65 & 22.68 & -19.80 + 084_4515 & 0.8108 & 136.35 & 1.53 & 18.46 & 2.79 & 23.72 & -18.96 + 083_1919 & 0.1406 & 19.32 & 2.90 & 2.88 & 0.51 & 21.37 & -17.28 + 102_6577 & 0.2080 & 15.32 & 1.22 & 5.01 & 0.51 & 22.16 & -17.37 + 072_4040 & 0.2908 & 31.22 & 7.45 & 4.03 & 0.86 & 20.72 & -19.57 + 172_5049 & 0.3564 & 32.35 & 3.17 & 5.72 & 1.00 & 22.36 & -18.39 + 173_5210 & 0.3570 & 37.06 & 3.83 & 5.15 & 0.95 & 20.63 & -20.13 + 213_6640 & 0.3656 & 27.88 & 1.60 & 3.58 & 0.88 & 20.76 & -20.05 + 072_2372 & 0.4258 & 9.79 & 1.12 & 5.93 & 0.47 & 22.98 & -18.19 + 092_1962 & 0.4261 & 20.35 & 1.36 & 4.09 & 0.55 & 19.98 & -21.19 + 113_0815 & 0.4279 & 6.78 & 1.00 & 3.88 & 0.98 & 20.76 & -20.42 + 094_1054 & 0.4331 & 11.61 & 1.00 & 3.24 & 0.55 & 20.33 & -20.87 + 103_1520 & 0.4620 & 15.09 & 0.81 & 3.35 & 0.33 & 22.50 & -18.86 + 103_1811 & 0.4636 & 25.71 & 1.78 & 4.05 & 0.89 & 22.53 & -18.83 + 072_7471 & 0.4830 & 30.25 & 5.18 & 5.45 & 0.95 & 21.82 & -19.64 + 093_1519 & 0.5346 & 13.84 & 2.48 & 4.36 & 1.14 & 22.48 & -19.22 + 084_4521 & 0.7540 & 12.82 & 1.85 & 3.74 & 0.81 & 22.04 & -20.47 + 153_2622 & 0.8070 & mgii + 153_2422 & 0.8072 & mgii + 152_5051 & 0.8086 & mgii + 113_4933 & 0.8112 & mgii + 093_6661 & 0.9877 & mgii / feii + 104_4809 & 0.9911 & mgii + 072_2928a & 0.9972 & feii + 092_3127 & 1.0510 & mgii + 064_6177 & 1.1449 & mgii + 083_0815 & 1.2849 & mgii + 212_2577 & 1.2859 & mgii / feii + 223_7508 & 1.3658 & feii + 084_1720 & 1.6058 & mgii / feii + 113_6825 & 1.7217 & mgii / feii + 283_3452 & 0.6472 & 20.299 & 0.176 & 3.24 + 083_5407 & 1.2233 & 21.330 & 0.806 & 14.83 + 123_2458 & 1.2627 & 21.806 & 0.301 & 5.54 + 054_2631 & 0.3780 & 23.238 & 0.253 & 4.65 + 253_3121 & & 24.082 & 0.419 & 7.49 + 222_3954 & 0.2715 & 24.179 & 0.313 & 5.52 + 052_3372a & & 24.689 & 0.835 & 13.07 + 053_3424 & 1.5537 & 24.722 & 0.525 & 8.13 + 082_3939 & 1.1555 & 24.734 & 0.211 & 3.26 + 043_6232 & & 25.436 & 1.084 & 13.12 + 263_6544 & & 25.510 & 0.282 & 3.30 + 094_6133 & 0.8059 & 25.793 & 0.607 & 6.18 + 254_4642 & & 25.915 & 0.431 & 4.09 + 284_2213 & & 26.037 & 0.437 & 3.84 + 124_2138a & & 26.097 & 0.461 & 3.89 + 124_7332 & & 26.179 & 0.450 & 3.59 + 274_6339 & & 26.484 & 0.629 & 3.96 + 214_7315 & & 26.518 & 0.653 & 4.00 + 124_4831 & & 26.518 & 0.562 & 3.44 + 222_3930 & & 26.617 & 0.642 & 3.61 + 272_6943 & & 26.623 & 0.929 & 5.20 + 172_2078 & & 26.778 & 0.918 & 4.48 + 042_4373 & & 26.831 & 1.222 & 5.68 + 072_2372 & 0.4258 & 23.854 & 0.203 & 3.42 + 074_5765 & 0.5989 & 24.149 & 0.253 & 4.11 + 074_6236 & 0.9346 & 25.742 & 0.329 & 4.11 + 173_7369 & 0.50 & x11 & & 5.2 & 3.31 & + 083_5273 & 0.62 & x20 & j141741.9 + 522823 & 5.0 & 2.92 & 44.25 + 142_4838 & 0.55 & x10 & j141651.2 + 522047 & 3.0 & 2.32 & 43.69 + 144_2774 & 0.34 & x9 & & 2.8 & 0.71 & + 142_2752 & 1.23 & x22 & & 2.6 & 2.28 & 43.82 + 083_5407 & 0.45 & x8 & j141734.8 + 522810 & 2.1 & 2.06 & 43.94 + 062_2060 & 0.61 & x52 & j141745.9 + 523032 & 1.5 & 5.67 & 43.58 + 184_2148 & 1.29 & x55 & & 1.5 & 3.78 & + 164_6109 & 1.15 & x66 & & 1.2 & 19.17 & 43.30 + 123_2458 & 0.21 & x26 & j141715.0 + 522312 & 1.0 & 2.00 & 43.65 + 082_5240 & 0.33 & x44 & j141729.9 + 522747 & 0.92 & 2.64 & + 074_6236 & 1.21 & x69 & j141749.2 + 522811 & 0.90 & 3.59 & 43.37 + 053_4446 & 0.83 & x61 & j141758.9 + 523138 & 0.89 & 3.67 & 42.94 + 184_7960 & 1.16 & x83 & & @xmath10.86 & soft & + 142_2530 & 0.46 & x40 & & @xmath10.67 & soft & + 074_2638 & 1.23 & x46 & j141745.7 + 522801 & 0.61 & 2.59 & 42.41 + 053_3424 & 0.85 & x125 & j141756.8 + 523124 & 0.46 & 3.64 & 43.51 + 052_1037 & 1.07 & x130 & j141754.2 + 523123 & 0.32 & 1.70 & 42.87 + 073_7749 & 0.63 & x146 & j141745.4 + 522951 & 0.32 & 33.16 & 42.80 + 113_6354 & 0.03 & x133 & j141720.0 + 522500 & 0.31 & 4.57 & 42.58 + 053_1525 & 1.01 & & j141757.4 + 523106 & 0.19 & 4.39 & 43.77 + 064_2658 & 0.73 & & j141752.4 + 522853 & 0.17 & hard & 42.60 + 074_4426 & 1.27 & & j141747.0 + 522816 & 0.14 & 3.72 & + 063_6344 & 0.77 & & j141751.7 + 523046 & 0.11 & 2.10 & 42.46 + 074_6044 & 0.62 & & j141749.2 + 522803 & 0.11 & 10.00 & 42.46 + 072_3963 & 0.64 & & j141737.3 + 522921 & 0.11 & hard & 42.18 + 063_4661 & 1.07 & & j141753.9 + 523033 & 0.11 & 3.42 & 42.46 + 082_3769 & 0.61 & & j141730.8 + 522818 & 0.099 & hard & + 103_5524 & 0.42 & & j141723.6 + 522555 & 0.083 & 13.00 & + 073_2249 & 0.43 & & j141746.7 + 522858 & 0.076 & 2.24 & + 072_3223 & 0.64 & & j141739.0 + 522843 & 0.066 & soft & + 042_4373 & & & 5.68 & 0 + 043_6232 & & & 13.12 & 0 + 052_1037 & 0.9462 & & & 0.32 + 052_3372a & & & 13.07 & 0 + 053_1525 & 3.0260 & & 0.92 & 0.19 + 053_3424 & 1.5537 & & 8.13 & 0.46 + 053_4446 & 0.6370 & & 1.78 & 0.89 + 054_2631 & 0.3780 & & 4.65 & 0 + 062_2060 & 0.9853 & blagn - ir & & 1.5 + 063_4661 & 0.9976 & none & & 0.11 + 063_6344 & 0.8929 & & & 0.11 + 064_2658 & 0.9515 & & & 0.17 + 072_2372 & 0.4258 & nlagn - lo & 3.42 & 0 + 072_3223 & & & 0.23 & 0.066 + 072_3963 & 0.7421 & & 0.10 & 0.11 + 072_4040 & 0.2908 & nlagn - lo & & 0 + 072_7471 & 0.4830 & nlagn - lo & & 0 + 073_2249 & & & 0.03 & 0.076 + 073_7749 & 0.8730 & [ nev ] & 1.20 & 0.32 + 074_2638 & 0.4322 & none & 2.11 & 0.61 + 074_4426 & & & 0.26 & 0.14 + 074_6044 & 0.9966 & none & 1.77 & 0.11 + 074_6236 & 0.9950 & blagn - ir & 4.11 & 0.90 + 082_3769 & & & 0.85 & 0.099 + 082_3939 & 1.1555 & & 3.26 & 0 + 082_5240 & & & 1.39 & 0.92 + 083_1919 & 0.1406 & nlagn - lo & 0.22 & 0 + 083_5273 & 1.1476 & blagn & 0.46 & 5.0 + 083_5407 & 1.2233 & blagn & 14.83 & 2.1 + 084_4515 & 0.8108 & nlagn & 0.16 & 0 + 084_4521 & 0.7540 & nlagn - lo & 2.33 & 0 + 084_6809 & 0.7467 & nlagn & 2.67 & 0 + 092_1962 & 0.4261 & nlagn - lo & & 0 + 093_1519 & 0.5346 & nlagn - lo & 0.18 & 0 + 094_1054 & 0.4331 & nlagn - lo & & 0 + 094_6133 & 0.8059 & none & 6.18 & 0 + 102_6577 & 0.2080 & nlagn - lo & & 0 + 103_1520 & 0.4620 & nlagn - lo & & 0 + 103_1811 & 0.4636 & nlagn - lo & & 0 + 103_5524 & & & & 0.083 + 113_0815 & 0.4279 & nlagn - lo & 0.11 & 0 + 113_6354 & 0.7037 & & 0.95 & 0.31 + 123_2458 & 1.2627 & blagn & 5.54 & 1.0 + 124_2138a & & & 3.89 & 0 + 124_4831 & & & 3.44 & 0 + 124_7332 & & & 3.59 & 0 + 133_2106 & 0.5328 & nlagn & 0.50 & 0 + 142_2530 & & & & @xmath10.67 + 142_2752 & 0.9830 & blagn - ir & & 2.6 + 142_4838 & 0.8077 & blagn,[nev],[oiii ] & & 3.0 + 144_2774 & & & & 2.8 + 152_6235 & 1.5772 & blagn & & 0 + 154_3015 & 0.4736 & nlagn & & 0 + 164_6109 & 0.8084 & blagn - ir & & 1.2 + 172_2078 & & & 4.48 & 0 + 172_5049 & 0.3564 & nlagn - lo & 0.98 & 0 + 173_5210 & 0.3570 & nlagn - lo & 1.84 & 0 + 173_7369 & & & & 5.2 + 184_2148 & & & 1.31 & 1.5 + 184_7960 & & & & @xmath10.86 + 193_5426 & 0.4703 & nlagn & 1.82 & 0 + 213_6640 & 0.3656 & nlagn - lo & 1.11 & + 214_7315 & & & 4.00 & + 222_3930 & & & 3.61 & + 222_3954 & 0.2715 & none & 5.52 & + 223_6264 & 0.4715 & nlagn & 2.16 & + 253_3121 & & & 7.49 & + 254_4642 & & & 4.09 & + 263_6544 & & & 3.30 & + 272_6943 & & & 5.20 & + 273_4925 & 0.9853 & blagn & & + 274_6339 & & & 3.96 & + 283_3452 & 0.6472 & blagn,[nev ] & 3.24 & + 284_2213 & & & 3.84 & + 284_4709 & 0.4498 & nlagn & 1.10 & + 292_3076 & 0.6937 & [ oiii ] & 0.61 & +	we identify active galactic nuclei ( agn ) in the groth - westphal survey strip ( gss ) using the independent and complementary selection techniques of optical spectroscopy and photometric variability . we discuss the x - ray properties of these agn using chandra / xmm data for this region . from a sample of 576 galaxies with high quality spectra we identify 31 galaxies with agn signatures . seven of these have broad emission lines ( type 1 agns ) . we also identify 26 galaxies displaying nuclear variability in hst wfpc2 images of the gss separated by @xmath07 years . the primary overlap of the two selected agn samples is the set of broad - line agns , of which 80% appear as variable . only a few narrow - line agns approach the variability threshold . the broad - line agns have an average redshift of @xmath1z@[email protected] while the other spectroscopic agns have redshifts closer to the mean of the general galaxy population ( @xmath1z@[email protected] ) . eighty percent of the identified broad - line agns are detected in x - rays and these are among the most luminous x - ray sources in the gss . only one narrow - line agn is x - ray detected . of the variable nuclei galaxies within the x - ray survey , 27% are x - ray detected . we find that [email protected]% of gss galaxies to v@xmath5@xmath624 are broad - line agns , [email protected]% are narrow - line agns , and [email protected]% contain variable nuclei . the fraction of spectroscopically identified blagns and nlagns at z@xmath01 reveals a marginally significant increase of [email protected]% when compared to the local population .
equilibrium structures of a rigid rod inside a phospholipid bilayer are simulated within the scmf theory@xcite . a phospholipid molecule is represented at a coarse - grained level within the three beads model@xcite which is found to describe adequately the mechanical properties of the phospholipid bilayer at equilibrium . the infinite rod lying parallel to the bilayer represents the excluded area for the lipids and water molecules , while the interactions between phospholipids in the bilayer are described through the mean fields . the results of scmf calculations are presented in figure [ results10 ] and figure [ results6 ] . the free energy differences are calculated per unit cylinder length and neglect end effects . this approximation of infinitely long cylinder is reasonable for cylinders in which the lateral surface area is much larger than the end - cap area . the interaction of lipids with the cylinder are described by a single parameter @xmath1 that is the energy of interaction of phospholipid tails t with the surface of the cylinder : @xmath2 implies only steric repulsion , while negative values reflect the attraction of t - beads to the surface of the cylinder . insertion of a completely hydrophilic cylinder , @xmath2 , into a bilayer is energetically unfavorable ( black lines in figure [ results10](a ) and figure [ results6](a ) ) . the energy penalty increases with the length of the cylinder . the hydrophilic cylinder compresses the bilayer at small penetrations and breaks it at distances about @xmath3 nm from the center . the insertion free energy increases monotonously with depth of penetration and reaches a plateau of height @xmath4 kt / nm . the rupture of the bilayer is similar to a first order transition in the bulk . two solutions , one corresponding to the compressed bilayer and another to the broken bilayer , can coexist at some distance , while for other distances one solution is stable and another is metastable . thus , the cylinder can follow the metastable branch with higher free energy of compressed and intact bilayer before abrupt rupture , accompanied with sudden drop of the free energy . another reason for the dispersion of the points at the plateau are the error of calculations . thus , it can be used to estimate the accuracy of the calculations ( @xmath5 kt / nm ) . the presence of a parallel hydrophobic stripe on a hydrophilic cylinder makes the penetration of the cylinder inside the bilayer more favorable . this stripe is described by its width or projected angle ( in degrees ) in figure [ results10 ] and figure [ results6 ] . we explore only the orientation with the stripe faced down to the bilayer , since this is the lowest energy orientation , but these calculations can also be used to estimate the energy of rotation of the cylinder in the bilayer . for relatively narrow stripes ( with the projected angle in the range @xmath6 degrees ) phospholipid heads move apart from the cylinder to accommodate the insertion at low penetrations , which allows for the energetically favorable contact between the stripe and the hydrophobic core of the membrane , and leads to the wetting of the stripe . the contact line moves up the hydrophilic cylinder surface forming a rim and increases the contact with the bilayer core . at higher penetrations the cylinder compresses the bilayer , and , eventually , breaks the membrane . the breakage distance between the cylinder and the membrane centers decreases with an increasing width of the stripe from @xmath7 nm ( the stripe of @xmath8 degrees projected angle ) to @xmath9 nm ( @xmath10 degrees stripe ) and the free energy of penetration becomes lower . the equilibrium position of the cylinder within the bilayer ( yellow band in figure [ results10](c ) corresponding to the minimum of the energy in figure [ results10](a ) ) also shifts to the center of the bilayer with increasing width of the stripe . thus , the width of the stripe controls not only the hydrophobicity of the cylinder and the insertion energy into the bilayer , but also the equilibrium position with respect to the bilayer core ( figure [ results10](b ) ) . the mean field concentration profiles in figure [ results10](b ) and figure [ results10](c ) demonstrate that hydrophobic interactions of @xmath11 kt are high enough for the cylinder to partially adsorb phospholipids from the bilayer . the hydrophobic stripe is covered with the phospholipids even after the rupture of the bilayer . for larger widths of the hydrophobic stripe ( @xmath12 degrees ) , the rupture of both leaflets of the bilayer does not occur and the cylinder is covered by phospholipids in the entire range of penetrations from @xmath13 to @xmath14 nm from the bilayer center . the bilayer only bends around the cylinder at small penetrations @xmath15 nm , and covers the whole hydrophobic part of its surface at larger penetrations @xmath16 nm . the scmf method provides the probability of conformations of molecules in a given field . thus , it allows for visualization of the solutions of the mean field equations in the form of the most probable conformations of molecules in the mean fields that correspond to the solution of the equations@xcite . the example of most probable conformations of phospholipids which correspond to the equilibrium position of the amphiphilic cylinder in the bilayer is shown in figure [ snapshot ] . at lower hydrophobicity ( @xmath17 kt , figure [ results6 ] ) , the behavior of the system is different in many aspects . the compression of the bilayer for relatively small ( @xmath6 degrees ) widths of the hydrophobic stripe leads to the rupture of the membrane farther from the bilayer center than for the same stripes with higher hydrophobicity . similarly , the equilibrium position of the cylinder with the same stripes but lower hydrophobicity is shifted closer to the surface of the bilayer . the overall surface of narrow stripes is too small to allow for the contact with two edges of the broken layer . hydrophobic stripes of larger width and @xmath18 kt induce small perturbations to the structure of the bilayer due to rearrangements of lipids around the stripe at relatively small penetrations . phospholipid heads pushed apart by the inserted cylinder enable contact of the stripe with the hydrophobic core of the bilayer . penetration of the cylinder to @xmath19 nm from the bilayer center , induces the rupture of the bilayer into two parts . each part of the broken layer contacts with the stripe at their edges . the equilibrium position of the cylinders with large hydrophobic stripes ( @xmath12 degrees ) is in the center of the bilayer . however , the interaction energy @xmath18 is insufficient for phospholipids to cover the cylinder from all sides and the cylinder forms a pore in the bilayer ( the central cross - section concentration profiles in figure [ results6](b ) are different from the corresponding profiles in figure [ results10](b ) for @xmath12 degrees stripes ) . the main difference between @xmath11 kt and @xmath18 kt cases is the difference in the free energy of insertion of the cylinder with large widths of the hydrophobic stripe ( @xmath12 degrees ) ( [ results10](a ) and [ results6](a ) ) . the more hydrophobic cylinder ( @xmath11 kt ) is fully covered with lipids tails and thus its free energy of insertion is much lower than that of the partially covered cylinder with @xmath18 kt . furthermore , the free energy plots for more hydrophobic cylinder @xmath11 kt have pronounced minima , while the plots corresponding to @xmath18 kt have wide regions of relatively constant energies around their minima which correspond to the equilibrium position . taking into account that the free energy calculations presented here have an accuracy of about @xmath20 kt / nm and the minima are shallow , one can plot the areas where the free energy cost of the cylinder insertion inside the membrane differs from its minimum value by less than @xmath21 kt / nm ( figure [ eqareas ] ) . the region of equilibrium positions is wider and is on average shifted from the center of the bilayer for lower hydrophobicity . thus , our results suggest that the cylinders with wide and low hydrophobicity stripes are able to change their position within a large range around the core of the bilayer with little or no energy cost . the shallow minima of less hydrophobic ( @xmath18 kt ) and wide ( @xmath12 degrees ) stripes can also be due to a particularity of the methodology used for the calculations . the vertical size of the simulation box is fixed for both cases and the walls restrict the movement of the bilayer , and so the absence of complete covering of the hydrophobic part of the cylinder with the phospholipids in case of @xmath18 kt leaves more free space inside the simulation box to bend the bilayer . as a result , small shifts of the cylinder inside the box for less than @xmath22 nm from the equilibrium , lead to bending of the bilayer together with the cylinder without reorganization of the bilayer structure . in contrast , there is almost no free space in the case of the cylinder with @xmath11 kt and small deviations of the position lead to significant reorganization of lipids covering the hydrophobic stripe , which has a higher free energy cost . however , this difference is only significant for two points of the free energy corresponding to wide stripes . although the minimum free energy orientation is with the stripe faced down to the bilayer , an estimate can be made of the free energy of the cylinder rotating around its axis . in the cases of completely hydrophilic , and completely hydrophobic cylinders , there is no changes to the free energy . rotation of the cylinder with the stripe , leads to contact of the bilayer edges and the hydrophobic core of the cylinder as well as contact between the heads of the bilayer phospholipids with a hydrophilic part of the cylinder . thus , there is an energy barrier for rotation , which can be estimated as follows . the maximum height of the barrier for rotation of cylinder at position @xmath13 nm from the bilayer center for @xmath10 degrees will be equal , approximately , to the average of the free - energy costs of insertion into this position of completely hydrophobic and completely hydrophilic cylinders ( from which we can estimate the energy of rotation for each position ) . in all others cases the equilibrium positions found for the orientation stripe - down will be true equilibrium position for any rotation of the cylinder , because in all these positions the hydrophobic core of the bilayer has the maximum possible contact with the hydrophobic stripe of the cylinder . hence , any rotation of the cylinder will not increase the contact area , but will significantly deform the bilayer , causing the free energy loss . our results demonstrate physical principles of the association of rod - shaped objects with a specific surface pattern and self - assembled phospholipid bilayers . in particular , a hydrophobic stripe along the axis of the cylinder can mimic self - assembled peptides with a secondary periodic structure such as @xmath0-helices . the position of the cylinder in the bilayer as well as the free energy of association depend on the geometry of the stripe , _ i.e. _ the width and the hydrophobicity . this microscopic information , in turn , can be used for design of artificial polymers with enhanced association with phospholipid bilayers . within the scmf theory@xcite the phospholipid molecule is modeled at a coarse - grained level , while the interactions between the molecules in the bilayer are represented by the mean fields . the three - beads model@xcite describes a phospholipid molecule as a sequence of three spherical beads of a radius @xmath23 , one hydrophilic ( h ) and two hydrophobic ( t ) , joined consequently with the bond length @xmath24 . two t - beads of neighboring molecules interact with the energy @xmath25 kt if the distance between the centers of the molecules is smaller than @xmath26 . the h - beads interact with implicit solvent molecules with the energy @xmath27 kt if the distance between the centers of the molecules is smaller than @xmath26 . the solvent molecules are considered as spherical beads of the same radius @xmath23 . the phospholipid molecule is free to bend around the central bead . a cylinder of diameter @xmath28 and longitudinal hydrophobic stripe on the surface is inserted into phospholipid bilayer at different positions in the simulation box . the position of the cylinder is fixed parallel to the bilayer with the orientation stripe - down . the hydrophobic part of the surface of the cylinder interacts with the t - beads with the energy @xmath29 , if the centers of the molecules are at the distance closer than @xmath30 , while the hydrophilic stripe has only steric repulsion interactions with phospholipids and water molecules . this allows to describe the hydrophobicity of the cylinder with a stripe by its width and a single interaction parameter , @xmath29 . we consider an infinite cylinder , where the cylinder length is equal to the size of the simulation box in the corresponding direction , and use the periodic boundary conditions . in cases the cylinder penetration is unfavorable , the phospholipid bilayer tends to bend and escape the inserted cylinder to minimize the equilibrium free energy . to restrict the bending of the phospholipid bilayer , we introduce hard walls at the top and bottom of the simulation box . this allows to estimate the free energy of insertion at a given position according to procedure described in ref .. garay - arroyo , a. ; colmenero - flores , j. m. ; garciarrubio , a. ; covarrubias , a. a. highly hydrophilic proteins in prokaryotes and eukaryotes are common during conditions of water deficit . _ j. biol . _ * 2000 * , _ 275 _ , 56685674 browne , j. ; tunnacliffe , a. ; burnell , a. anhydrobiosis : plant desiccation gene found in a nematode . _ nature _ * 2002 * , _ 416 _ , 38 tolleter , d. ; jaquinod , m. ; mangavel , c. ; passirani , c. ; saulnier , p. ; manon , s. ; teyssier , e. ; payet , n. ; avelange - macherel , m .- h . ; macherel , d. structure and function of a mitochondrial late embryogenesis abundant protein are revealed by desiccation . _ plant cell _ * 2007 * , _ 19 _ , 15801589 wise , m. j. ; tunnacliffe , a. popp the question : what do lea proteins do ? _ trends in plant science _ * 2004 * , _ 9 _ , 1317 goyal , k. ; tisi , l. ; basran , a. ; browne , j. ; burnell , a. ; zurdo , j. ; tunnacliffe , a. transition from natively unfolded to folded state induced by desiccation in an anhydrobiotic nematode protein . _ j. biol . chem . _ * 2003 * , _ 278 _ , 1297712984 popova , a. v. ; hundertmark , m. ; seckler , r. ; hincha , d. k. structural transitions in the intrinsically disordered plant dehydration stress protein lea7 upon drying are modulated by the presence of membranes . _ biochimica et biophysica acta _ * 2011 * , _ 1808 _ , 18791887 panchal , r. g. ; smart , m. l. ; bowser , d. n. ; williams , d. a. ; petrou , s. pore - forming proteins and their application in biotechnology . _ current pharmac . _ _ 3 _ , 99115 yang , l. ; harroun , t. a. ; weiss , t. m. ; ding , l. ; huang , h. w. barrel - stave model or toroidal model ? a case study on melittin pores . _ biophys . j. _ * 2001 * , _ 81 _ , 14751485 illya , g. ; deserno , m. coarse - grained simulation studies of peptide - induced pore formation . _ biophys . j _ * 2008 * , _ 95 _ , 41634173 gkeka , p. ; sarkisov , l. interactions of phospholipid bilayers with several classes of amphiphilic alpha - helical peptides : insights from coarse - grained molecular dynamics simulations . _ j. phys . b _ * 2010 * , _ 114 _ , 826839 tunnacliffe , a. ; wise , m. j. the continuing conundrum of the lea proteins . _ naturwissenschaften _ * 2007 * , _ 94 _ , 791812 pogodin , s. ; slater , n. k. h. ; baulin , v. a. surface patterning of carbon nanotubes can enhance their penetration through a phospholipid bilayer . _ acs nano _ * 2011 * , _ 5 _ , 11411146 pogodin , s. ; baulin , v. a. coarse - grained models of phospholipid membranes within the single chain mean field theory . _ soft matter _ * 2010 * , _ 6 _ , 22162226 pogodin , s. ; baulin , v. a. can a carbon nanotube pierce through a phospholipid bilayer ? _ acs nano _ * 2010 * , _ 4 _ , 52935300 pogodin , s. ; baulin , v. a. equilibrium insertion of nanoscale objects into phospholipid bilayers . _ cnano _ * 2011 * , _ 7 _ , 721726	under dehydration conditions , amphipathic late embryogenesis abundant ( lea ) proteins fold spontaneously from a random conformation into @xmath0-helical structures and this transition is promoted by the presence of membranes . to gain insight into the thermodynamics of membrane association we model the resulting @xmath0-helical structures as infinite rigid cylinders patterned with hydrophobic and hydrophilic stripes oriented parallel to their axis . statistical thermodynamic calculations using single chain mean field ( scmf ) theory show that the relative thickness of the stripes controls the free energy of interaction of the @xmath0-helices with a phospholipid bilayer , as does the bilayer structure and the depth of the equilibrium penetration of the cylinders into the bilayer . the results may suggest the optimal thickness of the stripes to mimic the association of such protein with membranes . journal link : http://pubs.acs.org/doi/abs/10.1021/nn204736b certain biological molecules interact with cell membranes to affect some functional property . late embryogenesis abundant ( lea ) proteins are an example of such molecules . these proteins are expressed in plant and some animal cells in response to desiccation conditions and they increase the ability of such cells to withstand desiccation stress@xcite . they are widespread among higher plants but have also been found in some microorganisms and nematodes exposed to water deficit@xcite . lea proteins may exist in the form of random coils@xcite and can spontaneously self - assemble into @xmath0-helices upon desiccation stress@xcite . such an @xmath0-helix structure in turn provides an ordered pattern of alternating hydrophilic and hydrophobic stripes that run along the axis of the helix@xcite which may facilitate the interaction and insertion of the protein into the phospholipid membrane . recently , popova _ _ et al.__@xcite have shown that the fraction of lea7 protein that takes up the @xmath0-helix conformation is increased when it is dried in the presence of liposomes , indicating that phospholipid membranes associate closely with lea proteins and play a critical role in stabilizing their @xmath0-helix conformation . lea proteins are amphiphilic and when in the @xmath0-helix conformation can be approximately described as solid rods with a distinct axial arrangement of hydrophilic and hydrophobic stripes . an entirely distinct class of membrane - active peptides , pore - forming peptides@xcite , also have surface patterning with a hydrophilic stripe along the backbone of an @xmath0-helix . however , the mechanism of association of these proteins with the phospholipid membrane is very different . membrane pores are formed due to the concerted self - assembly of several peptides on the membrane with a preferential orientation of peptides perpendicular to the bilayer plane ; the so - called `` barrel and stave '' mechanism@xcite . to allow for such an orientation , the length of the pore - forming peptides does not usually exceed the thickness of the bilayer@xcite . in contrast , lea proteins adopt a more random structure and are much longer@xcite . as a consequence they lie as single molecules parallel to the surface of the membrane bilayer . the hydrophobic groups on the side of the backbone control the depth of insertion of the lea proteins into the bilayer . to gain insight into the factors that influence the interaction of such a rod with a phospholipid bilayer we have calculated the free energy of association as a function of stripe geometry and position of the rod relative to the center of the bilayer . in a previous paper@xcite we showed that single chain mean field ( scmf ) theory could be used to calculate the equilibrium structures for insertion of a hydrophobically patterned rod into a phospholipid bilayer with the rod axis oriented perpendicular to the layer . these calculations aimed to determine the effect of patterning upon the puncture of the bilayer by the rod . here , equilibrium structures of association of a hydrophobically patterned rod with a bilayer are simulated with the rod axis oriented parallel to the layer . these calculations aim to mimic the behaviour of moieties such as lea proteins that may associate with , but not puncture , bilayers .
in this section , we briefly review the basic symmetry properties of the @xmath0 quantum link model of eq . ( 3 ) of the main text . 1 . as in any gauge theory , the hamiltonian is invariant against local symmetry transformations . in this case , it commutes with the infinitesimal @xmath0 gauge generators ( additional details are given in the second supplementary material file ) : @xmath112 - e_{x , x+1 } + e_{x-1,x}.\end{aligned}\ ] ] 2 . the parity transformation p is implemented as @xmath113 3 . while charge conjugation c acts as @xmath114 4 . for @xmath106 the hamiltonian also has a @xmath115 chiral symmetry which shifts all fields by one lattice spacing , @xmath116 however , this symmetry is explicitly broken when one imposes the gauss law @xmath117 . in this section , we sketch the main steps to reduce the _ microscopic _ hubbard model , eq . ( 6 ) of the main text , to an _ effective _ quantum link model at low energies using second order perturbation theory . we are interested in the scenario where the largest energy scale @xmath118 is given by the diagonal hamiltonian @xmath119 \nonumber \\ & + & \left(u + m\right ) \sum_x \left(-1 \right)^x \psi^\dagger_x \psi_x \nonumber \\ & = & u \sum_x \widetilde{g}_x^2 + \frac{g^2}{2 } \ ! \sum_x e_{x , x+1}^2 + m \sum_x \left(-1 \right)^x \psi^\dagger_x \psi_x.\end{aligned}\ ] ] the values @xmath120 and @xmath43 are small compared to @xmath121 , i.e. @xmath122 , but they are still relevant in the induced quantum link model . the term to be generated in second order perturbation theory is the correlated hopping of fermions mediated by the quantum link ( represented by a quantum spin ) . it appears as an effective interaction induced by the previous hamiltonian and the perturbation terms @xmath123 \nonumber \\ & & - t_b \sum_{x \ \text{even } } \left[b^{2 \dagger}_x b^2_{x+1 } + b^{2 \dagger}_{x+1 } b^2_{x } \right].\end{aligned}\ ] ] to second order in @xmath124 and @xmath125 , the effective hamiltonian reads @xmath126 \nonumber \\ & & \!\!\!\!\!\!\ ! - \frac{t_f^2}{u } \sum_x \psi^\dagger_x \psi_x \left(1 - \psi^\dagger_{x+1 } \psi_{x+1 } \right).\end{aligned}\ ] ] the last term proportional to @xmath127 was not present in the original quantum link model hamiltonian . this is no problem , because this term is also gauge invariant , and could have been added to the quantum link hamiltonian from the beginning . a ) pictorial representation of four gauge invariant states coupled in second order perturbation theory in the @xmath39 case . b ) expectation values of the charge difference @xmath128 and the electric flux @xmath129 as a function of the real - time @xmath130 , starting from one of the four states , under the evolution of the microscopic hamiltonian for @xmath131 ( solid line ) . the effective quantum link model ( dashed line ) yields an expectation value @xmath132 $ ] for the product of both operators ( left plot ) , while the product of both expectation values is @xmath133 $ ] ( right plot ) , signaling the collective dynamics of the coherent fermion hopping mediated by the quantum link . in the plots @xmath134.,scaledwidth=48.0% ] to test the reduction of the microscopic hubbard - type model to the effective quantum link model , we have considered a minimal setup of four lattice sites in the @xmath39 case , as illustrated in fig . [ 4state]a . we have compared the low - energy spectrum of the microscopic and the corresponding effective hamiltonian . the spectra coincide for @xmath135 , and even in the intermediate interaction regime @xmath136 the deviations are of order @xmath137 . in view of experimental realizations , another relevant question is to what extent the hamiltonian preserves the gauss law . we have considered several initial gauge invariant states , evolving in time according to @xmath138 . for @xmath139 ( @xmath140 ) , the probability to leave the gauge invariant subspace is below @xmath141 ( @xmath142 ) even for time scales of order @xmath143 . moreover , as demonstrated in fig[4state]b , the expectation values of @xmath144 and @xmath129 display oscillations typical of a coherent two - body process , in direct analogy with the double well experiments in @xcite . at the many - body level , we have also studied the flux configuration in the ground state of the microscopic hamiltonian compared to the emergent qlm using exact diagonalization with open boundary conditions in the @xmath36 representation . as illustrated in fig . 3a of the main text , the microscopic model compares very favorably with its qlm analog , and gauge invariance is effectively realized ( see fig . 3b of the main text ) . an experimental implementation will require three basic steps : preparation of an initial gauge invariant state , evolution via quantum link dynamics , and measurement of relevant physical observables . the first step can be implemented by preparation of mott insulator states of bosonic and fermionic species on different lattice sites via loading in deep optical lattice potentials . subsequently , the ground state or quench dynamics can be realized via adiabatic or rapid lowering of the depth of the optical lattices . coherent evolution according to the qlm may be validated at the few - body level by performing double well experiments ( corresponding to a single quantum link ) along the lines of ref . a numerical case study is presented in the supplementary information . finally , _ in - situ _ site - resolved imaging of bosonic particle number distributions @xcite allows one to measure @xmath129 and to reconstruct the spin - flux configuration and , thus to quantitatively probe the system . as far as 1d implementations are considered , the statistics of all basic components does not play a fundamental role ( this is not true in 2d and 3d , where fermi statistics is a key - ingredient of matter fields ) . in particular , the gauge bosons may be substituted by fermions ( as far as @xmath145 representations are concerned ) and matter fields may be represented by hard - core bosons . such flexibility may be relevant when considering specific experimental implementations , consistently enlarging the set of candidate systems . here we illustrate an alternative route toward realizing @xmath0 quantum link models where a _ dipolar bosonic species is sufficient to realize a gauge covariant link structure . for simplicity , we focus on the @xmath36 setup , although larger spins can , in principle , be achieved by considering on - site bosonic interactions . while this construction extends to higher dimensions in a straightforward manner , here we discuss the simpler 1d implementation . the microscopic model studied here uses a mixture of fermionic and bosonic particles in the presence of strong dipolar interactions @xcite . possible experimental realizations are quantum gases of magnetic atoms like cr @xcite , er @xcite , or dy @xcite , and dipolar molecules @xcite . as sketched in fig . [ si ] , the mixture is confined to a lattice , where fermions occupy sites labeled by @xmath3 and bosons are defined on the link sites @xmath146 and @xmath147 , to the left and to the right of @xmath3 . bosons can hop only between sites @xmath147 and @xmath148 , and serve as natural link variables when expressed in terms of schwinger bosons @xmath149 note that here the bosonic index @xmath150 is related to the lattice configuration , and , in contrast to eq . ( 6 ) of the main text , it is not associated with an internal degree of freedom . by identifying @xmath151 , @xmath152 ( for @xmath3 even ) and @xmath153 , @xmath154 ( for @xmath3 odd ) , we can relate the bosons with spatial indices @xmath146 and @xmath147 to the bosonic species @xmath155 that arise in model i discussed in the main text . the microscopic hamiltonian of model ii takes the form @xmath156 n_{y,\beta } \nonumber \\ & & + 2w_{fb } \sum_x n_x^f [ n_{x , l } + n_{x , r}],\end{aligned}\ ] ] where @xmath157 $ ] is given by the dipolar interaction between the particles . its strength is normalized such that @xmath158 = 2 v_{lr}[x , x ] = 1 $ ] . the bose - fermi interaction is a combined effect of both dipolar and short - range potentials . the latter stems from the partial overlap between the single - site wannier functions of bosons and fermions in @xmath159 and @xmath3 , respectively . in analogy with model i , and taking into consideration the fast spatial decay of dipolar interactions , one can reformulate the hamiltonian of eq . ( [ h_dip ] ) as @xmath160 , where @xmath161.\ ] ] by choosing @xmath162 , @xmath163 , one then obtains a quantum link model with @xmath164 , with additional gauge invariant diagonal terms generated by the dipolar interaction beyond nearest - neighbor sites . a ) lattice configuration of model ii . fermions hop between sites @xmath3 and @xmath63 , while bosons hop in double wells on each link . b ) lattice configuration including hamiltonian parameters . c ) alternative scheme employing raman assisted tunneling rates : fermions with spin are loaded into a species - dependent optical lattice , while bosons are confined in a periodic potential with half the wavelength.,scaledwidth=40.0% ] in case of _ magnetic _ atoms @xcite , the interaction regime @xmath165 may be achieved by properly tuning the interspecies scattering length and the optical lattice depth , leading to typical energy scales of the order @xmath166 for , e.g. , dy bosonic gases confined in an optical lattice with a lattice spacing of about @xmath167 @xcite . polar molecules have large _ electric _ dipole moments which can be aligned by using external electric fields providing sufficiently strong constraint energies @xmath118 @xcite when loaded into optical lattices @xcite . a clear advantage of this setup is that it can be straightforwardly adapted to 2d , since , in contrast to model i , just one bosonic species is required regardless of the dimensionality . furthermore , dipolar interactions are isotropic once the dipole moments are aligned perpendicular to the 2d plane @xcite . the above realization of gauge fields is based on bosonic atoms in a double well potential corresponding to the familiar _ external _ josephson effect . another intriguing possibility provided by dipolar interactions makes use of an _ internal _ josephson effect @xcite , where ( instead of two wells coupled by a tunnel coupling ) one considers atoms with two internal states coupled by a rabi frequency . such an internal josephson junction provides an alternative realization of quantum link spins . the setup that we propose is illustrated in fig . the gauge invariant dynamics are realized by combining a fermionic species _ with spin _ moving in a _ spin - dependent _ optical lattice , and a bosonic dipolar species confined in a deep optical lattice of half the wavelength . tunneling of fermions between adjacent lattice sites is implemented by a raman assisted transition in the spirit of @xcite , coupled to the _ internal _ josephson system with the corresponding bosonic link site ( c.f . [ si ] ) , leading to an effective hamiltonian term of the form @xmath168 . gauge invariance is then implemented by considering state - dependent dipole - dipole interactions between the bosonic links , and a finite detuning shift for the raman transition @xmath169 . a detailed study of this implementation scheme will be reported elsewhere . string breaking is an important dynamical mechanism in quantum chromodynamics ( qcd ) . a quark and an anti - quark separated by a distance @xmath171 are connected by a string of color electric flux . the string costs energy in proportion to its length , and thus leads to a linearly rising quark - anti - quark potential @xmath172 , where @xmath77 is the string tension . when the quark mass is taken to infinity , i.e. when one considers an @xmath173 yang - mills theory of pure gluons , in which quarks and anti - quarks appear just as external sources of color flux , the confining string is unbreakable . in qcd with dynamical quarks , however , the string can break by quark - anti - quark pair creation , as soon as the energy of the string is larger than the mass of the mesons generated in the pair creation process . the dynamics of string breaking , i.e. its evolution in real time , is inaccessible to classical simulations using lattice gauge theory , which is usually formulated in euclidean time . as for other quantum systems with a large number of degrees of freedom , studying the real - time evolution of a lattice gauge theory leads to a very severe sign problem . for this reason , understanding the real - time dynamics of gauge theories from first principles remains inaccessible to the traditional methods of theoretical physics . quantum simulators are a very promising tool for overcoming this problem . string breaking is only one example of a dynamical mechanism whose real - time evolution is interesting but very difficult to investigate . the real - time evolution of heavy ion collisions is another example , which may serve as a strong motivation and an ultimate long - term goal for the development of quantum simulators for gauge theories . another interesting phenomenon of great relevance in particle physics is the spontaneous breaking and restoration of symmetries . for example , in the early universe , chiral symmetry was intact before it got spontaneously broken at sufficiently low temperature . in heavy ion collisions , one attempts to recreate a droplet of the early unbroken phase and observes its evolution back into the vacuum state . the time evolution of metastable false vacuum states has been discussed in doomsday scenarios of cosmic evolution , in which the rapid expansion of a bubble of true vacuum wipes out the preexisting false vacuum . in one spatial dimension , string breaking by pair creation and false vacuum decay are similar phenomena , with the dynamical quark and anti - quark playing the role of the walls of the true vacuum bubble . in fact , as we will see , we can use the same model hamiltonians to quantum simulate both string breaking and false vacuum decay . in wilson s formulation of lattice gauge theory , the fundamental gauge degrees of freedom are parallel transporter matrices taking values in the gauge group , which are associated with the links connecting neighboring lattice sites on a 4-dimensional euclidean space - time lattice . in lattice qcd the link variables are @xmath174 unitary matrices with determinant 1 , which constitute an 8-parameter family of continuously varying classical gauge degrees of freedom per link . as a consequence , in wilson s formulation of lattice gauge theory , there is an infinite - dimensional hilbert space already for each individual link , which largely complicates the construction of quantum simulators . quantum link models are an alternative formulation of lattice gauge theory , in which the fundamental gauge degrees of freedom are represented by discrete quantum variables , so - called quantum links , which have a finite - dimensional hilbert space per link . in the simplest case of a @xmath0 gauge theory , a quantum link is simply represented by a quantum spin . when one chooses spin @xmath175 , a single qubit per link is hence sufficient to represent the gauge field . in an @xmath33 gauge theory the quantum link is a generalized quantum spin with a continuous gauge symmetry , which has at least 4 discrete degrees of freedom . hence , two qubits per link are required to represent a non - abelian gauge field . a @xmath176 quantum link model requires 6 and an @xmath173 quantum link model requires 20 discrete states per link , which amounts to 3 or 5 qubits per link , respectively . for simplicity , in this supplementary material we limit ourselves to @xmath0 quantum link models in one spatial dimension , considering both the spin @xmath175 representation ( one qubit ) and the spin 1 representation , with 3 discrete states per link . in this supplementary material , we discuss quantum link models with a small number of degrees of freedom , which still display the physical phenomenon of string breaking or false vacuum decay . the proposed model hamiltonians are not yet fully realistic from a particle physics point of view . in particular , they address the physics of the schwinger model ( i.e. qed in @xmath177 space - time dimensions ) rather than qcd in @xmath178 dimensions . for simplicity , the schwinger model is formulated with so - called staggered fermions , which have only one degree of freedom per lattice site . it should be pointed out that we are not yet addressing the schwinger model in the continuum limit . hence , several steps will have to be taken in order to turn the proposed quantum simulations into something that becomes truly useful for particle physics . still , the pathway towards this ultimate goal is clearly visible . in order to better understand the theoretical framework of quantum link models , let us begin to discuss the familiar situation of fermions moving in a classical background magnetic field described by a vector potential @xmath4 with @xmath179 . for simplicity , we consider spinless fermions hopping on a lattice with sites @xmath3 . the fermion creation and annihilation operators obey standard anti - commutation relations @xmath180 the term in the hamiltonian that describes gauge covariant hopping from a site @xmath7 to a nearest - neighbor site @xmath3 is then given by @xmath5 . here @xmath181 is a classical gauge parallel transporter associated with the link connecting @xmath3 and @xmath7 that is obtained by integrating the vector potential along the link @xmath182 after a gauge transformation @xmath183 one obtains @xmath184 and hence the parallel transporter transforms as @xmath185 in order to render the hopping hamiltonian gauge invariant , the gauge transformation of the classical background field must be accompanied by the gauge transformation of the fermion creation and annihilation operators @xmath186 which leaves their anti - commutation relations unchanged . in this work , we are not limiting ourselves to classical background fields . instead the electromagnetic field is treated as a dynamical entity with its own quantum dynamics that is intimately connected to the dynamics of the fermions . a dynamical lattice gauge field @xmath14 is no longer given in terms of a classical background field @xmath4 . instead , the link variable @xmath14 is an independent quantum degree of freedom , whose canonically conjugate `` momentum '' variable @xmath187 represents an electric field operator that obeys the commutation relation @xmath188 = u_{xy}.\ ] ] in a gauge theory one uses redundant gauge variant variables to describe the gauge invariant physics . in order to eliminate the unphysical gauge variant states , one must impose gauss law . physical states @xmath189 are gauge invariant , i.e. @xmath190 , where @xmath191 is the infinitesimal generator of gauge transformations . a general ( non - infinitesimal ) gauge transformation with parameters @xmath192 is described by the unitary transformation @xmath193 which indeed acts as @xmath194 the hamiltonian of a lattice gauge theory with a dynamical @xmath0 gauge field coupled to dynamical fermions takes the form @xmath195 here @xmath42 is the hopping amplitude and @xmath44 is the gauge coupling that determines the electric and the magnetic field energies . by construction , the hamiltonian is gauge invariant , i.e. @xmath28 = 0 $ ] . in the hamiltonian formulation of wilson s @xmath0 lattice gauge theory , the link variables are still complex phases @xmath29 , but the link angles @xmath196 are now independent degrees of freedom ( unrelated to a background field @xmath4 ) . correspondingly , the canonically conjugate electric field operators are given by @xmath30 , such that the commutation relation of eq.([commutator ] ) is indeed satisfied . since wilson s parallel transporter is a continuous variable , the corresponding hilbert space is infinite - dimensional even for just a single link . in the quantum link formulation of @xmath0 lattice gauge theory , on the other hand , the dimension of the link hilbert space is finite , and given by the @xmath35 states of an integer or half - integer quantum spin @xmath34 on each link . in this case , the electric field operator is given by @xmath32 with eigenvalues @xmath197 , while the gauge field is represented by the quantum link operators @xmath198 which act as raising and lowering operators of electric flux . it is to be noted that the operators @xmath199 satisfy a local su(2 ) algebra at each link ( see eq . [ commutator ] ) . this allows the realization of a local u(1 ) gauge invariance of the hamiltonian in eq . [ ham ] . [ [ lattice - hamiltonian - for - the - schwinger - model - with - staggered - fermions ] ] lattice hamiltonian for the schwinger model with staggered fermions ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the hamiltonian for a simple quantum link model is given by @xmath200 \nonumber \\ & & + m \sum_x ( -1)^x \psi_x^{\dag}\psi_x + \frac{g^2}{2}\sum_{x } e_{x , x+1}^2.\end{aligned}\ ] ] this model describes a single species of staggered fermions ( i.e. , the fermion field operator @xmath201 has only a single component ) minimally coupled to a @xmath0 gauge field . the electric flux operator defined on each link is @xmath202 , and the gauge field is represented by quantum links @xmath203 and @xmath204 , which satisfy the commutation relation @xmath205= \delta_{xy } u_{x , x+1}.\ ] ] in this supplementary material , we will consider quantum links both in the spin @xmath175 and in the spin 1 representation . in the minimal spin @xmath175 representation , the allowed electric flux eigenvalues are @xmath206 , i.e. it is impossible to have zero flux . this may seem unnatural from the point of view of particle physics . still , as we will see , the resulting physics resembles that of a non - zero vacuum angle @xmath207 . the spin 1 representation allows the three flux eigenvalues 0 and @xmath208 . the factor @xmath209 in the fermion mass term is due to the use of staggered fermions . it explicitly breaks the translation symmetry by one lattice spacing . this symmetry plays the role of a discrete @xmath115 chiral symmetry for staggered fermions . while in the spin 1 case the dynamics of the gauge field is governed by the term proportional to the gauge coupling @xmath210 , this term has a trivial contribution in the spin @xmath175 case . * gauge invariance : the hamiltonian commutes with the infinitesimal generators of local @xmath0 transformations @xmath211 * chiral symmetry : the model admits the following discrete chiral symmetry transformation : @xmath212 this transformation preserves all terms in the hamiltonian except the mass term . * charge conjugation c : this is again a discrete symmetry , which is implemented as : @xmath213 this transformation leaves all terms invariant . * parity p : the parity transformation is implemented as : @xmath214 * cp symmetry : the combined symmetry takes the form : @xmath215 before we discuss the implementation of the gauss law , note that the staggered fermions introduce a `` staggering '' in the occupation numbers of the vacuum , which resembles a half - filled state in condensed matter physics , even in the absence of the gauge field . because of the mass term , in the vacuum ( i.e. the dirac sea ) the fermions occupy the odd sites ( for @xmath216 , @xmath217 ) . this means that the gauss law can not be satisfied in the usual way : even with equal amounts of electric flux leaving and entering a site , the flux divergence @xmath218 depends on whether the site is even or odd . this is an unavoidable feature of the staggered fermion formulation . we define the gauss law such that the configuration with the staggered occupation in the absence of electric flux is a physical state that satisfies it , although not in the usual form . all physical states @xmath189 are then required to satisfy @xmath219 the string formed between an external static charge - anti - charge pair is an interesting physical object to be studied using quantum simulation . let us consider string breaking in the case of spin 1 quantum links . figure [ fig4]b shows an initial state which has a string joining an external static charge @xmath220 and an anti - charge @xmath221 placed at both ends . as the fermions become lighter , the sequence of figures [ fig4]c [ fig4]e shows the breaking of the string due to the creation of fermion - anti - fermion pairs , and the evolution to the final state of two mesons at the ends . for the static @xmath46 case , the vacuum energy is given by @xmath51 , while the energies of the initial unbroken string state and the final two - meson state take the form @xmath222 the critical distance for string breaking is determined by the condition @xmath223 and thus results in @xmath60 . ( 0,30)(20,0)2 ( 30,30 ) ( 60,30 ) ( 10,30)(0,-1)(0,1)2 ( 40,30)(0,-1)(0,1)2 ( 50,30)(20,0)2(0,-1)(0,1)2 ( 3.0,30)(10,0)3(-1,0)0 ( 5.0,30)(10,0)3(-1,0)0 ( 43.0,30)(10,0)3(-1,0)0 ( 45.0,30)(10,0)3(-1,0)0 ( -5,29)@xmath227 ( -1.5,33)@xmath55 ( 29,33)@xmath228 ( 39,33)@xmath229 ( 68.5,33)@xmath220 ( 0,15 ) ( 10,15)(20,0)3 ( 20,15)(20,0)3(0,-1)(0,1)2 ( 70,15)(0,-1)(0,1)2 ( 3.0,15)(20,0)4(-1,0)0 ( 5.0,15)(20,0)4(-1,0)0 ( -5,14)@xmath230 ( -1.5,18)@xmath55 ( 9,18)@xmath228 ( 19,18)@xmath229 ( 29,18)@xmath228 ( 39,18)@xmath229 ( 49,18)@xmath228 ( 59,18)@xmath229 ( 68.5,18)@xmath220 the minimal model with @xmath36 quantum links may seem unnatural from a particle physics point of view , because it does not allow zero flux . however , it resembles a non - zero vacuum angle @xmath61 . in order to explore these dynamics , let us consider the hamiltonian @xmath200 \nonumber \\ & & + m \sum_x ( -1)^x \psi_x^{\dag}\psi_x - 2 \sigma \sum_x ( -1)^x e_{x , x+1}.\end{aligned}\ ] ] the last term favors a staggered flux pattern and is charge conjugation and parity invariant . it does , however , explicitly break chiral symmetry . since chiral symmetry is anyways explicitly broken by the gauss law implementation , this additional breaking is unimportant . terms linear in the electric field are characteristic of a non - zero vacuum angle @xmath207 . for @xmath234 , charge conjugation and parity are both explicitly violated . at @xmath235 , on the other hand , charge conjugation and parity remain exact symmetries , as it is the case here . interestingly , in the schwinger model at @xmath61 , both charge conjugation and parity are spontaneously broken . as we will see , this is also the case in the model that we have just introduced , at least for sufficiently large values of the mass @xmath43 ( and for @xmath236 ) , while for small @xmath43 , charge conjugation and parity are restored . hence , the model with the staggered flux term proportional to @xmath77 mimics the physics of @xmath61 , which is again very difficult to simulate with a classical computer , due to a very severe sign problem . let us finally notice that , after implementing gauss law , the staggered electric field term can be recast into a renormalization for the fermionic mass term , that is @xmath237 , so that one recovers exactly the model in eq . ( 3 ) of the main text . let us consider the physics in the vacuum sector , using periodic boundary conditions of a finite system with an even number of lattice sites @xmath50 . for @xmath46 , there are three candidate vacuum states , which are depicted in figures [ fig6 ] ( 80,6)(0,-5 ) ( 0,0)(1,0)80 ( 0,0)(20,0)5 ( 10,0)(20,0)4(0,-1)(0,1)2 ( 6.5,0)(20,0)4(1,0)0 ( 14.0,0)(20,0)4(-1,0)0 ( -1,-5)@xmath50 ( 9,-5)@xmath238 ( 19,-5)@xmath239 ( 29,-5)@xmath240 ( 65,-5)@xmath241 ( 79,-5)@xmath50 first of all , it is instructive to convince oneself that these three states indeed satisfy the gauss law . the state shown in figure [ fig6 ] is invariant under both charge conjugation and parity , while the two states in figure [ fig7 ] are charge conjugation and parity partners of each other . at @xmath46 the energies of the three states are given by @xmath242 such that @xmath243 . hence , for @xmath244 ( with @xmath245 ) the state depicted in figure [ fig6 ] is the true vacuum ( i.e. the hamiltonian eigenstate of lowest energy respecting the gauss law ) . in this case , both c and p are unbroken . on the other hand , for @xmath246 , the two states shown in figure [ fig7 ] have a lower energy and thus c and p are then spontaneously broken . this is reminiscent of the situation in the schwinger model at @xmath61 . it would be interesting to quantum simulate the decay of the false vacuum of figure [ fig6 ] into the true vacua of figure [ fig7 ] in real time . the quench dynamics discussed in the main text is directly related to this phenomenon . as we have seen , a simple ( 1 + 1)-d quantum link model with staggered fermions allows us to address interesting dynamical questions using quantum simulation . along the way towards addressing similar problems in qcd , one faces the challenges of higher dimensions , non - abelian gauge fields , and multi - component fermions . in the quantum link formulation of qcd all these elements are present , and we see no fundamental obstacle against implementing then with ultra cold matter in optical lattices . as one goes beyond ( 1 + 1 ) dimensions , gauge field acquire transverse degrees of freedom and display a richer dynamics . since quantum link models are gauge invariant by construction , these dynamics are those of qcd in the continuum limit . how to best realize this in quantum simulations will be addressed in forthcoming publications .	using a fermi - bose mixture of ultra - cold atoms in an optical lattice , we construct a quantum simulator for a @xmath0 gauge theory coupled to fermionic matter . the construction is based on quantum links which realize continuous gauge symmetry with discrete quantum variables . at low energies , quantum link models with staggered fermions emerge from a hubbard - type model which can be quantum simulated . this allows us to investigate string breaking as well as the real - time evolution after a quench in gauge theories , which are inaccessible to classical simulation methods . recently , the condensed matter and atomic physics communities have mutually benefited from synergies emerging from the quantum simulation of strongly correlated systems using atomic setups @xcite . in particular , physically interesting quantum many - body systems , which can not be solved with classical simulation methods , are becoming accessible to analog or digital quantum simulation with cold atoms , molecules , and ions . in the future , quantum simulators may also enable us to address currently unsolvable problems in particle physics , including the real - time evolution of the hot quark - gluon plasma emerging from a heavy - ion collision or the deep interior of neutron stars @xcite . the challenge on the atomic physics side is to find a physical implementation of gauge theories with cold atoms , and to identify possible atomic setups representing dynamical gauge fields coupled to fermionic matter . below we provide a toolbox for a @xmath0 lattice gauge theory ( lgt ) using atoms in optical lattices @xcite . here fermionic atoms represent matter fields . they hop between lattice sites and interact with dynamical gauge fields on the links embodied by bosonic atoms . the lgt to be implemented is a so - called quantum link model ( qlm)@xcite , where the fundamental gauge variables are represented by quantum spins . qlms extend the concept of wilson s lgt @xcite . in particle physics they provide an alternative non - perturbative formulation of dynamical abelian and non - abelian gauge field theories @xcite . qlms are also relevant in condensed matter contexts , like spin liquids and frustrated systems @xcite . their hamiltonian formulation provides a natural starting point for quantum simulation protocols based on atomic gases in optical lattices @xcite . we will illustrate atomic quantum simulation of an abelian qlm in a 1d setup , demonstrating both dynamical string breaking and the real - time evolution after a quench , which are also relevant in qcd . the quantum simulator discussed below makes the corresponding real - time dynamics , which is exponentially hard for classical simulations based on wilson s paradigm @xcite , accessible to atomic experiments . cold quantum gases provide a unique experimental platform to study many - body dynamics of isolated quantum systems . in particular , cold atoms in optical lattices realize hubbard dynamics for both bosonic and fermionic particles , where the single particle and interaction terms can be engineered by external fields . the remarkable experimental progress is documented by the quantitative determination of phase diagrams in strongly interacting regimes , the study of quantum phase transitions , and non - equilibrium quench dynamics @xcite . one of the most exciting recent developments are _ synthetic gauge fields with atoms _ , which promises the realization of strongly correlated many - body phases , such as , e.g. , the fractional quantum hall effect with atoms @xcite . a fermion that is annihilated by @xmath1 and recreated by @xmath2 at a neighboring site @xmath3 , which propagates in the background of a classical abelian vector potential @xmath4 gives rise to the hopping term @xmath5 with @xmath6 . hopping between the adjacent lattice sites @xmath3 and @xmath7 accumulates the phase @xmath8 . the hopping term is invariant against @xmath0 gauge transformations @xmath9@xcite . when a fermion hops around a lattice plaquette @xmath10 , it picks up a gauge invariant magnetic flux phase @xmath11 , with @xmath12 . we emphasize that these synthetic gauge fields are _ @xmath13-numbers _ mimicking an external magnetic field for the ( neutral ) atoms . instead , here we are interested in _ dynamical _ gauge fields as they arise in particle physics @xcite . the corresponding fundamental bosonic degrees of freedom @xmath14 are no longer related to an underlying classical background field @xmath4 , but represent quantum operators associated with the lattice links . the hopping of the fermions is now mediated by the bosonic gauge field via the term @xmath15 , which is invariant under local changes of matter and gauge degrees of freedom @xmath16 , @xmath17 , and @xmath18 . here @xmath19 is an electric field operator associated with the link connecting @xmath3 and @xmath20 , where @xmath21 is a unit - vector in the @xmath22-direction . @xmath23 is the generator of gauge transformations ( see @xcite for a detailed discussion ) . gauge invariant physical states must obey gauss law , @xmath24 , which is the lattice variant of @xmath25 . to ensure gauge covariance of @xmath14 , it must obey @xmath26 = u_{xy}$ ] . the hamiltonian representing the electric and magnetic field energy of a compact @xmath0 lgt , @xmath27 , is gauge invariant , i.e. @xmath28 = 0 $ ] . in wilson s lgt , the link variables @xmath29 are still complex phases , and @xmath30 . since @xmath14 is a continuous variable , which implies an infinite - dimensional hilbert space per link , it is not clear how to implement it in ultra - cold matter , where one usually deals with discrete degrees of freedom in a finite - dimensional hilbert space . _ quantum link models _ offer an attractive framework for the quantum simulation of dynamical gauge fields @xcite . they extend the concept of a lgt to systems of discrete quantum degrees of freedom with only a finite - dimensional hilbert space per link . in contrast to the wilson formulation , qlms resemble a quantum rather than a classical statistical mechanics problem . the relation @xmath26 = u_{xy}$ ] is then realized by a quantum link operator @xmath31 which is a raising operator for the electric flux @xmath32 associated with the link connecting neighboring lattice sites @xmath3 and @xmath7 . a local @xmath33 algebra is generated by a quantum spin @xmath34 with just @xmath35 states per link @xcite . we will consider quantum links with @xmath36 or 1 . in the classical limit @xmath37 qlms reduce to the hamiltonian formulation @xcite of wilson s lgt . a ) correlated hop of a fermion assisted by @xmath38 consistent with gauss law in a qlm with spin @xmath39 . b ) realization of the process in a ) with bosonic and fermionic atoms in an optical super - lattice ( see text ) . c ) breaking of a string connecting a static @xmath40 pair : from an unbroken string ( top ) , via fermion hopping ( middle ) , to two mesons separated by vacuum ( bottom ) . d ) from a parity - invariant staggered flux state ( top ) , via fermion hopping ( middle ) , to the vacuum with spontaneous parity breaking . ] _ the implementation of quantum link models _ in ultra - cold matter requires the realization of a gauge invariant hamiltonian accompanied by the corresponding gauss law . here , we present a general procedure to obtain @xmath0 qlms including both gauge and matter fields . to illustrate our method , we focus on a simple example , a 1d @xmath0 qlm coupled to so - called staggered fermions with the hamiltonian @xmath41 \nonumber \\ & + & m \sum_x ( -1)^x \psi_x^\dagger \psi_x + \frac{g^2}{2 } \sum_x e_{x , x+1}^2.\end{aligned}\ ] ] here @xmath42 is the hopping parameter ( see fig . [ cartoon]a ) , @xmath43 is the fermion mass , and @xmath44 is the gauge coupling . in this case , the gauge generator is given by @xmath45 $ ] . staggered fermions are analogous to spinless fermions at half - filling in condensed matter physics . the corresponding vacuum represents a filled dirac sea of negative energy states . for @xmath39 , @xmath46 , and @xmath47 the vacuum state has @xmath48 and @xmath49 $ ] . the corresponding vacuum energy of a system with @xmath50 sites is @xmath51 . the above hamiltonian resembles the schwinger model @xcite . for @xmath39 it shares the non - perturbative phenomenon of string breaking by dynamical @xmath52 pair creation with qcd @xcite . an external static quark - anti - quark pair @xmath40 ( with the gauss law appropriately taken into account ) is connected by a confining electric flux string ( fig . [ cartoon]c , top ) , which manifests itself by a large value of the electric flux . for @xmath46 , the energy of this state is @xmath53 , and the flux is given by @xmath54 . at sufficiently large @xmath50 , the string s potential energy is converted into kinetic energy by fermion hopping , which amounts to the creation of a dynamical quark - anti - quark pair @xmath52 . in this process , which is known as string breaking , an external static anti - quark @xmath55 pairs up with a dynamical quark to form a @xmath56 meson . for @xmath46 , the resulting two - meson state of fig . [ cartoon]c ( bottom ) has an energy @xmath57 and a small flux @xmath58 . the energy difference @xmath59 determines the length @xmath60 at which the string breaks . another non - perturbative process of interest in particle physics is the real - time evolution after a quench . in particular , the quark - gluon plasma created in a heavy - ion collision quickly returns to the ordinary hadronic vacuum . this is accompanied by the spontaneous breakdown of the quark s chiral symmetry . the dynamics after a quench can be quantum simulated by using the @xmath36 representation for the electric flux ( which mimics the schwinger model at vacuum angle @xmath61 @xcite ) . in that case , like chiral symmetry in qcd , for @xmath47 parity is spontaneously broken , at least for small @xmath42 , for more details see @xcite . a quenched parity - invariant staggered flux state , which evolves into the true vacuum with spontaneous parity breaking , is schematically illustrated in fig . [ cartoon]d . in this case , the electric flux represents an order parameter for spontaneous parity breaking , which is expected to perform coherent oscillations . this is similar to the time evolution after a quench starting from a disoriented chiral condensate in qcd @xcite . the realization of an atomic lgt simulator requires : _ ( i ) _ the identification of physical degrees of freedom to represent fermionic particles and bosonic quantum link variables ; _ ( ii ) _ to impose the gauss law in order to remove the gauge variant states ; and _ ( iii ) _ to design the desired dynamics in the gauge invariant subspace . below we develop a rather general atomic toolbox to implement @xmath0 lattice gauge models coupled to matter fields based on mixtures of cold fermionic and bosonic atoms in optical lattices . within this toolbox , we consider two different microscopic realizations in terms of hubbard models , model i and ii . below we present in some detail the conceptually simpler model i ( see fig . [ implement ] ) , which assumes two - component bosons representing gauge fields . model ii , discussed in the @xcite , assumes one component bosons with magnetic or electric dipolar interactions ; it offers better scalability and experimental feasibility . our concepts generalize immediately to experiments in 2d and 3d , and to fermions with spin @xcite . schematic view of the optical super - lattices for one fermionic and two bosonic species 1 and 2 ( model i ) . a ) species 1 can hop between an even site @xmath3 and the odd site @xmath62 , while species 2 can hop between @xmath3 and @xmath63 . b ) illustration of various contributions to the hamiltonian . fermions and two - component bosons have on - site repulsions @xmath64 , while bosons of the same species have @xmath65 . the offsets of the bosonic and fermionic super - lattices are @xmath66 and @xmath67 , respectively . if the fermion hops to the left , it picks up the energy offset @xmath68 from a boson of species 2 which simultaneously tunnels to the right.,scaledwidth=34.0% ] _ ( i ) _ the spin @xmath69 representing the quantum link can be realized with a fixed number @xmath70 of bosonic atoms in a double well potential with tunnel coupling ( fig . [ cartoon]b ) . an optical super - lattice @xcite ( fig . [ implement ] ) provides an array of double wells with different depths , and a mott insulator phase of bosons allows loading with the desired number of atoms @xmath71 . for two neighboring sites @xmath3 and @xmath63 , with @xmath72 and @xmath73 denoting the boson destruction operators in the corresponding wells , we define a schwinger representation for the quantum link @xmath74 the electric flux is related to the population difference of the two sites . here the bosonic species index @xmath75 distinguishes between links originating from even and odd sites @xmath3 . eq . ( [ bosons ] ) requires that each boson can tunnel only to one specific neighboring site , based on a term @xmath76 the number of bosonic atoms is conserved locally on each link . in @xcite we discuss model ii with just a single bosonic species , by encoding @xmath77 in the geometric location of the bosons to the left or to the right of the site @xmath3 . we now also add spinless fermionic atoms at half - filling to our super - lattice setup , which can hop between neighboring sites based on the term @xmath78 . _ ( ii ) _ gauss law : using @xmath79 , the gauge generator reduces to @xmath80.\ ] ] here @xmath81 counts the atoms of type @xmath82 . up to an @xmath3-dependent constant , @xmath83 thus counts the total number of atoms at the site @xmath3 . to impose the gauss law , we will consider interaction terms which can be rewritten in the form @xmath84 as the dominant term in the hamiltonian , so that all gauge variant states are removed from the low - energy sector . this is reminiscent of the repulsive hubbard model for a mott insulator @xcite . in this sense , the gauge invariant states ( which obey @xmath85 $ ] ) can be viewed as `` super - mott '' states . _ ( iii ) _ it is well known that , for large on - site repulsion , the hubbard model reduces to the @xmath42-@xmath86 model @xcite . we now induce the dynamics of a @xmath0 qlm in a similar manner , by considering the 1d microscopic hamiltonian @xmath87 . up to an additive constant , it can be expressed as @xmath88 the last two terms describe repulsive on - site interactions as well as super - lattice offsets , and form the basic building block for the gauss constraint @xmath89 . the various contributions to the hamiltonian are illustrated in fig . [ implement]b . the qlm of eq . ( [ hamilton ] ) with @xmath90 emerges in second order perturbation theory , if one tunes the parameters to the values listed in fig . [ implement]b . the offsets @xmath91 give rise to an alternating super - lattice for both the fermions and the bosons . in analogy to super - exchange interactions @xcite , energy conservation enforces a _ correlated hop _ of the fermion with the spin - flip on the link , thus realizing the term @xmath92 . this is the key ingredient for the coupling of fermions and quantum links . additionally , a gauge invariant term @xmath93 $ ] is also generated @xcite . the reduction of the microscopic model of eq . ( [ h_micro ] ) to the qlm of eq . ( [ hamilton ] ) has been verified both at the few- and many - body level , is schematically illustrated in fig . [ results]a - b and extensively discussed in @xcite . _ a ) _ flux configuration in the ground state of eq.[h_micro ] compared to the qlm for @xmath36 obtained by exact diagonalization of an @xmath94 site system . the parameters of the qlm ( in units of @xmath95 ) are @xmath96 ( see @xcite ) , and @xmath97 ( squares , crosses , and circles ) . the corresponding microscopic parameters are @xmath98 and @xmath99 ( dashed - dotted , dashed , and solid lines ) . _ b ) _ accuracy of the effective gauge invariance parameter @xmath100 in the microscopic realization as a function of @xmath101._c , d ) _ real - time evolution of the total electric flux @xmath102 obtained by exact diagonalization of the qlm with @xmath103 . _ c ) _ for @xmath39 ( solid line ) string breaking is illustrated , starting from the initial state at the top of fig . [ cartoon]c , and approaching the corresponding vacuum expectation value ( dashed - dotted line ) of @xmath104 ( @xmath105 , @xmath106 , @xmath107 ; critical breaking length @xmath108 when @xmath109 ) . _ d ) _ for @xmath36 we show the evolution after a quench , starting from the initial state at the top of fig . [ cartoon]d . the flux order parameter performs coherent oscillations whose period and strength strongly depends on @xmath43 ( @xmath110 for dashed ( thick ) line , @xmath111).,scaledwidth=42.0% ] we have performed exact diagonalizations on small system sizes to quantitatively show the physical phenomena of string breaking and the dynamics after a quench which can be observed in an experiment . the main results are presented in fig . [ results]c - d . for @xmath39 , we evolve a _ string state _ initially prepared as in fig . [ cartoon]c under hamiltonian parameters such that the separation between charge and anti - charge is larger than the characteristic scale for string breaking @xmath60 . indeed , the large negative electric flux initially stored in the string quickly approaches its vacuum value , illustrating the string breaking mechanism . for @xmath36 , fig . [ results]d also shows the time evolution after a quench , starting from the parity - invariant state at the top of fig . [ cartoon]d . in fact , the electric flux , which is an order parameter for spontaneous parity breaking , displays coherent oscillations , reminiscent of a disoriented chiral condensate in qcd @xcite . a general experimental implementation , which will require three basic steps ( preparation of an initial gauge invariant state , evolution via quantum link dynamics , and measurement of relevant physical observables ) , is discussed in the supplementary materia@xcite . in the present work , we have proposed a quantum simulator of lattice gauge theories , where bosonic gauge fields are coupled to fermionic matter , allowing demonstration experiments for phenomena such as time - dependent string breaking and the dynamics after a quench . while the basic elements behind our model have been demonstrated individually in the laboratory , the combination of these tools and the extension to higher dimensions remain a challenge to be tackled in future generations of optical lattice experiments . while building a qcd quantum simulator to address questions related to non - zero baryon density and real - time evolution remains a long term goal , we see no fundamental obstacles on the atomic physics side , but rather a long list of challenges such as incorporation of multi - component quark fields and non - abelian plaquette terms in higher dimensions . a realistic pathway will be the investigation of increasingly complex ( quantum link ) models in an interplay between theory and experiment , with the short term goals of extending the present study to higher dimensions and in particular non - abelian gauge field models . _ acknowledgment : _ we thank d. b. kaplan , m. lewenstein , b. pasquiou , f. schreck , and m. zaccanti for discussions . pz and md thank the joint quantum institute for hospitality . work at bern is supported by the schweizerischer nationalfonds . work at innsbruck is supported by the integrated project aqute , the austrian science fund through sfb f40 foqus , and by the darpa ole program . mm is supported by quitemad s2009-esp-1594 , picc : fp7 2007 - 2013 ( grant nr . 249958 ) and micinn grant fis2009 - 10061 . authors are listed in alphabetical order . _ note added : _ while completing the present work , we became aware of two preprints @xcite on atomic quantum simulation of @xmath0 gauge theories ( without coupling to fermions ) .
minimal surfaces in the euclidean space @xmath4 are fundamental forms in nature , mathematics and physics in particular @xcite . their elegant shape is due to the local minimum area property which yields zero mean curvature everywhere . one special class of surfaces arises by experimenting with the soap bubbles framed between two or more jordan curves @xcite . for example , the minimal surface joining two coaxial circles in parallel planes has the shape of a catenoid , a configuration that is extremal for numerous problems @xcite . theorem [ mods ] in the present paper shows that catenoid has also the largest conformal modulus among minimal graphs over a given annulus . an open doubly - connected surface @xmath5 in @xmath4 is a conformal image of either * punctured complex plane @xmath6 * punctured disk @xmath7 * or an annulus @xmath8 , @xmath9 in this latter case the _ conformal modulus _ of @xmath10 is defined by @xmath11 note that the ratio @xmath12 is independent of the choice of the conformal mapping , say @xmath13 . this is immediate from the classical result of schottky @xcite , 1877 . let us first establish the notation of two circular annuli in the complex plane that will remain standard throughout this paper @xmath14 + a theorem of schottky asserts that : [ schottky ] there exists a conformal homeomorphism @xmath15 if and only if the annuli have the same modulus ; that is , @xmath16 moreover , up to the rotation of the annuli , every such map takes the form + @xmath17 we shall later devote section [ sec3.1 ] to a fresh proof of schottky s theorem . it is this proof that reveals the novelty of our approach and , to some extent , leads to computational advances needed for harmonic mappings of annuli . the coordinate functions @xmath18 in the conformal representation of a surface , called _ isothermal parameters _ , satisfy the cauchy - riemann system : @xmath19 throughout this paper we take advantage of the complex variable @xmath20 and the two wirtinger differential operators @xmath21 in this notation the cauchy - riemann system takes the form of one complex equation @xmath22 now , the surface is minimal if and only if the isothermal parameters are harmonic or , equivalently , the complex vector field @xmath23 is holomorphic ; that is , @xmath24 let us factor the ambience of the surface into the complex plane and the real line @xmath25 . thus @xmath26 , where @xmath27 is a complex harmonic map . a simple direct computation shows that @xmath28 hence , all zeros of the holomorphic function @xmath29 must have even order . this allows us to determine the third isothermal parameter in terms of @xmath30 @xmath31 it is evident that every complex harmonic map @xmath32 , for which @xmath29 admits a continuous branch of square root , can be lifted to the isothermal parameters of a minimal surface . the surface is flat ; that is , @xmath33 , if and only if @xmath30 is holomorphic ( orientation preserving ) or antiholomorphic ( orientation reversing ) . let us consider now the case in which @xmath34 is a minimal graph over an annulus @xmath35 . thus @xmath36 is a harmonic homeomorphism . without losing any generality we can assume that @xmath30 preserves the orientation . this yields an elliptic beltrami type equation for @xmath37 @xmath38 where @xmath39 exists as a single valued analytic branch in @xmath40 . our basic example of this is an upper slab of a catenoid furnished by the parameters : @xmath41 this is a minimal graph over the annulus @xmath42 @xmath43 @xmath44 note that the width of the slab @xmath45 , distance from the base circle to the top one , equals its conformal modulus , @xmath46 suppose we are now given two frames ( jordan curves ) to be used to create a soap bubble of a doubly connected minimal surface . generally speaking , when these frames are moved farther apart , the conformal modulus of a surface increases and there is a moment when the bubble brakes down . the critical upper bound of the conformal modulus for which the minimal graph exists is the essence of the nitsche conjecture , which is now one of our results . [ mods ] let @xmath10 be a minimal graph over the annulus @xmath47 , then @xmath48 equality occurs if and only if @xmath49 is the upper slab of a catenoid . in 1962 j.c.c . nitsche announced , in a short article @xcite , that the existence of a harmonic homeomorphism @xmath50 , whether or not it comes from a minimal graph , yields a lower bound on @xmath51 in terms of @xmath52 . he conjectured that the necessary and sufficient condition for such a mapping to exist is the following inequality , now known as the _ nitsche bound _ @xmath53 subsequently , his conjecture appeared in monographs @xcite , @xcite , ( * ? ? ? 21.3.2 ) and surveys @xcite . various lower bounds for @xmath54 have been obtained by lyzzaik @xcite , weitsman @xcite , kalaj @xcite , and by nitsche himself ( see @xcite ) . worth noting is lyzzaik s bound in @xcite that exhibits linear growth of @xmath55 as the ratio @xmath56 approaches infinity . however , none of these results came close to . here we prove this inequality , in somewhat greater generality , which obviously implies theorem [ mods ] . one should note in advance that whenever @xmath50 is a homeomorphism , the function @xmath57 extends continuously up to the boundary of @xmath58 . moreover , there are two possibilities ; either @xmath59 or the other way round @xmath60 in the former case , we say that @xmath30 is consistent with the order of boundary components . we also have two additional possibilities depending on whether @xmath30 preserves or reverses the orientation . accordingly , there are four homotopy classes of homeomorphisms between annuli . without loss of generality we shall confine ourselves to studying the class @xmath61 the following notation will represent harmonic mappings in this class , @xmath62 our preeminent result , though not the most general , is : [ th1](existence of harmonic maps ) the annuli @xmath63 and @xmath47 admit a harmonic homeomorphism @xmath64 if and only if @xmath3 when equality occurs , then every @xmath65 takes the form @xmath66 various geometric and analytic properties of harmonic mappings are discussed , e.g. , in the books @xcite and articles @xcite . there are notable recent studies @xcite that also concluded with the nitsche bound in , though from a slightly different framework . this is the framework of the dirichlet energy @xmath67= \iint_{\mathbb{a}}{\lvertdh\rvert}^2 = 2 \iint_{\mathbb{a}}\left({\lverth_z\rvert}^2 + { \lverth_{\bar z}\rvert}^2\right)\ ] ] in general , minimizing the energy among homeomorphisms need not lead to the laplace equation . the reason is that passing to the weak limit of a minimizing sequence of homeomorphisms in @xmath68 may result in a noninjective mapping . harmonicity is lost exactly at the branch points near which the extremal mapping fails to be injective . outside the branch set the extremal mappings are indeed harmonic . this latter fact follows from rad - kneser - choquet theorem @xcite . the above recent studies of the minima of the dirichlet energy can be encapsulated as : the class @xmath69 of homeomorphisms @xmath64 admits an energy minimizer if and only if @xmath3 all minimizers are harmonic , in symbols : @xmath70 moreover , modulo rotation , they take the form @xmath71 note that the necessary condition is immediate from theorem [ th1 ] , though it was first established in @xcite by completely different idea ( the method of free lagrangians ) . indeed , beyond the nitsche bound there is no homeomorphism minimizing the energy because otherwise it would be a harmonic map , by rad - kneser - choquet theorem @xcite . it is interesting to know what happens beyond the inner boundary of @xmath72 . for simplicity , suppose that @xmath73 so in the critical configuration we have @xmath74 . the nitsche map @xmath75 , so extended to the annulus @xmath76 , becomes a harmonic double cover of @xmath77 . folding takes place along the unit circle where the jacobian vanishes . on the other hand , passing to a weak limit of a minimizing sequence of homeomorphisms @xmath78 results in squeezing the inner portion @xmath79 of the annulus @xmath80 onto the unit circle . the minimizer , found in @xcite , is unique modulo rotation and takes the form : @xmath81 the critical nitsche map @xmath75 , extended harmonically beyond the unit circle and furnished with the vertical isothermal coordinate @xmath82 for @xmath83 , gives rise to a symmetric slab of the catenoid . this result strongly suggests nonexistence of any harmonic homeomorphism @xmath78 . but _ priori _ , there might exist harmonic homeomorphisms which do not minimize the energy . to show that this is not the case we found that the integral means @xmath84 over the circles @xmath85 are better suited than the energy integrals . it should be observed that harmonicity of a function @xmath86 is invariant under conformal change of the @xmath87-variable . therefore , the nitsche bound remains valid for harmonic homeomorphisms defined on any doubly connected domain whose conformal modulus coincides with that of @xmath58 . it is therefore of interest to look at the role of the boundary curves in the target annulus as well . although we need the inner boundary of the target to be a circle , the circular shape of the outer boundary turns out to be inessential , there remains a substitute of the nitsche bound in terms of the integral means @xmath88 . in our generalized form of the nitsche bound the target will be a half circular annulus ; that is , a doubly connected domain , denoted by @xmath89 , whose inner boundary is a circle @xmath90 . we do not specify the outer boundary of @xmath89 as it can be arbitrary . let @xmath91 denote the class of orientation preserving harmonic homeomorphisms @xmath92 which preserve the order of the boundary components . in particular , @xmath93 . [ th2 ] ( generalized nitsche bound ) for every @xmath94 we have @xmath95^\frac{1}{2 } \ge \frac{1}{2 } \left(\frac{\sigma}{r}+ \frac{r}{\sigma}\right)r_\ast\ ; , \quad r \le \sigma < r\ ] ] if equality occurs at some radius @xmath96 , then it holds for every @xmath97 and @xmath98 for some @xmath99 . such a more general statement not only strengthens the nitsche conjecture , but also is the key to the proof . theorem [ th2 ] should be viewed as a sharp lower estimate for the growth of integral means of harmonic mappings under certain initial constraints . these constraints concern topological behavior of @xmath30 near the inner boundary @xmath100 rather than its boundary values . it causes no loss of generality to assume that @xmath30 is smooth up to the inner boundary ; we shall reduce ourselves to smooth mappings via lemma [ lemappr ] . what one really needs for smooth mapping @xmath101 is its harmonicity in @xmath102 and the following three initial conditions : + a. [ in1 ] @xmath103 is a homeomorphism homotopic to the identity ; b. [ in2 ] @xmath104 at @xmath105 ; c. [ in3 ] @xmath106 . in fact these three conditions yield inequality . even more , it turns out that condition is redundant when @xmath107 . the proof of this case of theorem [ th2 ] is based upon the ideas from our earlier paper @xcite . however , example [ ex ] demonstrates that constraints are insufficient to deduce inequality when @xmath52 is large . a new approach is required in which we make use of the jacobian bound . the key ingredient is the following result concerning harmonic self - mappings of the unit disk , which is interesting in its own right . [ th3 ] let a harmonic homeomorphism @xmath108 be @xmath109-smooth in the closed unit disk @xmath110 . then @xmath111 the first inequality is strict unless @xmath112 is an isometry . we end this overview by stating an analogue of nitsche s conjecture for noninjective harmonic maps , which has geometric interpretation for more general doubly connected minimal surfaces , not necessarily graphs . let @xmath113 be a harmonic map ( not necessarily injective ) with nonzero winding number ; that is , @xmath114 then @xmath115 equality occurs for the double cover nitsche map : @xmath116 the solution of the nitsche conjecture and the methods presented here open new questions yet to be answered . the annuli @xmath58 and @xmath117 as well as the half circular annulus @xmath89 will henceforth be rescaled so their inner boundaries are the unit circles , still denoted by @xmath118 whenever the target of @xmath119 need not be specified we abbreviate this notation to @xmath120 . thus , from now on @xmath121 the complex derivatives @xmath122 and @xmath123 in polar coordinates are : @xmath124 hence one finds the hilbert - schmidt norm of the differential matrix @xmath125 the laplacian @xmath126 and the jacobian determinant @xmath127 we shall work with a number of circular means to which the following commutation rule will apply @xmath128 the main objects are the circular means of @xmath129 for mappings @xmath120 , @xmath130 we shall also discuss so - called mean radius of @xmath131 defined by the rule : @xmath132 here , as usual , dot over @xmath133 and @xmath134 stands for the @xmath135-derivative . thus , we aim to prove the inequality @xmath136 next observe that away from the outer boundary of @xmath137 the energy of @xmath120 is always finite . the proof is an exercise with green s formula ; first we have for @xmath138 @xmath139 hence we infer that @xmath140 is strictly increasing and then @xmath141 now the following limit is easily seen to exist for @xmath142 : @xmath143 the so defined quantity @xmath144 is called the _ initial speed _ , it represents the initial data of _ harmonic evolution of circles _ under the mapping @xmath101 . we just found that the initial speed is always nonnegative for @xmath145 ; later in proposition [ schii ] we shall see that it equals 1 for conformal mappings . the first derivative @xmath146 , now well defined in the left closed interval @xmath147 , is continuous . finally , formula remains valid for @xmath148 and reads as , @xmath149 this is what we wished to establish . + other circular means require assuming that @xmath101 has continuous normal and tangential derivatives at the unit circle . this technical obstacle will be overcome by the following approximation argument . [ lemappr](weak approximation ) given @xmath150 there exist annuli @xmath151 and harmonic mappings @xmath152 @xmath153 such that * each @xmath154 is @xmath155smooth up to the inner boundary ( the unit circle ) * each @xmath156 is a homeomorphism homotopic to the identity * @xmath157 * @xmath158[wlim ] because of harmonicity the sequence @xmath159 and its all derivatives converge uniformly on every circle @xmath160 , @xmath161 . choose and fix a sequence of radii @xmath162 close enough to 1 to have all circles @xmath163 contained in @xmath164 since @xmath165 is a real - analytic diffeomorphism , the level sets @xmath166 are real - analytic jordan curves in @xmath137 . they separate the boundary components . consider the doubly connected domain @xmath167 whose inner boundary is @xmath168 and outer boundary is that of @xmath169 . there is uniquely determined circular annulus @xmath170 which admits a conformal map @xmath171 such a map is @xmath172 - smooth and injective up to the closer of @xmath173 . actually , it extends conformally beyond the boundary of @xmath173 . relevant details can be found in @xcite and @xcite . we may assume that @xmath174 preserves the order of boundary components , that is @xmath175 for @xmath176 . the outer radius of @xmath177 is less than @xmath178 . in fact , we have @xmath179 passing to a limit with a suitable subsequence of @xmath180 results in a conformal map of @xmath169 onto itself . furthermore , composing @xmath174 with a rotation of the @xmath87-variable ( appropriately chosen for each function @xmath174 ) , we manage that @xmath181 , for every @xmath182 . actually the convergence is uniform together with all derivatives on every compact subset of the annulus @xmath169 . we are now ready to define the harmonic mappings in question @xmath183 to complete the proof of the lemma , we need only establish the existence of the weak limit in . consider an annulus @xmath184 with outer boundary @xmath185 . it is evident , by topology , that @xmath186 lies inside the jordan curve @xmath187 . even more , there is an annulus @xmath188 , @xmath189 such that @xmath190 for all @xmath191 . this is because @xmath192 converge uniformly on @xmath193 to the identity map . thus , again by topology , @xmath194 . now , in view of , the dirichlet energy of @xmath195 on @xmath196 is free from @xmath197 . indeed , by a conformal change of variables , we see that @xmath198 the proof of the lemma is completed by noting that @xmath159 converges weakly to @xmath30 in the sobolev space @xmath199 . here is the first taste and sample of the utility of circular averages . we use them for conformal mappings to prove theorem [ schottky ] . although the result is classical and shorter proofs can be given ( @xcite or @xcite ) , this proof and the comments following it indicate in some detail the underlying strategy to be used for harmonic mappings . [ schii ] the initial speed of conformal evolution of circles always equals 1 . that is , if a homeomorphism @xmath200 is conformal , then @xmath201 moreover , for each @xmath202 the circular means of @xmath129 satisfy : @xmath203 equality occurs , somewhere at @xmath204 , if and only if @xmath205 . the classical schottky theorem follows by imposing the outer boundary condition , @xmath206 for @xmath207 , to infer that @xmath208 . this can be reversed via consideration of the inverse conformal map , ascertaining theorem [ schottky ] . one compelling motive for studying the averages such as @xmath209 is to take advantage of convexity properties of holomorphic ( later harmonic ) functions . let us begin with the laurent expansion @xmath210 the system @xmath211 is orthogonal on every circle @xmath212 . thus , @xmath213 all that matters is to find a certain second order differential operator @xmath214 acting on @xmath215 that fits into the following scenario : @xmath216 \geqslant 0\,,\ ; \textit{with equality if and only if } \;\;\;h(z ) = \lambda z\,,\;\;\|\lambda\| = 1\ , .\ ] ] much of the essential properties of @xmath209 are contained in such inequality when combined with the topological behavior of @xmath101 near the inner boundary of @xmath169 . the interested reader may wish to consult @xcite for a fuller discussion of this topic . here , for conformal case , direct computation shows that @xmath217 \,:=\ ; \frac{1}{\rho } \,\frac{\textrm{d}}{\textrm{d}\rho}\left [ \rho^3\,\frac{\textrm{d}}{\textrm{d}\rho}\left(\frac{u}{\rho^2 } \right ) \right ] \ ; = \;4\sum_{n\in\mathbb z } n(n-1 ) \| a_n\|^2 \rho^{2n-2 } \geqslant 0\ ] ] hence @xmath218 now the cauchy - riemann equations in polar coordinates @xmath219 and topology ( winding number along the inner boundary ) come into play . but first note that @xmath101 is @xmath220-regular up to the inner boundary of @xmath137 . even more , since @xmath221 on @xmath222 , it extends as a conformal map slightly inside the unit circle . let us reveal in advance that at this point of the investigation of harmonic mappings we shall justify the assumption of @xmath220 - regularity by lemma [ lemappr ] . this makes it legitimate to perform the following computation , @xmath223 we just proved that @xmath224 . returning to we infer that the function @xmath225 is nondecreasing , and hence @xmath226 . we actually have slightly stronger inequality @xmath227 , which yields @xmath228 . then it follows again from that @xmath229 . finally , if equality occurs in one of for some @xmath230 , we infer from that @xmath231 , except for @xmath232 and @xmath233 , which gives a linear function @xmath234 . since @xmath221 on @xmath235 we conclude that @xmath236 , as desired . let us emphasize the principal features of this proof and indicate possible generalizations . first of all , the cauchy - riemann system was fundamental in the above proof . for , it let us replace the radial derivative @xmath237 by its angular derivative @xmath238 ; whereas for harmonic mappings our strategy must rely on the second order equation : @xmath239 the basic complex harmonic functions in the annulus @xmath58 are the integer powers @xmath240 , @xmath241 , @xmath242 , and the logarithm @xmath243 . if we combine these functions suitably in pairs , we obtain an analogue of the laurent expansion for harmonic functions @xmath244 where @xmath245 for @xmath246 and @xmath247 . these components are orthogonal on every circle @xmath248 , @xmath249 . we refer to @xmath250 and @xmath251 as the fourier coefficients of @xmath101 . the leading terms , corresponding to @xmath252 , will take control of the geometric behavior of the mappings @xmath253 . we conveniently scale them to become the identity map on the unit circle . so obtained functions , called the _ nitsche maps _ , form a one parameter family : @xmath254 it is worth noting that , in the above sum , the terms have vanishing neumann and dirichlet boundary values at the inner circle , respectively . moreover , the parameter @xmath255 represents the initial speed of the evolution of circles , @xmath256 the outer radius of the image of any annulus @xmath257 complies with the conjectural nitsche bound : @xmath258 of particular significance is the _ critical nitsche map _ with zero initial speed , @xmath259 it has been proven in @xcite that @xmath260 is exactly the mapping , unique up to rotation , that minimizes the dirichlet energy subject to all homeomorphisms @xmath261 between annuli @xmath262 and @xmath263 , with @xmath264 given by formula . for another extreme property , we consider a subclass of homeomorphisms @xmath265 ( harmonic evolutions of circles ) with initial speed @xmath266 and having zero first moment on the inner boundary ; precisely , @xmath267 in the expansion . such is the mapping @xmath268 . in this class of harmonic mappings the following generalization of the nitsche bound has been established in @xcite : @xmath269 equality occurs for @xmath270 , uniquely up to a rotation . the idea of the proof is very much the same as that for proposition [ schii ] . however , one has to remodel the operator @xmath271 , say @xmath272 in order to obtain @xmath273 = 0\,$ ] for @xmath274 . the precondition @xmath275 seems to be redundant regardless of the initial speed . in fact this has already been removed for @xmath276 sufficiently small , depending on the initial speed @xmath255 , by more advanced estimates in @xcite . unexpectedly , the redundancy of this seemingly minor precondition turned out to be the key difficulty in completing the proof of the nitsche conjecture . to each nitsche map @xmath278 , there corresponds a differential operator which tells us something about convexity properties of integral means of harmonic maps , with @xmath260 as extreme case . we have already seen such an operator for @xmath279 in the proof of proposition [ schii ] . but our interest here is in the critical nitsche map @xmath280 of initial speed @xmath281 . associated with @xmath282 is its operator @xmath283 , see @xcite , which we introduce here in three different ways as follows : @xmath284\,&:= \;\ddot{u}+ \frac{3-\rho^2}{\rho ( \rho^2 + 1 ) } \dot{u } - \frac{8}{(\rho^2 + 1)^2}u \label{l1 } \\ & = \frac{\rho^2 + 1}{\rho^3 } \frac{{\textnormal d}}{{\textnormal d}\rho } \left [ \rho^3 \frac{{\textnormal d}}{{\textnormal d}\rho } \left(\frac{u}{\rho^2 + 1 } \right ) \right ] \label{l2}\\ & = \dashint_{{\mathbb{t}}_\rho } \left [ 2 { \lverth_\rho\rvert}^2 + \frac{2}{\rho^2 } { \lverth_\theta\rvert}^2 - 2 \frac{\rho^2 - 1}{\rho ( \rho^2 + 1 ) } { \lverth^2\rvert}_\rho - \frac{8}{(\rho^2 + 1)^2 } { \lverth\rvert}^2 \right ] \label{l3}\end{aligned}\ ] ] the interested reader may wish to learn the following principal feature of this operator ; all the orthogonal components @xmath285 with @xmath286 give rise to nonnegative values of @xmath287\,$ ] , whereas @xmath287\equiv 0\,$ ] for @xmath282 . we shall not appeal to these properties explicitly , though they suggest us how to approach sharp estimates . but for the completeness of our arguments we still owe to the reader a verification of formula ; the one in is clear . it is interesting to note that contains no second derivatives of @xmath101 . let us begin with the integral means and their derivatives @xmath288 at this and only this stage we use the laplace equation , which yields : @xmath289 - \frac{2}{\rho^2 } \operatorname{re}\dashint_{{\mathbb{t}}_\rho } \bar h \ , h_{\theta \theta}\ ] ] the latter term , upon integration by parts along the circle @xmath290 , becomes @xmath291 . in this way we represent @xmath292 by using only first derivatives of @xmath30 , @xmath293\ ] ] substituting these formulas for @xmath133 , @xmath294 and @xmath295 into we arrive at . the generalized nitsche bound in is a straightforward consequence of a rather sophisticated identity for harmonic functions . we created this identity for the convenience of the reader in order to capture the essentials of the proof of theorem [ th2 ] when @xmath296 . [ proplesse ] let @xmath30 be a complex harmonic function in the annulus @xmath297 , @xmath298 , that is @xmath109-smooth up to the boundary . then @xmath299 \cdot \left\| \frac{\rho h_\rho -i h_\theta}{1+\rho^2 } - \frac{2 \rho^2\ , h}{(1+\rho^2)^2 } \right\|^2 \\ & \hskip0.4 cm + \frac{1}{\pi } \iint_{{\mathbb{a } } } \left[(r^2-\rho^2 ) - ( r^2 - 1 ) \log \frac{r}{\rho } \right ] \cdot \left\| \frac{\rho h_\rho + i h_\theta}{1+\rho^2 } + \frac{2 \ , h}{(1+\rho^2)^2 } \right\|^2 \end{split}\ ] ] the derivation of this identity is postponed until the end of this section . @xmath300 + no restriction for @xmath301 is needed in proposition [ proplesse ] . nevertheless , for the proof of the nitsche bound we shall have to restrict the outer radius to the interval @xmath302 in order to ensure that double integrals of are nonnegative . the first integrand is certainly positive for every @xmath303 . however , the second integral needs the restriction @xmath304 . we have @xmath305 to see this , note that the expression in the left hand side represents a concave function in @xmath306 with nonnegative values at the endpoints of the interval [ 1 , r ] . for the proof of we take @xmath307 , then choose and fix a radius @xmath308 between @xmath309 and @xmath178 , @xmath310 . consider the annulus @xmath311 . given @xmath312 we appeal to lemma [ lemappr ] to construct harmonic homeomorphisms @xmath313 that are @xmath109-smooth up to the inner boundary of @xmath314 and converge to @xmath30 weakly in @xmath315 . before proceeding to the identity we note three particulars concerning integral means over the unit circle : * @xmath316 , because @xmath317 on @xmath318 + * @xmath319 this is because homeomorphisms @xmath195 take circles @xmath290 , @xmath320 into jordan curves inside which there lies the unit disk . precisely , we have at every point of the unit circle : @xmath321 we also have the identity * @xmath322 which follows from the computation of the winding number of @xmath195 around @xmath318 , @xmath323 on substituting ( i - iii ) into , with @xmath178 replaced by @xmath309 , we obtain @xmath324 \cdot \left\| \frac{\rho h^k_\rho -i h^k_\theta}{1+\rho^2 } - \frac{2 \rho^2\ , h^k}{(1+\rho^2)^2 } \right\|^2 \\ & \hskip0.4 cm + \frac{1}{\pi } \iint_{a(1 , \sigma ) } \left[(\sigma^2-\rho^2 ) - ( \sigma^2 - 1 ) \log \frac{\sigma}{\rho } \right ] \cdot \left\| \frac{\rho h^k_\rho + i h^k_\theta}{1+\rho^2 } + \frac{2 \ , h^k}{(1+\rho^2)^2 } \right\|^2 \end{split}\ ] ] we are going to pass to the limit as @xmath325 . note that @xmath326 uniformly on @xmath327 and weakly in @xmath315 . passing to the limit in the double integrals results in the desirable estimate from below , due to lower semicontinuity of the double integrals , in which @xmath328 and @xmath329 converge weakly in @xmath330 . @xmath331 \cdot \left\| \frac{\rho h_\rho -i h_\theta}{1+\rho^2 } - \frac{2 \rho^2\ , h}{(1+\rho^2)^2 } \right\|^2 \\ & \hskip0.4 cm + \frac{1}{\pi } \iint\limits_{a(1 , \sigma ) } \left[(\sigma^2-\rho^2 ) - ( \sigma^2 - 1 ) \log \frac{\sigma}{\rho } \right ] \cdot \left\| \frac{\rho h_\rho + i h_\theta}{1+\rho^2 } + \frac{2 \ , h}{(1+\rho^2)^2 } \right\|^2 \end{split}\ ] ] hence @xmath332 or , equivalently @xmath333 this is the estimate we wished to obtain . for uniqueness statement in theorem [ th2 ] , we assume that equality occurs in . then we see from that @xmath30 must satisfy the following equations : @xmath334 adding and subtracting the equations we uncouple the @xmath135 and @xmath335 derivatives , @xmath336 the general solution takes the form @xmath337 where @xmath338 is any complex number . since @xmath339 on @xmath318 we conclude that @xmath340 . the proof of theorem [ th2 ] in case @xmath302 will therefore be accomplished once we establish the identity . the identity is obtained by integrating @xmath287\,$ ] against a weight over the interval @xmath341 . we shall integrate by parts using the divergence form of @xmath287\,$ ] in . the weight must be carefully crafted in order not to produce the derivative @xmath342 as a boundary term . this term is out of control ; it can even be infinity in some cases , exactly when the energy of @xmath101 over @xmath169 is infinite . there is essentially only one weight that suits well to this conception , namely @xmath343 . since @xmath287\,$ ] vanishes for the critical nitsche mapping , it is natural to try to simplify computation by making a substitution @xmath344 in order to express @xmath345 $ ] by means of @xmath346 we compute the terms under the integral sign in @xmath347 therefore , @xmath348&=\frac{(\rho^2 + 1)^2}{2\ ; \rho^4}\ , \dashint_{c_{\rho}}\left({\lvertg_{\theta}+ig\rvert}^2-{\lvertg\rvert}^2+\rho^2{\lvertg_{\rho}\rvert}^2\right ) \\ & = \frac{(\rho^2 + 1)^2}{2\ ; \rho^2}\ \dashint_{c_{\rho}}\left[\;{\lvertg_{\rho}\rvert}^2+\rho^{-2}{\lvertg_{\theta}\rvert}^2 + 2\rho^{-2}\operatorname{im}(\bar g g_{\theta})\;\right ] \end{split}\ ] ] in this way we arrive at somewhat simpler formula @xmath349 = \frac{(\rho^2 + 1)^2}{\rho^2 } \dashint_{{\mathbb{t}}_\rho } \left [ { \lvertg_z\rvert}^2 + { \lvertg_{\bar z}\rvert}^2 + \frac{1}{\rho^2 } \operatorname{im}\left(\bar g \ , g_\theta \right ) \right]\ ] ] we shall now exploit the divergence form of @xmath350 $ ] in . multiply both sides of by the weight @xmath343 and integrate from @xmath351 to @xmath352 . @xmath353 = \int_1^r \;\frac{\rho\,(r^2-\rho^2)}{\,\rho^2 + 1\;}\;\mathcal l[u ] \;\;\;= \\ & \\&= \int_1^r \frac{(r^2-\rho^2)(\rho^2 + 1)}{\rho } \dashint_{{\mathbb{t}}_\rho } \left [ { \lvertg_z\rvert}^2 + { \lvertg_{\bar z}\rvert}^2 + \frac{1}{\rho^2 } \operatorname{im}\left(\bar g \ , g_\theta \right ) \right ] \end{split}\ ] ] integration by parts of the lefthand side leaves only boundary terms @xmath354 here we split the righthand side for the purpose of integrating only the second term by parts . to achieve this objective , we represent the factor in front of @xmath355 as derivative of a function that vanishes at @xmath352 , namely : @xmath356\ ] ] integration by parts produces additional boundary term and formula becomes @xmath357 \frac{{\textnormal d}}{{\textnormal d}\rho } \dashint_{{\mathbb{t}}_\rho } \operatorname{im}\left ( \bar g \ , g_\theta \right)\\ & + ( r^2 - 1 ) \log r \ , \dashint_{{\mathbb{t } } } \operatorname{im}\left ( \bar g \ , g_\theta \right)\ \end{split}\ ] ] with the aid of commutation rule we pass the @xmath135-differentiation inside the mean integral and then integrate by parts along the circle @xmath358 to obtain @xmath359 now , the righthand side of takes the form @xmath360 \dashint_{{\mathbb{t}}_\rho } { \lvertg_z\rvert}^2\\ & + \int_1^r \left[\frac{(r^2-\rho^2)(\rho^2 + 1)}{\rho } + \frac{(r^2-\rho^2)(\rho^2 - 1)}{\rho } - ( r^2 - 1 ) 2 \rho \log \frac{r}{\rho } \right ] \dashint_{{\mathbb{t}}_\rho } { \lvertg_{\bar z}\rvert}^2\\ & \hskip2 cm + ( r^2 - 1 ) \log r \dashint_{{\mathbb{t } } } \operatorname{im}\left ( \bar g\ , g_\theta \right ) \end{split}\ ] ] which simplifies to @xmath361 \int_{{\mathbb{t}}_\rho } { \lvertg_z\rvert}^2 \\ & + \frac{1}{\pi } \int_1^r \left [ r^2-\rho^2 - ( r^2 - 1 ) \log \frac{r}{\rho } \right ] \int_{{\mathbb{t}}_\rho } { \lvertg_{\bar z}\rvert}^2\\ & \hskip2cm+ ( r^2 - 1 ) \log r \;\dashint_{{\mathbb{t } } } \operatorname{im}\left ( \bar g\ , g_\theta \right ) \end{split}\ ] ] finally , the entire formula reads as @xmath362 { \lvertg_z\rvert}^2 \\ & + \frac{1}{\pi } \iint_{{\mathbb{a } } } \left [ ( r^2-\rho^2 ) - ( r^2 - 1 ) \log \frac{r}{\rho } \right ] { \lvertg_{\bar z}\rvert}^2 \end{split}\ ] ] to conclude with the identity it only remains to observe that @xmath363 this ends the proof of proposition [ proplesse ] . recall from section [ sec 3.3 ] the orthogonal decomposition @xmath364 of a harmonic function , where @xmath245 for @xmath246 and @xmath247 . putting this in the identity breaks it up into independent identities , one for each term @xmath365 . this could be another way of proving proposition [ proplesse ] . precisely , for each individual term @xmath365 the problem reduces to showing that so obtained quadratic form with respect to the complex variables @xmath366 is nonnegative . however , it would not make the proof simpler ; one immediately encounters the difficulty in the simplest possible case @xmath367 . try it ! nevertheless , the orthogonal decomposition and the method of quadratic forms for the fourier coefficients of @xmath101 will prove extremely useful for the case @xmath368 , see proposition [ bigrprop ] in which a counterpart of the identity can not be even formulated without appeal to the coefficients @xmath369 . let us return to the initial conditions discussed in the introduction . an example below demonstrates that , in contrast to the case @xmath107 , condition can not be omitted when @xmath52 is large . this is the underlying reason why the method of section [ thin ] could not work for arbitrary values of @xmath52 . [ ex ] fix @xmath370 and let @xmath371 be a positive number to be chosen later . define for @xmath372 , @xmath373 it is clear that @xmath30 is harmonic in @xmath374 and satisfies with @xmath375 . we represent @xmath101 as a sum of two terms which are orthogonal on every circle @xmath376 , @xmath377 , @xmath378 in this way we accomplish the following computation @xmath379 it follows that holds if @xmath371 is chosen to be sufficiently large , depending only on the parameter @xmath338 , which remains fixed . however , the generalized nitsche bound fails on circles of large radius , because @xmath380 exhibits logarithmic growth as @xmath381 . the reader may wish to note that the average jacobian determinant over the unit circle is independent of @xmath382 and is negative , @xmath383 . in view of the above example , we need to take advantage of the additional information that the averaged jacobian is nonnegative at the unit circle , as stated in condition . the condition will come into play via the following inequality . [ bigrprop ] let @xmath384 be a harmonic function that is @xmath109-smooth up to the inner boundary circle @xmath318 . denote by @xmath385 the harmonic extension of @xmath30 to the closed unit disk . then for all @xmath386 we have @xmath387\ge 0 \end{split}\ ] ] since @xmath388 , proposition [ bigrprop ] covers all values @xmath389 . + our proof of involves the fourier coefficients of @xmath30 . these are complex numbers @xmath390 , @xmath391 , that appear in the orthogonal expansion @xmath392 first observe that the harmonic extension of @xmath30 inside the unit disk is expressed by two infinite sums @xmath393 this is certainly true for mappings @xmath394 that are continuous up to the inner boundary . then the terms in can be computed using orthogonality of the powers of @xmath395 . @xmath396 upon substituting these terms to the lefthand side becomes a quadratic form @xmath397 with respect to the complex variables @xmath398 , which we aim to show to be nonnegative for every complex numbers @xmath390 , @xmath391 . more precisely , @xmath399 splits into an infinite sum of quadratic forms , each of which depends only on two complex variables , @xmath400 where @xmath401 for example , @xmath402 which is positive definite as long as @xmath403 . a key to the extremal case lies in the property of the quadratic form @xmath404 , which is positive semidefinite , @xmath405 meaning that @xmath406 yields @xmath407 . the other quadratic forms will be shown to be positive definite . [ techlem ] for every @xmath408 and @xmath409 we have @xmath410 we postpone the proof of this technical lemma until appendix 1 . the second inequality in is immediate from @xmath411 integration of the jacobian determinant over the unit disk gives the area of the image , which equals @xmath412 . to prove the first part of , it suffices to consider the case when @xmath112 is sense - preserving . at every point @xmath413 of the unit circle @xmath318 we have @xmath414 and @xmath415 . now we compute the jacobian determinant at @xmath416 as follows . @xmath417={\lvertf_\theta\rvert}\operatorname{re}(\overline { f_\rho } f)={\lvertf_\theta\rvert}\cdot{\lvertf\rvert}_{\rho}\;.\ ] ] since @xmath112 is harmonic , we have @xmath418 . green s formula yields @xmath419 thus our goal is to prove @xmath420 let us write @xmath421 , where @xmath422)$ ] is increasing and satisfies @xmath423 . by poisson s formula , for @xmath424 and @xmath425 , we have @xmath426 multiply by @xmath427 and take the real part : @xmath428\,d\phi.\ ] ] now the normal derivative @xmath429 at the unite circle can be computed as follows . @xmath430}{1-\cos(\theta-\phi)}\,d\phi.\ ] ] the lefthand side of becomes a double integral with respect to both @xmath335 and @xmath431 @xmath432}{1-\cos(\theta-\phi)}\,\left(\xi'(\theta)-1\right)\;d\theta\;d\phi\,\ ] ] we are reduced to showing that [ le5 ] suppose @xmath422)$ ] is increasing and @xmath433 periodic ; that is @xmath434 . then @xmath435}{1-\cos(\theta-\phi)}\,\left(\xi'(\theta)-1\right)\;d\theta\;d\phi\ , \;\geqslant 0\ ] ] equality occurs if and only if @xmath436 . elementary , though lengthy proof of this lemma is given in appendix 2 . since the result of section [ thin ] applies to the restriction of @xmath30 to @xmath438 , we only concern ourselves with integral means of @xmath30 over the circles of radius @xmath439 . first assume that @xmath394 is @xmath440-smooth up to the inner boundary . thus we may compute the winding number of @xmath30 around @xmath318 , @xmath441 moreover , @xmath442 and @xmath443 let @xmath444 be the continuous extension of @xmath445 that is harmonic in @xmath446 . since @xmath447 is harmonic in @xmath448 and vanishes on @xmath318 , it has an extension as a harmonic function on @xmath449 . in particular , @xmath450 . by theorem [ th3 ] we have @xmath451 substituting these inequalities to yields the desired nitsche bound @xmath452 there is no difficulty to relax the @xmath109-smoothness assumption . let us go into this in detail with the aid of lemma [ lemappr ] , in much the same way as in section [ subsec55 ] . we fix @xmath307 , then choose and fix a radius @xmath308 so that @xmath453 . consider the annulus @xmath311 . given @xmath312 we appeal to lemma [ lemappr ] to construct harmonic homeomorphisms @xmath313 that are @xmath109-smooth up to the boundary of @xmath314 and converge to @xmath30 weakly in @xmath315 . this time we have four particulars concerning integral means over the unit circle : * @xmath454 + * @xmath455 + * @xmath456 + * @xmath457 for each @xmath195 the nitsche bound holds , @xmath458 it is essential that before passing to the limit we have ignored , based on ( i)-(iv ) , the integral means of @xmath30 and their derivatives over the inner boundary of @xmath169 . now , because of harmonicity the sequence @xmath159 converges uniformly on every circle @xmath290 , @xmath459 but not necessarily for @xmath460 . passing to the limit we obtain the nitsche bound for every @xmath120 . + the uniqueness statement is somewhat delicate , it takes into account a stronger variant of , which in terms of the quadratic forms in reads as @xmath461 here @xmath462 are the associated fourier coefficients of @xmath195 ; the quadratic forms with @xmath463 , being positive definite , are omitted . we can now pass to the limit @xmath464 passage to the limit in the finite sum of quadratic forms is justified because for every fixed @xmath465 we have @xmath466 and @xmath467 . finally , for the uniqueness , suppose that @xmath468 , where @xmath469 . then yields @xmath470 for each integer @xmath465 . hence , by lemma [ techlem ] , all the coefficients @xmath471 and @xmath472 vanish except for the case @xmath473 . this leaves us with the only possible function @xmath474 for some @xmath475 , because @xmath476 on @xmath318 . the proof of theorem [ th2 ] is complete , modulo lemma [ techlem ] . the quadratic form @xmath477 in the decomposition is obtained by putting @xmath478 into the left hand side of [ trueforall ] . we first do the case of positive indices ; that is , + * case @xmath479 . * with the aid of formulas we find the coefficients : @xmath480 we need to show that @xmath481 and @xmath482 are positive and @xmath483 . ignoring the term @xmath484 in @xmath482 , we obtain the estimate @xmath485 next , estimate @xmath481 from below as follows . @xmath486 regarding @xmath487 , note that @xmath488 for all @xmath479 , and @xmath489 with @xmath490 , inequality yields @xmath491 , which together with and imply @xmath492 the latter inequality holds for all @xmath493 because @xmath494 . thus @xmath495 is positive definite . when @xmath496 , we simplify further by ignoring a negative term as follows @xmath497 in light of and the inequality @xmath498 will follow once we show that @xmath499 to this end , we use @xmath500 when @xmath501 , a direct computation shows that implies . when @xmath502 , we use the fact that @xmath503 is increasing in @xmath465 to obtain @xmath504 from which follows . * case @xmath505 . * for convenience we set @xmath506 , where @xmath507 is a positive integer . with this new notation the coefficients of @xmath508 are @xmath509 organizing in powers of @xmath135 , we find @xmath510 similarly , @xmath511 it is clear by that @xmath512 is positive , so @xmath513 with @xmath514 , inequalities yield @xmath515 when @xmath516 , we ignore the last term in and obtain @xmath517 in light of and the inequality @xmath518 will follow once we show that @xmath519 for this we return to . since @xmath520 , it follows that @xmath521 where the latter is the minimum value of the former expression in @xmath135 , attained at @xmath522 . it equals @xmath523 when @xmath524 , while for @xmath525 we have @xmath526 proving . this completes the proof of lemma [ techlem ] . the double integral in lemma [ le5 ] is the set @xmath527 \times [ 0 , 2\pi]\,$ ] . we can write @xmath528 , where @xmath529 is a @xmath109-smooth @xmath530-periodic function . observe that @xmath531 in particular , @xmath532 for all @xmath335 . the left hand side of is equal to @xmath533}{1-\cos(\theta-\phi)}\,\zeta'(\theta)\;d\theta\ , d\phi\ ] ] with the help of the trigonometric identity @xmath534 we represent as the sum @xmath535 , where @xmath536}{1-\cos(\theta-\phi)}\zeta'(\theta)\;d\theta\ , d\phi\\ b&=\iint_{\mathbb q } \frac{\sin(\theta-\phi)\sin\big[\zeta(\theta)-\zeta(\phi)\big]}{1-\cos(\theta-\phi)}\zeta'(\theta)\;d\theta\ , d\phi . \end{split}\ ] ] since @xmath537 is periodic , the integral of @xmath538 over @xmath539 is equal to @xmath540 . let us subtract this integral from @xmath541 to write it as : @xmath542}{1-\cos(\theta-\phi)}\zeta'(\theta)\;d\theta\ , d\phi.\ ] ] we divide the set @xmath543 into @xmath544 accordingly , the integral @xmath541 splits into @xmath545 and @xmath546 . the @xmath545 part can be estimated from below using the inequality @xmath547 , @xmath548}{1-\cos(\theta-\phi)}\zeta'(\theta)\;d\theta\ , d\phi \\ & \ge \iint_{\mathbb q^+ } \frac{-\cos(\theta-\phi)\big[1-\cos(\zeta(\theta)-\zeta(\phi))\big]}{1-\cos(\theta-\phi)}\;d\theta\ , d\phi \end{split}\ ] ] in @xmath546 part we perform integration by parts with respect to @xmath335 . no boundary terms appear because @xmath537 is periodic and @xmath549 vanishes on the common boundary of @xmath550 and @xmath551 . @xmath552}{1-\cos(\theta-\phi)}\zeta'(\theta)\;d\theta\ , d\phi \\ & = \iint_{\mathbb q^- } \frac{\cos(\theta-\phi)}{1-\cos(\theta-\phi)}\frac{d}{d\theta}\left\ { \zeta(\theta)-\zeta(\phi)-\sin\big[\zeta(\theta)-\zeta(\phi)\big]\right\}\;d\theta\ , d\phi \\ & = \iint_{\mathbb q^- } \frac{\sin(\theta-\phi)}{(1-\cos(\theta-\phi))^2}\left\{\zeta(\theta)-\zeta(\phi)-\sin\big[\zeta(\theta)-\zeta(\phi)\big]\right\}\;d\theta\ , d\phi . \end{split}\ ] ] we also integrate @xmath553 by parts with respect to @xmath335 @xmath554\right\ } \;d\theta\ , d\phi \\ & = \iint_{\mathbb q } \frac{1-\cos\big[\zeta(\theta)-\zeta(\phi)\big]}{1-\cos(\theta-\phi)}\;d\theta\ , d\phi \end{split}\ ] ] the latter integral splits as @xmath555 , where @xmath556 is taken over @xmath550 and @xmath557 over @xmath551 . the sum @xmath558 is estimated by combining and . @xmath559\right\}\,d\theta\,d\phi\ge 0\ ] ] finally , we write the sum @xmath560 using shorthand notation @xmath561 and @xmath562 . @xmath563 the definition of @xmath551 and inequalities imply @xmath564 we claim that the integrand in is nonnegative , that is , @xmath565 under the conditions . from this will follow by adding up and . it remains to verify . replacing the pair @xmath566 with @xmath567 if necessary , we may assume that @xmath568 . now , if @xmath569 , then @xmath570 , and follows . suppose @xmath571 . then @xmath572 is increasing with respect to @xmath573 $ ] . for a fixed @xmath574 $ ] , the minimal admissible value of @xmath575 under is @xmath576 . this leads us to consider two cases . u. dierkes , s. hildebrandt , a. kster , and o. wohlrab , _ minimal surfaces . i. boundary value problems _ , grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences ] , 295 . springer - verlag , berlin , 1992 . u. dierkes , s. hildebrandt , a. kster , and o. wohlrab , _ minimal surfaces . boundary regularity _ , grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical sciences ] , 296 . springer - verlag , berlin , 1992 . r. osserman , _ isoperimetric and related inequalitites _ , differential geometry ( proc . pure math . xxvii , part 1 , stanford univ . , stanford , calif . , 1973 ) , soc . , providence , r.i . , 1975 .	the nitsche conjecture is deeply rooted in the theory of doubly connected minimal surfaces . however , it is commonly formulated in slightly greater generality as a question of existence of a harmonic homeomorphism between circular annuli @xmath0 in the early 1960s , while attempting to describe all doubly connected minimal graphs over a given annulus @xmath1 , j.c.c . nitsche observed that their conformal modulus can not be too large . then he conjectured , in terms of isothermal coordinates , even more ; + + _ a harmonic homeomorphism @xmath2 exists if an only if : _ 0.2 cm @xmath3 + 0.5 cm this fascinating and engaging problem remained open for almost a half of a century . in the present paper we provide , among further generalizations , an affirmative answer to his conjecture
level correlations in the semiclassical spectra of classically integrable systems have recently received a renewed attention . the most important development was the realization of the long - range nature of such correlations @xcite , which was explored for rectangular billiards . while it had been previously known that the short - range correlations are absent , the fact reflected by the poisson statistics of the nearest neighbor level spacings @xcite , the evidence for the long - range correlations was indirect , namely , through the saturation property of the spectral rigidity @xcite,@xcite . in @xcite the correlation function of the level density was obtained , which explicitly describes the long - range correlations in the energy spectrum . furthermore , in terms of an easily measured quantity , the variance of the number of levels on an energy interval was investigated and was shown to have very unusual properties . namely , for the interval width narrower than the energy scale associated with the inverse time of the shortest periodic orbit ( traversal along the smaller side of the rectangle ) , the variance equals , in the lowest approximation , to the mean number of levels in the interval , indicating the absence of correlations in level positions . for intervals wider than such energy scale , the variance exhibits non - decaying oscillations around the `` saturation value '' with the amplitude smaller , yet parametrically comparable to the latter and with the `` period '' of the same order as the above mentioned scale ( the width of the interval at which the transition from the uncorrelated to correlated behavior occurs ) . while such behavior of the variance had being previously predicted via a formal mathematical approach @xcite , ref . @xcite established that it is a direct consequence of the long - range correlations between energy levels . two independent analytical derivations were produced @xcite : one based on the direct use of quantum mechanical expressions for the energy levels for a particle in a box @xcite and the other based on semiclassical periodic orbit theory @xcite . within the latter , it was shown that the oscillations of the variance can be explained by just a few shortest periodic orbits . these results were confirmed numerically with the use of an ensemble averaging procedure wherein the rectangles of the same area , but varying aspect ratios , were used . the reason why the oscillations of the variance can be considered counter - intuitive is because with the increase of the interval width , and the corresponding increase of the mean number of levels , the fluctuation of the number of levels in the interval may actually decrease . it should be pointed out that the only reason that the variance does not become zero for a rectangular box is that the harmonics that correspond to the shortest periodic orbits have incommensurate frequencies and thus add incoherently@xcite . if one could find a system where variance would be dominated by a single periodic orbit harmonic , it could be near zero for certain widths of the energy interval . this would be even more counter - intuitive as statistically independent systems would produce different level structures yet the total number of levels for such intervals would be nearly the same ! in this work we report that we found just such a system - modified coulomb problem - and , in addition to examining the variance , we also show that the saturation value of the spectral rigidity @xcite exhibit quantum jumps associated with the change in the winding number ratio of radial and angular motions of periodic orbits . in what follows , we first discuss level correlations in the semiclassical spectrum and illustrate it by a classically integrable problem of a particle in a rectangular box @xcite . next , we give a detailed analytical description of the modified coulomb problem and derive expressions for the saturation spectral rigidity and for the level number variance using the periodic orbit theory . we then proceed with the numerical evaluation of these quantities for a model spectrum , that captures key features of our model , where we observe the quantum jumps in saturation rigidity and single - harmonic oscillations of the variance . we will consider intervals @xmath0 $ ] , @xmath1 , where the states with energies near @xmath2 have large quantum numbers and can be described semiclassically . denote by @xmath3 the cumulative number of levels ( or spectral staircase)@xcite @xmath4 where @xmath5 is unit step function and @xmath6 labels the energy eigenstates . a universal ( flattened ) representation of the ladder data is obtained by rescaling the energy variable so as to eliminate the particular shape of the average @xmath7 from the ladder@xcite . to do this , define the new scaled _ dimensionless _ energy variable @xmath8 by @xmath9 here @xmath10 denotes the ensemble average . in particular , to a computed eigenvalue @xmath11 the value @xmath12 is assigned . since @xmath8 is a monotone function of @xmath2 @xmath13 so that the mean level density ( and the mean level spacing - with caveats explained in ref . @xcite ) is _ unity _ in the scaled variable @xmath14 since @xmath7 is used as the scaled energy axis variable in order to give the ladders an approximately @xmath15 slope , it is important to have a fair idea of its functional form for presenting the numerical data . in what follows , we will omit the `` prime '' for the scaled energy variable . since the present work concentrates on the long - range correlations in the spectrum of the eigenvalues , we will not discuss per se the distribution of nearest - neighbor level spacings , except to state that our computations for the modified coulomb problem below give the poisson statistics @xcite , as ought to be the case for a classically integrable system . for the statistics of large numbers of levels , the following standard measures will be used . the first is the spectral rigidity @xmath16 , defined in @xcite,@xcite as the best linear fit to the spectral staircase in the interval @xmath17 $ ] @xmath18 ^{2}\right\rangle \label{delta3}\ ] ] the explicit form of which is @xmath19 ^{2}-\frac{12}{e^{4}}\left [ \int_{\varepsilon -e/2}^{\varepsilon+e/2}d\varepsilon\varepsilon\mathcal{n}\left ( \varepsilon\right ) \right ] ^{2}\right\rangle \label{delta3 - 1}\ ] ] for the number of levels @xmath20 on the interval @xmath21 $ ] @xmath22 the variance @xmath23 is another measure of the fluctuations . notice that in the flattened spectrum ( [ delta ] ) , considered here , @xmath24 . the fluctuation measures @xmath25 and @xmath16 can be expressed in terms of the correlation function of the density of levels , @xcite , @xmath26 regardless of the form of @xmath27 , for instance , @xmath28 using these relationships one can further show that @xmath25 supersedes @xmath16 via an integral relationship@xcite @xmath29 in the periodic orbit theory , the correlation function ( [ k ] ) can be expressed as a sum over classical periodic orbits @xcite . the important energy scale in the system is that associated with the period of the shortest periodic orbit @xmath30 for instance , in classically chaotic systems @xmath31 and for classically integrable systems @xmath32 @xcite . for energies @xmath33 , the levels are uncorrelated and one finds @xmath34 in the opposite limit , @xmath35 , the properties of spectral correlations are very different for the classically chaotic and classically integrable systems @xcite . for the former , they are well known and are described by random matrix theory @xcite and supersymmetric non - linear sigma model @xcite . for the latter , it was believed that they lead to the saturation rigidity given by @xcite @xmath36 where @xmath37 and @xmath38 are the amplitudes and the periods of the periodic orbits and @xmath39 is the dimension of phase space . it turns out however , that the more precise formulae , up to the leading terms @xmath40 , are as follows : @xcite,@xcite @xmath41 \label{delta3inf - p2}\\ \sigma^{\infty}\left ( \varepsilon;e\right ) & \simeq\sum_{j}\frac { 8a_{j}^{2}}{\hbar^{n-1}t_{j}^{2}}\sin^{2}\left ( \frac{et_{j}}{2\hbar } \right ) = \overline{\sigma}^{\infty}\left ( \varepsilon;e\right ) \left [ 1-\frac{4}{\hbar^{n-1}\overline{\sigma}^{\infty}}\sum_{j}\frac{a_{j}^{2}}{t_{j}^{2}}\cos\left ( \frac{et_{j}}{\hbar}\right ) \right ] \label{sigmainf - p}\ ] ] where @xmath42 in the above equations , the superscript `` @xmath43 '' refers to `` saturation behavior '' and the overbar to averaging over the oscillations . note that both @xmath37 and @xmath38 depend explicitly on the position of the center of the interval @xmath44 . for instance , in a rectangular , with the aspect ratio of its sides @xmath45 , one finds @xcite that the periods are integers ( representing the number of retracings ) of irreducible cycles @xmath46 @xmath47 where @xmath48 and @xmath49 are coprime winding numbers of classical periodic orbits such that @xmath50 @xmath51 being the periods of of motion along the sides @xmath52 . expressions for @xmath53 , and resulting formulae for the quantities of interest , can be found in refs . @xcite and @xcite . the key consequences of the above results are as follows . first , the `` amplitude '' of oscillations around @xmath54 decays with the width of the interval @xmath55 . conversely , the `` amplitude '' of oscillations around @xmath56 does not decay with the increase of @xmath55 ; furthermore , this amplitude is of the order of @xmath56 . second , the amplitudes of oscillations decreases rapidly with the period of periodic orbits . in a rectangle , for instance , @xmath57 and , using eq . ( [ tj ] ) , it is easy to see that just a few terms with smallest winding numbers should dominate the sums in the above equations ; this was indeed confirmed numerically @xcite . to further appreciate these consequences , consider the contribution from a single harmonic only and compare the result with the known behavior of @xmath58 in completely uncorrelated system , in an almost rigid spectrum ( gaussian ensemble ) and completely rigid spectrum ( harmonic oscillator ) . it is convenient to consider the derivative @xmath59 , for which we find @xcite,@xcite @xmath60{ccccc } & $ \genfrac{}{}{0pt}{}{\text{uncorrelated}}{\text{(integrable short range)}}$ & $ \genfrac{}{}{0pt}{}{\text{nearly rigid}}{\text{(gaussian ensembles)}}$ & $ \genfrac{}{}{0pt}{}{\text{rigid}}{\text{(harmonic oscillator)}}$ & $ \genfrac{}{}{0pt}{}{\text{oscillatory}}{\text{(integrable long range)}}$\\ $ \frac{\partial\sigma^{\infty}\left ( \varepsilon;e\right ) } { \partial e}\propto$ & $ 1 $ & $ e^{-1}$ & $ 0 $ & $ \frac{4e}{\hbar^{n-2}}\sum _ { j\text{:short}}\frac{a_{j}^{2}}{t_{j}}\sin\left ( \frac{et_{j}}{\hbar } \right ) $ \end{tabular } \ \ ] ] where the summation is limited to the few shortest periodic orbits ( and the corresponding harmonics ) . clearly , depending on the interval width @xmath55 , the oscillatory @xmath58 exhibits the range of behaviors , from uncorrelated to rigid . moreover , @xmath61 can become _ negative _ , implying a seemingly paradoxical result where the fluctuation of the number of levels decreases as the average interval width ( and the mean number of levels ) increases . finally , because the frequencies of harmonics are incommensurate , @xmath58 ordinarily does not reach zero ; it oscillates between @xmath62 and @xmath63 , each typically of order of @xmath64 parametrically ( for a particle in the box see @xcite ) . however , if one can find a system where the shortest periodic orbit @xmath65 dominates the sum , @xmath58 can be reduced , as per eqs . ( [ sigmainf - p]-[delta3sigmainfav - p ] ) , to @xmath66 and can become effectively zero for @xmath67 . the same effect would be also achieved if the periods of other periodic orbits are integer multiples of @xmath65 . we found such a system in a modified coulomb problem that we proceed to discuss below . we consider a particle in the central potential @xmath68 classically , the trajectory of the motion is given by @xcite @xmath69 where@xmath70 using the canonical action variables @xcite@xmath71 we can express the energy as @xmath72 and rewrite expressions for @xmath73 and @xmath74 as @xmath75 respectively . the frequencies of radial and angular motion are given by@xcite : @xmath76 where the notation @xmath77 was introduced . for any energy @xmath2 , the motion is conditionally periodic except for the following two circumstances . first , for the values of the angular momentum @xmath78 such that @xmath79 the motion becomes periodic with the periods of radial and angular motions related by @xmath80 here @xmath81 is the period of an irreducible cycle @xmath82 ( @xmath83 and @xmath84 are coprime ) , whose orbital and angular winding numbers are , respectively , @xmath83 and @xmath84 . in other words , the orbit closes for the first time after @xmath83 periods of radial motion and @xmath84 periods of angular motion . second , from@xmath85 the motion becomes circular for @xmath78 such that @xmath86 in which case @xmath87 . from eqs . ( [ gamma ] ) and ( [ ithetair ] ) the corresponding frequency is found as @xmath88 where @xmath89 is still given by ( [ omegar ] ) but , since the distance from the center remains fixed , does not have the meaning of a radial frequency . it is very important to notice that @xmath89 depends only on the energy @xmath2 , and does not depend on the angular momentum @xmath78 . as was already mentioned , at any energy @xmath2 the conditionally periodic motion becomes periodic for such values of @xmath78 that @xmath90 is rational ; these values , however , do not depend on @xmath2 , except for the constraint @xmath91 that follows from eqs . ( [ gamma ] ) and ( [ gamma - cir ] ) . consequently , the following picture of the periodic orbits emerges . in addition to circular orbits , whose period @xmath92 is given by eqs . ( [ omegar ] ) and ( [ gamma - cir ] ) and changes continuously as a function of energy , there are irreducible orbits such that @xmath93 + i \label{mr}\ ] ] where @xmath94 $ ] is the floor function and @xmath95 are integers such that @xmath83 and @xmath84 are coprime . these correspond to rational @xmath90 s ( [ gamma - wn ] ) and their period is given by @xmath96 in view of inequality ( [ gamma - lmt ] ) , new rational values of @xmath90 become possible at discrete ( quantized ) values of @xmath2 . in particular , the shortest periodic orbits @xmath97 + 1\text { , } i<\gamma^{(cir ) } \label{mr - min}\ ] ] are especially important . the key observation here is that as @xmath2 increases ( @xmath98 decreases ) , the smaller values of @xmath83 become possible , with quantum jumps occurring for energies such that @xmath99 is integer . this fact will prove to be crucial in evaluation of the spectral rigidity and level number variance . finally , for either type of the periodic orbit , it can be subsequently retraced with the period of @xmath100 , where @xmath101 is the number of retracings . the quantum spectrum is given by @xcite @xmath102 which clearly follows from ( [ ecanon ] ) via born - zommerfeld quantization of the action variables @xmath103 and the second of eqs . ( [ elp ] ) is the limit of large quantum numbers , @xmath104 , that in the semiclassical approximation used here . further , we make yet another approximation that concerns with the fact that the standard kepler problem ( bohr atom in the quantum limit ) is a so called `` supersymmetric '' or `` resonant '' problem @xcite . this is because the frequency of radial and angular motions coincide in the kepler ( so that the motion is periodic ) , which is indicative of extra symmetry in the problem , as well as additional conserved quantities - runge - lenz vector in the present case . in the bohr atom , the latter is associated with the @xmath105-fold degeneracy of the @xmath101 s energy eigenstate . conversely , in a `` generic '' ( non - resonant ) integrable system , the motion is ordinarily conditionally periodic , except for specific values of certain parameters upon which the motion may become periodic . thus , in order to achieve the greatest possible difference with the standard kepler problem , a large parameter @xmath106 must be considered in ( [ v - r ] ) . accordingly , we assume that the condition @xmath107 holds , where @xmath108 is the bohr radius , that is @xmath109 . this condition translates , as follows form ( [ elp ] ) , to that of the quantum numbers @xmath73 , @xmath110 being limited from above which , combined with the semiclassical approximation , yields @xmath111 using ( [ constraint ] ) to expand eq . ( [ elp ] ) , we obtain @xmath112 and , with a substitution , @xmath113 we find @xmath114 classically , the condition corresponding to ( [ constraint ] ) would be @xmath115 leading to @xmath116 where @xmath117 and @xmath118 @xmath119 are now dimensionless . obviously , in both quantum and classical circumstances , @xmath120 in the zeroth order . the latter leads to simplified formulas for the frequencies since in this approximation @xmath121 and , from eqs . ( [ gamma - cir ] ) and ( [ e - exp]),@xmath122 as follows from ( [ epsli ] ) . due to the linear relationship between @xmath2 and @xmath123 in ( [ elp - exp ] ) , spectral properties of the two are identical . consequently , in what follows , it is spectrum ( [ epslp ] ) that is studied numerically . similarly to the rectangle , where ensemble averaging was understood in terms of variations of the rectangle s aspect ratio @xcite , here ensemble averaging is understood in terms of the variations of @xmath106 . flattening of the spectrum is achieved via ( [ epsilon_reduced ] ) with @xmath124 which follows immediately from eqs . ( [ cn ] ) and ( [ epslp ] ) . this is equivalent to starting with the scaled hamiltonian @xmath125 for which @xmath126 in what follows , we will drop subscript `` @xmath127 . '' the frequencies are now given by @xmath128 where @xmath129 obtained by solving ( [ epsilon - sc ] ) for @xmath130 . ( for convenience , we carried over the notation @xmath99 ) we now turn to evaluation of the level correlation function ( [ kinf - p ] ) . using the result obtained in appendix , we find @xmath131 to reduce this to the sum on @xmath83 only , we notice that from ( [ gamma - sc ] ) and ( [ mr - min ] ) , @xmath132 + 1 \label{mr - min2}\ ] ] second , for each @xmath83 , there are @xmath133 $ ] possible values of @xmath84 , and , consequently , we obtain @xmath134 \cos\left ( \frac{2\pi\left ( \varepsilon_{1}-\varepsilon_{2}\right ) m_{r}}{\omega_{r}}\right ) \label{kinf - mc2}\ ] ] for very small @xmath135 , is possible to reduce this sum to an integral . neglecting the difference between the function and its floor , including time - reverse of each orbit , and using eqs . ( [ omega - sc ] ) and ( [ gamma - sc ] ) , we find @xmath136 where @xmath137 and @xmath138 is the period of the shortest periodic orbit . this is in complete analogy to the approximate form of the correlation function found for a rectangular box@xcite where the @xmath139-function term corresponds to the absence of level correlations and the second term the onset of thereof . similarly , saturation spectral rigidity ( [ delta3sigmainfav - p ] ) is given by @xmath140 \frac{1}{m_{r}^{3}}=\frac{\sqrt{2\beta}}{\pi^{2}}\sum_{m_{r}=2}\left [ \frac{m_{r}}{\gamma^{(cir)}}\right ] \frac{1}{m_{r}^{3}}\label{delta3inf - final}\ ] ] where @xmath55 is now understood as interval width in the spectrum @xmath123 . the second equality follows from the fact that the floor function in the sum automatically takes care of summation starting with @xmath141 . together , ( [ delta3inf - final ] ) , ( [ gamma - sc ] ) and ( [ mr - min2 ] ) transparently predict quantum jumps in saturation level rigidity . as the energy increases , @xmath142 decreases and , as it takes on smaller integer values , a transition @xmath143 takes place leading to the jump in saturation rigidity . we observe such jumps in numerical simulations , discussed in the next section . finally , the level number variance is given by@xmath144 \frac{1}{m_{r}^{3}}\sin^{2}\left ( \frac{\pi em_{r}}{\omega_{r}}\right ) \label{sigmainf - mc}\ ] ] assuring that @xmath145 . we conduct numerical simulations on the spectrum ( [ epslp ] ) for central values of @xmath146 and @xmath147 . as previously mentioned , `` ensemble averaging '' is accomplished for each @xmath106 by taking @xmath148 values of @xmath106 around the central value . these @xmath106 s must be sufficiently close to the central value , as to eliminate the systematic dependence on @xmath106 , yet sufficiently far to ensure proper sampling . as a first step , we verified the poisson ( exponential ) distribution for the nearest level spacings , which should be the case for an integrable system without extra degeneracies@xcite . the numerical result for @xmath149 vis - a - vis ( [ delta3inf - final ] ) are shown in fig . 1 [ ptb ] figs / fig1.eps while ( [ delta3inf - final ] ) provides a wonderful fit to numerical data on the top plateau , it predicts much bigger jumps between the plateaus than is seen numerically . as a results , to fit lower plateaus we empirically added an appropriate empirical constant via substitution @xmath150 \rightarrow\left [ \frac{m_{r}}{\gamma^{(cir)}}\right ] + \sum_{n=1}^{\left [ m_{r}/\gamma^{(cir)}\right ] -1}\frac{1}{2^{2n-1}}\label{correction}\ ] ] at this time we do not have a good explanation for this discrepancy . further , the theoretical formula predicts additional noticeable jump further up in the spectrum , when @xmath151 , which is not seen on the experimental curve . with the same modification ( [ correction ] ) in eq . ( [ sigmainf - mc ] ) , we plot @xmath152 as a function of @xmath55 for two different values of @xmath123 [ ptb ] figs / fig2.eps [ ptb ] figs / fig3.eps clearly , while the periodicity is superbly predicted by the analytical expression , there is a striking difference in the intermediate structure in comparison with the numerical results . we predict analytically , and observe numerically , quantum hall like jumps of the averaged saturation level rigidity in a modified coulomb problem . these are explained semiclassically in terms of jumps in the winding numbers of the shortest periodic orbits as the position of the interval center moves through the energy spectrum . also , analytically and numerically , we predict sinusoidal oscillations of the saturation level number variance with the interval width . this is a striking result which indicates that while distribution of the levels on a interval varies greatly from sample to sample , the total number of levels in the interval may be nearly identical for certain values of the interval width . this is because all the higher harmonics are integer fractions of the shortest periodic orbit and add up coherently . the latter explains the difference with other systems , such as rectangular box , where the oscillations may be large but the variance does nt reach a near - zero value . the amplitude of an irreducible cycle @xmath46 is given by@xcite @xmath153 this equation can be simplified by differentiating ( [ omega_m ] ) on @xmath154 , which gives @xmath155 whereof @xmath156 + \omega _ { 1}\omega_{2}\left ( \partial\omega_{1}/\partial i_{2}+\partial\omega _ { 2}/\partial i_{1}\right ) } { t\det\left\ { \partial\omega_{i}/\partial i_{k}\right\ } } \label{dotprod2}\ ] ] so that @xmath157 + \omega _ { 1}\omega_{2}\left ( \partial\omega_{1}/\partial i_{2}+\partial\omega _ { 2}/\partial i_{1}\right ) \right\vert _ { \mathbf{m } } } \label{a_m - simpl}\ ] ] which shows a universal dependence of @xmath158 on @xmath81 .	we discover quantum hall like jumps in the saturation spectral rigidity in the semiclassical spectrum of a modified kepler problem as a function of the interval center . these jumps correspond to integer decreases of the radial winding numbers in classical periodic motion . we also discover and explain single harmonic dominated oscillations of the level number variance with the width of the energy interval . the level number variance becomes effectively zero for the interval widths defined by the frequency of the shortest periodic orbit . this signifies that there are _ virtually _ _ no variations from sample to sample _ in the number of levels on such intervals .
in this short note , we study the following modified smoluchowski equation ( also called fokker planck equation ) , for an orientation distribution @xmath0 defined for a time @xmath1 and a direction @xmath2 ( the unit sphere @xmath3 of @xmath4 ) as follows : @xmath5 where @xmath6 , @xmath7 , and @xmath8 are the laplace beltrami , divergence , and gradient operators on the sphere . the vector @xmath9 is the first moment associated to @xmath10 ( the measure on the sphere is the uniform measure such that @xmath11 ) , and @xmath12 represents the mean direction of the distribution @xmath10 . the first term of the right - hand side of is an alignment term towards @xmath13 , and the function @xmath14 represents the strength of this alignment . the function @xmath15 is the intensity of the diffusion on the sphere . when @xmath15 is a constant and @xmath16 , we can recover the standard smoluchowski equation on the sphere , with dipolar potential @xcite . indeed , the dipolar potential is given by @xmath17 , and the equation can be recast as : @xmath18 other classical kernels @xcite for the study of semi - dilute and concentrated suspensions of polymers are the so - called maier saupe potential @xmath19 , or the original onsager potential @xmath20 . in these cases , there are a lot of studies regarding the phase transition phenomenon for equilibrium states @xcite , and in particular a hysteresis phenomenon occurs for the maier saupe potential in dimension @xmath21 . the case of the dipolar potential has also been studied precisely @xcite , with an analysis of the rates of convergence to the equilibrium as time goes to infinity . in this case , there exists a so - called continuous phase transition for a critical threshold @xmath22 : when @xmath23 the solution converges exponentially fast to a non - isotropic equilibrium ; when @xmath24 it converges exponentially fast to the uniform distribution . at the critical value @xmath25 , the solution converges to the uniform distribution at a rate @xmath26 . here we study the modifications arising when @xmath14 and @xmath15 depend on @xmath27 , which can be motivated by some biological modeling ( see @xcite and references therein ) . in that case , we can not use @xmath15 as a bifurcation parameter anymore . instead , we will use the initial mass @xmath28 ( a conserved quantity ) as the key parameter to study the phase transition . we will assume that : [ hyp - regularity ] 1 . the functions @xmath14 and @xmath15 are @xmath29 , with @xmath30 and @xmath31 . the function @xmath32 is an increasing function . we denote by @xmath33 its inverse , i.e. @xmath34 the first part of this hypothesis implies that we do not have any singularity of @xmath35 in as : if @xmath36 , we simply have @xmath37 . the second part is made for the sake of simplicity . it leaves enough flexibility to to reveal key behaviors in terms of phase transitions . it would be easy to remove it at the price of an increased technicality . additionally , it means that when @xmath10 is more concentrated in the direction of @xmath13 , the relative strength of the alignment force compared to diffusion is increased as well . this can be biologically motivated by the existence of some social reinforcement mechanism . we will see that we can observe a wealth of phenomena , including hysteresis . the purpose of this note is to summarize the analytical results , as well as some numerical simulations which illustrate this phenomena . all the proofs are detailed in @xcite , where arises as the spatially homogeneous version of a space - dependent kinetic equation , obtained as the mean - field limit of a self - propelled particle system interacting through alignment . this spatially homogeneous study is crucial to determine the macroscopic behavior of this space - dependent kinetic equation . we first state results about existence , uniqueness , positivity and regularity of the solutions of . under hypothesis [ hyp - regularity ] , we have the following [ theorem - existence - uniqueness ] given an initial finite non - negative measure @xmath38 in @xmath39 , there exists a unique weak solution @xmath10 of such that @xmath40 . this solution is global in time . moreover , @xmath41 , with @xmath42 for all positive @xmath43 , and we have the following instantaneous regularity and uniform boundedness estimates ( for @xmath44 , the constant @xmath45 being independent of @xmath10 ) , for all @xmath46 : @xmath47 for later usage , we define @xmath48 as an anti - derivative of @xmath49 : @xmath50 . in this case , the dynamics of corresponds to the gradient flow of the following free energy functional : @xmath51 indeed , if we define the dissipation term @xmath52 by @xmath53 we get the following conservation relation : @xmath54 we now define the von mises distribution which provides the general shape of the non - isotropic equilibria of @xmath35 . [ def - vonmises ] the von mises distribution of orientation @xmath55 and concentration parameter @xmath56 is given by : @xmath57 the order parameter @xmath58 is defined by the relation @xmath59 and has expression : @xmath60 the concentration parameter @xmath58 defines a one - to - one correspondence @xmath61 . the case @xmath62 corresponds to the uniform distribution , while when @xmath63 is large ( or @xmath58 is close to @xmath64 ) , the von mises distribution is closed to a dirac delta mass at the point @xmath65 . some comments are necessary about the interval of definition of @xmath33 . first note that , under hypothesis [ hyp - regularity ] , @xmath49 is defined from @xmath66 , with values in an interval @xmath67 , where we may have @xmath68 . so @xmath33 is an increasing function from @xmath67 onto @xmath69 . moreover , for later usage , we can define @xmath70 where @xmath71 may be equal to @xmath72 , and where we recall that @xmath73 denotes the dimension . the equilibria are given by the following proposition : [ prop - equilibria ] the following statements are equivalent : 1 . @xmath74 and @xmath75 . 2 . @xmath76 and @xmath77 . 3 . there exists @xmath78 and @xmath55 such that @xmath79 , where @xmath56 satisfies the compatibility equation : @xmath80 let us first remark that the uniform distribution , corresponding to @xmath81 is always an equilibrium . indeed , we have @xmath82 and is satisfied . however , proposition [ prop - equilibria ] does not provide any information about the number of the non - isotropic equilibria . indeed , equation can be recast into : @xmath83 which is valid as long as @xmath84 . we know that @xmath33 is an increasing unbounded function from its interval of definition @xmath67 onto @xmath66 , and thanks to hypothesis [ hyp - regularity ] and to , we know that @xmath85 as @xmath86 ( if @xmath87 ) . so since @xmath88 as @xmath86 ( see for instance @xcite ) , we have the two following results ( also valid in the case @xmath89 ) : @xmath90 we deduce that this function reaches its maximum , and we define @xmath91 for @xmath92 , the only solution to the compatibility condition is @xmath81 , and the only equilibrium is the uniform distribution @xmath93 . except from these facts , we have no further direct information of this function @xmath94 , since @xmath95 and @xmath33 are both increasing . figure [ fig - shapes ] depicts some examples of the possible shapes of the function @xmath94 . we see that depending on the value of @xmath28 , the number of families of non - isotropic equilibria , given by the number of positive solutions of the equation , can be zero , one , two or even more . we now turn to the study of the stability of these equilibria , through the study of the rates of convergence . the main tool to prove convergence of the solution to a steady state is lasalle s principle , that we recall here ( the proof follows exactly the lines of @xcite ) . by the conservation relation , we know that the free energy @xmath96 is decreasing in time ( and bounded from below since @xmath97 is bounded ) . lasalle s principle states that the limiting value of @xmath96 corresponds to an @xmath98-limit set of equilibria : [ prop - lasalle]lasalle s invariance principle . let @xmath38 be a positive measure on the sphere @xmath3 . we denote by @xmath99 the limit of @xmath100 as @xmath101 , where @xmath10 is the solution to the modified smoluchowski equation with initial condition @xmath38 . then the set @xmath102 is not empty . furthermore @xmath103 converges in any @xmath104 norm to this set of equilibria ( in the following sense ) : @xmath105 since we know the types of equilibria , we can refine this principle to adapt it to our problem : [ prop - lasalle - refined ] let @xmath38 be a positive measure on the sphere @xmath3 , with initial mass @xmath28 . if no open interval is included in the set @xmath106 , then there exists a solution @xmath107 to the compatibility solution such that we have : @xmath108 this last proposition helps us to characterize the @xmath98-limit set by studying the single compatibility equation . when @xmath81 is the unique solution , then this gives us that @xmath10 converges to the uniform distribution . otherwise , two cases are possible , either @xmath109 , and @xmath10 converges to the uniform distribution , or @xmath110 , and the only unknown behavior is the one of @xmath111 . if we are able to prove that it converges to @xmath112 , then @xmath10 converges to a fixed non - isotropic steady - state @xmath113 . however , proposition [ prop - lasalle - refined ] does not give information about quantitative rates of convergence of @xmath27 to @xmath114 , and of @xmath115 to @xmath116 , as @xmath101 . so we now turn to the study of the behavior of the difference between the solution @xmath10 and a target equilibrium @xmath117 . this study consists in two types of expansion . if we expand the solution around the uniform equilibrium , some simple energy estimates give us exponential convergence when @xmath118 . but when we expand the solution around a non - isotropic equilibrium @xmath117 , we see that the condition of stability is related to the monotonicity of the function @xmath94 . hence we can see directly on the graph of this function ( see examples on figure [ fig - shapes ] ) both the number of family of equilibria and their stability : if the function is decreasing , the family is stable . by contrast it is unstable when the function is increasing . when the difference between @xmath10 and @xmath117 converges exponentially fast to @xmath116 ( on the stable branch ) , we are able to control the displacement of @xmath119 , which gives convergence to @xmath112 . we then have convergence of @xmath10 to a given equilibrium@xmath113 . all these results are summarized in the following two theorems . in what follows , we say that a constant is a universal constant when it does not depend on the initial condition @xmath38 ( that is to say , it depends only on @xmath28 , @xmath73 and the coefficients of the equation @xmath14 and @xmath15 , and on the exponent @xmath120 of the sobolev space @xmath104 in which the result is stated ) . we have the following instability and exponential stability results around the uniform equilibrium : [ thm - uniform ] * suppose that @xmath118 . we define @xmath121 there exists a universal constant @xmath45 , such that if @xmath122 , then for all @xmath1 , we have @xmath123 * if @xmath124 , and if @xmath125 , then we can not have @xmath109 in proposition [ prop - lasalle - refined ] : the solution can not converge to the uniform equilibrium . to study the stability around a non - isotropic equilibrium , we fix @xmath28 , and we denote by @xmath63 a positive solution to the compatibility equation ( we will not write the dependence of @xmath95 and @xmath33 on @xmath63 when there is no possible confusion ) . we denote by @xmath126 the value of @xmath127 ( independent of @xmath55 ) . [ thm - nonisotropic ] we have the following instability and exponential stability results when starting close to a non - isotropic equilibrium : * suppose @xmath128 . for all @xmath129 , there exist universal constants @xmath130 and @xmath131 , such that for any initial condition @xmath38 satisfying @xmath132 for some @xmath55 , there exists @xmath112 such that @xmath133 where the rate is given by @xmath134 the constant @xmath135 is the best constant for the following weighted poincar inequality ( see the appendix of @xcite for more details on this constant , which does not depend on @xmath65 ) : @xmath136.\ ] ] * suppose @xmath137 . then any equilibrium of the form @xmath138 is unstable , in the following sense : in any neighborhood of @xmath138 , there exists an initial condition @xmath38 such that @xmath139 . consequently , in that case , we can not have @xmath140 in proposition [ prop - lasalle - refined ] . let us now focus on the case where we always have @xmath141 for all @xmath142 ( see for example the lowest two curves of figure [ fig - shapes ] ) . in this case , the compatibility equation has a unique positive solution for @xmath124 . with the results of the previous subsection about stability and rates of convergence , we obtain the behavior of the solution for any initial condition @xmath38 with initial mass @xmath28 . * if @xmath118 , then the solution converges exponentially fast towards the uniform distribution @xmath143 . * if @xmath144 , the solution converges to the uniform distribution . * if @xmath124 and @xmath125 , then there @xmath145 such that @xmath10 converges exponentially fast to the von mises distribution @xmath146 , where @xmath142 is the unique positive solution to the equation @xmath147 . the special case where @xmath148 leads to the heat equation @xmath149 . its solution converges exponentially fast to the uniform distribution , but this solution is not stable under small perturbation of the initial condition . let us remark that for some particular choice of the coefficients , as in @xcite , it is also possible to get an algebraic rate of convergence in the second case @xmath144 . for example when @xmath150 , we have @xmath151 for @xmath43 sufficiently large . so we can describe the phase transition phenomena by studying the order parameter of the asymptotic equilibrium @xmath152 , as a function of the initial density @xmath28 . we have @xmath153 if @xmath154 , and @xmath95 is a positive continuous increasing function for @xmath124 . in the common situation where @xmath155 when @xmath86 , it is easy to see , since @xmath156 when @xmath86 , that we have @xmath157 since @xmath158 is lipschitz , we always have @xmath159 . so the first derivative of @xmath95 is discontinuous at . this is the case of a second order phase transition ( also called continuous phase transition ) . the critical exponent @xmath160 can take arbitrary values in @xmath161 $ ] , as can be seen by taking @xmath162 such that @xmath163 . in general , we have the following practical criterion , which ensures a second order phase transition . [ lemma-2nd - order ] if @xmath164 is a non - increasing function of @xmath97 , then we have @xmath141 for all @xmath142 . in this case , the critical exponent @xmath160 in , if it exists , can only take values in @xmath165 $ ] . we now turn to a specific example , where all the features presented in the stability study can be seen . we focus on the case where @xmath166 , as in @xcite , but we now take @xmath167 . from the modeling point of view , this occurs in the vicsek model with vectorial noise ( also called extrinsic noise ) @xcite . in this case , we have @xmath168 , so the assumptions of lemma [ lemma-2nd - order ] are not fulfilled , and the function @xmath33 is given by @xmath169 . expanding @xmath158 when @xmath63 is large or @xmath63 is close to @xmath116 , we get @xmath170 consequently , there exist more than one family of non - isotropic equilibria when @xmath28 is close to @xmath171 ( and @xmath124 ) . the function @xmath172 can be computed numerically . the results are displayed in figure [ fig-2d-3d - hysteresis ] in dimensions @xmath173 and @xmath174 . we observe the following features : * there exists a unique critical point @xmath175 for the function @xmath158 , corresponding to its global maximum @xmath176 ( in dimension @xmath177 , we obtain numerically @xmath178 and @xmath179 , in dimension @xmath21 we get @xmath180 and @xmath181 ) . * the function @xmath158 is strictly increasing in @xmath182 and strictly decreasing on @xmath183 . from these properties , it follows that the solution associated to an initial condition @xmath38 with mass @xmath28 can exhibit different types of behavior , depending on the three following regimes for @xmath28 . * if @xmath92 , the solution converges exponentially fast to the uniform equilibrium @xmath143 . * if @xmath184 , there are two families of stable solutions : either the uniform equilibrium @xmath93 or the von mises distributions of the form @xmath138 , for @xmath55 where @xmath63 is the unique solution with @xmath185 of the compatibility equation . if @xmath38 is sufficiently close to one of these equilibria , there is exponential convergence to an equilibrium of the same family . + the von mises distributions of the other family ( corresponding to solution of such that @xmath186 ) are unstable in the sense given in theorem [ thm - nonisotropic ] . * if @xmath187 and @xmath125 , then there exists @xmath112 such that @xmath10 converges exponentially fast to the von mises distribution @xmath188 , where @xmath63 is the unique positive solution to the compatibility equation @xmath147 . at the critical point @xmath189 , the uniform equilibrium is stable ( and for any initial condition sufficiently close to it , the solution converges exponentially fast to it ) , but the stability of the family of von mises distribution @xmath190 is unknown .. at the critical point @xmath191 , the family of von mises distribution @xmath192 is stable , where @xmath193 is the unique positive solution of . for any initial condition sufficiently close to @xmath194 for some @xmath55 , there exists @xmath145 such that the solution converges exponentially fast to @xmath195 . however , in this case , the stability of the uniform distribution @xmath196 is unknown . as previously , in the special case @xmath148 , the equation reduces to the heat equation and the solution converges to the uniform equilibrium . since @xmath58 is an increasing function of @xmath63 , we can invert this relation @xmath197 into @xmath198 and express the density @xmath199 as a function of @xmath95 . the result is depicted in figure [ fig - phase - diagram ] for dimension @xmath177 or @xmath21 . with this picture , we recover the phase diagram in a conventional way : the possible order parameters @xmath95 for the different equilibria are given as functions of @xmath28 . the dashed lines corresponds to branches of equilibria which are unstable . we can also obtain the corresponding diagrams for the free energy and the rates of convergences . for this particular example , the free energies @xmath200 and @xmath126 ( we recall that they correspond respectively to the free energy of the uniform distribution and of a von mises distribution @xmath138 for a positive solution @xmath63 of the compatibility equation , including both stable and unstable branches ) are given by @xmath201 the plots of these functions are depicted in dimensions @xmath177 and @xmath21 are depicted on the left part of figure [ fig - free - energy ] . since the functions are very close in the figure for some range of interest , we depict the difference @xmath202 in a more appropriate scale , in the right part of figure [ fig - free - energy ] . the dashed lines correspond to unstable branches of equilibria . we observe that the free energy of the unstable non - isotropic equilibria ( in dashed line ) is always above that of the uniform distribution . there exist @xmath203 and a corresponding solution @xmath204 of the compatibility solution ( with @xmath205 , corresponding to a stable family of non - isotropic equilibria ) such that @xmath206 . if @xmath207 , the global minimizer of the free energy is the uniform distribution , while if @xmath208 , then the global minimum is reached for the family of stable von mises equilibria . the physical relevance of this value is not clear though , as we will see in the numerical illustration of the next subsection . the rates of convergence to the stable equilibria , following theorems [ thm - uniform ] and [ thm - nonisotropic ] , are given by @xmath209 where @xmath210 is the rate of convergence to the uniform distribution @xmath28 , and @xmath211 is the rate of convergence to the stable family of von mises distributions @xmath138 , where @xmath63 is the unique solution of the compatibility condition such that @xmath185 . details for the numerical computation of the poincar constant @xmath135 are given in the appendix of @xcite . the computations in dimensions @xmath177 and @xmath21 are depicted in figure [ fig - rates ] . in order to highlight the role of the density @xmath28 as the key parameter for this phase transition , we introduce the probability density function @xmath212 and we get @xmath213 when @xmath28 is constant , this equation is equivalent to . we now consider @xmath28 as a parameter varying slowly with time ( compared to the time scale of the convergence to equilibrium , see figure [ fig - rates ] ) . if this parameter starts from a value @xmath92 , and increases slowly , the only stable distribution is initially the uniform distribution @xmath214 , and it remains stable . so we expect that the solution stays close to it , until @xmath28 reaches the critical value @xmath215 . for @xmath124 , the only stable equilibria are the von mises distributions , and the solution converges to one of these equilibria . the order parameter defined as @xmath216 , then jumps from @xmath116 to @xmath217 . if then the density @xmath28 is further decreased slowly , the solution stays close to a von mises distribution , and the order parameter slowly decreases , until @xmath28 reaches @xmath218 back . for @xmath92 , the only stable equilibrium is the uniform distribution , and the concentration parameter jumps from @xmath219 to @xmath116 . this is a hysteresis phenomenon : the concentration parameter describes an oriented loop called hysteresis loop . let us now present some numerical simulations of the system in dimension @xmath173 . we start with a initial condition which is a small perturbation of the uniform distribution , and we take @xmath220 , with @xmath221 . we use a standard central finite different scheme ( with @xmath222 discretization points ) , implicit in time ( with a time step of @xmath223 ) . the only problem with this approach is that the solution converges so strongly to the uniform distribution for @xmath118 , so after passing @xmath215 , the linear rate of explosion for @xmath224 is given by @xmath225 , and is very slow when @xmath28 is close to @xmath215 . so since @xmath224 is initially very small when passing the threshold @xmath144 we would have to wait extremely long in order to see the convergence to the stable von mises distribution . to overcome this problem , we adding a threshold @xmath226 and strengthen @xmath227 when @xmath228 , by @xmath229 we note that this transformation that we still have @xmath228 if it was the case before applying the transformation . figure [ fig - numerics - kinetic-2d ] depicts the result of a numerical simulation with a threshold @xmath230 . we clearly see this hysteresis cycle , which agrees very well with the theoretical diagram . the jumps at @xmath189 and @xmath144 are closer to the theoretical jumps when @xmath231 is very large . we were not able to see any numerical significance of the value @xmath232 ( for which uniform and non - isotropic distributions have the same free energy ) in all these numerical simulations . in particular , @xmath232 is close to @xmath218 ( see figure [ fig - free - energy ] ) , so in most of the cases where both uniform and non - isotropic distribution are stable , the uniform distribution is not the global minimizer of the free energy , but in practical , meta - stability is very strong , and the solution still converges to the uniform distribution . in this note , we have given the summary of strong results on the stability and instability of the equilibrium states of the modified smoluchowski equation . this allows to have a precise description of the dynamics of the solution when time goes to infinity : it converges exponentially fast to a fixed equilibrium , with explicit formulas for the rates of convergence . we have also exhibited a specific example in which we observe a first order phase transition with a hysteresis loop ( in contrast with the second order phase transition of the original smoluchowski equation with dipolar potential @xcite ) . the details of the proofs will be found in a longer paper @xcite as well as numerical comparisons between the particle and kinetic models to confirm that the hysteresis is really intrinsic to the system and not simply an artifact of the kinetic modeling .	in this note , we study the phase transitions arising in a modified smoluchowski equation on the sphere with dipolar potential . this equation models the competition between alignment and diffusion , and the modification consists in taking the strength of alignment and the intensity of the diffusion as functions of the order parameter . we characterize the stable and unstable equilibrium states . for stable equilibria , we provide the exponential rate of convergence . we detail special cases , giving rise to second order and first order phase transitions , respectively . we study the hysteresis diagram , and provide numerical illustrations of this phenomena . 1-universit de toulouse ; ups , insa , ut1 , utm ; + institut de mathmatiques de toulouse ; + f-31062 toulouse , france . + 2-cnrs ; institut de mathmatiques de toulouse umr 5219 ; + f-31062 toulouse , france . + email : [email protected] , [email protected] 3- ceremade - umr cnrs 7534 + universit de paris - dauphine + 75775 paris cedex 16 , france + email : [email protected] 4- department of physics and department of mathematics + duke university + durham , nc 27708 , usa + email : [email protected]
magnetic thin film systems and multilayer systems have been studies very actively because of their magnetic properties and possible application for practical devices , such as magnetic recording media technologies . samples were grown on smooth , low - stress , 160 nm . thick sinx membranes by magnetron sputtering ; they all had 20 nm . thick pt buffer layers and 3 nm . thick pt caps . between the buffer layer and the cap , the samples had 50 repeating units of a 0.4 nm . thick cobalt layer and a 0.7 nm . thick pt layer . experiments used linearly polarized x rays from the advanced light source at lawrence berkeley laboratory ( supported by usdoe ) , the thirteenth harmonic of the beamline 9 , undulator gap of 54 mm . near the resonant cobalt l edge . to achieve transverse coherence , the raw undulator beam was passed through a 35 micron diameter pinhole before being scattered in transmission by the sample . the distance from the sample to the ccd is 118 cm . the resonant magnetic scattering was collected by the princeton soft x - ray ccd camera 1024 x 1024 pixels in an area one inch by one inch . the intensity of the raw undulator beam was 2 x 10 ` * * ` 14 photons / sec . , the intensity of the coherent beam was 2 x 10 ` * * ` 12 photons / sec . , and the intensity of the scattered beam , was 2 x 10 ` * * ` 7 photons / sec . each speckle pattern was measured for 30 to 100 seconds , so the total number of photons in each ccd image 1024 x 1024 pixels is about 10 ` * * ` 9 . the speckle patterns may be used to reconstruct the magnetic domain structure of the sampl@xmath0 ; this is but one of a general class of the old inverse or phase retrieval problem@xmath1 . the data appears as a series of ccd image files 726axxx.spe ( of increasing photon energy ) that is read into the data processing computer program . the file numbers can be converted to photon energy since the difference between the two maxima files 122 ( photon energy 779.2 ev . ) and 145 is 15.8 ev . figure 1 ( left ) shows the magnetic resonance has two peaks . each piece of data has an image file associated and that file has a 1024 x 1024 matrix containing intensity values ( if plotted in 3d it has the shape of a centered crown ) . the princeton ccd camera file is read into the freely available matlab data analysis program by the following code fragment : .... % auto - ignore fid = fopen('nameoffile.spe','r ' ) ; header = fread(fid,2050,'uinit16 ' ) ; % half of 4100 bytes immat = fread(fid,1024 * 1024,'uint16 ' ) ; z = reshape(immat,1024,1024 ) ; fclose(fid);z = double(z ) ; [ x , y]=meshgrid(1:1024,1:1024 ) ; mesh(x , y , z ) ; % display 3d plot axis square ; axis tight ; view(90,90 ) ; print -djpeg99 nameoffile.jpg % highest resolution saved % send email [email protected] .... figure 1(right ) shows the variation of the intensity in file 122 in the radial direction that has been azimuthally averaged ( consider slicing through the center out past the edge of the crown ) . the data analysis calculates the radius of each pixel , places it into appropriate bins and finds the average intensity ; since each bin is only one pixel wide the bin number is the radius rounded to integer . the profile was fitted to a gaussian using non - linear levenberg - marquardt least squares . the difference between the maximum and minimum calculated values is the height , the distance of the maximum from the origin is called the peak center , and the fwhm ( full width at half maximum ) is called the width . more about the variation of the height , width and center with filenumber will appear at the end . the series of files 726axxx.spe are analysed for normalized cross - correlatio@xmath2 using 2d matrices of the image files from which the average pixel values have been subtracted . the sum of the product of pixel values from two images divided by the square root of the product of the sum of the pixel values squared of each image ( makes it normalized ) is gamma ( normalized cross - correlation ) of the two images . if gamma is 1 they are correlated , if zero uncorrelated and if -1 anti - correlated . the normalized cross - correlation function , gamma , is related to the coherence functio@xmath3 . table 1 shows the values of gamma for four of the files selected as file 117(middle of first peak ) , file 122 ( top of first peak ) , file 133 ( minimum between the two peaks ) and file 145 ( top of second peak ) . the ccd images were processed so as to remove the burns , remove anomalous charge scattering , remove the blocker arm , centered in the image and the central disk darkened to remove the burns in that area also . the bright spots on the image are fixed burns in the ccd camera that cause a few pixels to be unusable ; these are the ones that appear very bright . at the end of a series of spe image files a ccd burns only image was taken and a burns and anomalous charge scattering image was taken . the difference of these two images was used to subract off the anomalous charge scattering pisels from each image file pixels ( provided they are above the minimum background value of the image ) . after the anomalous charge scattering is removed , the burns ( or hot spots ) are removed . the hot spots are masked off , and after rotating around the middle of the 1024 x 1024 image , new background replaces the hot spots . next , the circle of maximum intensity is found by having an imaginary turtle going out from beyond the edge of the blocker disc in rays every 1 degree finding maximum intensity . the intensities are sorted , the lowest 20 points out of 360 dropped and a least squares fit of the 340 points is done yielding the circle of maximum intensity . the approximate edges of the blocker arm ( about 7 degrees ) are located by finding those points on the circle of maximum intensity where the second derivative changes sign . a mask is created , a rotation picks up a patch , and the pixels of the blocker arm are replaced . missing background is fille in on the edges to make 1024 x 1024 . all pixels further out from the center , beyond the circle of maximum intensity were used to find the centroid and then the whole pattern was relocated to the center of the image matrix . due to the use of coherent light from synchrotron radiation , the scattering patterns produced are highly speckled . each speckle is the sum of light scattered from all the illuminated magnetic domains . so small changes in the microscopic orientation of the magnetic domains can have a large effect on the speckle pattern . the bulk magnetization of cobalt usually is in the x - y plane but when there are only a few cobalt atoms in the layer the magnetication becomes perpendicular to the plane of the layers . the observed intensity is related to the charge density , polarization , photon energy , magnetization , atomic scattering factor and scattering geometry ; there are terms due to magnetic resonance and electronic structur@xmath4 . the azimuthal variation of intensity is calculated by looking at the angle from the center for each pixel , classifying it as belonging to bins ( 1 to 360 ) and finding the average of each bin ; this is plotted versus the bin number with each bin being one degree wide . in fig . 2 the difference between the maxima and the lowest minimum is v1 , v2 . the azimuthal variation of the intensity is described by an equation for magneto - crystalline anisotropy energy that includes a term proportional to the sine of twice the azimuthal angle square@xmath5 . the amplitude has real and imaginary charge and magnetic anomalous scattering factors which are tensors in the general cas@xmath6 . the right side of fig . 2 pretty much follows the shape of average intensity versus file number in fig . 1(left ) and illustrates the variation of the magnetic anisotropy difference ( maximum minus lowest minimum ) with photon energy(increases with file number ) . the average of the value of first maximum and second maximum would be a reasonable estimate of the variable . the reason the magnetic anistropy has these variations is because the atomic factor is a tenso@xmath7 . the autocorrelation has a maximum at the origin . you could think of it as the convolution of the complex conjugate of f(-x ,- y ) and f(x , y ) . if f(x , y ) has a fourier transform f(s , r ) , then its autocorrelation function has the transform absolute value squared of f(s , r ) and has no phase information(wiener - khinchin theorem ) . in the data analysis , the autocorrelation is evaluated by taking the fourier transform of the reverse complex conjugate of the image and the fourier transform of the image , then taking the real part of the inverse fourier transform of the product of these two fourier transforms . the three dimensional plot of the autocorrelation looks like a mountain with a narrow spike in the middle . the threshholded ( 25 - 30% ) contours(5 ) in the x - y plane of the different files show more variation and different symmetries . figure 3(upper left ) file 117 , halfway up the first peak , looks like it has four 2-fold axes . figure 3(upper right ) file 122 , at the top of the first peak , may have one 4-fold and two 2-fold axes . figure 3(lower left ) file 133 , the minimum between the peaks , appears to have one 4-fold axis and two 2-fold axes . figure 3(lower right ) file 145 , top of the second peak , seems to have four 2-fold axes . in fig . 4(top ) , the height of the peak verus file number is plotted and is in agreement with fig . the correlation length is related to the inverse of the width . in fig . 4(bottom ) , the inverse of the peak center is plotted versus file number ; the spacing of the magnetic domains is inverse to the peak center . in all cases the curves follow fig . 1(left ) more or less ; the last one does nt follow the second peak very well . it is possible to see azimuthal intensity variation through data anaylsis of ccd images . because of the odd behavior of the second peak and the differences in the autocorrelation functions(reflecting differences in electronic configurations ) of the first and second peaks , there appears to be a difference in the nature of the first and second peaks . 0 t. o. mentes , c. sanchez - hanke and c. c. kao , _ * 9 * , 90 ( 2002 ) . f. yakhou , a. letoublon , f. livet , m. de boissieu and f. bley , _ j. magn . magn . mater _ * 233 * , 119 ( 2001 ) . a. rahmim , m.sc . thesis , univ . of british columbia , vancouver , 2001 .	data analysis of the ccd files from x ray magnetic resonance scattering linearly polarized in transmission geometry produces information about the radial and azimuthal intensities . in a series of measurements of increasing photon energies trends in data are analyzed .
magnetic fields are known to play a significant role in determining the equilibrium state of the plasma in the solar atmosphere and solar wind ( e.g. , parker 1975 , 1991 ; narain and ulmschneider , 1990 , 1996 ; priest 1999 ) . much of the magnetic flux in the `` quiet '' photosphere seems to be concentrated into small ( 100200 km ) intergranular flux tubes . the physical processes that heat the chromosphere and corona have not yet been identified definitively , but there is little doubt that magnetohydrodynamic ( mhd ) effects are prevalent . even many proposed nonmagnetic mechanisms depend on the underlying properties of the magnetically structured atmosphere . the outflowing solar wind is fed by open magnetic flux tubes , and many mhd processes have been proposed to deposit heat and momentum at locations ranging from the extended corona to interplanetary space . the continually evolving convection below the photosphere gives rise to a wide spectrum of mhd fluctuations in the magnetic atmosphere and wind . the propagation of waves through the solar atmosphere has been studied for more than a half century ( alfvn 1947 ; schwarzschild 1948 ; biermann 1948 ) . although compressible ( e.g. , acoustic and magnetoacoustic ) mhd waves are likely to be dynamically and energetically important in some regions of the atmosphere , it is the mainly incompressible alfvn mode that has been believed for many years to be dominant in regions that are _ open to the heliosphere _ ( e.g. , osterbrock 1961 ; kuperus , ionson , & spicer 1981 ) . indeed , the mhd fluctuations measured by spacecraft in the solar wind have a strongly alfvnic character ( belcher & davis 1971 ; hollweg 1975 ; tu & marsch 1995 ; goldstein , roberts , & matthaeus 1995 ) . even though much has been learned about the generation , propagation , reflection , and damping of alfvn waves in the solar atmosphere , most earlier studies have focused on only a finite range of heights and treated the interactions with regions above or below as boundary conditions . this necessarily involved the approximation that the relevant phenomena are mainly local , i.e. , that they do not depend on the conditions very far away from the region being modeled . there are circumstances , however , for which this approximation breaks down . for example , the properties of reflecting alfvn waves with long periods ( i.e. , of order 1 day ) in the solar wind depend formally on the conditions `` at infinity , '' since they behave asymptotically as standing waves ( e.g. , heinemann & olbert 1980 ) . in this paper we construct a comprehensive model of the radially evolving properties of alfvnic fluctuations in a representative open magnetic region of the solar atmosphere and fast solar wind . the model takes account of nonlocal interactions by tracing the waves from their origin as transverse flux - tube oscillations in the photosphere all the way out to the interplanetary medium ( truncated for convenience at 4 au ) . this is done with the smallest possible number of free parameters . there are two overall aims of this paper : 1 . we wish to understand better the global `` energy budget '' of alfvn waves , including relative amplitudes of inward and outward propagating waves , along the open flux tubes that feed the solar wind . 2 . in order to determine how mhd turbulence contributes to the heating of the extended solar corona , we need to pin down precisely how alfvn waves provide the natural preconditions for a turbulent cascade . the second aim above was motivated by a recent study of the small - scale dissipation of mhd turbulence in the extended corona ( cranmer & van ballegooijen 2003 ) . this kinetic dissipation depends strongly on how the turbulence is excited at its largest scales , and in this paper we attempt to put the `` outer scale '' wave dynamics on firmer footing so that the resulting `` inner scale '' can be better understood . the work described by this paper builds on prior studies by hollweg ( 1973 , 1978a , 1981 , 1990 ) , heinemann & olbert ( 1980 ) , spruit ( 1981 , 1984 ) , an et al . ( 1990 ) , barkhudarov ( 1991 ) , velli ( 1993 ) , lou ( 1993 , 1994 ) , lou & rosner ( 1994 ) , macgregor & charbonneau ( 1994 ) , kudoh & shibata ( 1999 ) , matthaeus et al . ( 1999 ) , hasan et al . ( 2003 ) , and many others . one unique aspect of this paper is that the photospheric spectrum of transverse fluctuations which constrains the alfvn wave amplitudes at all larger radii is specified directly from detailed observations of magnetic bright point ( mbp ) motions and is not ( as in many other models ) given as an arbitrary boundary amplitude . however , a complete physical description of the fluctuations in the photosphere ( e.g. , the relative inward / outward amplitudes and the kinetic / magnetic energy partition ) is obtained only after the fully nonlocal wave reflection has been computed for all radii . another way this work differs from many previous models is that the expansion and successive merging of flux tubes on granular and supergranular scales is described using a two - dimensional model of a magnetic network element in the inhomogeneous solar atmosphere . despite the attempted comprehensiveness of this model , we needed to make three specific approximations in order to render the calculations tractable . these approximations are summarized here , but they are also discussed further below and justified for certain regimes of applicability . first , we ignore all effects of compressible fluctuations ( e.g. , acoustic waves ; fast and slow magnetosonic waves ) despite their importance in understanding observed intensity oscillations and chromospheric heating . this is done in order to fairly assess the relative importance of the incompressible alfvn mode before resorting to more involved models . second , the alfvn wave model is mainly linear , which limits its applicability in regions where the wave amplitudes become large in comparison to background field strengths and characteristic speeds . ( some effects of nonlinearity are examined , though , in 6 . ) third , we do not model explicitly the back - reaction of the waves on the mean properties of the solar atmosphere and solar wind . we do , however , compute quantities such as the wave pressure acceleration and turbulent heating rate for use in future models of this back - reaction . the remainder of this paper is organized as follows . in 2 we give an overview of the physical processes to be incorporated in the wave models together with a `` cartoon '' description of the steady - state magnetic field topology . in 3 we describe the adopted steady - state ( i.e. , zero - order ) plasma conditions in detail . the specification of the photospheric frequency spectrum of transverse fluctuations is given in 4 , and the wave equations to be solved are given in solutions of these equations , including some with various prescriptions for damping , are presented in 6 . we conclude with a summary of major results ( 7 ) and a discussion of remaining issues ( 8) . appendix a contains supplementary equations describing analytic solutions of the kink - mode wave equations for isolated flux tubes in an isothermal atmosphere . appendix b compares various published formalisms for the non - wkb transport equations for alfvn waves in an accelerating wind . appendix c summarizes the properties of the fully developed anisotropic mhd turbulence spectrum discussed by , e.g. , cranmer & van ballegooijen ( 2003 ) . figure 1 illustrates the basic magnetic field geometry that we believe is representative of flux tubes that feed the high - speed solar wind . a key feature of the adopted configuration is the successive merging of strong - field magnetic flux tubes between granules ( figure 1a ) and supergranules ( figure 1b ) . on the largest scales , figure 1c shows the more or less axisymmetric magnetic field that is characteristic of solar minimum ( using the field model of banaszkiewicz , axford , & mckenzie 1998 ) , but nearly all of the work presented in this paper can also be applied straightforwardly to open - field regions at other phases of the solar cycle . we assert that most of the plasma that eventually becomes the time - steady solar wind originates in intergranular magnetic flux tubes known observationally as g - band bright points , network bright points , or in groups as `` solar filigree . '' this assertion is seemingly uncontroversial ( knowing what we do about solar magnetic fields ) , though it is seldom stated . adopting a ripening convention in nomenclature , we refer to these 100200 km size photospheric features as _ magnetic bright points _ bright points in the chromosphere ( rutten & uitenbroek 1991 ) or the still larger x - ray bright points in the low corona ( e.g. , golub et al . 1977 ; parnell 2002 ) . ] high - resolution observations reveal the presence of mbps in the dark lanes between granules , and these features are associated with regions of strong ( 12 kg ) magnetic field believed to be contained within nearly vertical flux tubes ( e.g. , sheeley 1967 ; dunn & zirker 1973 ; muller 1983 , 1985 ; piddington 1978 ; rabin 1992 ; solanki 1993 ; berger & title 2001 ) . there is increasing evidence for magnetic structures on even smaller scales than 100200 km , but we leave the study of these structures to future work ( see , e.g. , stein & nordlund 2002 ; snchez almeida , emonet , & cattaneo 2003 ; trujillo bueno , shchukina , & asensio ramos 2004 ) . as magnetic flux rises stochastically from the convection zone to the photosphere ( e.g. , priest , heyvaerts , & title 2002 ) , field lines are simultaneously jostled horizontally by fluid motions on granular ( 12 mm ) scales . magnetic flux is concentrated into thin tubes by some combination of `` flux expulsion '' from the upflowing granule centers to the downflowing lanes ( parker 1963 ) , the rapid evacuation of these downflowing superadiabatic regions ( i.e. , convective collapse ; parker 1978 ; spruit 1979 ) , and enhanced radiative cooling leading to thermal relaxation ( snchez almeida 2001 ) . mbps are observed frequently to merge with neighboring flux elements and to spontaneously fragment into several pieces ( e.g. , berger et al . once formed , mbps continue to be shaken back and forth by the underlying convective motions ( kulsrud 1955 ; osterbrock 1961 ; van ballegooijen 1986 ; huang , musielak , & ulmschneider 1995 ) , which results in various kinds of wavelike fluctuations describable in the `` thin - tube '' mhd limit ( spruit 1981 , 1982 , 1984 ; see further references in 5.1 below ) . oscillatory motions can also be induced by impulsive reconnection events ( e.g. , moore et al . 1991 ) or , conversely , the random wave trains may display observational time signatures that could be misinterpreted as small - scale flaring ( moriyasu et al . 2004 ) . there is a great variety of mhd wave activity expected and observed in the inhomogeneous solar atmosphere . in addition to isolated mbp fluctuations , there is ambient acoustic wave energy excited by convection , shock steepening in the chromosphere and transition region , and both driven and free oscillations in sunspots and coronal loops ( see recent reviews by axford et al . 1999 ; roberts 2000 ; bogdan 2000 ; hirzberger 2003 ) . a significant fraction of the sun s magnetic flux may also be distributed _ outside _ the mbps , network , and active regions ( e.g. , schrijver & title 2003 ) , and acoustic waves in `` field - free '' regions may in fact be magnetoacoustic . damping of mhd waves and turbulence has been a key ingredient in many proposed models of chromospheric and coronal heating . our main focus , though , is on the incompressible waves that eventually escape from the atmosphere into the solar wind . magnetic flux tubes rooted in mbps undergo both transverse ( kink - mode ) and longitudinal ( sausage - mode ) oscillations that can propagate upward from the photosphere . because the strong - field flux tubes are in horizontal pressure equilibrium with the surrounding weak - field material , they have a lower density and thus are susceptible to buoyancy effects and evanescence for long enough periods . like acoustic waves , the compressible longitudinal modes steepen into shocks and damp over a few scale heights , while the incompressible kink modes can propagate into the corona relatively undamped . nonlinear effects can lead to mode conversion between kink and longitudinal modes ( e.g. , ulmschneider , zhringer , & musielak 1991 ) . somewhere in the low chromosphere , the thin flux tubes are believed to expand laterally to the point where they merge with one another into a homogeneous network field distribution ( spruit 1984 ; pneuman , solanki , & stenflo 1986 ; gu et al . 1997 ) . at this point , the thin - tube description breaks down and standard mhd wave theory becomes more applicable ( i.e. , kink modes become transverse alfvn waves ) . the merged network flux bundles have horizontal scale lengths of 2 to 6 mm and are probably maintained by large - scale convective flows that push the field to the edges of supergranular cells . at a larger height still below the chromosphere - corona transition region the network magnetic field expands laterally to 1030 mm scales and is thought to merge again into a large - scale `` canopy '' ( e.g. , kopp & kuperus 1968 ; gabriel 1976 ; giovanelli 1980 ; anzer & galloway 1983 ; dowdy , rabin , & moore 1986 ) . the spatial scale of the canopy is set by the typical distance between network flux bundles ( i.e. , the size of supergranulation cells ) in the chromosphere . observational evidence for preferential wind acceleration in the rapidly expanding network `` funnels '' is growing ( rottman , orrall , & klimchuk 1982 ; hassler et al . 1999 ; peter & judge 1999 ; aiouaz et al . 2004 ; see also martnez - galarce et al . 2003 ) but is still not definitive ( e.g. , dupree , penn , & jones 1996 ) . as waves propagate upward into the corona , the radially varying alfvn speed allows for gradual linear reflection ( ferraro & plumpton 1958 ) . the transition region can also act a sharp `` reflection barrier '' to alfvn waves with wavelengths exceeding the local scale length of the alfvn speed in that thin zone ( see 6 ) . it is thus possible for a time - steady superposition of upward and downward propagating alfvn waves to be maintained ( e.g. , hollweg 1981 , 1984 ) . strong reflection is not necessarily an impediment to there being a substantial upward wave flux ; one merely needs more power in the upward modes than in the downward modes . somewhere in the solar atmosphere the mhd fluctuations become turbulent , but it is unclear whether the turbulent cascade becomes energetically important in the photosphere ( petrovay 2001 ) , in the chromosphere and transition region ( chae , schhle , & lemaire 1998 ) , or in the corona ( e.g. , dmitruk & matthaeus 2003 ) . in the extended corona , the high - speed solar wind begins to accelerate supersonically above a heliocentric distance of 2 to 3 solar radii ( @xmath0 ) . in the radially inhomogeneous wind , the dissipationless propagation of mhd waves does work on the mean fluid and provides an added wave - pressure acceleration ( e.g. , belcher 1971 ; jacques 1977 ; leer , holzer , & fl 1982 ) . the presence of the wind also modifies how waves propagate , and above the alfvn critical point where the wind speed equals the local alfvn speed at about 10 @xmath0both the inward and outward modes are advected outward with the wind . large coronal holes are the most probable source regions for the fast solar wind ( for a review of observations and theoretical models , see cranmer 2002 ) . some flux tubes in coronal holes have a higher density than the surrounding open - field plasma ; these `` polar plumes '' seem to trace out the superradial expansion of the merged - canopy magnetic field in the corona ( e.g. , deforest , lamy , & llebaria 2001 ) . in addition to alfvn waves , there is some evidence for both fast - mode and slow - mode magnetosonic waves in the corona ( ofman , nakariakov , & deforest 1999 ; nakariakov et al . 2004 ) , but they have been observed only in relatively confined regions such as loops and plumes . fast and slow modes are believed to be more attenuated by collisional damping processes than alfvn waves before they reach the corona . however , once some fraction of the energy flux of alfvn ( and possibly fast - mode ) waves escapes into the solar wind , the classical transport theory of collisional damping begins to break down and collisionless wave - particle interactions should dominate the damping ( e.g. , hollweg & isenberg 2002 ; marsch , axford , & mckenzie 2003 ; cranmer 2000 , 2001 , 2002 , 2004 ; and many references therein ) . studies of wave - particle resonance damping have seen a recent resurgence of interest because of their potential importance in producing the preferential ion heating and acceleration seen both in the extended corona ( kohl et al . 1997 , 1998 , 1999 ) and in situ ( marsch 1999 ) . as alfvn waves propagate into interplanetary space , their velocity and magnetic - field amplitudes grow to nonlinear magnitudes ( i.e. , @xmath1 becomes of order unity ) . higher order ponderomotive effects begin to dominate the wave propagation equations ( lau & siregar 1996 ) and collisionless wave - wave resonances can create additional mode coupling and damping ( lee & vlk 1973 ; lacombe & mangeney 1980 ; lou 1993 ) . at some point it may even be inappropriate to use a `` wavelike '' paradigm to discuss the increasingly turbulent fluctuations ( e.g. , goldstein et al . in the outer heliosphere , the description of magnetic flux tubes as relatively closed systems implicit in the above discussion breaks down as well , since processes such as stream - stream interactions , interstellar pickup ion injection , and cosmic ray transport increasingly dominate the physics . in this paper we purposefully examine only a subset of the many processes listed above . this is done mainly to keep the modeling tractable , but it also helps clarify the extent to which the _ specific wave modes that we study _ can account for various observations . models of linear waves depend sensitively on the assumed background zero - order plasma properties ( e.g. , density , flow speed , magnetic field strength , flux tube geometry ) . in this section we describe the empirically constrained time - steady plasma along the central axis of a radially pointed ( but superradially expanding ) magnetic flux tube , from the photosphere to a distance of 4 au . the empirical description consists of two parts : a two - dimensional magnetostatic model of thin mbp flux tubes that expand into the supergranular network canopy ( 3.1 ) and a one - dimensional analytic continuation of the plasma parameters in the extended corona , assumed here to be along the axis of symmetry of a polar coronal hole at solar minimum ( 3.2 ) . we first develop a two - dimensional numerical model of a supergranular network element as a collection of thin flux tubes . the gas pressure in the atmosphere decreases with increasing height , causing a lateral expansion of the flux tubes . neighboring mbp flux tubes within the network element merge into a monolithic structure at some height @xmath2 . ( heights @xmath3 are measured from the optical - depth - unity photosphere ; radii @xmath4 are measured from sun center . ) for a thin , isothermal flux tube in pressure equilibrium with its surroundings , the interior magnetic field strength varies with height as @xmath5 , where @xmath6 is the pressure scale height ( e.g. , spruit 1981 , 1984 ) . thus we determine the so - called _ merging height _ by solving for @xmath7 where @xmath8 is the field strength inside the flux tube at the base of the photosphere ( @xmath9 ) and @xmath10 is the average flux density in the network patch . observations of the mbp flux - tube field strength range between 1000 and 2000 g , and the network - averaged field strength varies between about 20 and 300 g ( e.g. , gabriel 1976 ; giovanelli 1980 ) . in order to determine a representative value for the merging height we use @xmath11 km , @xmath12 g , and @xmath13 g to obtain @xmath14 km , a height in the low chromosphere ( see also pneuman et al . 1986 ) . above the merging height the network element consists of a single thick flux tube that further expands with height . the outer edge of this tube forms a magnetic canopy that overlies the neighboring supergranular cells . a second merging occurs when neighboring network elements come together at a `` canopy height '' @xmath15 above the supergranular cell centers ; we set @xmath16 mm ( see also hasan et al . figure 2 shows the three - part structure of the network element : 1 . the region below the merging height ( @xmath17 mm ) is described as a collection of thin flux tubes embedded in a field - free medium . the field strength @xmath18 is assumed to be the same for all flux tubes , and their cross sections and other plasma properties are consistent with the thin - tube approximation . 2 . between heights of 0.6 and 1 mm , the merged network flux element expands laterally to the edges of the supergranular cell , which overlies a field - free cell - center chromosphere . 3 . between 1 and 12 mm , the `` fully merged '' magnetic field fills the supergranular cell volume and expands primarily in the vertical direction . in the remainder of this paper , the term `` merging height '' refers specifically to the merging of the thin flux tubes at @xmath19 mm . the total magnetic flux @xmath20 of the network element is constrained empirically to be @xmath21 mx . the mbp field strengths at the photosphere and at the merging height are 1430 g and 120.4 g , respectively . the upper , monolithic part of the magnetic structure ( @xmath22 ) is assumed to be cylindrically symmetric and is described by a modified version of the magnetostatic flux tube model of hasan et al . ( 2003 ) . in the present case , the flux tube is contained within a cylinder of transverse radius @xmath23 mm , which simulates the effect of neighboring network elements . the internal and external gas pressures are taken from semiempirical val / fal models by vernazza , avrett , & loeser ( 1981 ) and fontenla , avrett , & loeser ( 1990 , 1991 , 1993 , 2002 ) . the magnetic field components @xmath24 and @xmath25 in cylindrical geometry ( using @xmath26 as the perpendicular distance from the central flux tube axis ) are computed by varying the shapes of the field lines until a minimum - energy state is obtained ( for details see hasan et al . 2003 ) . the horizontal distribution of @xmath27 at the merging height is adjusted iteratively such that the magnitude of the field @xmath28 is independent of @xmath26 . this is needed for consistency with the constant field strength of the thin flux tubes just below the merging height . figure 2 shows the final iterated magnetic structure . the cross sections of the flux tubes below the merging height ( @xmath29 mm ) were computed by matching @xmath30 and @xmath31 at the merging height and assuming that @xmath30 is independent of height within each flux tube ( different tubes have different @xmath30 ) . the transverse radius of the `` network patch '' in the photosphere is about 3 mm . we can estimate the number @xmath32 of mbps inside the approximately circular patch , but we note that the wave analysis below does not depend on the value of @xmath32 . from the conservation of magnetic flux , we know that the filling factor of mbp flux tubes at the photosphere is given by @xmath33 . observationally , the transverse radii of mbps at the photosphere are 50 to 100 km . the number of flux tubes required to fill an area that is 8% of the total patch area ( @xmath34 , with @xmath35 mm ) ranges between @xmath36 ( for the upper limit on mbp radius ) and @xmath37 ( for the lower limit ) . we can also define a mean distance between nearest neighbors @xmath38 by dividing up up the network patch into @xmath32 equal - area circles and defining @xmath38 as twice their radii . thus , @xmath39 , which gives values of 350 and 700 km for the above limiting cases . before continuing , it is worthwhile to ask the question : `` why is it not possible to extend the ` monolithic ' flux - tube model all the way down to the photosphere ? '' observationally , the photosphere is far from monolithic in its distribution of magnetic fields . granulation causes both the formation of thin mbp flux tubes and their spreading out ( via random walk ) . the larger - scale supergranulation pushes the magnetic elements back together into network lanes and vertices . this competition between mbp spreading and convergence leads to a _ dynamical equilibrium _ described by the different assumptions applied above and below the merging height . a more complete model must contain physics that naturally captures this equilibrium state , but hopefully without the abrupt transition assumed here at @xmath2 . here we describe the plasma parameters along the central axis of the modeled network element at heights ranging from 12 mm ( the top of the magnetostatic model ) to interplanetary space . we adopt a slightly modified version of the polar magnetic field configuration of banaszkiewicz et al . ( 1998 ) : @xmath40}\ ] ] where @xmath41 is now defined as the radial component of the field along the axis of symmetry , @xmath42 , and the above description applies only for @xmath43 , with @xmath44 ( i.e. , the top height [ 12 mm ] of the magnetostatic grid ) . the model above uses the same current - sheet constant ( @xmath45 ) used by banaszkiewicz et al . ( 1998 ) , but with a 5% modification to their preferred quadrupole constant . we also add an exponential correction term ( which drops rapidly to nearly zero for @xmath46 ) to ensure that the value and slope of @xmath41 match those of the magnetostatic model at @xmath47 . figure 3a plots the product @xmath48 versus height both below and above @xmath47 . this quantity is proportional to the inverse of the traditional superradial divergence factor @xmath49 , and it shows the outer monopolar expansion region ( @xmath50 ) as a constant . for comparison we also plot the analytic functions used in the funnel models of hackenberg , marsch , & mann ( 2000 ) and li ( 2003 ) . for the radial dependence of the electron density , we use a function motivated by fits to white - light polarization brightness measurements in the extended corona : @xmath51 which applies only for @xmath43 . the overall scale set by the inverse square term was adjusted to match in situ density measurements at 1 au ( i.e. , @xmath52 ) . the middle three terms above were adjusted to produce agreement with measurements by , e.g. , guhathakurta & holzer ( 1994 ) , fisher & guhathakurta ( 1995 ) , doyle , teriaca , & banerjee ( 1999 ) , esser & sasselov ( 1999 ) , and figure 10 of lie - svendsen , hansteen , & leer ( 2003 ) . the last term was set to match the value and slope of the magnetostatic model density at @xmath47 . figure 3b plots the hydrogen number density @xmath53 , computed assuming a helium - to - hydrogen number density ratio of 0.05 ( i.e. , @xmath54 , with the total mass density @xmath55 given by @xmath56 ) . figure 3c shows several velocity quantities along the central axis of the flux tube . the hydrogen outflow speed @xmath57 was computed by mass flux conservation ( i.e. , @xmath58 constant ) , with the mass loss rate set by _ ulysses _ measurements in interplanetary space ( goldstein et al . 1996 ) . at 1 au , the product @xmath59 @xmath60 s@xmath61 , and thus we compute @xmath62 km s@xmath61 at 1 au . at an infinite distance , @xmath57 approaches a constant value of 781.9 km s@xmath61 . the alfvn speed , defined as @xmath63 is nearly constant below the merging height ( and would be precisely constant for an isothermal atmosphere ) and rises to two successive maxima : 2530 km s@xmath61 at @xmath64 , and 2890 km s@xmath61 at @xmath65 . the alfvn speed drops to 31.3 km s@xmath61 at 1 au and decreases nearly exactly as @xmath66 after that . the alfvn critical point , where @xmath67 , is at @xmath68 . the lower boundary condition for our model of alfvnic fluctuations is the power spectrum of transverse mbp motions in the photosphere . the dynamical behavior of g - band bright points has been studied observationally by a number of groups ( see , e.g. , muller et al . 1994 ; berger & title 1996 ; van ballegooijen et al . 1998 ; berger et al . 1998 ; krishnakumar & venkatakrishnan 1999 ; nisenson et al . 2003 ) . in this section we present an empirical description of mbp dynamics as a linear superposition of two types of motion : ( 1 ) the `` random walk '' undertaken by isolated flux tubes , and ( 2 ) a series of rapid `` jumps '' that occur when individual flux tubes merge , fragment , or reconnect with surrounding magnetic field . the general procedure for specifying the power spectrum of horizontal mbp kinetic energy ( as a function of frequency ) is illustrated in figure 4 . the primary measured quantity is a time series of discrete position measurements for the mbps which can be differentiated to obtain horizontal velocity components @xmath69 and @xmath70 , with the @xmath3 direction being normal to the solar surface . ( observations support the rational assumption that there is no preferred global direction on granular scales , so below we discuss just the @xmath71 component and assume that @xmath72 is statistically equivalent . ) the timescale dependence of mbp fluctuations is encapsulated in the velocity autocorrelation function , which we define for a time sequence of @xmath32 measured velocities @xmath73 as @xmath74 for the time index @xmath75 between 1 and @xmath32 and an arbitrary delay time represented by index @xmath76 . this expression is the discrete version of the more general definition @xmath77 for delay time @xmath78 . the unidirectional power spectrum @xmath79 ( i.e. , @xmath80 per unit interval of frequency @xmath81 ) is the fourier transform of the autocorrelation function , with @xmath82 ( e.g. , the wiener - khinchin theorem ; see also van ballegooijen et al . note that @xmath83 is defined for both positive and negative @xmath81 . for the present applications , all functions are symmetric about zero frequency , and we will thus consider only positive frequencies ( simultaneously doubling the normalization of @xmath79 to conserve total energy ) . in general , we specify the kinetic energy power spectrum @xmath84 , which is defined in such a way as to integrate to the total kinetic energy density @xmath85 in transverse motions : @xmath86 with @xmath87 \ , = \ , 2\rho p_{x } ( \omega ) \,\,\ , . \label{eq : pxy}\ ] ] the factors of two inside the square brackets take account of the negative frequencies , and the last expression assumes @xmath88 . in the analysis below we derive power spectrum components for the two assumed phases of the mbp motion : the random walk ( subscript @xmath89 ) of isolated flux tubes , and the occasional discrete jumps ( subscript @xmath90 ) caused by merging , fragmenting , or reconnecting . specifying the power separately for these two phases is not the ideal solution , but it is all that can be done at present . ideally , the proper observational procedure would be to determine the _ complete _ time series @xmath69 for the walk and jump phases taken together , then compute the autocorrelation function and total power spectrum consistently . unfortunately , because mbps fragment or merge during the jump phases , it is extremely difficult to `` follow '' a single feature during these times to determine the complete time series . the random - walk component of mbps , during the times they exist as separate entities , was studied by van ballegooijen et al . ( 1998 ) and nisenson et al . these observations yielded the result that the discretely derived autocorrelation functions can be fit well by lorentzian functions ( see figure 4e ) , with @xmath91 and the more precise observations of nisenson et al . ( 2003 ) gave values of @xmath92 km@xmath93 s@xmath94 and @xmath95 s. we adopt these values in the models below . thus , the walk - component of the kinetic energy spectrum is given by @xmath96 this is plotted in figure 4 g , but note that all power spectra are plotted as the product @xmath97 because this denotes the power per decade of frequency ( i.e. , the energy density per unit @xmath98 ) . maxima in this quantity highlight the frequencies that contribute most to the total wave energy . the impulsive `` jump '' phase of mbp motions is described by choudhuri , dikpati , & banerjee ( 1993 ) , berger et al . ( 1998 ) , hasan , kalkofen , & van ballegooijen ( 2000 ) , and others . the potential for rapid transitions in the locations of thin flux tubes is also indicated in empirical models of the quasi - equilibrium evolution of granular magnetic fields ( van ballegooijen & hasan 2003 ) , in which slow motions of separatrix surfaces in the photosphere are amplified at larger heights due to flux - tube expansion . we model an impulsive mbp event as narrow gaussian enhancement in @xmath69 centered on an arbitrary @xmath99 . a series of these events is assumed to occur with a mean time interval @xmath100 between events . captures the essential nature of these jumps and their contribution to the energy spectrum , a more accurate way of modeling them would be to sample from an empirically derived `` waiting - time '' probability distribution . ] the gaussian jump , with velocity amplitude @xmath101 and @xmath102 half - life @xmath103 , is described by @xmath104 between times @xmath105 and @xmath106 , and the limit @xmath107 in eq . ( [ eq : autoint ] ) is not taken . the autocorrelation function is thus @xmath108 ( see figure 4f ) and the resulting kinetic energy spectrum is given by @xmath109 ( see figure 4h ) . in the models below we adopt @xmath110 s and @xmath111 s , which are consistent with the observations of berger et al . the velocity amplitude of the jump @xmath101 is known with much less certainty because it represents the `` tail '' of the observed distribution of speeds . berger & title ( 1996 ) found speeds up to 5 km s@xmath61 , so this seems to be a rough upper limit to @xmath101 . this is our only true free parameter . because the walk and jump phases of mbp motion seem to be statistically uncorrelated , we compute the full kinetic energy spectrum @xmath84 as the simple sum of @xmath112 and @xmath113 ( see figure 4i ) . it is then possible to integrate this spectrum over frequency ( see eq . [ [ eq : ukearly ] ] ) to obtain the transverse velocity variance of mbp motions in the photosphere : @xmath114 \ , \mbox{km}^{2 } \ , \mbox{s}^{-2 } \label{eq : dvphot}\end{aligned}\ ] ] where the latter expression uses the values adopted above for @xmath115 , @xmath103 , and @xmath100 . note that even for a large impulsive velocity of @xmath116 km s@xmath61 , the root - mean - squared velocity is significantly smaller than this value ( @xmath117 km s@xmath61 ) because the jumps occur infrequently . finally , we also compute the _ total _ energy spectrum @xmath118 ( i.e. , including contributions from both kinetic and magnetic energy ) at the photosphere using the analytic relations for isothermal thin flux tubes given in appendix a ( specifically , eq . [ [ eq : isopart ] ] ) . for very high frequencies the kinetic and magnetic energy components are in equipartition , and @xmath118 is just twice @xmath84 . for very low frequencies the kink - mode waves are evanescent and the physically realistic solution contains much more kinetic energy than magnetic energy ( thus , @xmath119 ) . the resulting total energy spectrum , plotted as a solid line in figure 4i , is essentially our lower boundary condition on the amplitudes of alfvn waves of various frequencies . we describe how this information is folded into the global solutions in 5.3 . we model the transverse wave properties in the open magnetic regions described in 3 as purely linear perturbations to the assumed zero - order background plasma state . the basic mhd equations that need to be solved are the mass and momentum conservation equations and the magnetic induction equation , given by @xmath120 @xmath121 \label{eq : momcon0}\ ] ] @xmath122 where the velocity @xmath123 and magnetic field @xmath124 are not yet separated into zero - order and first - order parts , @xmath125 is the gas pressure , and @xmath126 is the gravitational acceleration . we apply these equations in two regions with very different physics : 1 . _ below the merging height _ ( 5.1 ) we model the waves as incompressible lagrangian perturbations of the central axis of a thin , strong - field flux tube that expands superradially and is surrounded by a field - free region . in this region we assume the outflow speed of the solar wind is negligibly small . in addition to the properties within the thin tube , we also specify the density external to the tube @xmath127 in the field - free region . 2 . _ above the merging height _ ( 5.2 ) we model the waves as incompressible eulerian perturbations filling the volume of the expanding flux tube , which is assumed to be surrounded by similar tubes . all background properties , including a nonzero solar wind speed , vary only in the radial direction . for both regions we transform the mhd equations above into wave equations with an assumed @xmath128 time dependence . the equations are then solved `` monochromatically '' for a grid of frequencies , also assuming that in each solution the frequency remains constant as a function of height . we do not make the wkb ( i.e. , eikonal ) approximation that wavelengths are small compared to background scale lengths ; indeed we do not even need to define the concept of wavelength because the complete spatial oscillation pattern is computed numerically as a function of height . the solutions for individual frequencies are subsequently assembled into a full radially varying power spectrum , normalized by the empirically derived power spectrum at the photosphere ( 4 ) . the following subsections present the specific equations that we solve in the two regions outlined above . below the merging height , the mbp flux tubes are shaken transversely and kink - mode waves are excited ( see also wilson 1979 ; spruit 1981 ; ulmschneider et al . 1991 ) . for incompressible perturbations about the equilibrium state , the density @xmath55 is a zero - order quantity , the velocity @xmath123 is horizontal and a first - order quantity , and the magnetic field @xmath124 has a zero - order vertical component and a first - order horizontal component . following the motion of the thin tubes , we write the lagrangian forms of the momentum and induction equations as follows : @xmath129 @xmath130 where the advective derivative @xmath131 follows the motion of the tube s central axis . above we have assumed that @xmath132 as a statement of mass conservation for incompressible flows . we write the scalar horizontal perturbations in velocity and magnetic field as @xmath133 and @xmath134 , and we implicitly assume linear polarization of the waves in a single transverse dimension . in many ways , though , the equations to be derived are degenerate with toroidal alfvn waves at larger heights ( e.g. , heinemann & olbert 1980 ) and our assumption should not appreciably limit the generality of the results . ( other polarization modes have been studied by , e.g. , spruit 1982 , 1984 ; lou 1993 ; roberts 2000 ; noble , musielak , & ulmschneider 2003 ; ruderman 2003 . ) in order to derive the wave equations we note three aspects of thin tubes in the solar atmosphere : 1 . for an oscillating flux tube , the direction perpendicular to its instantaneous axis will , generally , be inclined with respect to the radial direction away from the sun . the components of the above equations parallel to the flux tube axis are uninteresting , and will be considered to be `` solved '' by the zero - order background state described in 3 . we assume transverse total pressure balance between the flux tube and the surrounding field - free region , and that the field - free region is in simple hydrostatic equilibrium . thus , @xmath135 3 . the motion of the tube induces motions in the surrounding field - free region , which in turn must have a back - reaction on the tube s original motion . spruit ( 1981 ) took this into account by increasing the apparent inertia of the tube . thus , for the perpendicular component of eq . ( [ eq : momconlag ] ) , the factor of @xmath55 in the advection term on the left - hand side must be replaced by @xmath136 . this is equivalent to the assumption that the tube `` carries along '' a parcel of the surrounding fluid with equal kinetic energy density to that of the tube itself . osin , volin , & ulmschneider ( 1999 ) reviewed different approaches to the inclusion of this back - reaction effect and found that spruit s ( 1981 ) approach adequately describes the physics for transverse oscillations of a nearly vertical flux tube . with the above considerations , spruit ( 1981 ) showed that the perpendicular component of eq . ( [ eq : momconlag ] ) can be expressed as a linearized wave equation @xmath137 where the magnitude of the gravitational acceleration @xmath138 is nearly constant over the heights we consider . in spruit s ( 1981 ) ideal limit that the flux tubes are completely isolated from one another , the density quantities introduced above are defined as @xmath139 @xmath140 and @xmath141 is a modified kink - mode phase speed that takes the surrounding inertia into account . note that in the lagrangian picture , @xmath133 is the time derivative of the horizontal displacement @xmath142 . the first term on the right - hand - side of eq . ( [ eq : momspruit ] ) is due to the _ buoyancy _ of the low - density flux tube . the second term on the right - hand - side is the magnetic tension restoring force due to the curvature of the flux tube . the lagrangian induction equation is given simply by @xmath143 analytic solutions to the above equations are possible when the radial derivative terms have constant coefficients ; this occurs in an exponential isothermal atmosphere ( see appendix a for details ) . traditionally , the solutions to these equations display evanescence for frequencies below a critical cutoff value @xmath144 . for the adopted background state at the photosphere ( @xmath9 ) , @xmath145 and @xmath146 km s@xmath61 , and appendix a gives an analytic estimate for the corresponding critical period ( @xmath147 ) of about 12.5 minutes . this period is significantly longer than the acoustic cutoff frequency of 36 minutes in the photosphere and chromosphere . thus , the kink mode has been suspected for several decades as being able to transport more convective wave energy up to the corona than acoustic waves . before discussing our numerical solutions to eqs . ( [ eq : momspruit])([eq : indspruit ] ) between @xmath9 and @xmath2 , one simplification assumed above must be reexamined . just below the merging height , the flux tubes can not be considered truly isolated from one another . the enhanced inertia assumed by spruit ( 1981 ) assumed that the surrounding fluid carried along by a given tube is all field - free , but near the merging height this is not the case . spruit ( 1982 ) gave the equations for selected kink - mode wave properties for the general case where the surrounding medium has a nonzero field strength , but here we deal with the encroachment of neighboring flux tubes in a simpler manner . in the above equations we express @xmath148 and @xmath149 by @xmath150 @xmath151 where @xmath152 is essentially a statistical filling factor of neighbor tubes within the near - tube region that gets carried along with a tube s oscillatory motion . the isolated tube limit is @xmath153 , but at the merging height ( and above ) , @xmath154 . the specific form of the modifications above were constrained by the need for both @xmath155 and for the buoyancy term in eq . ( [ eq : momspruit ] ) to vanish in the `` merged '' limit of @xmath154 . the reduction of @xmath156 also reduces the critical frequency for evanescence ( see eq . [ [ eq : isoomc ] ] ) , and at the merging height @xmath157 . we derive @xmath158 by using magnetic flux conservation together with the assumption that the overall area subtended by the full network patch is constant between the photosphere and the merging height . thus , @xmath159 where @xmath160 g. because @xmath41 increases rapidly with decreasing height below @xmath2 , @xmath152 rapidly drops from 1 ( at @xmath161 km ) to 0.5 at @xmath162 km , then more slowly down to 0.08 at @xmath163 . in figure 3 we plot the resulting values of @xmath148 ( as a hydrogen number density ) and @xmath164 below the merging height . the wave equation ( eq . [ [ eq : momspruit ] ] ) is simplified by assuming an @xmath128 time dependence with a real frequency @xmath81 . we solve numerically for the radial dependence of @xmath133 by expressing the second - order wave equation as two coupled first - order ordinary differential equations in @xmath133 and @xmath165 ( with both quantities assumed to be complex ) , and using fourth - order runge - kutta integration ( e.g. , press et al . the upper boundary conditions at @xmath2 are specified by the solutions of the wave equations _ above _ @xmath2 ( see next section ) , and the numerical integration proceeds from @xmath2 down to the photosphere . the coupling of solutions below and above the merging height is discussed in 5.3 . above the merging height , the transverse incompressible fluctuations act as mhd alfvn waves , and our solution procedure largely follows that of heinemann & olbert ( 1980 ) and barkhudarov ( 1991 ) . formally , the mechanisms of wkb theory can be extended to describe linear wave reflection ( e.g. , hollweg 1990 ) , but we follow the usual non - wkb formalism in order to solve for the radial dependence of the transmitted and reflected wave properties from the merging height all the way to a distance of 4 au . the monochromatic non - wkb wave transport equations are derived from the mass , momentum , and induction equations listed above ( eqs . [ [ eq : mascon0]][[eq : induct0 ] ] ) in the limits that all background quantities vary only in radius and that the velocity and magnetic field perturbations are perpendicular to the zero - order field direction . khabibrakhmanov & summers ( 1997 ) showed how to treat general vector operations in a superradially expanding flux tube . we express the wave properties in terms of elsasser ( 1950 ) variables , defined here as @xmath166 ( see also tu & marsch 1995 ) , with @xmath167 representing outward propagating waves and @xmath168 representing inward propagating waves . in terms of these variables , the incompressible first - order equations are expressed as two coupled transport equations : @xmath169 where the ( signed ) scale heights are @xmath170 and @xmath171 . these equations are valid for superradial divergence and for all values of the zero - order outflow speed @xmath57 . note that linear reflection arises because of the @xmath172 term on the right - hand side . in appendix b we discuss additional details about these equations and how they are equivalent to other versions given in earlier work . to our knowledge , the `` compact '' form of eq . ( [ eq : zpm ] ) has not been recognized fully , although the expressions of heinemann & olbert ( 1980 ) and khabibrakhmanov & summers ( 1997 ) were closely related . the above form of eq . ( [ eq : zpm ] ) is particularly useful in showing the large - scale density dependence of the wave amplitude in two limiting cases when the outward propagating waves are dominant ( i.e. , @xmath173 ) . near the sun , where @xmath174 , we can approximate the lower - sign version of the above equation as @xmath175 and thus @xmath167 is proportional to @xmath176 as predicted by wkb theory ( see also moran 2001 ) . similarly , far from the sun , where @xmath177 , @xmath167 is seen to be proportional to @xmath178 . more general properties of the non - wkb solutions of eq . ( [ eq : zpm ] ) are discussed in 6 , and also by macgregor & charbonneau ( 1994 ) and krogulec et al . ( 1994 ) . as with the solutions below the merging height , we assume an oscillatory time dependence of the elsasser variables ( i.e. , @xmath128 with a real , constant frequency ) , and we solve for their radial dependence numerically . we restrict our solutions to positive frequencies and note that taking the negative of a given frequency produces solutions to eq . ( [ eq : zpm ] ) that are the complex conjugates of the analogous solutions obtained with @xmath179 . thus , the radial evolution of physical quantities ( i.e. , real wave amplitudes ) is unaffected by the sign of @xmath81 . the existence of the alfvn critical point complicates the numerical solution of eq . ( [ eq : zpm ] ) , but we follow the general solution procedure outlined by barkhudarov ( 1991 ) . once the oscillatory time dependence has been assumed , the complex amplitudes @xmath168 and @xmath167 are expressed as the products of real amplitudes and phase factors of unit magnitude . there are then four ordinary differential equations for these quantities that are solved first at the alfvn critical point @xmath180 ( analytically , using certain physicality constraints such as the requirement that the outward wave energy always exceed the inward wave energy ) , then we integrate numerically using the fourth - order runge - kutta method both upwards and downwards from @xmath180 . some of barkhudarov s ( 1991 ) expressions had to be modified to take account of the superradial divergence of the magnetic field . the linear amplitudes @xmath181 are specified only to within an arbitrary normalization factor , although both the phase factors and all _ relative _ quantities ( such as ratios of elsasser amplitudes at different radii ) do not depend on this normalization . our baseline model consists of a grid of 300 frequencies , evenly spaced in @xmath182 over five orders of magnitude with periods ranging from 3 to 300,000 s. the discrete radial grid extends from just above the photosphere ( @xmath183 km ) to 4 au ( @xmath184 ) and contains 11809 grid points distributed mainly logarithmically , but with some regions ( like the transition region ) sampled more finely . in the photosphere , chromosphere , and low corona ( below @xmath47 ) , the relative grid separation @xmath185 is 0.00064 . within the most rapidly changing 20 km of the transition region , @xmath185 is decreased to 0.0001 . in the extended corona and solar wind , @xmath185 is made to gradually increase to 0.016 at the outer boundary . the runge - kutta algorithm also has an adaptive stepsize that subdivides the above grid zones until a relative accuracy of @xmath186 is achieved in the integration variables . this degree of accuracy is needed to follow the oscillatory behavior of the waves . the non - wkb wave equations are solved first for each frequency in the upper corona / wind region as described in 5.2 , and the resulting elsasser variables at the lower boundary ( i.e. , the merging height ) are used to compute the complex values of @xmath133 and @xmath134 at that height . these quantities , still with an arbitrary degree of normalization , are used as the upper boundary conditions for the numerical solution of the flux - tube wave equations given in 5.1 . the induction equation , eq . ( [ eq : indspruit ] ) , is used only to convert the boundary condition for @xmath134 into a condition for @xmath165 . we make the assumption that all of the alfvnic wave energy in the upper region is converted smoothly into kink - mode wave energy in the lower region . after the transport equations are solved in both regions , the photospheric mbp power spectrum ( derived in 4 ) is used to renormalize the wave power quantities at all heights . to show how this is done we must first define the kinetic and magnetic energy densities for each monochromatic model : @xmath187 and also the supplementary quantities @xmath188 where below the merging height we use @xmath148 for the above densities . the energy densities defined above do not depend on time @xmath189 since they contain products of @xmath128 and its complex conjugate . these definitions also ensure that the total fluctuation energy for each frequency satisfies @xmath190 . in the simple wkb theory ( i.e. , all outward propagating waves ) , @xmath191 , @xmath192 , and @xmath193 , and the departure from wkb theory can be assessed roughly by departures from these ideal energy identities . in 4 we derived the total power @xmath194 at the photosphere . the numerical integrations described above gave us the various @xmath195 quantities on a two - dimensional grid in @xmath81 and radius ( with the energies for each frequency known up to an arbitrary multiplicative constant ) . we thus compute `` renormalized '' power spectra on the discrete grid as @xmath196 once the power spectra have been defined we can then specify various frequency - averaged energy densities . in general , we define @xmath197 with analogous definitions for @xmath85 , @xmath198 , and @xmath199 ( see also eq . [ [ eq : ukearly ] ] ) . for ease of interpretation we also define the frequency - averaged velocity , magnetic , and elsasser amplitudes as the quantities in angle brackets below : @xmath200 @xmath201 the procedures outlined above resulted in a large amount of numerical data ( @xmath202350 megabytes ) describing the behavior of non - wkb kink - mode and alfvn waves as a function of frequency and radius . in this section we attempt to distill and present the salient results in three gradual steps : ( 1 ) in figures 57 we present frequency - dependent wave properties that have not yet been renormalized to the photospheric power spectrum ( and thus are plotted as dimensionless ratios ) , ( 2 ) in figure 8 we show the total power spectrum as a function of frequency for selected radii , and ( 3 ) in figures 918 we present various frequency - integrated quantities that depend on the photospheric power normalization . in order to determine to what degree the waves in various regions depart from ideal wkb theory , we show in figure 5 the outward propagating wave action flux @xmath203 ( defined in appendix b ) for a selection of periods , each normalized to the value of @xmath203 in the photosphere . for fluctuations having wavelengths so small that the local plasma parameters are approximately constant over several wavelengths , wave action is conserved and the quantity @xmath49 is constant . note from figure 5 that in the extended corona and solar wind ( e.g. , @xmath204 ) alfvn waves with periods shorter than a few hours obey wave action conservation , but for periods exceeding @xmath20210 hours this breaks down . note that for _ all _ computed periods the wave action is not conserved below the transition region ( @xmath205 ) ; this region acts as a sufficiently sharp `` barrier '' to induce significant reflection in the chromosphere ( see also wentzel 1978 ; hollweg 1981 ; campos & gil 1999 ) . in much of the previous work on non - wkb alfvn wave reflection , the departures from wkb theory have been characterized as a function of frequency . for frequencies exceeding a critical value @xmath206 , the resulting wavelengths are so short that the wave propagates as if it were in a homogeneous medium and there is negligible reflection . for frequencies lower than @xmath206 there is significant reflection , and as @xmath207 the oscillation approaches the properties of a standing wave with equal inward and outward power . in a magnetized hydrostatic atmosphere , the critical frequency is given by the local value of @xmath208 , or more precisely for arbitrary expansion factors , @xmath209 ( e.g. , ferraro & plumpton 1958 ; an et al . 1990 ) . in a supersonic wind , though , the radial dependence of the alfvn speed is no longer the dominant factor in determining how much reflection takes place . at large distances from the sun , heinemann & olbert ( 1980 ) and barkhudarov ( 1991 ) showed that @xmath206 is given approximately by @xmath210 , where @xmath211 is the asymptotic outflow speed and @xmath180 is the alfvn radius . ( we use this expression as a fiducial definition of @xmath206 , but we also note that it neglects several order - unity correction factors that depend on the flow tube geometry . ) for the zero - order solar wind model defined in 3.2 we find an associated critical period @xmath212 of about 30 hours , or 1.25 days . figure 6 shows contours of the alfvn ratio i.e . , the ratio of kinetic to magnetic energy density @xmath213 as a function of period and height . for ideal mhd alfvn waves in a homogeneous medium this ratio is 1 and the waves are in energy equipartition ( waln 1944 ) . figure 6 can be broken up into four `` quadrants '' that have the following limiting properties : 1 . _ upper left : _ for long periods below the transition region , most of the wave energy is kinetic with only a negligible magnetic energy density . this is consistent with the predictions of kink - mode wave theory for the so - called `` shallow '' evanescent solution . in appendix a we discuss several reasons why the solar atmosphere is suspected to naturally prefer this solution . lower left : _ for short periods below the transition region , the alfvn ratio rapidly fluctuates above and below 1 . this occurs because the wavelengths are small compared to the photospheric and chromospheric scale heights , but they are large compared to the scale heights in the transition region . thus significant reflection occurred and there is a superposition of upward and downward waves . the fluctuations are well described by standing waves having fixed nodes in @xmath133 and @xmath134 ( hollweg 1981 , 1984 ) . if one could separate the solutions into the component upward and downward waves , their individual alfvn ratios would be @xmath2021 as predicted by the kink - mode theory in appendix a. ( this can also be seen roughly by averaging over several nodes . ) _ upper right : _ for long periods above the transition region , most of the wave energy is magnetic as was found by other non - wkb solar wind models ( e.g. , heinemann & olbert 1980 ) . these long periods correspond to `` quasi - static '' motions of the field lines and have a similarity to the motions invoked in dc theories of coronal heating ( e.g. , kuperus et al . 1981 ; van ballegooijen 1986 ; milano , gmez , & martens 1997 ) . lower right : _ for short periods above the transition region , the plasma appears homogeneous to the relatively small - wavelength fluctuations and the ideal mhd equipartition holds . figure 7 shows contours of the ratio @xmath214 which can be thought of as an effective local reflection coefficient . the ideal wkb theory corresponds to a ratio of zero ( i.e. , all outward propagation ) . there is significant reflection below the transition region , with a maximum value of this ratio of 0.99825 at the photosphere for a period of @xmath20240 minutes . as discussed in 5.2 , we chose regularity conditions at the alfvn critical point that ensured more outward than inward power at all heights , and thus @xmath214 must always be less than one . below the transition region , the behavior of @xmath214 versus frequency resembles analytic solutions that take account of the chromospheric stratification and the strong reflection at the transition region ( e.g. , hollweg 1978a ; schwartz , cally , & bel 1984 ) . because of the finite size of the atmosphere below the transition region , a mild resonance structure is seen in the amount of reflected wave flux ( not apparent in figure 7 ) . the resonances are not as sharp as in the isothermal models of schwartz et al . ( 1984 ) , though , because the nonzero temperature gradient results in a `` smearing '' of the preferred resonance frequencies . above the transition region , the solutions plotted in figure 7 are very similar to those of heinemann & olbert ( 1980 ) and barkhudarov ( 1991 ) ; there is significant reflection when @xmath215 and much less reflection for higher frequencies ( lower periods ) . for frequencies above @xmath206which we see below encompasses most of the dominant part of the power spectrum we can use the analytic regularity conditions presented in 4 of barkhudarov ( 1991 ) to estimate the ratio @xmath214 at the alfvn critical point : @xmath216 by comparing to the numerical results in figure 7 we verify that this relation is accurate for periods less than @xmath202200 minutes . it may be possible to also estimate the radial dependence of this ratio using similar analytic formulae . we defer this to future work , but note that this would be useful to models of coronal heating via turbulent cascade ( see 6.2.2 ) . figure 8 shows the result of convolving the above results with the photospheric power spectrum derived in 4 . in this figure we plot the total power @xmath217 for selected heights , and with a specific choice for the parameter @xmath101 of 3 km s@xmath61 ( see below ) . the amount of power loss between the photosphere and the merging height is consistent with the analytic predictions of isothermal kink - mode theory . for high frequencies ( i.e. , periods less than 35 minutes ) , @xmath118 decreases from @xmath9 to @xmath2 by a factor of 15 to 20 ; this is approximately the level of the decrease of the background magnetic field between those heights , implying that the propagating kink - mode relation @xmath218 applies . for low frequencies ( presumably evanescent ) , @xmath118 decreases by a factor of 190 , which is very nearly equal to the drop in density from @xmath9 to @xmath2 . this is consistent with the prediction of nearly constant @xmath133 for the shallow evanescent solution discussed in appendix a ( and thus @xmath219 ) . the somewhat irregular structure that develops in the spectrum between the photosphere and the merging height ( and is passively advected outwards above @xmath2 ) is not numerical noise . because of the nonisothermal temperature structure of the low chromosphere and the use of a radially varying filling factor @xmath152 , the properties of the waves below the merging height depend on frequency in a more complicated way than in the ideal case described in appendix a. if the upward and downward propagating waves had identical strengths and dispersive properties , the standing - wave nodes in @xmath220 and @xmath221 would cancel out exactly in @xmath222 . however , @xmath222 exhibits a weak nodal structure because of incomplete cancellation and thus contributes to the irregularity in the power spectrum ( see also schwartz et al . 1984 ) . as in figure 4 , we plot in figure 8 the product @xmath223 versus wave period to more clearly show the periods that make the greatest contribution to the overall wave energy . we quantify this concept by defining the averaged , or first - moment frequency as @xmath224 we plot this quantity versus height in figure 14 below , but here we just note that the effective period @xmath225 is about 3.5 minutes in the photosphere , then it decreases to about 1.8 minutes above the merging height . periods of a few minutes are natural to expect in the photosphere and chromosphere , and possibly in the corona as well ( e.g. , chashei et al . 1999 ) . however , it is reasonable to ask if these periods are expected to dominate in interplanetary space . in situ spacecraft generally measure fluctuations in velocity , density , and the magnetic field with the most power at periods of a few hours ( e.g. , goldstein et al . however , a spacecraft sitting still in the ecliptic plane at 1 au would see our model network flux tube rotate past in about 4 to 5 hours , with its component flux tubes ( each originating in a different mbp ) rotating by in substantially less than one hour . thus it is possible that if the in situ power spectra actually do sample a `` fossil '' spectrum from the sun , the dominant periods of order 1 hour could be due to the passage of many uncorrelated flux bundles past the spacecraft , and not the waves within any one flux bundle . ( the ideal test would be to see if a spacecraft corotating with the solar rotation period measures a significantly different fluctuation spectrum than has already been observed . ) figure 9 , a key condensation of results for this paper , plots the frequency - integrated velocity amplitude @xmath226 as a function of height for a selection of @xmath101 parameters and compares it to several different measurements of wave amplitudes . we discuss each set of measurements briefly , by number , below : 1 . the dotted line shows a best - fit height dependence for the microturbulence needed to match photospheric and chromospheric line widths in the semiempirical val / fal models ( e. avrett , 2003 , personal communication ; see also fontenla et al . 1993 , 2002 ) . the filled circles show similar `` nonthermal '' line - broadening velocities measured ( on the solar disk ) in the transition region and low corona by the sumer ( solar ultraviolet measurements of emitted radiation ; wilhelm et al . 1995 ) instrument on _ soho _ ( the _ solar and heliospheric observatory _ ) . the height of each point has been estimated by matching the height - dependence of temperature in the val / fal model with the assumed formation temperatures of the ions that correspond to each point ( chae et al . the crosses show nonthermal velocities inferred by sumer measurements made above the solar limb ( banerjee et al . 1998 ) . off - limb doppler broadening observations are better suited for measuring the properties of transverse alfvn waves than observations on the solar disk . 4 . the gray region shows lower and upper limits on the nonthermal velocity ( esser et al . 1999 ) as computed from off - limb measurements made by the uvcs ( ultraviolet coronagraph spectrometer ) instrument on _ soho _ ( kohl et al . 1995 , 1997 ) . the stars show early measurements ( armstrong & woo 1981 ) of the random wavelike component of the solar wind velocity from interplanetary scintillation observations of radio signals passing through the corona , with the advecting diffraction pattern being measured by more than one receiver . the error bars show a more recent determination of velocity fluctuations specifically transverse to the radial direction from radio scintillations ( canals et al . 2002 ) in the fast solar wind . 7 . the _ helios _ and _ ulysses _ probes measured time - averaged elsasser amplitudes that we converted to a representative velocity amplitude using eqs . ( [ eq : ukub])([eq : upum ] ) and assuming @xmath227 in the heliosphere . the data from both spacecraft were summarized by bavassano , pietropaolo , & bruno ( 2000 ) . measured line - of - sight ( one - dimensional ) velocities have been multiplied by @xmath228 to take account of equivalent fluctuations in both transverse dimensions ( e.g. , eq . [ [ eq : pxy ] ] ) . because measurements ( 1 ) and ( 2 ) above refer to motions mainly in the radial direction , we do not expect them to correspond to transverse kink - mode or alfvn waves ; they are plotted mainly for heuristic comparison . note , though , that in figure 9 we also plot ( for the @xmath229 km s@xmath61 case ) a dashed curve that shows the frequency - averaged amplitude of the magnetic fluctuations in velocity units : @xmath230 which for ideal mhd equipartition would be exactly equal to @xmath226 . it is possibly a coincidence that this quantity so closely matches the chae et al . ( 1998 ) observations , but this agreement may contain information about the _ mode coupling _ between transverse and longitudinal waves in the transition region . the three choices for the mbp - jump velocity amplitude @xmath101 used in figure 9 are 0 , 3 , and 6 km s@xmath61 . the middle value seems to best match the off - limb nonthermal line broadening measurements , and we will use this as a baseline value in most subsequent calculations and plots . note that the in situ measurements fall well below all of the reasonable choices for @xmath101 ( even the lower limit assuming no jumps whatsoever ) and they exhibit a steeper gradient than the undamped models . the heliospheric `` deficit '' of wave power , compared to most prior assumptions about the wave power in the solar atmosphere , is well known ( see , e.g. , roberts 1989 ; lou 1993 ; mancuso & spangler 1999 ) . in 6.2.2 we propose a potential solution of these discrepancies based on a particular theory of turbulent wave damping in the solar wind . figure 10 shows additional details about the height dependence of the frequency - integrated wave properties . the ratio @xmath231 shows the region of strong departure from energy equipartition around the transition region , and that kinetic energy exceeds magnetic energy by a slight amount even down to the photosphere . the dimensionless magnetic amplitude @xmath232 is less then 1 over most of the computed range of heights ( thus justifying the linear approximation ) , but exceeds 1 above @xmath233 . nonlinear calculations ( e.g. , lau & siregar 1996 ) find that @xmath232 does not grow so high in the heliosphere and may saturate at values close to 1 . figure 11 shows the height dependence of the two elsasser amplitudes . above the transition region , the frequency - integrated outward propagating amplitude @xmath234 is much larger than the inward amplitude @xmath235 . below the transition region , the amplitudes approach a constant ratio @xmath236 . we also plot the height dependence of @xmath237 that would be expected for ideal wkb wave action conservation ( i.e. , @xmath238 ) and we scale it to the computed value of @xmath234 at 4 au . the wkb value departs slightly from the non - wkb value between the transition region and a height of @xmath2020.01 @xmath0 and is substantially smaller below the transition region where there is strong inward wave power . the undamped linear curves all disagree markedly with the in situ measurements of the elsasser amplitudes ( see , however , 6.2.2 ) . lastly , in figure 12 we plot the frequency - integrated energy flux density @xmath239 for the baseline @xmath229 km s@xmath61 case . below the merging height this quantity applies to the energy flux carried upward by the mbp kink modes , and above the merging height it applies to the volume - filling alfvn waves . the wave flux is defined generally as @xmath240 which is based on the monochromatic definition given by heinemann & olbert ( 1980 ) . for outward propagating waves obeying wave action conservation , one can write @xmath241 . in the lowest layers of the atmosphere one can also ignore the factors above that depend on @xmath57 and thus the net flux depends mainly on the difference between @xmath242 and @xmath243 . because of the strong reflection below the transition region there is strong cancellation ; i.e. , the magnitude of the purely outward wave flux @xmath244 exceeds the net flux by a factor of about 20 . this means that @xmath20295% of the kink - mode wave energy below the transition region is `` trapped , '' with only 5% escaping . analysis of a series of models having a range of @xmath101 values yields a potentially useful explicit expression for the photospheric wave flux in the mbps . recalling that the random - walk and impulsive parts of the power spectrum are assumed to be linearly independent ( e.g. , eq . [ [ eq : dvphot ] ] ) , we have fit the numerically computed photospheric fluxes with the following function : @xmath245 with less than 0.1% uncertainty in the fit ( mainly due to roundoff error in the above expression having only three significant figures ) . thus , for the reasonable range of @xmath246 between 0 and 6 km s@xmath61 , @xmath247 ranges between about 10@xmath248 and 10@xmath249 erg @xmath60 s@xmath61 ( see also musielak & ulmschneider 2001 ) . note that for any given value of @xmath101 the jump - term in the above expression has a greater relative contribution to the total flux than the corresponding jump - term in eq . ( [ eq : dvphot ] ) has to the photospheric velocity variance . this occurs because the bulk of the power in the jump motions is at higher frequencies than the walk motions and is less susceptible to evanescent decay . the flux is a frequency - integrated quantity that incorporates this aspect of the solutions , whereas the velocity variance does not . we also plot in figure 12 several other flux quantities . the `` granule - averaged '' wave flux is given by the quantity @xmath250 ( for the flux tube filling factor @xmath152 , see eq . [ [ eq : phidef ] ] ) , and it differs from @xmath239 itself only below the merging height . the quantity @xmath250 spreads the wave energy out evenly within the modeled network patch , and thus is a more relevant quantity to compare with observations and predictions of wave fluxes that arise in the network from the convection . note that the photospheric value of @xmath251 erg @xmath60 s@xmath61 is of similar magnitude as the _ acoustic _ wave flux of @xmath252 erg @xmath60 s@xmath61 predicted by musielak et al . we also plot a `` supergranule - averaged '' flux @xmath253 , where @xmath254 is a filling factor for the large - scale funnel / canopy magnetic structure shown in figure 2 . ( we define @xmath254 only below the top height [ 12 mm ] of the magnetostatic grid , as the ratio of the field strength at that top height [ 11.8 g ] to the field strength along the central axis of the flux tube at lower heights . ) the quantity @xmath253 is what one would use in order to compute the total wave power ( in erg s@xmath61 ) integrated over the entire coronal hole . in this section we discuss two separate mechanisms that have been proposed to damp alfvn waves in the solar atmosphere and solar wind : viscosity ( 6.2.1 ) and mhd turbulent cascade ( 6.2.2 ) . for clarity we do not treat these mechanisms together , though in a completely self - consistent model all damping mechanisms should be included and allowed to interact with one another . the first damping mechanisms proposed for mhd waves in the solar corona were collisional in nature ( alfvn 1947 ; osterbrock 1961 ) . in the high - density plasma near the sun , mhd waves can be damped by viscosity , thermal conductivity , electrical resistivity ( i.e. , joule / ohmic heating ) , and ion - neutral friction . for alfvn and fast - mode waves propagating parallel to the magnetic field , though , the dominant collisional dissipation channel in the corona is believed to be proton viscosity . here we estimate the viscous damping expected for the background plasma conditions described above and find it to be negligible . we compute a collisional damping length @xmath255 as a function of height @xmath3 , and there should be appreciable damping only where @xmath256 . generally , the product of the damping length and a linear damping rate @xmath257 ( i.e. , the imaginary part of the frequency ) is the wave group velocity @xmath258 in the inertial frame , given here for dispersionless alfvn waves as @xmath259 . for viscous damping , @xmath260 , where @xmath261 is the proton kinematic viscosity and @xmath262 is the radial wavenumber ( van de hulst 1951 ; osterbrock 1961 ) . thus , for alfvn waves propagating upward and parallel to the background magnetic field , this leads to the general expression @xmath263 ( tu 1984 ; whang 1980 , 1997 ) . in a strongly collisional plasma the kinematic viscosity is given by the classical braginskii ( 1965 ) formalism ( see also hollweg 1986a ) . in the collisionless solar wind , though , there is no clear prescription for the effective viscosity . we thus define the proton kinematic viscosity phenomenologically as @xmath264 where @xmath265 is the proton most - probable speed , @xmath266 is boltzmann s constant , @xmath267 is the proton temperature , @xmath268 is the proton mass , and @xmath269 is an effective viscous timescale . we adopt a fiducial proton temperature law for this analysis that comes from the semiempirical val / fal model ( used for the magnetostatic flux tube model ) below @xmath270 , and we use @xmath271 above 1.0172 @xmath0 . this expression reasonably reproduces the proton temperatures measured by _ helios _ and _ ulysses _ in the highest - speed solar wind ( tu , freeman & lopez 1989 ; goldstein et al . 1996 ) and it has a peak in the extended corona of @xmath2022.2 mk in general agreement with uvcs/_soho _ ly@xmath272 measurements ( e.g. , kohl et al . 1998 ; esser et al . 1999 ) . for strong collisional coupling , the effective viscous time scale @xmath269 is given by the mean time between collisions ( braginskii 1965 ) , @xmath273 where @xmath274 is the proton / electron charge and @xmath275 is the coulomb logarithm , taken here to be 21 . the above timescale applies to the viscous damping of motions _ along _ the magnetic field . for shear motions transverse to the field , though , the appropriate viscous time ( even in the limit of strong collisions ) is reduced . the off - diagonal braginskii coefficients in the stress tensor are obtained approximately by dividing @xmath276 by dimensionless factors that take account of the magnetic field . thus , @xmath277 where @xmath76 is either 1 or 2 , and @xmath278 is the proton larmor frequency @xmath279 . the @xmath280 terms are roughly analogous to the direct , hall , and pedersen components of the mhd conductivity in an ionized plasma . in the low - density collisionless limit , the classical @xmath281 case can not apply because it would imply the viscosity becomes infinitely large as the mean time between collisions becomes infinite ( the `` molasses '' limit ) . williams ( 1995 ) argued essentially that the general viscosity in a collisional or collisionless plasma is found by taking the shorter of either the @xmath281 or the @xmath282 timescales . we assert , though , that the @xmath283 case seems to be the most appropriate for the viscous damping of transverse alfvn waves in the collisionless solar wind . consider the viscosity as an effective diffusion coefficient ( @xmath284 ) describing scattering events with mean - free - path length and time scales of @xmath285 and @xmath189 , respectively . the actual energy losses arise from the interparticle collisions with timescale @xmath286 , but spatially the _ transverse _ structure would be dominated by features with sizes of order the proton thermal larmor radius , and thus @xmath287 . for these values the viscosity reproduces the @xmath283 case . figure 13 shows the radial dependence of the linear damping length @xmath255 for the three cases , @xmath281 , 1 , and 2 , computed for a wave period of 1 minute . for heights below about 0.3 @xmath0 above the photosphere , it is unlikely that any damping would take place because all three cases have damping lengths much longer than the local height . this mid - corona distance ( @xmath288 to 2 @xmath0 ) is approximately where collisions start to become unimportant in coupling together electron , proton , and heavy ion plasma properties ( e.g. , cranmer , field , & kohl 1999 ) . above this height , then , the appropriate damping length should transition to either the @xmath282 or the @xmath283 case , both of which are substantially longer than the local height in the corona . in the far solar wind , we believe the @xmath283 case is the most realistic , and it always remains several orders of magnitude larger than the local height . thus , our preliminary conclusion is that for both waves near the peak of the power spectrum ( periods of a one to a few minutes ) , and for longer periods , linear viscous damping is _ unimportant _ as a significant attenuation mechanism for alfvn waves in the corona and fast solar wind . for periods much shorter than 1 minute , though , this kind of damping could be important and may have already been responsible for the sharp drop in power inferred from photospheric mbp motions between 0.1 and 1 minute . the second type of wave damping we consider is turbulent dissipation ; i.e. , a nonlinear cascade of energy from large to small scales that terminates in an irreversible conversion of wave energy into heat . the need to include some kind of nonlinear damping or saturation can be seen in figure 10 , where above a distance of @xmath20220 @xmath0 the magnetic fluctuation amplitude begins to exceed the background field strength ( in opposition to in situ observations ) and the linear assumption breaks down . addressing the full problem of anisotropic mhd turbulence in the solar corona and solar wind , though , is well beyond the scope of this paper ( see , though , cranmer & van ballegooijen 2003 for a summary of many related issues ) . here we include only one specific aspect of this theory : a phenomenological damping rate that depends on the properties of the largest scales in the turbulence and not on the details of the cascade . the presence of both outward and inward propagating alfvn waves is a necessary prerequisite for the nonlinear coupling to higher wavenumber that drives the cascade ( see , e.g. , matthaeus et al . 1999 ; dmitruk , milano , & matthaeus 2001 ) . any expression for the turbulent energy decay must then depend on both the @xmath167 and @xmath168 elsasser amplitudes , and thus the inclusion of turbulence introduces further coupling between these two modes . a straightforward phenomenological form for the nonlinear transport has been suggested by zhou & matthaeus ( 1990 ) from the standpoint of `` reduced mhd '' ( rmhd ; see also strauss 1976 ; montgomery 1982 ) . effectively , the transport equation is the same as eq . ( [ eq : zpm ] ) , but with the following term added to the right - hand side : @xmath289 where @xmath290 is a transverse length scale representing an effective correlation length of the turbulence i.e . , a similarity length scale associated with the energy transport from large to small eddies . ( it may be imprecise to refer to this length scale as an `` outer scale , '' but for simplicity we treat this as a synonym . ) we assume that @xmath290 scales with the transverse width of the open flux tube ; i.e. , that it remains proportional to @xmath291 . specifically , we normalize @xmath290 by specifying its value at the merging height @xmath2 . we also define the constant @xmath292 as the ratio of @xmath290 at the merging height to the transverse radius of the network flux bundle at that height : 3 mm ( see figure 2 ) . equivalently , we can express @xmath293 if @xmath41 is measured in gauss ( see also hollweg 1986b ) . in order to know how best to implement the nonlinear term given above , we need to understand over what range of heights this term acts to appreciably damp the waves . we do not expect a strong rmhd cascade to develop in the region below the merging height because despite the stochastic nature of the fluctuation spectrum the flux tubes do not strongly interact with one another . for heights above @xmath2 we can compare the effective damping length @xmath290 to the local height @xmath3 above the photosphere . for @xmath294 , we find that @xmath295 only above a height of @xmath296 . it makes sense that the chromosphere and transition region are not expected to undergo much damping from this rmhd mechanism ; there seems to be insufficient time for the turbulent cascade to develop in such small volumes . dmitruk & matthaeus ( 2003 ) discussed a hierarchy of timescales that should be satisfied in order for this turbulent damping process to be strong . a simplified version of this hierarchy is as follows : @xmath297 where @xmath298 is a nonlinear outer - scale eddy cascade time , @xmath299 is a timescale for alfvn wave reflection , and @xmath300 is representative of the main driving period of the waves . in figure 14 we compare several of these quantities to see where these conditions are met . in figure 14 we plot a representative reflection time @xmath299 which we define as @xmath301 . this is probably of the same order of magnitude as the dmitruk & matthaeus ( 2003 ) `` alfvn wave crossing time '' @xmath302 , since the latter can be defined as @xmath303 , or the wave travel time over a representative alfvn - speed scale height @xmath304 . thus , @xmath305 . for reference we also show the total wave travel time from the photosphere to a given height , defined as @xmath306 we equate the driving period @xmath300 to the spectrum - integrated first - moment period @xmath225 ( see eq . [ [ eq : omavg ] ] ) and we plot its value for the baseline @xmath229 km s@xmath61 model . interestingly , @xmath307 only _ below _ @xmath308 , and the hierarchy is not satisfied at large distances where the damping length is relatively small . note , though , that the demand in the above hierarchy for @xmath300 to be the largest of the three timescales comes only from the well - known non - wkb result that long periods are the easiest to reflect , and thus they provide the most turbulent mixing between inward and outward modes . however , the amount of reflection in our model is a known quantity ( see , e.g. , figures 7 and 11 ) , and it is irrelevant whether it could have been maximized with longer - period waves . therefore we do not need to consider the last inequality in eq . ( [ eq : tineq ] ) as a precondition for turbulent damping . the key piece of the dmitruk & matthaeus ( 2003 ) timescale hierarchy is that the nonlinear driving time @xmath298 must be short compared to the reflection time @xmath299 . only when this condition is satisfied can the turbulent cascade develop to the point where the wave damping can occur efficiently . in appendix c we describe one way of using a model for the anisotropic turbulent power spectrum to derive @xmath298 ( see also oughton , dmitruk , & matthaeus 2004 ) . this timescale depends inversely on the outer - scale length @xmath290 , and in figure 14 we plot the radial dependence of @xmath298 for two extreme values of the normalizing constant @xmath292 . an upper limit to the outer - scale length is given by @xmath294 ( i.e. , that the `` stirring '' takes place on the spatial scale of the entire network element ) . a reasonable lower limit for @xmath290 at the merging height is the radius of the smallest expected mbp flux tube . at the merging height , the flux tubes have expanded to fill the volume of the network patch , and thus the mbp radius is given by half the nearest - neighbor distance @xmath38 . for the lower - limit value of @xmath309 km ( see 3.1 ) we obtain a lower limit of @xmath310 . we find in figure 14 that @xmath299 begins to exceed @xmath298 somewhere above a height of 0.1 to 1 @xmath0 . we thus solve the nonlinearly modified wave transport equation only above 0.1 @xmath0 . this restriction allows us to make a useful simplifying assumption . as seen in figure 11 , above this height there is only relatively weak inward wave power ; i.e. , @xmath311 . we solve a modified time - steady wave action conservation equation in the limit of pure outward propagation . note also that turbulence does not act in a straightforward `` monochromatic '' way , as was assumed in the constant - frequency linear solutions to eq . ( [ eq : zpm ] ) . we thus ignore any frequency ( and wavenumber ) transport and compute the overall effects of the turbulent damping on the frequency - integrated wave energy ( see , however , verdini & velli 2003 ) . thus , we solve the following modified spectrum - averaged wave action conservation equation in the limits of no time dependence and dominant outward propagation : @xmath312 \ , = \ , - \frac{\langle z_{- } \rangle^{2 } \langle z_{+ } \rangle}{l_{\perp } } \label{eq : zpmdamp0}\ ] ] ( see appendix b ) . we use the linear solution for @xmath235 obtained above in order to solve for the nonlinearly damped value of @xmath234 : @xmath313 l_{\perp } ( r ' ) } \right\ } \,\ , , \label{eq : zpmdamp}\ ] ] where all quantities with subscript ` 0 ' are given at the effective lower boundary @xmath314 . this is not a completely self - consistent method of obtaining the damped wave power , since no back - reaction on @xmath235 is computed , but it gives an adequate order - of - magnitude result for the total power in the regions dominated by outward propagating waves . figure 15 shows the result of integrating eq . ( [ eq : zpmdamp ] ) for three choices of @xmath292 ( 0.1 , 0.35 , and 1 ) , as well as the solution for no damping ( @xmath315 ) . we plot these solutions as velocity amplitudes in order to facilitate comparison with the observational data , which are also plotted as in figure 9 . for each value of @xmath292 , the resulting damped @xmath242 energy density was added to @xmath243 ( which remains undamped ) to obtain @xmath316 , and this total energy density was split into kinetic and magnetic parts assuming that the linear undamped ratio @xmath317 remains the same for the damped models . the resulting @xmath85 was then used to compute @xmath226 as in eq . ( [ eq : ukub ] ) . the value of @xmath292 that produces the best agreement with the in situ measurements is approximately 0.35 , implying a transverse outer scale at the merging height of about 1100 km . this value seems appropriate and consistent for motions excited between granules of the same spatial scale . if @xmath292 were much smaller than this value , it would imply that the turbulence was dominated by internal motions in the mbp flux tubes , which we essentially ignore in the thin - tube models . furthermore , note from figure 15 that for @xmath318 there is so much damping that the approximation @xmath319 breaks down above @xmath202100 @xmath0 and we would have needed to solve a more complicated set of damping equations . the observational constraints which imply @xmath320 allow us to avoid both of the above consistency problems that arise for smaller values . properties of the @xmath320 solutions were also plotted in several other figures in order to compare with the linear results . figure 10 shows that the damped magnetic amplitude ratio @xmath232 is smaller in interplanetary space than the undamped ratio . for the damped model , @xmath232 exceeds 1 only above 135 @xmath0 and does not exceed a value of 1.7 within the computed range of distances ( see also lau & sireger 1996 ) . figure 11 shows the better agreement between the damped elsasser amplitude @xmath234 and the measured outward elsasser energy . the inward amplitude @xmath235 still does not agree with the measured inward elsasser energy ( see also hollweg 1990 ) , but note that the linear value was not modified in the damping calculations done here . full simulations of the turbulence would probably see an enhancement of @xmath235 and possibly even an approach to inward / outward energy equipartition ( e.g. , oughton & matthaeus 1995 ) . a self - consistent treatment of waves in the solar atmosphere and solar wind would consider the impact of fluctuations on the mean , time - steady plasma properties . incorporating this back - reaction into the model steady state is beyond the scope of this paper , but below we compute a subset of the necessary source terms ( e.g. , the heating rate and the wave pressure acceleration ) for use in future models . the nonlinear wave damping discussed in 6.2.2 produces gradual heat deposition along an open magnetic flux tube . the rate of energy conversion is derivable essentially from eq . ( [ eq : zpmdamp0 ] ) , and here we give a slightly more general version of the energy conversion rate based on a comparison between phenomenological turbulence models and numerical simulations . the time rate of change in the general elsasser variance @xmath321 is given by @xmath322 ( e.g. , hossain et al . 1995 ; matthaeus et al . 1999 ; dmitruk et al . 2001 , 2002 ) . comparison with numerical simulations found that the dimensionless parameter @xmath323 is of order unity . we thus assume @xmath324 and define the total heating rate @xmath325 as @xmath326 in our models this quantity is defined consistently only above @xmath314 , where the damping is applied , but below we calculate @xmath325 for all heights along the flux tube . we caution that the heating rates below 0.1 @xmath0 are only estimates based on the undamped elsasser amplitudes at those heights . figure 16 plots the heating rate per unit mass ( @xmath327 ) for the range of @xmath292 values shown in figure 15 in addition to an even weaker - damping case of @xmath328 . for the regions where the nonlinear damping does not change the radial profiles of @xmath329 from their undamped values , @xmath325 scales with @xmath330 as expected . however , for heights where damping occurs the heating rate is lowered because of the cumulative loss of wave power up to that height . thus , for each height , there is a maximum heating rate that defines a critical value @xmath331 . for @xmath332 the cumulative damping grows weaker , and the decrease in @xmath325 comes from the increase in the denominator of eq . ( [ eq : dmit ] ) . for @xmath333 , there is more cumulative damping and the decrease in @xmath325 comes from the loss of local wave power in the numerator of eq . ( [ eq : dmit ] ) . for a distance of @xmath334 in the extended corona , @xmath331 is about 0.07 , and the maximum heating rate per unit mass at that value is @xmath335 erg s@xmath61 g@xmath61 . we also plot in figure 16 an empirical heating rate ( summed from proton and electron contributions ) that was derived in order to reproduce a large number of remote - sensing and in situ measurements of fast solar wind conditions . we use the heating rates from the sw2 model of allen , habbal , & hu ( 1998 ) . for comparison with the above numbers , the empirically constrained heating rate at @xmath334 is about @xmath336 erg s@xmath61 g@xmath61 ( see also dmitruk et al . 2002 ; cranmer 2002 ) . in the extended corona ( @xmath3373 @xmath0 ) , figure 16 shows that the empirical heating rate agrees best with the @xmath320 case that also produces the right amount of in situ damping ( figure 15 ) . this rough agreement provides additional support for the outer - scale normalization specified by this value . it is worthwhile also to compare the heating rates computed above with those computed using a damping rate appropriate for isotropic kolmogorov ( 1941 ) turbulence ( see , e.g. , hollweg 1986b ; chae et al . 1998 ; li et al . 1999 ; chen & hu 2001 ) . in order to derive a kolmogorov heating rate appropriate for quantitative comparison with the above results , we make both the quasi - isotropic approximation @xmath338 and we assume mhd energy equipartition @xmath227 . thus , eq . ( [ eq : dmit ] ) becomes @xmath339 where we use the damped velocity amplitude @xmath226 as plotted in figure 15 . the above expression differs by a factor of @xmath228 from the simpler formulae given in many other papers ; this factor is required in order to remain consistent with the isotropic / equipartition limit of eq . ( [ eq : dmit ] ) . removing this factor would be equivalent to specifying @xmath340 in eq . ( [ eq : hossain ] ) , but we note that the _ relative _ comparison between eqs . ( [ eq : dmit ] ) and ( [ eq : kolm ] ) does not depend on the value of @xmath323 . in figure 17 we plot the comparison between the complete heating rate @xmath325 and the kolmogorov approximation @xmath341 . for simplicity we show only the case having the preferred @xmath320 outer - scale normalization . the curves are substantially different from one another nearly everywhere , which indicates that the inward / outward imbalance generated by non - wkb reflection is probably a very important ingredient in alfvn wave heating models of the solar wind . the differences are small in the photosphere and low chromosphere , where strong reflection leads to nearly equal inward and outward wave power . in the extended corona , though , the kolmogorov heating rate begins to exceed the anisotropic turbulent heating rate by as much as a factor of 30 . the isotropic kolmogorov assumption assumes the maximal amount of possible mixing between inward and outward modes , which is inconsistent with the relatively weak reflection computed for the corona in our models . figure 17 also shows the radial dependence of an analytic scaling relation for the turbulent heating rate given by dmitruk et al . ( 2002 ) for the region above the coronal maximum in @xmath342 . this scaling relation gives @xmath343 , and we show it for an arbitrary normalization . the analytic heating rate per unit mass has a radial gradient less steep than the other curves in figure 17 , but it has a general shape that is similar . waves that propagate radially through an inhomogeneous medium exert a nondissipative net force on the gas . this net momentum deposition has been studied for several decades for both acoustic and mhd waves and is generally called either `` wave pressure '' or the ponderomotive force ( e.g. , bretherton and garrett 1968 ; dewar 1970 ; belcher 1971 ; alazraki and couturier 1971 ; jacques 1977 ) . initial computations of the net work done on the bulk fluid have been augmented by calculations of the acceleration imparted to individual ion species ( e.g. , isenberg & hollweg 1982 ; mckenzie 1994 ; li et al . 1999 ) , estimates of the departures from maxwellian velocity distributions induced by the waves ( goodrich 1978 ; hollweg 1978b ) , and extensions to nonlinearly steepened wave trains ( gail , ulmschneider , & cuntz 1990 ; pijpers 1995 ) . for non - wkb alfvn waves propagating in a radial flux tube , heinemann & olbert ( 1980 ) gave the general second - order expression for the wave pressure acceleration : @xmath344\ ] ] which has the radial component @xmath345 in the limit of equipartition between the kinetic and magnetic wave energy densities , the term in parentheses is zero . for a spherical geometry ( e.g. , macgregor & charbonneau 1994 ) , we note that the quantity @xmath346 can be written simply as @xmath347 . in our non - wkb model we have computed the frequency - integrated kinetic and magnetic energy densities @xmath85 and @xmath198 . we find that the spectrum - weighted wave pressure acceleration can be expressed in terms of these variables as @xmath348 figure 18 compares the computed values of @xmath349 for the undamped ( @xmath229 km s@xmath61 ) and damped ( @xmath320 ) cases , as well as an effective wkb wave pressure computed using just the first term on the right - hand side of eq . ( [ eq : awpavg ] ) . there is some numerical noise in the plots that comes from the necessity to take numerical derivatives . departures from the idealized wkb form arise only below @xmath350 and thus may not be important for the acceleration of the solar wind . in the photosphere , though , the non - wkb wave pressure acceleration exceeds the approximate wkb acceleration by a factor of 100 . the wkb approximation also gives an unrealistically large spike in @xmath349 at the transition region that does not appear in the exact non - wkb solution . it remains to be seen if these differences would significantly affect various observational constraints placed on the properties of the transition region ( e.g. , woods , holzer , & macgregor 1990a , 1990b ) . figure 18 also shows the magnitude of the sun s gravitational acceleration @xmath351 as a function of height . presumably , the wave pressure acceleration has little net effect on the corona and solar wind in regions where @xmath352 . for closed coronal loops , though , laming ( 2004 ) computed a similar non - wkb wave pressure acceleration and found that its strength can rival that of gravity even in the chromosphere . the alfvn wave reflection in these models is strong enough to produce ion abundance variations that may explain the observed dependence on first ionization potential ( fip ) . the goal of this work was to produce a detailed and self - consistent description of the global energy budget of transverse incompressible waves in open magnetic regions of the solar atmosphere . here we list the unique features of the model , key results , and some of the insights gained about the overall wave - plasma system . 1 . measurements of g - band bright points in the photosphere were used to set the power spectrum of transverse waves as the lower boundary condition of our model . the observationally inferred power spectrum was summed from two phases of mbp motion assumed to be statistically independent : isolated random walks and occasional rapid jumps due to mbp merging , fragmenting , or magnetic reconnection . the steady - state plasma density , magnetic field , and flow velocity were constrained empirically from the photosphere to a distance of 4 au . the successive merging of flux tubes on granular and supergranular scales in the atmosphere was described using a two - dimensional magnetostatic model of the magnetic network element . non - wkb wave transport equations , incorporating linear reflection terms , were solved for a grid of wave periods between 3 and @xmath353 s over the range of heights given above . below a mid - chromospheric _ merging height _ the waves were modeled as modified kink - mode flux tube waves , and above this height they were modeled as alfvn waves propagating parallel to the background magnetic field . the waves are reflected strongly at the transition region , with only about 5% of the wave energy transmitted and 95% reflected . above the transition region , most periods are reflected only weakly by the large - scale radial gradient of the alfvn speed , but periods exceeding 1 day are reflected strongly in interplanetary space . the period - averaged reflection coefficient from the extended corona ( @xmath334 ) to 1 au ranges between @xmath354 and @xmath355 . 5 . at the photosphere , the wave periods containing the most power are between 1 and 30 minutes , but the kink - mode waves with periods greater than about 12 minutes are evanescent below the merging height . only the shorter periods propagate to larger distances , and above the merging height the power spectrum is dominated by wave periods between about 1 and 6 minutes . in situ observations of power spectra that are strongest at periods of several hours may be explained by many uncorrelated flux tubes rotating past spacecraft in the solar wind on these longer timescales . the period - averaged transverse velocity amplitude of the waves agrees with observed off - limb nonthermal line widths from sumer and uvcs when the mbp jump amplitude @xmath101 is taken to be @xmath2023 km s@xmath61 . for all reasonable values of @xmath101 between 0 and 6 km s@xmath61 the modeled wave amplitudes at heights greater than 0.3 au are significantly larger than the in situ measurements , implying that large - scale damping is needed . we investigated the potential impact of the linear viscous damping of alfvn waves . if the effective viscous stress timescale undergoes a transition from its collisionally dominated form to either of the possible collisionless forms ( given by the @xmath282 or 2 cases of 6.2.1 ) , the resulting damping lengths are always much longer than the local height from the photosphere , and viscosity thus can not damp the waves . we also considered nonlinear turbulent damping using a phenomenological model of the energy loss terms in the wave transport equations . the one free parameter is the normalization of the outer - scale correlation length @xmath290 of the turbulence , which scales with height as @xmath291 . interestingly , a single choice for the constant @xmath320 , which specifies the value of @xmath290 at the merging height to be about 1100 km , produces both the right amount of damping above 0.3 au to agree with the in situ measurements and the right amount of heating in the extended corona to agree with empirically constrained solar wind acceleration models . because of the relatively weak degree of reflection in the corona , non - wkb effects are probably not important for computing the wave pressure acceleration in regions that contribute to the overall solar wind acceleration . however , non - wkb effects produce order - of - magnitude differences ( from the wkb approximation ) in the wave pressure acceleration at and below the transition region . the inclusion of damping also affects the total amount of wave pressure acceleration imparted to the solar wind . the model presented in this paper contains a great deal of the relevant physics of alfvn waves in the solar atmosphere and solar wind , but it is incomplete in several ways . in this section we discuss various ways to extend the model to improve its physical realism . our cylindrically symmetric magnetostatic model of the network flux tube is highly idealized . a fully three - dimensional model of a network `` patch , '' possibly using a high - resolution magnetogram image as the lower boundary condition , would have a more realistic geometry ( see , e.g. , van ballegooijen & hasan 2003 ) . such a model would contain a _ distribution _ of flux tube strengths , merging heights , and canopy heights . the resonant node structure seen in the power spectra ( e.g. , figure 8) would likely disappear in such a heterogeneous model . our adopted photospheric boundary condition for transverse flux - tube motions is also an idealized approximation . first , the observations used to derive the velocity autocorrelation functions are limited in time resolution . extremely high - frequency acoustic modes e.g . , with periods _ below _ 1 minute have been seen in the photosphere and chromosphere ( deubner 1976 ; wunnenberg , kneer , & hirzberger 2002 ; deforest , depontieu , & hassler 2003 ) , and the kink - mode spectrum may extend to these frequencies as well . second , much more needs to be learned about the statistics of the `` jumps '' that we modeled as identical gaussians . a well - defined algorithm for measuring the combined power spectrum of both isolated mbp motions and their mergings and fragmentings should be developed . it is possible that the transient nature of the motions may be better represented by waves having both complex frequency and complex wavenumber ( e.g. , wang , ulrich , & coroniti 1995 ) , or perhaps by suitably defined wavelet functions . also , new diagnostics such as polarization variability within line profiles may add better constraints to our understanding of flux - tube motions ( ploner & solanki 1997 ) . compressible mhd waves and shocks must also be considered alongside the incompressible waves modeled in this paper ( for recent work , see rosenthal et al . 2002 ; bogdan et al . 2002 , 2003 ; hasan et al . 2003 ; hasan & ulmschneider 2004 ; wedemeyer et al . 2004 ; bloomfield et al . 2004 ; suzuki 2004 ) . longitudinal flux - tube waves are likely to be excited at or below the photosphere , and these motions can provide additional heating ( via collisional damping ) and , in some regions , they can transfer some of their energy into _ new _ alfvnic fluctuations that were not taken into account in our models . in the solar wind , additional linear and nonlinear couplings need to be included to understand better the relatively large measured values of @xmath235 ( see figure 11 ) . the energy and momentum deposition terms computed in 6.3 need to be included in self - consistent solar wind models . extensive modeling has been done using isotropic kolmogorov turbulent heating terms ( effectively , eq . [ [ eq : kolm ] ] ) and the impact of a more realistic treatment of wave reflection is likely to be significant . models including these effects can also be extended to the slow solar wind associated with streamers , to coronal holes at other phases of the solar cycle ( e.g. , miralles , cranmer , & kohl 2002 ) , and possibly to quasi - steady structures like spicules and prominences . a substantial unresolved issue is how the wave - related heating and acceleration is apportioned to electrons , protons , and heavy ions in regions where these species become decoupled . our original motivation for this work was the determination of `` outer - scale '' ( i.e. , low - frequency ) alfvn wave properties in the corona and solar wind , for use as initial conditions in models of mhd turbulent cascade . the kinetic consequences of this turbulence have been studied recently by , e.g. , leamon et al . ( 1999 , 2000 ) , cranmer & van ballegooijen ( 2003 ) , voitenko & goossens ( 2003 , 2004 ) , gary & nishimura ( 2004 ) , and gary & borovsky ( 2004 ) . the ultimate goal of this work is to model the spatial evolution of a reflecting , cascading , and dissipating power spectrum of fluctuations as a function of @xmath356 and @xmath357 . observations of preferential ion heating and acceleration , as well as departures from isotropic maxwellian distributions , are key discriminators between competing models . this paper is dedicated to the memory of peter nisenson , who played a major role in the observations of g - band bright point motions that made this work possible . the authors would like to thank eugene avrett for the use of tabulated pandora model atmospheres . we are also grateful to john kohl , leo milano , william matthaeus , marco velli , wolfgang kalkofen , s. peter gary , and the anonymous referee for valuable and inspiring discussions . this work is supported by the national aeronautics and space administration ( nasa ) under grants nag5 - 11913 , nag5 - 12865 , nag5 - 10996 , and nng04ge84 g to the smithsonian astrophysical observatory , by agenzia spaziale italiana , and by the swiss contribution to the esa prodex program . this work has made extensive use of nasa s astrophysics data system . * appendices * in this appendix we discuss a class of analytic solutions to the incompressible thin - tube wave equations given in 5.1 . for constant coefficients in eq . ( [ eq : momspruit ] ) , we assume that @xmath133 depends on time and height as @xmath358 . for nonzero velocity amplitudes , this wave equation becomes a simple quadratic equation in @xmath262 that is equivalent in many ways to the acoustic - gravity wave equation for an isothermal hydrostatic atmosphere ( e.g. , lamb 1932 ; mihalas & mihalas 1984 ; wang et al . 1995 ) . for a real frequency @xmath81 , this equation is satisfied only for a complex @xmath262 , with @xmath359 where @xmath360 is the critical kink - mode cutoff frequency . the second expression above applies for an isolated flux tube in an isothermal atmosphere with scale height @xmath361 , where @xmath362 is the isothermal sound speed and @xmath363 is the ratio of gas pressure to magnetic pressure ( e.g , . spruit 1981 ) . in this case , the identity @xmath364 yields the standard form of eq . ( [ eq : isoksol ] ) . other useful identities in the isothermal limit , all derivable from transverse pressure balance , are @xmath365 only for @xmath366 can there be a real component of @xmath262 , and thus a propagating wave . a purely imaginary @xmath262 corresponds to evanescence . for the background plasma state described in 3 , @xmath145 and @xmath146 km s@xmath61 at the photosphere . with these values , the critical period ( @xmath147 ) is found to be 12.49 minutes . above the photosphere , the critical period computed with eq . ( [ eq : isoomc ] ) decreases to a minimum of 9.53 minutes at a height of @xmath202200 km , then begins to increase , formally diverging to infinity at the merging height where @xmath367 ( see eq . [ [ eq : delrho ] ] ) . this range of periods compares favorably with the chromospheric estimates of 11.7 minutes given by spruit ( 1981 ) , derived for @xmath368 , and 9 minutes given by hasan & kalkofen ( 1999 ) , derived for @xmath369 . for reference to other isothermal models , we use the above relations to derive effective isothermal values of the plasma beta and scale height at the photosphere ; these are @xmath370 and @xmath371 km , although the latter is about 20% larger than the actual density scale height at @xmath9 . the imaginary part of @xmath262 gives the time - averaged radial dependence of the velocity amplitude . for high frequencies corresponding to propagating kink - mode waves , @xmath372 is proportional to @xmath373 . ( because @xmath374 , this implies that @xmath375 as expected from wkb alfvn wave theory . ) thus , the kinetic energy density of the waves integrated over the cross - section of the tube ( i.e. , @xmath376 ) is constant . ) must remain proportional to the local gas pressure ( @xmath377 ) , so that @xmath378 . ] for low frequencies in the evanescent domain , let us define @xmath379 , and there are two solutions for the radial dependence of the amplitude : @xmath380 \,\,\ , .\ ] ] the upper - sign solution is steeper than the propagating solution , and the lower - sign solution is shallower . in the limit @xmath381 , the steep solution goes as @xmath382 and the shallow solution is constant . thus , for the steep [ shallow ] solution , the wave energy density integrated over the tube area grows [ decays ] with increasing height . in 6 we see that non - wkb low - frequency waves naturally `` find '' the shallow solution , probably because it is the more physical solution with decaying not diverging energy density ( see also wang et al . 1995 ) . an additional reason that a stellar atmosphere may `` choose '' the shallow evanescent solution instead of the steep solution was given by cranmer ( 1996 ) for acoustic - gravity waves in a nonmagnetized atmosphere . despite the fact that the solar wind is extremely _ subsonic _ in the deep photospheric and chromospheric layers we study here , the existence of a nonzero gradient of the outflow speed leads to additional terms in the wave dispersion relation . cranmer ( 1996 ) showed that these terms yield a nonzero real part of @xmath262 for all frequencies , even those below the evanescent cutoff . for a subsonic flow ( @xmath383 ) in the evanescent regime ( @xmath384 ) the real part of the radial wavenumber is given by @xmath385 and the shallow solution is the one with a positive ( i.e. , upward ) phase speed @xmath386 corresponding to the physically relevant situation of more upward then downward wave power . lastly we compute the frequency - dependent partition between kinetic energy density ( @xmath387 ) and magnetic energy density ( @xmath388 ) . the linearized induction equation ( eq . [ [ eq : indspruit ] ] ) is written in the isothermal limit as @xmath389 and thus the ratio of energies is @xmath390 using the analytic solution for @xmath262 above , the total energy density is given by @xmath391 where the upper [ lower ] sign gives the steep [ shallow ] evanescent solution as defined above . note that in the limit @xmath381 , the shallow solution is dominated by kinetic fluctuations and the steep solution is dominated by magnetic fluctuations . over the years there have been several versions of the the non - wkb alfvn wave equations published in the literature that superficially do not resemble one another . this introduces a potential difficulty in comparing the results of these past efforts . in this appendix we collect several of these differently appearing equations and confirm that they are essentially equivalent . the transport equations of heinemann & olbert ( 1980 ) are expressible in terms of our elsasser variables ( eq . [ [ eq : elsdefine ] ] ) as @xmath392 z_{\pm } \ , = \ , z_{\pm } \left [ ( { \bf u } \pm { \bf v}_{a } ) \cdot \nabla \right ] \ln \left ( \rho^{1/4 } \right ) - z_{\mp } \left [ ( { \bf u } \pm { \bf v}_{a } ) \cdot \nabla \right ] \ln \left ( \rho^{1/4 } b_{0}^{-1/2 } \right)\ ] ] where we have specified that the transverse length scale of the flux tube is proportional to @xmath291 . it is relatively easy to see the correspondence between the above form and our eq . ( [ eq : zpm ] ) when we assume that the vectors @xmath393 and @xmath394 are purely radial and that @xmath395 is perpendicular to the radial direction . for mass conservation in a flux tube , the quantity @xmath396 is proportional to @xmath397 , and thus the above logarithmic derivatives yield the scale heights defined in 5.2 . khabibrakhmanov & summers ( 1997 ) gave essentially the same equations , but allowing for more general wave polarization states and pressure anisotropy . the alfvnic transport equations of zhou & matthaeus ( 1990 ) contain nonlinear terms , and thus in some ways are more general than the equations discussed in this paper . however , when looking only at their linear terms , one obtains @xmath398 ( see also dmitruk , milano , & matthaeus 2001 ) . the above equation is exactly equivalent to eq . ( [ eq : zpm ] ) , but some algebraic manipulation is required in order to demonstrate this . among other steps , one needs to apply the auxiliary conditions given by zhou & matthaeus ( 1990 ) , @xmath399 together with the fact that @xmath400 . the non - wkb equations given by velli ( 1993 ) are similar to those of zhou & matthaeus ( 1990 ) . recasting velli s ( 1993 ) differently - defined elsasser variables into our conventional form , one obtains @xmath401 only the last term differs from eq . ( [ eq : zhoum90 ] ) above , but at least for the incompressible transverse waves we study here they are equivalent . this can be shown by noting that , for an arbitrary scalar function @xmath402 , one can state that @xmath403 only when two conditions are met : ( 1 ) the field is divergence - free ( @xmath404 ) , and ( 2 ) the fluctuations are purely transverse ( i.e. , when @xmath405 ) . in the above comparison , this condition is satisfied because @xmath406 depends only on the radius @xmath4 and the incompressible @xmath407 amplitudes have a zero @xmath4-component . barkhudarov ( 1991 ) derived the non - wkb transport equations in terms of nonstandard elsasser - like variables @xmath408 . also , in that paper the equations were given only for spherical symmetry . for a more general superradially expanding flux tube , the equations of barkhudarov ( 1991 ) become @xmath409 and these are the specific equations solved by the numerical runge - kutta code described in 5.2 . for completeness , we also give the full equation of _ wave action conservation _ ( i.e. , wave energy conservation measured in the comoving wind frame ) . we use the variable naming convention of barkhudarov ( 1991 ) , where @xmath49 and @xmath351 correspond to outward and inward waves , respectively , but we follow heinemann & olbert ( 1980 ) in defining these quantities in velocity units and for general superradial expansion : @xmath410 the wave action conservation equation is obtained by writing the four real transport equations ( for the real and imaginary parts of @xmath168 and @xmath167 ) and multiplying each term in the equations by their corresponding real or imaginary elsasser component . the sum of all four equations yields the wave action equation : @xmath411 + \nabla \cdot \left [ \rho { \bf u } ( f^{2 } - g^{2 } ) \right ] \ , = \ , 0 \,\,\ , .\ ] ] in regions dominated by outward propagating waves ( @xmath173 ) , where both the background properties and wave amplitudes are time - steady , this reduces to simply @xmath412 and the assumption of wave action conservation is equivalent to the assumption that @xmath49 is constant ( see also jacques 1977 ; barkhudarov 1991 ) . this is not necessarily a wkb result , but it applies only in regions of negligible reflection . in the wave transport models presented in this paper we have not been overly concerned with the direction of the alfvn wavevector @xmath413 relative to the background magnetic field @xmath414 . mainly this is because the dispersion relation of the waves in the mhd limit depends only on @xmath356 , the parallel component of the wavevector , @xmath415 for the @xmath416 modes , and not on the component @xmath357 perpendicular to the field . thus the monochromatic wave transport equations `` know '' about @xmath356 via their dependence on wave frequency @xmath81 but are independent of @xmath357 . in this appendix we investigate several issues concerning the expected distribution of propagation directions and its evolution as mhd turbulent cascade develops . from a linear standpoint , an initial distribution of propagation directions @xmath417 ( i.e. , the angle between @xmath413 and @xmath414 ) will evolve with heliocentric distance due to several physical processes . the radial variations of the inertial - frame phase speeds @xmath418 induce a radial `` stretching '' in wavelength and thus a large - scale decrease in @xmath356 with increasing height . the transverse spreading of the open flux tube should also cause the perpendicular wavenumber @xmath357 to decrease with increasing height as well . the overall sense of increase or decrease in @xmath417 depends on the time - steady properties of the magnetic field and outflow . other effects can contribute to a radial variation in @xmath417 . first , any transverse gradients of the alfvn speed can refract oblique waves toward the center of the flux tube , thus decreasing @xmath417 ( wentzel 1989 ) . second , viscous damping of high - frequency waves preferentially dissipates oblique fluctuations , since @xmath419 . the analysis in 6.2.1 assumed @xmath420 , but the impact of viscous damping can be enhanced significantly if the waves are sufficiently oblique . all of the above effects should be included if the radial variation of @xmath417 ( or @xmath421 averaged over the power spectrum ) is to be computed . as the turbulent cascade develops in the corona , an additional nonlinear evolution of the wave energy within `` wavenumber space '' ( @xmath422 ) occurs . it has been known for several decades that mhd turbulence in the presence of a steady - state magnetic field develops a strong anisotropy in wavenumber . both numerical simulations and rmhd analytic descriptions indicate that the spectral transport from large to small spatial scales proceeds mainly in the two - dimensional plane perpendicular to @xmath414 ( see , e.g. , higdon 1984 ; shebalin , matthaeus , & montgomery 1983 ; goldreich & sridhar 1995 , 1997 ; bhattacharjee & ng 2001 ; cho , lazarian , & vishniac 2002 ; oughton et al . that is , the cascade spreads out the power to successively larger values of @xmath357 while leaving @xmath356 relatively unchanged . here we follow cranmer & van ballegooijen ( 2003 ) and give an analytic solution for the turbulent power spectrum in the limit that the cascade is allowed to proceed to its final `` driven '' steady state in a small homogeneous volume of plasma . we then compare the frequency dependence of this power spectrum to the empirically constrained ( linear ) frequency spectrum as presented in figure 8 . let us define the three - dimensional total power spectrum as @xmath423 where we write the volume element @xmath424 in cylindrical coordinates as @xmath425 and thus assume symmetry in the two directions transverse to the background field . cranmer & ballegooijen ( 2003 ) , following the general development of rmhd anisotropic cascade theory , assumed that the turbulence eventually leads to a state of complete mixing between the inward and outward modes , and thus @xmath426 . this assumption is incompatible with the non - wkb wave reflection models computed in this paper , but we follow the earlier analysis in order to compare the _ shapes _ of the relevant power spectra and to obtain insight about how to better model the turbulent cascade . for fully developed anisotropic mhd turbulence , we assume the power spectrum to be a separable function of two variables : @xmath357 and a nonlinearity parameter @xmath427 defined as the ratio of the local wind - frame frequency @xmath428 to an assumed nonlinear eddy turnover rate @xmath429 ( see , e.g. , goldreich & sridhar 1995 ) . we use the notation of 2.3 of cranmer & van ballegooijen ( 2003 ) and define @xmath430 where @xmath431 is a reduced power spectrum , scaled to velocity - squared units and defined as @xmath432 for the mhd inertial range , @xmath433 is proportional to @xmath434 . for simplicity , let us define @xmath435 for a given outer - scale perpendicular wavenumber @xmath436 . the factor of @xmath437 above is needed to normalize the full power spectrum as defined in eq . ( [ eq : p3dint ] ) . the @xmath356 dependence of the power spectrum is contained in the dimensionless @xmath438 function in eq . ( [ eq : p3d ] ) . the condition @xmath439 is defined as `` critical balance '' by goldreich & sridhar ( 1995 ) , and their analysis only constrains the general form of @xmath438 , not its exact value . the bulk of the wave power is believed to reside at low values of @xmath427 ( e.g. , @xmath351 is nonzero only for @xmath440 ) , where we define @xmath441 this condition captures the highly nonlinear state of turbulence , for which a coherent wave survives for no more than about one period before nonlinear processes transfer its energy to smaller scales . one reasonable possibility for the parametric dependence of this function is a normalized gaussian , @xmath442 cho et al . ( 2002 ) analyzed numerical simulations of mhd turbulence and found that @xmath438 can be fit reasonably well by decaying exponential or castaing functions . cranmer & van ballegooijen ( 2003 ) solved a simple wavenumber diffusion equation to obtain the analytic form @xmath443^{-n } \label{eq : gkappa}\ ] ] which is normalized to unity when integrated over all @xmath427 , and with @xmath444 for the mhd inertial range , @xmath445 , and the three constants @xmath272 , @xmath363 , and @xmath257 describe the relative strengths of @xmath356 diffusion , @xmath357 advection , and @xmath357 diffusion , respectively . we assume @xmath446 , thus leaving the only `` free '' parameter to be the ratio @xmath447 . cranmer & van ballegooijen ( 2003 ) discussed the most realistic values of this ratio ; earlier models implied that @xmath448 , but one would need this ratio to be smaller than about 0.25 in order to produce enough parallel cascade in the corona to heat protons and heavy ions via cyclotron resonance . with the above definitions it becomes possible to derive the effective frequency power spectra @xmath449 that are consistent with fully developed anisotropic turbulence . noting that we define @xmath81 as the frequency measured in the inertial frame centered on the sun , we define @xmath450\ ] ] where the dirac delta function collapses the three - dimensional wavenumber spectrum into a one - dimensional frequency spectrum . this `` translation '' must be done differently for the inward and outward mode spectra , as indicated by the difference between the inertial - frame phase speeds of the two modes . the above expression is defined formally for both positive and negative frequencies ; below we multiply by 2 in order to consider only positive frequencies . for the purposes of this appendix , let us use the simpler form of eq . ( [ eq : ggauss ] ) for @xmath438 that gives the straightforward analytic solution @xmath451 \right\}\ ] ] where we define the effective outer - scale frequency @xmath452 and we assume @xmath453 . we note that the use of eq . ( [ eq : gkappa ] ) for @xmath438 would have yielded a solution similar to the above form of @xmath454 , only differing slightly in a small region of frequency near @xmath455 . this frequency sets the scale for the spectrum . we use the effective outer - scale frequency for outward propagating waves to define the nonlinear `` driving time '' @xmath456 plotted in figure 14 . figure 19 compares the above turbulent frequency power spectra @xmath457 with the empirically constrained linear frequency spectra @xmath458 as derived in 5 above . for simplicity we plot the spectra at only one heliocentric distance ( @xmath334 ) and note that one can use the radial dependences of the various timescales plotted in figure 14 in order to extend these results to other heights . the above assumption of full mixing for the nonlinear cascade ( i.e. , @xmath338 ) is far from the linear non - wkb result of @xmath459 in the extended corona . the inability of the analytic rmhd anisotropic cascade theory to take account of the strong imbalance between inward and outward wave amplitudes is a major shortcoming of this kind of model . we need a better understanding of the _ detailed wavenumber dependence _ of the inward - to - outward power ratio for mhd turbulence . in figure 19 we use the preferred normalization @xmath320 to set the scale of @xmath290 . with this normalization at @xmath334 , the outer - scale driving timescales @xmath460 are 5.7 and 4.1 minutes for the @xmath461 and @xmath462 signs , respectively . the peaks of the scaled power spectra @xmath463 occur at frequencies of @xmath464 , but the empirically constrained linear spectra @xmath465 contain most of their power at slightly higher frequencies ( i.e. , shorter periods ) . the fact that the two spectra exhibit maxima within the same order of magnitude of frequency , though , may be just a result of choosing the radius of @xmath334 . figure 14 shows that if the curves in figure 19 were computed instead at 1 au , the peaks of the linear @xmath466 curves would be in the same range of periods ( 1 to 5 minutes ) , but the peaks of the @xmath454 curves would be at periods between 30 and 40 minutes . at 1 au the bulk of the power in the linear frequency spectrum would be at an extremely high local frequency from the standpoint of the cascade ( i.e. , @xmath467 ) . this suggests that the sun emits a `` fossil '' frequency spectrum that eventually can advect into a region of the three - dimensional wavenumber spectrum that was previously believed to be strongly _ depleted _ due to the anisotropic cascade . the implications of this result will be investigated in future work . aiouaz , t. , peter , h. , lemaire , p. , & keppens , r. 2004 , in soho-13 : waves , oscillations , and small - scale events in the solar atmosphere , ed . h. lacoste ( noordwijk , the netherlands : esa ) , esa sp-547 , 375 stein , r. f. , & nordlund . 2002 , in solmag 2002 : proceedings of the magnetic coupling of the solar atmosphere euroconference , iau colloq . h. lacoste ( noordwijk , the netherlands : esa ) , esa sp-505 , 83	= 11.7pt we present a comprehensive model of the global properties of alfvn waves in the solar atmosphere and the fast solar wind . linear non - wkb wave transport equations are solved from the photosphere to a distance past the orbit of the earth , and for wave periods ranging from 3 seconds to 3 days . we derive a radially varying power spectrum of kinetic and magnetic energy fluctuations for waves propagating in both directions along a superradially expanding magnetic flux tube . this work differs from previous models in three major ways . ( 1 ) in the chromosphere and low corona , the successive merging of flux tubes on granular and supergranular scales is described using a two - dimensional magnetostatic model of a network element . below a critical flux - tube merging height the waves are modeled as thin - tube kink modes , and we assume that all of the kink - mode wave energy is transformed into volume - filling alfvn waves above the merging height . ( 2 ) the frequency power spectrum of horizontal motions is specified only at the photosphere , based on prior analyses of g - band bright point kinematics . everywhere else in the model the amplitudes of outward and inward propagating waves are computed with no free parameters . we find that the wave amplitudes in the corona agree well with off - limb nonthermal line - width constraints . ( 3 ) nonlinear turbulent damping is applied to the results of the linear model using a phenomenological energy loss term . a single choice for the normalization of the turbulent outer - scale length produces both the right amount of damping at large distances ( to agree with in situ measurements ) and the right amount of heating in the extended corona ( to agree with empirically constrained solar wind acceleration models ) . in the corona , the modeled heating rate differs by more than an order of magnitude from a rate based on isotropic kolmogorov turbulence .
in multiferroic materials such as @xmath10mno@xmath11 ( r : rare earths ) @xcite , ni@xmath11v@xmath12o@xmath13@xcite , and mnwo@xmath1@xcite , ferroelectricity is believed to be driven by magnetism because it emerges simultaneously with a spiral magnetic order upon cooling . these materials are known as `` type - ii '' multiferroics,@xcite and their ferroelectricity as `` improper.''@xcite even though such multiferroicity has to date been found only at low temperatures , it has attracted very significant research interest due to the possibilities it offers for tunable multiferroic devices . phenomenologically , the breaking of magnetic inversion symmetry and strong dzyaloshinskii - moriya ( dm ) interactions are believed to be essential ingredients for understanding the coupling between ferroelectricity and magnetism.@xcite theoretically , inversion symmetry - breaking is thought to be caused by strong magnetic frustration . experimentally , while magnetic frustration is certainly suggested by the available susceptibility data,@xcite the coupling between ferroelectricity and magnetism has been difficult to quantify , and more spectroscopic studies are required to confirm its existence and explore its origin . ( color online ) crystal structure ( triclinic ) of fevo@xmath1 viewed along the @xmath14 axis . the three different v sites are indicated , each occurring as a vo@xmath1 tetrahedron.,width=283 ] fevo@xmath1 is a multiferroic compound that has been characterized mostly in polycrystalline@xcite or thin - film form,@xcite although single crystals have also been synthesized.@xcite it crystallizes in a triclinic structure , with each unit cell containing 6 fe@xmath15 , 6 v@xmath16 , and 24 o@xmath17 ions , as shown in fig . [ struc1 ] . neutron scattering studies of polycrystal powders@xcite indicate a collinear incommensurate antiferromagnetic ( afm ) order below @xmath18 22 k and a non - collinear incommensurate , or spiral , afm order below @xmath19 15 k. ferroelectricity occurs only below @xmath20 , a result found also in other type - ii multiferroic materials.@xcite to date , however , most studies of fevo@xmath1 have focused on this ferroelectricity,@xcite and there has been only limited investigation of its magnetic states and their relation to frustration . the availability of fevo@xmath1 as high - quality single crystals makes it an excellent target material for nmr studies of its magnetic properties and of their coupling with ferroelectricity . nmr is a low - energy , local probe , which is in principle an ideal tool for the study of multiferroics . it is sensitive both to ferroelectricity , through the coupling between the nuclear quadrupole moment and the electric field gradient ( efg ) , and to magnetism , through the hyperfine coupling between the nuclei and the magnetic moments . in fevo@xmath1 , however , only the @xmath0v signal could be observed within our available nmr window and the weak quadrupole moment ( @xmath21 barns ) of this nucleus makes the coupling to the efg too weak to be detected . therefore we focus primarily on the low - energy magnetic properties , which reveal strong magnetic frustration in this system , and thus also provide essential information for understanding the nature of multiferroicity . in this paper we present all of the information about magnetism in fevo@xmath1 that can be deduced from @xmath0v nmr . the high - temperature spectrum is single - peaked in the paramagnetic phase , where the knight shift , like the susceptibility , follows a curie - weiss form with a large weiss constant . on cooling below 65 k , the spectrum splits into two peaks , which indicates a local symmetry - breaking effect on the nmr time scale , or the onset of short - range magnetic order far above the magnetic transition . from the spin - lattice relaxation time we identify the two magnetic transitions at @xmath9 = 19 k and @xmath22 13 k , consistent with other results reported on single crystals , but lower than those for powder samples . below @xmath20 , we find a spectrum with no applied field , which reflects a complicated distribution of hyperfine fields transferred from the incommensurate magnetic moments of the fe ions . however , the upper bound for the ordered moment is only 1.95@xmath7/fe , far less than the paramagnetic local moment , while the temperature dependence of @xmath4 below @xmath20 suggests strong low - energy spin fluctuation effects . we discuss how all of these features are the fingerprints of strongly competing magnetic interactions . the structure of our report is as follows . in sec . ii we describe briefly the nmr methods as applied to fevo@xmath1 . in sec . iii we present all of the results we obtain for @xmath0v spectra over a range of temperatures , for the relaxation times @xmath23 and @xmath24 , for the knight shift @xmath25 , and at zero magnetic field . in sec . iv we discuss the implications of our results for frustrated magnetism in fevo@xmath1 and it connection to multiferroicity . section v contains a brief summary . single crystals of fevo@xmath1 were synthesized by the flux - growth method , using v@xmath12o@xmath26 as the self - flux.@xcite the crystal structure is shown in fig . [ struc1 ] and the crystal alignment , required for nmr , was determined by von laue measurements . the complete @xmath0v nmr spectra were measured by field sweeps of the spin - echo signal , using a fixed frequency of 111.7 mhz and sweeping the magnetic field with its orientation parallel to the ( 0 1 -1 ) plane of the crystal . the spin - echo pulse sequence was @xmath27 , with @xmath28 @xmath29s and respective @xmath30 and @xmath31 pulse lengths of 2.5 and 4 @xmath29s . the knight shift was determined in the paramagnetic state from @xmath32 , where @xmath33 is the resonance frequency , @xmath34 is the applied field , and @xmath35 = 11.197 mhz / t is the @xmath0v gyromagnetic ratio . the spin - lattice relaxation rate @xmath36 was measured by the spin - inversion method , and the spin - spin relaxation rate @xmath37 from the spin echoes , both of which were found to follow standard , single - exponential spin recovery / decay functions . two comments are in order on the @xmath0v spectrum at @xmath38 = 140 k , shown in fig . [ spec2](a ) , which has a finite knight shift @xmath39 . first , from the triclinic crystal structure shown in fig . [ struc1 ] , three @xmath0v resonance peaks are expected . each unit cell in fevo@xmath1 contains three pairs of non - identical v@xmath16 sites , each linked by lattice inversion symmetry and labeled by v1 , v2 , and v3 . all three types of site have different bond lengths to all of their neighboring o@xmath17 and fe@xmath15 ions , and therefore should appear as separate peaks . however , in our data only one of the three site pairs is detected within the nmr window . as we will show in sec . iii , the spin - spin relaxation time @xmath40 of our signal is very short ( less than 50 @xmath29s ) due to the dominance of magnetic fluctuations . thus we believe our inability to observe the other two @xmath0v pairs arises because these sites have a stronger hyperfine coupling and therefore even shorter @xmath40 times , which move the corresponding signals out of our measurement window . second , we did not find measurable quadrupolar effects on the resonance frequency , which is reasonable for the center site of the local vo@xmath1 tetrahedra and for the low quadrupole moment of @xmath0v . as noted in sec . i , the absence of significant quadrupole effects prevents us from studying the lattice structure , and therefore the ferroelectricity , directly . ( color online ) ( a ) field - sweep @xmath0v spectra over the full range of temperatures probed , with the field applied parallel to the crystal ( 0 1 -1 ) planes . red arrows indicate the fields used for the @xmath4 and @xmath41 measurements shown in fig . the solid line at 25 k is a double - gaussian fit to the spectrum . ( b ) fwhm of the nmr spectra measured under the same field - sweep conditions . the temperatures @xmath42 , @xmath9 , and @xmath20 are introduced in the text . vertical bars on the data points indicate the standard deviations of the fitting curves.,width=283 ] we begin our investigation of the nmr response of fevo@xmath1 single crystals by considering the high - field @xmath0v spectra at elevated temperatures . spectra starting at 140 k are shown in fig . [ spec2](a ) . between 140 k and 80 k they show a single peak at a field of 9.66 t , which is a regular paramagnetic signal . as we decrease the temperature to 1.5 k , it is clear that the line shape , line width , and peak frequency change dramatically , reflecting a rich variety of magnetic properties for a single material . we note that a spurious @xmath0v signal is also observed at 10.0 t ( @xmath43 ) , which has a @xmath44 value more than three orders of magnitude higher than the other peaks and does not sense the magnetic transitions on cooling , and thus is probably the consequence of a weak impurity phase or an inclusion of the crystal - growth flux . a weak and asymmetric shoulder feature at 9.65 t is also visible at temperatures between 40 and 80 k before being lost as the spectrum broadens , and this may reflect some dilute local disorder . figure [ spec2](a ) shows representative spectra in all of the different magnetic phases of fevo@xmath1 . below a temperature @xmath45 k , the @xmath0v spectrum broadens and develops a prominent double - peak feature suggestive of strong spin correlations , or short - range magnetic order on the nmr time scale ; we discuss this interpretation in detail below . the spectra at 15 k and at 1.5 k show such significant broadening that they again appear to have a single peak . this form is typical for systems with spatially inhomogeneous magnetic order when subject to an applied field , because the nuclei respond to both the external field and the varying internal field in the crystal . we recall here ( sec . i ) that at 15 k the system is in the collinear incommensurate afm phase while at 1.5 k it is in the incommensurate spiral afm phase . the full width at half - maximum height ( fwhm ) of the peak , shown in fig . [ spec2](b ) , quantifies the increase in broadening with cooling as the sample steps through the sequential changes of magnetic properties . ( color online ) ( a ) spin - lattice relaxation rate @xmath4 and spin - spin relaxation rate @xmath41 as functions of temperature , measured at the peak position of the spectra . ( b ) temperature dependence of the boltzmann - normalized echo intensity at the peak position of the @xmath0v spectrum.,width=302 ] to identify the magnetic transition temperatures , we measured the spin - lattice relaxation rate @xmath4 and the spin - spin relaxation rate @xmath41 . we show results for both quantities , measured for each temperature at the peak positions on the low - field side of the spectra [ red arrows in fig . [ spec2](a ) ] ; below @xmath20 , @xmath4 is lowest at the peak frequency and increases by approximately @xmath46 across the frequency range . the temperature dependence of @xmath4 , shown in fig . [ t1vst3](a ) , first displays a slow decrease with temperature upon cooling below 100 k. however , there is a prominent increase below 30 k leading to a peak at 19 k , which indicates the first magnetic phase transition at @xmath9 . upon further cooling , @xmath4 decreases again , although a small second peak is resolvable at @xmath47 k , indicating the second magnetic transition ( @xmath20 ) . both transitions can also be resolved in the @xmath41 data , also shown in fig . [ t1vst3](a ) , where @xmath41 exhibits the same behavior as @xmath4 ; two peaks in the relaxation rate are clearly formed at the two magnetic transition temperatures @xmath48 k and @xmath49 k. while @xmath23 in a magnetic system is expected to be controlled by spin fluctuations , this is in general not so clear for @xmath24 . however , the fact that @xmath41 peaks at the transitions and shows a temperature dependence so similar to that of @xmath4 indicates that @xmath41 is indeed dominated by the magnetic fluctuations in fevo@xmath1 . we comment again here that the @xmath37 values shown in fig . [ t1vst3](a ) are very short ( under 50 @xmath29s over much of the temperature range ) , a result we ascribe to these very strong magnetic fluctuations and which we believe prevents our detection of the other two v site pairs ( sec . ii ) . to verify the nature of both transitions we have also measured the echo intensity @xmath50 at the peak frequency , integrated over a finite frequency range . the boltzmann - corrected echo intensity , @xmath51 , is shown in fig . [ t1vst3](b ) and is inversely proportional to both fwhm and @xmath52 , where @xmath53 is the spin - echo refocusing time . as the temperature is lowered from 30 k to @xmath9 , @xmath51 decreases strongly due to the combination of inhomogeneous line - width broadening [ fig . [ spec2](b ) ] and the temperature dependence of @xmath54 [ fig . [ t1vst3](a ) ] . in fact @xmath55s is very short at @xmath9 [ fig . [ t1vst3](a ) ] because of the strong magnetic fluctuations , and this causes a large signal loss in our spin - echo measurements . the rise in echo intensity below @xmath9 is caused by the fall in @xmath41 [ fig . [ t1vst3](a ) ] . however , the increase below @xmath20 [ fig . [ t1vst3](b ) ] occurs despite the strong spectral broadening at these temperatures [ fig . [ spec2](b ) ] and can not be explained by @xmath40 effects alone . while the kink directly below @xmath20 is caused by the sharp fall in @xmath41 [ fig . [ t1vst3](a ) ] , the continued increase in echo intensity at low temperatures , once @xmath40 is very long again , is due to rf enhancement of the nmr signal . this phenomenon is typical in ordered magnetic systems and has also been reported in multiferroic materials such as tbmn@xmath12o@xmath26;@xcite in this compound it shows a hysteresis effect at high fields that the authors proposed may originate in a coupling between the afm domain walls and ferroelectric domain walls . in fevo@xmath1 we do not find evidence for hysteresis effects , and we suggest that the rf enhancement below @xmath20 is intrinsic in a spiral magnet . ( color online ) ( a ) temperature dependence of the knight shift @xmath25 ( left axis ) measured at the peak frequency of the @xmath0v spectrum in the paramagnetic phase . knight - shift data shown as open diamond , circle , and triangle symbols are measured under a field of 9.5 t , data shown as solid symbols at 4 t. the bulk susceptibility ( star symbols , right axis , adapted from ref . ) is shown for comparison . inset : @xmath25 as a function of the dc susceptibility with temperature as the implicit parameter . solid lines in both panels are fits to the knight - shift data using the curie - weiss function . ( b ) zero - field @xmath0v nmr spectrum measured at @xmath38 = 1.5 k. @xmath56 and @xmath57 label the resonance frequencies of the high- and low - frequency peaks . inset : @xmath38 dependence of @xmath56 . the arrows indicate the two magnetic transition temperatures.,width=302 ] in this section we discuss the distinctive magnetic properties of fevo@xmath1 revealed by nmr in the different regimes of temperature . we focus first on the properties in the truly paramagnetic phase above @xmath45 k. figure [ knvst4](a ) shows the knight shift , @xmath25 , deduced from the peak frequency of the spectrum , as a function of temperature . the knight shift measures the spin susceptibility of the system , and so the monotonic increase of @xmath25 on cooling is consistent with paramagnetic behavior . in fact @xmath58 can be fitted very well by a curie - weiss form , @xmath59 , with the weiss constant @xmath60 k ; the fit is shown in fig . [ knvst4](a ) . in the inset we show @xmath25 against the dc susceptibility ( adapted from ref . ) with temperature as the implicit parameter , from which we estimate the hyperfine coupling constant to be @xmath61 koe/@xmath62 . values of @xmath63 reported in the literature show some considerable variation , with curie - weiss fits to susceptibility data from polycrystals giving @xmath64 125 k@xcite and from single - crystal measurements giving @xmath3 97 k.@xcite this spread of results may reflect an important role for domain - wall effects , a topic to which we return below . two incontrovertible statements are that our measurements are fully consistent with previous studies and that they are consistent with @xmath65 . such a large value of @xmath63 compared to @xmath9 is typical for a number of type - ii multiferroic systems,@xcite and is a key piece of evidence for strong magnetic frustration . we turn next to the double - peak feature in the spectrum at temperatures between @xmath42 and @xmath9 , where the intensities of the two peaks are rather similar and can be fitted rather well by a double - gaussian function [ fig . [ spec2](a ) ] . the knight shifts calculated from both peaks are shown in fig . [ knvst4](a ) , where it is clear that @xmath42 represents a bifurcation in behavior . @xmath58 in this temperature range deviates both from the high-@xmath38 curie - weiss form and from the high-@xmath38 linear scaling with the bulk susceptibility [ fig . [ knvst4](a ) ] . as a consequence of this line - splitting , the overall line width of the spectral features also broadens significantly below 65 k , as shown in fig . [ spec2](b ) . we confirmed ( data not shown ) that the splitting of the peaks is proportional to the external field , which indicates a varying local susceptibility rather than any static magnetic ordering . indeed , such a line - splitting is clearly a local symmetry - breaking effect , and far the most probable interpretation of our data is that the double - peak spectra are caused by strong spin correlations , or equivalently short - range magnetic ordering on the time scale of nmr . in particular , the hyperfine fields of the two v sites linked by lattice inversion symmetry may be different in this short - range - ordered state , splitting the spectrum into two peaks with equal intensity as observed . while the splitting we observe could also be caused by a breaking of crystal symmetry , no structural measurements have yet detected such a process at temperatures as high as @xmath42 . further evidence in favor of a short - range ordering scenario for temperatures @xmath66 in fevo@xmath1 can be found by comparison with the situation when @xmath67 , where the magnetic order is long - ranged and static , magnetic inversion symmetry is broken,@xcite and our zero - field nmr spectrum also resolves a double - peak feature ( below ) . strong spin correlations or short - ranged magnetic order above @xmath9 in fevo@xmath1 have been also been proposed to interpret specific - heat measurements , where a significant absence of magnetic entropy ( or an `` entropy recovery '' ) is observed over a broad temperature range above @xmath9.@xcite while magnetic heat - capacity measurements are complicated by issues including phonon subtraction , the line splitting and broadening we observe by nmr provide direct evidence for short - range magnetic ordering . our results also give the first accurate measurement of the onset temperature @xmath42 . short - range ordering of this type has also been reported by nmr , on the basis of inhomogeneous line - width broadening , in the materials licuvo@xmath1@xcite and lamn(o@xmath68f@xmath69)@xmath11.@xcite the appearance of short - range magnetic order in the paramagnetic state is a first indication of magnetic frustration effects , which are necessary to suppress a static afm order in this regime . the fact that @xmath42 is over three times the size of @xmath9 , making the short - range - ordered region remarkably broad , suggests that frustration is very strong in fevo@xmath1 . this observation is fully consistent with the magnetic frustration revealed by the large weiss constant in the knight shift . the tendency towards short - ranged magnetic order may be an important ingredient in explaining the discrepancies between nel temperatures reported in the literature . the values we obtain from nmr , @xmath48 k and @xmath49 k , are consistent with the magnetization measurements also performed on single crystals.@xcite however , values reported for powder samples@xcite are @xmath70 k and @xmath71 k , respectively 3 k and 2 k higher , which represent a discrepancy well beyond the expected error bars of the individual measurements . we suggest that the transition temperatures in powder samples can be enhanced both by grain - boundary and domain - wall effects and by strain effects . given that short - range magnetic ordering already occurs at 65 k , both sets of effects provide a ready source of pinning for fluctuating magnetic moments , particularly when spiral orientations are favored . strain effects have been found to be very effective in enhancing multiferroic properties in a number of compounds.@xcite next we discuss the magnetic properties in the ordered phase . in addition to our high - field nmr studies , we have also performed zero - field nmr measurements to study the magnetic structure in the non - collinear ( spiral ) phase below @xmath20 . in fig . [ knvst4](b ) we show the zero - field @xmath0v spectrum at 1.5 k , which has a clear double - peak structure with the two maxima centered at @xmath56 ( high - frequency ) and @xmath57 ( low - frequency ) . the broad spectrum around each peak is caused by the distribution of hyperfine fields on the v sites transferred from the fe moments , all of which are different due to their incommensurate order ( neutron scattering measurements in this phase reveal a spiral magnetic modulation period of approximately 100 nm.@xcite ) . the fwhm of the high - field spectra at @xmath72 k , shown in fig . [ spec2](b ) , is approximately 8 mhz , which is considerably less than @xmath56 and therefore indicates that the hyperfine field on the v sites is almost perpendicular to the applied external field . the nmr spectrum in incommensurate magnetically ordered states usually has a characteristic `` double - horn '' feature,@xcite but this is obtained when the applied field is not perpendicular to the internal field . thus in our present field configuration we are not able to distinguish between an incommensurate spin structure and other forms of modulation that also give rise to a distribution of hyperfine fields , and can state only that our broad line shapes are consistent with the known incommensurate order . this lack of specificity applies also in the short - range - ordered phase between @xmath9 and @xmath42 , where we can not probe the commensurate or incommensurate nature of the spin fluctuations . we comment that the incommensurate `` double - horn '' shape is not similar to the double - peak structures we find in either our high - field or zero - field nmr measurements [ figs . [ spec2](a ) and [ knvst4](b ) ] . although the spectral intensity is higher at @xmath56 than at @xmath57 , we believe this difference is due primarily to the sensitivity of the nmr pick - up . this splitting of the spectrum is probably caused by the breaking of inversion symmetry in the hyperfine field on the v sites , similar to the situation we discussed ( on the nmr time scale ) in the short - range - ordered state , and the default expectation would be peaks of equal weight . we have also measured @xmath56 as a function of temperature , finding [ inset , fig . [ knvst4](b ) ] that it increases significantly on cooling from 10 k down to 1.5 k , reflecting the development of the ordered moment . by using the value of @xmath73 measured in the paramagnetic phase , the high - frequency resonance peak ( @xmath56 ) at zero field sets an upper bound for the ordered moment , of 1.95 @xmath62/fe at @xmath38 = 1.5 k. in the paramagnetic phase , however , the magnetization data give a local moment of 5.83 @xmath62/fe.@xcite thus the ordered moment below @xmath20 is only 1/3 of the net moment , indicating again the effects of strong magnetic frustration even at the lowest temperatures . below @xmath20 , the temperature dependence of the spin - lattice relaxation rate is different from a conventional antiferromagnet,@xcite where @xmath74 due to relaxation by gapless spin waves , and the temperature dependence is stronger still in the presence of magnetic anisotropy . as shown in fig . [ t1vst3](a ) , @xmath4 , measured at high field , has power - law behavior below @xmath20 with @xmath75 . similar unconventional behavior and anomalously slow spin dynamics have been measured in other frustrated magnetic systems , such as volborthite,@xcite where they were ascribed to a very high density of available low - energy excitations . the low power - law temperature dependence found in fevo@xmath1 would seem to indicate the presence of persistently strong low - energy spin fluctuations on top of the spiral ordered state below @xmath20 . the small ordered moment and the strong low - energy spin fluctuations in the spiral magnetic phase reflect once again the effects of magnetic frustration in suppressing the ordered moment while enhancing spin fluctuations . combined with the large weiss constant and the short - range magnetic ordering at high temperatures , fevo@xmath1 shows explicit evidence of strong frustration all across the phase diagram . magnetic frustration in fevo@xmath1 is clear from the structure shown in fig . [ struc1 ] , where there are multiple inequivalent fe o fe and fe o o fe paths in the system . these paths give rise to effective magnetic coupling processes , referred to respectively as superexchange and super - superexchange in the structural and magnetic study of ref . , and it is reasonable to assume that these interactions compete strongly . our data provide independent evidence reinforcing the presence of strong magnetic frustration in fevo@xmath1 , and by extension its importance for multiferroicity in the form of ferroelectric incommensurate sdw phases . the origin of magnetically - driven ferroelectricity in improper multiferroics is discussed in ref . . unlike the case of proper multiferroics , it does not depend on a `` @xmath76-type '' polar distortion mechanism despite the absence of orbital moments@xcite on both v@xmath16 and fe@xmath15 ( which is @xmath77 ) . in fact the breaking of magnetic inversion symmetry may in itself not be a sufficient condition , as this is broken at @xmath9 in fevo@xmath1 , i.e. in the non - ferroelectric collinear incommensurate sdw phase.@xcite instead a genuine spiral magnetic order is required to sustain a polar structure,@xcite with the polarity vector required to lie in the plane of the spiral.@xcite as for the microscopic mechanism responsible for this interaction , the strong coupling between the charge and spin sectors in fevo@xmath1 has been described as a magnetoelastically mediated magnetostriction@xcite and as a magnetoelectric coupling@xcite whose primary origin was proposed to lie in trilinear spin - phonon interactions.@xcite in the magnetic sector , one of the most important terms leading to frustration and incommensurate ordered phases is the dzyaloshinskii - moriya interaction , which arises from spin - orbit coupling in non - inversion - symmetric bonding geometries . these interactions are generic in systems of low crystal symmetry , exactly the situation encountered in fevo@xmath1 , where the triclinic structure has six fe@xmath15 ions ( three structurally inequivalent ) in each unit cell . in combination with superexchange terms , which favor collinear order unless strongly frustrated , dzyaloshinskii - moriya interactions often act to produce spiral magnetic order . the resulting exchange striction , or lattice relaxation in the spin - ordered state , drives a polar charge state , i.e. a ferroelectric . finally , we comment once again that unfortunately we were not able to perform a direct investigation of the ferroelectric properties of fevo@xmath1 in this study . the weak quadrupole moment of @xmath0v combined with the low efg at the centers of the vo@xmath1 tetrahedra result in a coupling between @xmath0v and the crystal lattice that is too small for us to detect . ideally , future studies of fevo@xmath1 would perform @xmath78o nmr measurements on @xmath78o - enriched crystals ; because the o ions bridge the fe ions and mediate the magnetic superexchange and super - superexchange interactions , they can be expected to have a much stronger quadrupolar coupling to the lattice distortion in the ferroelectric phase . in summary , we have performed @xmath0v nmr measurements on single crystals of fevo@xmath1 with both zero and high applied magnetic fields . we confirm both magnetic transitions to phases of collinear incommensurate ( @xmath9 ) and spiral incommensurate ( @xmath20 ) magnetic order , both occurring at values lower than those found for polycrystalline samples . our data reveal a temperature @xmath45 k marking the onset of short - ranged magnetic order on the nmr time scale . we observe a large weiss constant ( @xmath63 ) in the knight shift , a prominent spectral splitting accompanying the short - range correlations , small magnetic moments in the ordered phases ( deduced from the hyperfine field ) , and strong low - energy spin fluctuations in this regime ( deduced from spin - lattice relaxation times ) . these results provide explicit evidence for strongly frustrated exchange interactions in fevo@xmath1 , and thus underline the importance of magnetic frustration for the occurrence of improper ferroelectricity in multiferroic materials . we thank dr . c. y. wang and prof . x. j. zhou for x - ray ( von laue ) measurements of the crystal alignment . work at north china electric power university was supported by the nsf of china ( grant no . 11104070 ) and by the scientific research foundation for the returned overseas chinese scholars , state education ministry . work at renmin university of china was supported by the nsf of china ( grant nos . 11374364 , 11174365 , and 11222433 ) and by the national basic research program of china ( grant nos . 2010cb923004 , 2011cba00112 , and 2012cb921704 ) . work at fujian institute of research on the structure of matter was supported by the national basic research program of china ( grant no .	we report @xmath0v nuclear magnetic resonance ( nmr ) studies on single crystals of the multiferroic material fevo@xmath1 . the high - temperature knight shift shows curie - weiss behavior , @xmath2 , with a large weiss constant @xmath3 116 k. however , the @xmath0v spectrum shows no ordering near these temperatures , splitting instead into two peaks below 65 k , which suggests only short - ranged magnetic order on the nmr time scale . two magnetic transitions are identified from peaks in the spin - lattice relaxation rate , @xmath4 , at temperatures @xmath5 19 k and @xmath6 13 k , which are lower than the estimates obtained from polycrystalline samples . in the low - temperature incommensurate spiral state , the maximum ordered moment is estimated as 1.95@xmath7/fe , or 1/3 of the local moment . strong low - energy spin fluctuations are also indicated by the unconventional power - law temperature dependence @xmath8 . the large weiss constant , short - range magnetic correlations far above @xmath9 , small ordered moment , significant low - energy spin fluctuations , and incommensurate ordered phases all provide explicit evidence for strong magnetic frustration in fevo@xmath1 .
one of the remarkable special classes of the exact solution of the nonvacuum einstein equations @xmath0 available to date was discovered by szekeres @xcite . they are obtained by solving einstein equations with irrotational dust @xmath1 as the source , for the line element @xmath2 relative to comoving coordinates . here @xmath3 is a velocity ( i.e. unit timelike ) vector field and @xmath4 is the energy density of the system . one can say that szekeres solutions are obtained when the spherical symmetry orbits in the lemaitre - tolman - bondi ( ltb ) model @xcite are made nonconcentric to destroy the symmetry , but the energy - momentum tensor is still that of dust . therefore , the szekeres spacetime is often called as quasispherical , for definiteness we shall name it as quasispherical szekeres ( qsz ) solutions . being an exact model of the spacetime geometry , the qsz solutions have primarily been adapted with regards to studies of nonspherical collapse of inhomogeneous dust cloud @xcite in four dimensions ( 4d ) , and in higher dimensions @xcite , in cosmology @xcite and also in observational cosmology @xcite . they are very important because they admit no killing vectors @xcite and hence they are lacking symmetry . in any attempt to perturbatively quantize gravity as a field theory , higher - derivative interactions must be included in the action . such terms also arise in the effective low - energy action of string theories . among the higher curvature gravities , the most extensively studied theory is the so - called einstein - gauss - bonnet ( egb ) gravity @xcite . the egb gravity is a special case of lovelocks theory of gravitation , whose lagrangian contains just the first three terms . gauss - bonnet gravity provides one of the most promising frameworks to study curvature corrections to the einstein action in supersymmetric string theories , while avoiding ghosts and keeping second order field equations . in einstein - gauss - bonnet gravity less number of exact solutions have been known so far . static and spherically symmetric black hole solutions with or without a cosmological constant , as well as a topological black hole in an anti - de - sitter spacetime were obtained @xcite . the effects of gauss - bonnet terms on the vaidya solutions have been investigated in @xcite , and on the ltb solutions in @xcite . here , we consider the five - dimensional ( 5d ) action with the gauss - bonnet terms for gravity and give an _ exact model _ of the quasispherical gravitational collapse . using our new solution , we investigate the nature of singularities of such a spacetime in terms of its being hidden within a black hole , or whether it would be visible to outside observers and compare it with analogous relativistic case . thus , the aim of this paper is to extend the previous studies on the quasispherical gravitational collapse of inhomogeneous dust , including the second order perturbative effects of quantum gravity solutions : a gauss - bonnet generalization of the qsz solutions in 5d spacetime , namely , 5d - qsz - egb . there are several issues that motivate our analysis : how does the gauss - bonnet term affect the final fate of collapse ? what is the horizon structure in the presence of the second order perturbative effects of quantum gravity ? whether such solutions lead to naked singularities ? do they get covered due to departure from spherically symmetry ? does the nature of the singularity changes in a more fundamental theory preserving cosmic censorship @xcite ? in this section , we derive the relevant equations for szekeres model for irrotational dust to the 5d gauss - bonnet extended einstein equation 5d - qsz - egb solutions . we begin with the following 5d action : @xmath5 + s_{matter},\ ] ] where @xmath6 is 5d the ricci scalar , and @xmath7 is 5d gravitational constant which is set to unity . the gauss - bonnet lagrangian is of the form @xmath8 where @xmath9 is the coupling constant of the gauss - bonnet terms . this type of action is derived in the low - energy limit of heterotic superstring theory @xcite . in that case , @xmath9 is regarded as the inverse string tension and positive definite , and we consider only the case with @xmath10 in this paper . the action ( [ action ] ) leads to following set of field equations : @xmath11 where @xmath12 is the einstein tensor and @xmath13 \nonumber \\ & & -\frac{1}{2}g_{ab}l_{gb},\end{aligned}\ ] ] is the lanczos tensor . to find qsz spacetime , szekers @xcite imposed no _ a priori _ symmetry assumption , but instead postulated the metric in a special form which , for our 5d case , can be written as : @xmath14 the geodesic fluid flow vector is @xmath15 and the coordinates ( @xmath16 ) are comoving spatial coordinate constant along each world - lines so that @xmath17 . @xmath18 and @xmath19 are functions of @xmath20 to be determined from the einstein equations . the energy - momentum tensor for dust is @xmath21 it is seen that there exist solutions according as @xmath22 or as @xmath23 ( throughout this paper @xmath24 and @xmath25 ) . in 4d , the family of the solution corresponding to @xmath26 is a coincidental generalization of the friedman and kantowski - sachs models . @xcite pointed out , in 4d , that the case @xmath22 found no useful application in astrophysical cosmology . hence , we shall confine our discussion to the case @xmath23 . after the einstein equations are solved , it is required that there must exists two functions @xmath27 and @xmath28 such that @xmath29 @xmath30 here , @xmath31 is an arbitrary function of @xmath32 with restriction that @xmath33 . the solution for @xmath34 reads @xmath35 with the free functions @xmath36 , @xmath37 , @xmath38 , @xmath39 and @xmath40 satisfying an algebraic equation @xmath41 the factor @xmath4 determines whether the 2-surface is spherical ( @xmath4 = + 1 ) , pseudospherical [ hyperbolic ] ( @xmath4 = -1 ) or planar ( @xmath4 = 0 ) @xcite . upon introducing transformations @xmath42 the 3-metric yields the more familiar spherical form @xmath43 as can be seen , if @xmath44 = const , and @xmath32= const , the above metric becomes the metric of the three - dimensional sphere . hence , every @xmath44 = const , and @xmath32= const slice of the szekeres spacetime is a sphere of radius @xmath6 . thus the term quasispherical adapted . however , the spheres under consideration are not concentric . thus , the qsz solutions considered here are a generalization of the 5d - ltb solutions in which sphere of constant mass are made nonconcentric . the @xmath36 , @xmath37 , @xmath38 , @xmath39 and @xmath40 determine how the center of a 3-sphere changes its position in a @xmath45 constant space when the radius of the sphere is changed . when the spheres are concentric , the metric ( [ metric ] ) becomes the line element of the 5d - ltb model @xcite which arises when we set @xmath46 , with no loss of generality . the acceleration equation is given by the @xmath47 component of the field equations @xmath48 - \frac{(w^2 - 1)}{r^2}+ \frac{\dot{r}^2}{r^2 } = 0 . \label{dyn01}\ ] ] thus , it is necessary that @xmath6 satisfies the _ friedmann_-like equation @xmath49 = ( w^2 - 1 ) + \frac{f}{r^2 } - 2 \alpha \frac{\dot{r}^4}{r^2}.\label{eq : fe}\ ] ] here , @xmath50 is an arbitrary function of @xmath32 and is referred to as mass function . in a newtonian limit @xmath51 is equal to the mass inside the shell of radial coordinate @xmath32 , assuming that the mass function has no angular dependence . the function @xmath52 must be positive , because @xmath53 implies the existence of negative mass . equation ( [ eq : fe ] ) is the master equation of the system which governs the dynamical properties of the system , which is the same as one obtains in a spherically symmetric collapse of dust in egb @xcite . hence , we can refer to the the 5d - qsz - egb discussed here as quasispherical collapsing space - times . it is now straight forward to calculate the energy density @xmath4 , which we obtain by substituting the solutions for @xmath18 and @xmath19 . the mass density is @xmath54 this solution has in general no symmetry . for the spherical symmetry these solution corresponds to more conversant 5d tolman - bondi models @xcite . in many instance , it may be preferential to use in our work not density @xmath4 , but mass @xmath55 . @xmath56 is mass within sphere ( @xmath32=constant ) at given time @xmath44 , which we compute as follows @xmath57 upon using the density eq.([density ] ) in ( [ mass1 ] ) , we get @xmath58\end{aligned}\ ] ] now @xmath59 is a metric on the unit 3-sphere , so @xmath60 so that @xmath61 and @xmath62 this is a generalization of a formula obtained for mass function in 4d qsz @xcite . in addition , @xmath63 implies @xmath64 or @xmath65 . it is easy to see that as @xmath66 the master solution ( [ eq : fe ] ) of the system reduces to the corresponding 5d - qsz solution in @xcite @xmath67 rotation and acceleration of the dust source are zero , the expansion is @xmath68 and the shear is @xmath69.\ ] ] the 5d - ltb limits of these scalars can be obtained by replacing @xmath19 by @xmath6 in the above expressions . we assume , as in qsz models @xcite , the following regularity conditions:- 1 . the metric is @xmath70 . there are no shell crossing singularities ( @xmath71 ) . the metric is locally euclidean , i.e. , we must have @xmath72 the dynamical properties of the model discussed above is governed by eq . ( [ eq : fe ] ) , which is the same as one gets in the corresponding spherical symmetric collapsing solutions @xcite . to get the 5d homogeneous dust limit , we put @xmath73 , where @xmath74 and k is constant . then one gets three type of solutions - parabolic , elliptic or hyperbolic according as @xmath75 , @xmath76 or @xmath77 . in particular , the 5d friedmann - robertson - walker metric follows when @xmath78 and @xmath79 , where @xmath80 is the curvature index of 5d friedmann - robertson - walker models and @xmath81 . then the metric ( [ metric ] ) obtains usual form @xmath82 it may be noted that in the general relativistic case ( @xmath84 ) , eq . ( [ eq : fe ] ) has three types of solutions and @xmath85 , @xmath86 or @xmath87 determines the type of evolution . from eq . ( [ eq : fe ] ) , we obtain @xmath88 there are two families of solutions that correspond to the sign in front of the square root in eq . ( [ dyn0 ] ) . we call the family that has the minus ( plus ) sign the minus ( plus ) branch solution . in the general relativistic limit @xmath89 , we recover the @xmath90-qsz solution in einstein gravity @xcite . maeda @xcite has analyzed ltb models near center ( @xmath91 ) in egb and pointed out the occurrence of major changes in the final fate of collapse ( see also , @xcite ) . here we present the 5d - qsz - egb exact solution in close form , which facilitates us to analyze the final fate of gravitational collapse . the condition @xmath83 , is the marginally bound condition , meaning the collapsing shell is at rest at spatial infinity @xmath92 with zero energy in the infinite past . in the present discussion , we are concerned with gravitational collapse , which requires @xmath93 , i.e. , we assume that all portion of the dust cloud are momentarily collapsing . ( [ dyn0 ] ) can be integrated to @xmath94^{1/2 } } \right ] \nonumber \\ + \sqrt{\frac{\alpha r^2}{\sqrt{r^4 + 8\alpha f}-r^2}}\ ; , \quad\end{aligned}\ ] ] where @xmath95 is an arbitrary function of integration . here we note that the solutions @xmath27 are same as the 5d - ltb - egb models @xcite and are not affected by the dependence of the szekeres model on the @xmath96 coordinates . as it is possible to make an arbitrary relabeling of spherical dust shells by @xmath97 , without loss of generality , we fix the labeling by requiring that , on the hypersurface @xmath98 , @xmath32 coincide with the area radius @xmath99 this corresponds to the following choice of @xmath100 : @xmath101^{1/2}}\right ] \nonumber \\ + \sqrt { \frac{\alpha}{\sqrt{1 + 8\alpha { \tilde f}}-1}}\ ; , \quad\end{aligned}\ ] ] where @xmath102 . now , we discuss the nature and occurrence of curvature singularities in the 5d - qsz - egb models . it follows from eq . ( [ density ] ) that the curvature singularities occurs in 5d - qsz - egb solutions when @xmath103 szekeres @xcite analyzed the singularities of the first kind and second kinds according to the above two conditions . as is the case of the ltb , in 4d , @xmath104 corresponds to a shell crossing , however it is qualitatively different from that which occurs in ltb models ( see @xcite for details ) . in this paper we shall restrict ourselves to the singularities of the first kind which we call the central singularity . let us assume at @xmath105 , we have @xmath106 , which is the time when the matter shell @xmath32 = constant hits the physical singularity , i.e. , @xmath107 is defined by @xmath108 . the central singularity curve can be obtained using eq . ( [ solution ] ) as @xmath109 which represents the proper time for the complete collapse of a shell with coordinate @xmath32 . interestingly , positive @xmath9 delays the formation of singularity . in the limit of vanishing @xmath9 we recover the crunch time for relativistic 5d - qsz - egb . the eight arbitrary functions @xmath36 , @xmath37 , @xmath38 , @xmath39 , @xmath40 @xmath51 , @xmath110 and @xmath111 completely specify the dynamics of collapsing shells . however , only seven are independent because relation ( [ abc ] ) is between them . with the choice of @xmath32 , one can fix one more function and thus the total degree of freedom is six . the corresponding 5d - ltb - egb models have only three free functions : @xmath51 , @xmath112 and @xmath111 . hence the 5d - qsz - egb is functionally more generic . in order to study the collapse of a finite spherical body in egb , we have to match the 5d - qsz - egb solution along the timelike surface at some @xmath113 to the @xmath90-egb schwarzschild exterior discovered by boulware and desser @xcite , and wheeler @xcite . the analysis is similar to the case of matching 5d - ltb - egb to the @xmath90-egb schwarzschild exterior @xcite and , will not be discussed here . bonnor @xcite was the first to proved that szekeres spacetime can be matched to schwarzschild vacuum spacetime . the apparent horizon ( ah ) is the outermost marginally trapped surface for the outgoing photons . the ah can be either null or spacelike , that is , it can `` move '' causally or acausally . the main advantage of working with the apparent horizon is that it is local in time and can be located at a given spacelike hypersurface . the event horizon instead is nonlocal . moreover , the event horizon is a an outer covering surface of apparent horizon and they coincide in case of static or stationary spacetime . trapped surface is defined as a compact spacelike 2-surface both of whose future pointing null geodesics families are converging . physically , it captures the notion of trapping by implying that if 2-surface @xmath114 ( @xmath115 = contant ) is a trapped surface then its entire future development lies behind a horizon . to obtain criterion for existence of such @xmath114 , let @xmath116 be the tangent vector to the null geodesics . it follows that along null geodesics , we have @xmath117 and @xmath118 . for the metric ( [ metric ] ) the above defined null congruence satisfies the following condition : @xmath119 on @xmath114 . a choice of affine parameter @xmath120 with @xmath121 , clearly satisfies the condition ( [ null_geo ] ) . the positive ( negative ) sign of invariant @xmath122 is determines the divergence ( convergence ) of the null geodesic . also @xmath123 corresponds to the outward and inward geodesic respectively . now , @xmath124 upon taking the time derivative of the null condition @xmath125 , we get @xmath126 in the second condition @xmath127 , consider the @xmath128 component ; @xmath129 eliminating @xmath130 from the previous two equations we can rewrite eq . ( [ cov_div ] ) as @xmath131 thus the apparent horizon must satisfy @xmath132 in absence of shell crossings ( @xmath23 ) , and using eqs . ( [ el ] ) and ( [ ew ] ) , we have @xmath133 considering eq . ( [ eq : fe ] ) , the apparent horizon condition ( [ app1 ] ) becomes @xmath134 one can also employ the usual condition for the existence of the apparent horizon @xmath135 considering eqs . ( [ el]),([ew ] ) and ( [ eq : fe ] ) , the apparent horizon condition ( [ app ] ) again gives the same result ( [ app - cond ] ) . in the relativistic limit , @xmath66 , @xmath136 @xcite . further , eq . ( [ app - cond ] ) has a mathematical similarity for the analogous situation in null fluid collapse where the expression for apparent horizon is @xmath137 @xcite . in this section , we analyze end state of the collapse of 5d - qsz - egb dust collapse in terms of the given regular initial density and velocity profile . we denote @xmath138 as the initial density of the dust cloud at @xmath139 for a fixed radial direction ( @xmath140 constant ) . it is assumed that @xmath141 , @xmath34 and @xmath52 to be expandable around @xmath142 on the initial hypersurface . further the function @xmath34 is restricted by eqs . ( [ enu ] ) and ( [ abc ] ) . henceforth , in this section , we adopt here a method similar to @xcite which we modify here to accommodate the higher dimension spacetime . we take @xmath143 @xmath144 where @xmath145 and the regularity condition implies that @xmath146 @xcite , and @xmath147 therefore , @xmath148 substituting @xmath149 and @xmath150 in the @xmath141 equation one gets , @xmath151 this implies that @xmath152 can not have angular dependence whereas @xmath153 ( for @xmath154 ) have angular dependence provided @xmath155 . the results obtained here are consistent with the earlier work @xcite , but with some changes in the coefficients due to 5d spacetime . since earlier analysis were done in qsz spacetime , as a result , one may conclude that the 5d - qsz - egb spacetime has same local nakedness behavior as the qsz spacetime . to conserve space , we shall avoid replication of all other analysis being similar to qsz spacetime @xcite . to further analyze the horizon curve , we combine eqs . ( [ solution ] ) and ( [ sing - curve ] ) giving @xmath156^{1/2 } } \right ] .\end{aligned}\ ] ] the apparent horizon in the interior of the dust ball lies at @xmath157 . the corresponding time @xmath158 is given by @xmath159 \nonumber \\ + \frac{1}{2 } \sqrt{f-2\alpha } \ ; . \quad\end{aligned}\ ] ] as mentioned above , at @xmath105 , we have @xmath160 , which is the time when the matter shell @xmath32 = constant hits the physical singularity . the singularity is at least locally naked if @xmath161 , and if @xmath162 , it is a black hole , and in case of the equality one has to compare the slopes . we consider the following three cases : [ [ case - i ] ] case i + + + + + + homogeneous case , i.e. , the density profile is homogenous and has no angular dependence , thus @xmath163 using eq . ( [ fc ] ) , it can be deduced that @xmath164 ( see also fig.1 ) , and hence singularity is naked . [ [ case - ii ] ] case ii + + + + + + + next , we assume that only @xmath165 implies that the density profile is inhomogeneous and has no angular dependence @xmath166 it is seen that @xmath164 which leads to formation of a naked singularity . [ [ case - iii ] ] case iii + + + + + + + + finally , let @xmath167 which means the density profile is inhomogeneous and has angular dependence @xmath168 in this case also singularity is naked because @xmath164 ( see also fig.1 ) . clearly , for a positive @xmath9 , the central shell does not get trapped , and the untrapped region around the center increases with increasing @xmath9 , for both homogeneous and inhomogeneous models . the center ( @xmath142 ) remains untrapped , since for nonzero values of the gauss - bonnet parameter @xmath169 , above eq . ( [ app - cond ] ) admits no solution . interestingly , the theory demands @xmath9 to be a positive number , which forbids the apparent horizon from reaching the center thereby making the singularity massive and eternally visible , which is forbidden in corresponding general relativistic scenario . it is seen here that the gauss - bonnet term modifies the time of formation of singularity , and the time lag between singularity formation and the apparent horizon formation , in contrast to the 5d dust models . indeed , the time for the occurrence of the central shell focusing singularity for the collapse is increased in comparison to the 5d - qsz case . the introduction of a gauss - bonnet term slows down the collapse process @xcite , which can be explained as follows . the contribution from the gauss - bonnet term leads , when we think in terms of a newtonian potential , to a repulsive term . ( [ eq : fe ] ) , for the marginally bound case , becomes @xmath170 in the @xmath171 limit we recover the 5d - qsz solution , and in this limit , eqs . ( [ eq : fe ] ) and ( [ eq : fe1 ] ) lead to the following expression for acceleration : @xmath172 in the 5d - qsz - egb spacetime the acceleration expression ( [ eq : fe ] ) can now be expressed as @xmath173^{-1/2 } .\ ] ] for a comparison with newton like force equation obtained in the 5d - qsz case @xmath174 , clearly a positive value of @xmath9 diminishes the acceleration in the collapse process . as we move out to larger shells and for large values of @xmath9 the detailed dynamics should depend on higher order terms in the expansion . the other reason may be , for slow down of the collapse process in 5s - qsz - egb , there is relatively less mass energy [ see eq . ( [ eq : m3 ] ) ] collapsing in the 5d - qsz - egb spacetime as compared to the 5d - qsz case . this can be seen from the mass function @xmath175 , which is given by @xmath176 using eqs . ( [ el ] ) , ( [ ew ] ) and ( [ eq : fe ] ) into eq . ( [ eq : m2 ] ) we get @xmath177 the quantity @xmath51 can be interpreted as energy due to the energy density @xmath4 . finally , for the reason mentioned above , the presence of the coupling constant of the gauss - bonnet terms @xmath9 produces a change in the location of these horizons . such a change could have a significant effect in the dynamical evolution of these horizons . for nonzero @xmath9 the structure of the apparent horizon is nontrivial . in general relativity noncentral singularity is always covered @xcite ( see also @xcite ) . however , in the presence of the gauss - bonnet term we find that even the noncentral singularity is naked , in spite of being massive ( @xmath178 ) . the szekers models , depending on the sign of @xmath4 , are subdivided in to the quasispherical , quasiplane and quasihyperbolic @xcite . the geometry of the later two is not really understood @xcite . on the other hand the quasispherical has been rather well investigated @xcite , and it has found important applications in cosmology and gravitational collapse . the qsz metric is a dust model which has no killing vector @xcite , but contains ltb model as a spherically symmetric special case . it has been found that the ltb metric admits both naked singularities and black holes depending upon the choice of initial data . indeed , both analytical @xcite and numerical results @xcite in dust indicate the critical behavior governing the formation of black holes or naked singularities . a similar situation also occurs in higher dimension the ltb models @xcite , and these results also carry over to qsz spacetime @xcite . maeda @xcite and we @xcite have shown that in spherically symmetric inhomogeneous dust collapse , the effect of adding a positive @xmath9 does radically alter the final fate and leads to formation of a massive timelike singularity which is prohibited in general relativity . in this work , we have obtained an exact solution in closed form in egb , which represents the quasispherical collapse of irrotational dust in 5d spacetime namely 5d - qsz - egb . the solution is nonsymmetric generalization of the spherically symmetric 5d - ltb - egb solutions . it can be reduced to 5d - qsz or 5d - ltb in the general relativistic case ( @xmath66 ) . our analysis also supports the earlier results @xcite , deducing that under physically reasonable initial conditions naked singularities do develop in the 5d - qsz - egb models , which are not spherically symmetric , and admit no killing vectors . the second order curvature corrections changes the final fate of gravitational collapse and the nature of singularity that occurs in 5d general relativistic dust models . however , mild departure spherical collapse can not alter the standard picture of the structure and formation singularity , since the 5d - qsz - egb solutions discussed here are qualitatively similar to the analogous spherical solutions . it is seen here that the gauss - bonnet term : ( i ) decelerates the collapse process ( ii ) alters the time of formation of singularities and the time lag between singularity formation , and ( iii ) modifies the apparent horizon formation and the location of apparent horizons . our analysis to examine the nature of singularities ( naked or hidden by horizon ) is based on the comparison of @xmath179 ( time for the formation of the apparent horizon or trapped surface ) and @xmath180 ( time for the formation of a central singularity ) . in qsz the singularity can be directionally naked @xcite . hence , it would be necessary to investigate in further details the final fate of the inhomogeneous dust collapse in the egb theory in order to bring out explicitly the difference in global nakedness when we depart from spherically symmetric @xcite . the conjecture that such a singularity from a regular initial surface must always be hidden behind an event horizon , called cosmic censorship conjecture ( ccc ) was proposed by penrose @xcite . the ccc forbids the existence of naked singularities . despite almost 30 years of effort , we are far from a general proof of ccc ( for recent reviews and references , see @xcite ) . but , significant progress has been made in trying to find counter examples to ccc . our analysis , as in the 5d - ltb - egb @xcite , shows that there exists a regular initial data which leads to a naked singularity and hence in our nonspherical case also the ccc is violated . the usefulness of these models is that they do offer an opportunity to explore the properties of singular spacetime . the investigation of a mild departure from standard spherical symmetry models may be valuable in attempts to put ccc in concrete mathematical form . finally , studying such models which lacks symmetry is important , so that one can check which properties of gravitational collapse are preserved . this may also helps us to bring out some universal features in the theory of gravitational collapse @xcite . we have shown here that there exist regular initial data which leads to a naked singularity violating ccc . however , this may not be a serious threat to ccc because of the following two reasons viz . ( i ) the matter considered here is dust which is only an effective , macroscopic approximation to a fundamental description of matter @xcite . ( ii ) in general relativity the energy - momentum tensor given by eq . ( [ eq : emt ] ) satisfies the weak energy condition . however , this may not be true in egb because the gauss - bonnet term itself violates the energy condition . it is our pleasure to thank n. dadhich for helpful comments and discussion . sgg would also like to thank iucaa , pune for hospitality while part of this work was done . sj acknowledges support under dst project no.sr/s2/hep-002/2008 . 99 p. szekeres , commun . 41 , 55 ( 1975 ) ; p. szekeres , phys . d 12 , 2941 ( 1975 ) . a. krasi@xmath181ski , inhomogeneous cosmological models cambridge university press , cambridge , england , 1997 ) . p. s. joshi , _ gravitational collapse and spacetime singularities _ ( cambridge university press , cambridge , england , 2007 ) . p. s. joshi and a. krolak , class . * 13 * , 3069 ( 1996 ) . s. s. deshingkar , s. jhingan and p. s. joshi , gen . * 30 * , 1477 ( 1998 ) u. debnath , s. nath and s. chakraborty , mon . not . * 369 * , 1961 ( 2006 ) s. m. c. goncalves , class . grav . * 18 * , 4517 ( 2001 ) . w. b. bonnor , commun . 51 , 191 ( 1976 ) ; w. b. bonnor , a.h . sulaiman , and n. tomimura , gen . gravit . 8 , 549 ( 1977 ) ; w. b. bonnor , classical quantum gravity 3 , 495 ( 1986 ) ; w. b. bonnor and d. j. r. pugh , south afr . j. phys . * 10 * , 169 ( 1987 ) . s. g. ghosh and a. beesham , phys . rev.d * 64 * , 124005 ( 2001 ) . s. g. ghosh and a. banerjee , int d * 12 * , 639 ( 2003 ) . s. g. ghosh and d. w. deshkar , astrophys . space sci . * 310 * , 111 ( 2007 ) . d. christodoulou , commun . phys . * 93 * , 171(1984 ) . f. cooperstock , s. jhingan , p. s. joshi and t. p. singh , class . quantum grav . * 14 * , 2195 ( 1997 ) . i. h. dwivedi and p. s. joshi , classical quantum gravity * 9 * , l69 ( 1992 ) ; p. s. joshi and t. p. singh , gen . * 27 * , 921 ( 1995 ) ; _ ibid . d * 51 * , 6778 ( 1995 ) . p. s. joshi and i. h. dwivedi , phys . d * 47 * , 5357 ( 1993 ) . s. jhingan , p. s. joshi and t. p. singh , class . quant . grav . * 13 * , 3057 ( 1996 ) d. m. eardley and l. smarr , phys . d * 43 * , 2239 ( 1979 ) ; d. m. eardley , in _ gravitation in astrophysics , asi series , proceeding of the nato advanced study institute , cargese , france , 1986 _ , edited by b. carter and j. b. hartle ( proceeding of the nato advanced study institute , series b : physics , vol 156)(plenum press , new york , 1986 ) , p 235 . r. penrose , riv . nuovo cimento * 1 * , 252 ( 1969 ) ; in _ general relativity - an einstein centenary volume _ , edited by s. w. hawking and w. israel ( cambridge university press , cambridge , england , 1979 ) pp 581 - 638 . p. s. joshi , _ global aspects in gravitation and cosmology _ ( clendron , oxford , 1993 ) ; c. j. s. clarke , class . quantum grav . * 10 * , 1375 ( 1993 ) ; r. m. wald , arxiv : gr - qc/9710068 ; s. jhingan and g. magli , arxiv : gr - qc/9903103 ; t. p. singh , j. astrophys . astron . * 20 * , 221 ( 1999 ) ; p. s. joshi , pramana * 55 * , 529 ( 2000 ) ; t. harada , pramana * 63 * , 741 ( 2004 ) .	we obtain a general five - dimensional quasispherical collapsing solutions of irrotational dust in einstein gravity with the gauss - bonnet combination of quadratic curvature terms . these solutions are generalization , to einstein - gauss - bonnet gravity , of the five - dimensional quasispherical szkeres like collapsing solutions in general relativity . it is found that the collapse proceed in the same way as in the analogous spherical collapse , i.e. , there exists a regular initial data such that the collapse proceed to form naked singularities violating cosmic censorship conjecture . the effect of gauss - bonnet quadratic curvature terms on the formation and locations of the apparent horizon is deduced . = 5000
the discovery that the current universe is in an accelerating expansion phase , obtained from the observations of type ia supernovae @xcite , inaugurate an era of intense theoretical research to understand the mechanism driving this accelerating expansion . an immediate consequence is that gravity behaves differently on cosmological distance scales . a variety of possible solutions to the cosmic acceleration puzzle have been debated during this decade including the cosmological constant , exotic matter and energy , modified gravity , anthropic arguments , etc . the most favored ones are dark energy based models and modified gravity theories such @xmath1 gravity and dgp gravity with an equation of state parameter @xmath2 ( where @xmath3 and @xmath4 are the energy density and pressure of the cosmic fluid , respectively ) . the dark energy component is usually described by an equation - of - state parameter @xmath5 . the simplest and traditional explanation for dark energy is a cosmological constant , which is usually interpreted as the vacuum energy . however , the value required to explain the cosmic expansion , which is of the order of @xmath6 can not be explained by current particle physics . this is known as the cosmological constant problem . on the other hand , the first year wmap data combined with the 2df galaxy survey and the supernova ia data favor the phantom energy equation - of - state of the cosmic fluid @xmath7 over the cosmological constant and the quintessence field . a candidate for phantom energy is usually a scalar field with the wrong sign for kinetic energy term @xcite . one crucial fate of an expanding universe driven by phantom energy is that the phantom energy density and the scale factor diverge in finite time , ripping apart all bound systems of the universe ( galaxies , stars , atoms , nuclei ) , before the universe approaches the big - rip singularity @xcite . in this work we are interested by the effect of cosmological expansion on local systems , namely a black hole embedded in a an expanding friedmann - lemaitre - robertson - walker ( flrw ) universe driven by phantom energy . the study of non - stationary event horizons and their thermodynamical parameters , relevant for a quantum theory of gravity , are currently attracting a great interest @xcite-@xcite . in the pure expanding flat flrw universe dominated by phantom energy , the radius of the observer s event horizon diminishes with time and consequently the horizon entropy , @xmath8 an other example is the quasi - de sitter space , where the event horizon and apparent horizon are different @xcite . it was found that the first law and second law of thermodynamics can not hold when both the horizons are considered . other authors discussed the effect of the presence of black holes in expanding phantom energy - dominated universe on the validity of the generalized second law of gravitational thermodynamics ( gsl ) @xcite . in this paper , we present a detailed calculation of the entropy of the solution of einstein s equations recently found and representing a black hole embedded in an expanding friedmann - lemaitre - robertson - walker ( flrw ) universe@xcite . our approach is based on the brick - wall method ( bmw ) invented by thooft @xcite . in this method a brick wall around the horizon prevents the divergence of the free energy ( or entropy ) of the black hole , which is identified with the canonical ensemble statistical - mechanical free energy ( or entropy ) due to quantum excitations in thermal equilibrium with the black hole . however , the bmw can not be used for non - equilibrium systems , like black holes with multi - horizons , where each horizon can be considered as an isolated thermodynamical system . the solution to the problem , known as the thin - layer improved method , consists in considering a thin layer near the horizon as a local equilibrium system and invoking thermodynamics of composite systems , such that the entropy of a multi - horizons black hole is the sum of the contributions arising from each horizon @xcite . then in the improved bwm , the global equilibrium has been replaced by a local equilibrium on microscopic scales . this method has been applied to a wide range of black holes with multi - horizons like schwarzschild - de sitter black hole , kerr - de sitter black hole , vaidya black hole , the 5d ricci - flat black string and recently to the calculation of the entropy in the pure expanding flrw universe @xcite . the organization of the paper is as follows : in section ii we review the exact solution describing a black hole embedded in an expanding friedmann - lemaitre - robertson - walker ( flrw ) universe . in section iii , the statistical mechanical free energy , thermal energy and entropy due to a scalar field are computed using the improved thin - layer bwm . then we discuss the conditions under which the d - bound conjecture @xcite is protect in expanding universe in the presence of black holes . in section iv , our attention will be focused on the conditions under which the second law of thermodynamics and the gsl are satisfied when the universe is driven by phantom energy . finally , we discuss and summarize our results in section v. the first solution of einstein s theory of general relativity describing a black hole like object embedded in an expanding universe was introduced by mcvittie in 1933 @xcite , and is given in isotropic coordinates by @xmath9 where @xmath10 is the scale factor and @xmath11 is the mass of the black hole in the static case . in fact , when @xmath12 it reduces to the schwarzschild solution . when the mass parameter is zero , the mcvittie reduces to a spatially flat flrw solution with the scale factor @xmath10 . the global structure of ( [ eq : mcvittie ] ) has been studied and particularly it has been shown that the solution possesses a spacelike singularity on the 2-sphere @xmath13 and can not describe an embedded black hole in an expanding spatially flat flrw universe @xcite . in the following we adopt the new solution describing a black hole embedded in a spatially flat flrw universe @xcite @xmath14 using the areal radius @xmath15 the metric takes the following suitable painlevel@xmath16-gullstrand form ds^2= & -dt^2+(1-)^-1dr^2[eq : metric02 ] + & -2rh(1-)^-1dtdr+r^2d^2 , where @xmath17 is the hubble parameter and overdot stands for derivative with respect to the cosmic time . we observe that the term @xmath18 plays the role of variable cosmological constant . in order to write the metric in the nolan gauge , we introduce the time transformation @xmath19 to remove the @xmath20 term @xmath21,\label{eq : time - transform}\ ] ] where the integrating factor @xmath22 satisfy @xmath23.\label{eq : diff - exact}\ ] ] then substituting @xmath24 into @xmath25 and replacing @xmath26 , we obtain @xmath27 where the two - dimensional metric is @xmath28f^{2}dt^{2}+\left[\left(1-\frac{2m_{0}a}{r}\right)-\frac{r^{2}h^{2}}{\left(1-\frac{2m_{0}a}{r}\right)}\right]^{-1}dr^{2}.\label{eq : metric - nolan}\ ] ] the apparent horizons ( ah ) are solutions of the equation @xmath29 which leads to @xmath30 where we have used the hawking - hayward quasi - local mass @xmath31 this mass is always increasing in an expanding universe . discarding the unphysical branch with the lower sign , we obtain r_c= & ( 1+),[eq : ahc ] + r_b= & ( 1-),[eq : ahb ] where @xmath32 and @xmath33 are the cosmic and the black hole ah , respectively . we observe that the ah coincides at a time @xmath34 for which @xmath35 this coincidence takes place in a future or past universe depending on the kind of matter accretion onto the black hole . now we consider the statistical mechanical entropy that arises from a minimally coupled quantum scalar field of mass @xmath36 in thermal equilibrium at temperature @xmath37 using the thin layer improved bwm method . the field equation on the background given by ( [ eq : metric02 ] ) is @xmath38 where @xmath39 . using the relation @xmath40 , the klein - gordon equation takes the form ( r , t ) & = 0 . in the semi - classical approximation , the wave - function is obtained by the ansatz @xmath41 . neglecting higher orders contributions and substituting in eq.@xmath42 with the ansatz @xmath43 , we obtain b^2 + 2hrb+^2[eq : momentum ] + -(1-(1-r_ah))(^2+r^-2l(l+1 ) ) & = 0 . in our derivation we have assumed the constancy of the frequency @xmath44 near the ah and that @xmath45 . solving for @xmath46 we obtain @xmath47 where we have set @xmath48 which is just @xmath49 . the sign ambiguity in eq.@xmath50 is related to outgoing @xmath51 or ingoing @xmath52 particles . the number of radial modes with energy less that @xmath44 is defined by @xmath53 where we used the average of the radial momentum . the second term in ( [ n ] ) is caused by a different direction such that the outgoing and ingoing particles are taken into account . using eq.@xmath50 we get @xmath54 now , according to the canonical ensemble theory the free energy is defined by @xmath55 integrating by parts we obtain @xmath56 where @xmath57 is the total number of modes with energy less than @xmath44 given by @xmath58 substituting the expression of @xmath59 and restricting the integration over @xmath60 to the range @xmath61,\ ] ] the free energy takes the form @xmath62 the integral over the radial variable is determined by the improved thin - layer bwm boundary conditions @xmath63 where @xmath64,@xmath65 are ultraviolet cutoffs and @xmath66 are thickness of the thin layer , near the bh horizon and cosmomlogical horizon , respectively . then , the integrals over @xmath67 are performed by noting that @xmath68 in the near vicinity of the ah . in fact @xmath1 can be approximated by @xmath69 then using the near horizon approximation @xmath70 , and the following integral @xmath71 the free energy is written as @xmath72 with f_b= & -_r_b+h_b^r_b+h_b+_bdr , + f_c= & -_r_c - h_c^r_c - h_c-_cdr , where @xmath73 and @xmath74 iare the hawking temperatures evaluated at the bh and cosmic ah , respectively . using @xmath75 we obtain @xmath76.\ ] ] let us now introduce the invariant distances @xmath77 and @xmath78 defined by _ b & = _ r_b^r_b+h_b=,[eq:-1 ] + _ b & = _ r_b+h_b^r_b+h_b+_b=,[eq : invariant bhah ] the invariant distances near the cosmic ah , @xmath79 and @xmath80 , are obtained by the substitutions @xmath81 and @xmath82 in the above relations . then , the free energy takes the form @xmath83.\ ] ] the internal energy and entropy defined by @xmath84 and @xmath85 , respectively , are then obtained as u= & , + s= & . using the area of the bh horizon and cosmic ah , @xmath86 and @xmath87 , respectively , we rewrite the entropy as @xmath88.\label{eq : entropy03}\ ] ] now the inverse temperature is calculated at the ah using @xmath89 , where the dynamical surface gravity associated with the ah is @xmath90@xcite , and where the metric @xmath91 is defined by @xmath92 . the calculation yields@xmath93 where @xmath94 , @xmath95 . we also show that the misner - sharp energy defined by @xmath96 is given by @xmath97 writing @xmath98 is terms of the ah , the temperature and the misner - sharp energy become @xmath99 @xmath100 the temperature associated with dynamical horizons is not only related to the misner - sharp mass on the ah , but also to the matter content of the universe @xcite . here , we point that in a series of papers devoted to thermodynamics of expanding flrw universe with dark energy it is assumed the usual law @xmath101 for stationary horizons@xcite , which certainly invalids the results obtained @xcite . the expression of temperature ( [ eq : temp ] ) is similar with the temperature in york theory of black holes confined in isothermal cavities @xcite . in this context , a local observer at rest will measure a local temperature which scale as @xmath102 for a self - gravitating mass in thermal equilibrium with the boundary of the cavity . the effect of the boundary is reflected in the relation ( [ eq : misner - sharp ] ) , which state that the hawking - hayward quasi - local mass is the sum of the thermal energy and a gravitational self - energy due to the cavity . in our case , the black hole is embedded in an expanding universe , and the cosmic apparent horizon plays the role of the boundary of the cavity . on the other hand the behavior of temperature with time is contrasted with the naive expectation that the temperature is redshifted by the scale factor @xmath10 relative to its value in static background . the additional correction terms to the temperature become important when the density energy of the universe is negligible compared to the expansion rate . the same remark apply to the entropy which will show a behavior with time different from the expected one @xmath103 . here some comments are appropriate about the assumption of local equilibrium used in the calculation , even we are considering dynamical horizons . the improved thin - layer bmw method is based on the fact that hawking radiation is mainly due to the vacuum fluctuations near the black hole horizon(s ) , and then allows to identify the black hole entropy with the statistical - mechanical entropy associated with quantum fields outside the black hole and confined in small region(s ) in the vicinity of the horizon(s ) . this cure the problem of maintaining thermal equilibrium at large scales of the original bwm of thooft , without taking into account the back - reaction of the background . however , in small regions we are dealing with the assumption of local equilibrium and the validity of the statistical laws we used in the calculation , and therefore need a physical justification . the local equilibrium state is achieved if thermodynamics properties of the system are slowly varying . the last condition when applied to the case of the case of a black hole immersed in an expanding flrw universe is equivalent to the following statement : `` local equilibrium is achieved if local thermalization of the system is faster than the expansion rate of the universe '' . in fact assuming that @xmath104 and using eq.(39 ) we obtain @xmath105 this condition can be rewritten as @xmath106 which is an expression of the last statement in the pure flat flrw case where only the cosmic ah survives , @xmath107 , we obtain @xmath108 which is proportional to the de sitter temperature , @xmath109 now , substituting eq.@xmath110 in @xmath111 the entropy takes the form @xmath112.\label{eq : entropy}\ ] ] in order to compare our results with the standard flrw universe , we set the cutoffs as @xmath113 and introduce the following factors @xmath114 then the final expressions of thermal energy and entropy take the forms @xmath115 @xmath116 we assume now that matter in the universe is in the form of an imperfect fluid with a radial heat flux , described by the stress - energy tensor @xmath117 where @xmath118 is the fluid four velocity@xmath119 , @xmath120 a spacial vector describing radial heat current , and @xmath121 the general solution of the einstein equations are given by @xcite , 8g=3 & ( ) ^2h^2,[eq : ae1 ] + 8gp= & -()^2.[eq : ae2 ] assuming @xmath122 and radial energy inflow ( @xmath123 ) , the accretion rate can be written as@xmath124 where @xmath125 . in terms of the comoving ah , these equations reduce to h= & r_a,[eq : e3 ] + 3h+= & -8gr_ap.[eq : e4 ] at this stage we would like to comment about the law of temperature @xmath126 widely used in the literature . with the aid of eqs.([eq : e3],[eq : e4 ] ) we observe that the later law is only valid if the energy density of the universe is significant relative to the expansion rate @xmath127 now , it is instructive to rewrite the dynamical surface gravity as an invariant . in fact using the misner - sharp mass @xmath128 and defining the projection of the ( 3 + 1)-dimensional stress - energy tensor in the normal direction of the 2-sphere , @xmath129 , we obtain @xmath130 where we have set @xmath131 . the later invariant relation has been recently proved in different gauges @xcite . on the other hand this equation leads immediately to the analogue of first law for dynamical black holes . in fact using @xmath132 and @xmath133 as the ah area and volume associated with the ah , we obtain @xmath134 assuming now an equation of state of the form @xmath135 and using eqs.([eq : e3],[eq : e4 ] ) , we rewrite the factors @xmath136 as functions of the eos parameter @xmath137 , @xmath138 where we defined the reduced ah @xmath139 . the factor in the standard flrw universe , already given in @xmath140 , follows by setting @xmath141 in the second relation @xmath142 which is exactly the entropy factor obtained in @xcite . in this case the temperature associated with the cosmic ah becomes @xmath143 since @xmath144 in a universe driven by phantom energy , this result put some doubts on the claim that the gsl in the presence of a black hole in a phantom energy - dominated universe is protected if the temperature is written as @xmath145 with @xmath146 @xcite . in this work , the backreaction of the phantom fluid on the black hole has been ignored as the authors assumed the usual pure flat flrw metric . in figure 1 we plotted the behavior of the factors @xmath147 and @xmath148 as functions of the eos parameter @xmath137 for different values of the bh and cosmic ah , respectively . it is interesting to note that the factors @xmath147 and @xmath148 do not depend on the ah when @xmath149 and @xmath148 with the eos parameter @xmath137 for different reduced ah @xmath150 and @xmath151 on the left and right panel , respectively.,title="fig:",width=302,height=302 ] and @xmath148 with the eos parameter @xmath137 for different reduced ah @xmath150 and @xmath151 on the left and right panel , respectively.,title="fig:",width=302,height=302 ] in the following let us check if the expression of the total entropy meets the d - bound conjecture @xcite generalized to dynamical black holes . the later conjecture asserts that the entropy of a system within a boundary is less than or equal to the gravitational entropy , @xmath152 we consider the special ranges of values of @xmath150 and @xmath151 used in figure 1 , and look for the possible values of the eos parameter @xmath137 for which the d - bound conjecture is protected . in fact , from the relations giving the ah in eqs.([eq : ahc]-[eq : ahb ] ) , and assuming that @xmath153 , we have @xmath154 for @xmath155 and @xmath156 , and @xmath157 for @xmath158 and @xmath159 . then , within these ranges of the eos parameter and the ah , it is immediate to verify that the entropy associated with the bh and cosmic ah satisfy the d - bound conjecture , @xmath160 respectively . when @xmath161 the bh horizon disappears and we obtain @xmath162 for @xmath163 to verify the d - bound conjecture for the total entropy @xmath164 , we have to verify that the following condition @xmath165 holds , where @xmath166 corresponds to the ah in the pure flat flrw universe . using @xmath167 , the condition ( [ eq : d - bound ] ) is rewritten as @xmath168 where we have used the reduced ah . since we have @xmath169 , the condition ( [ eq : dec ] ) becomes @xmath170 we easily verify that the above condition is verified for an eos parameter in the interval @xmath171 , and @xmath158 . and @xmath155 . on the other hand , if we restrict the values of the reduced cosmic ah to @xmath172 , the condition ( [ eq : cond ] ) is satisfied for @xmath173 the later values of @xmath151 corresponds to a small hawking - hayward quasi local mass . finally , we point that even the phantom field is unstable and violate the null , strong and dominant energy conditions ( null , sec , dec ) @xcite , the entropy of dynamical black holes immersed in phantom energy - dominated universe , meets the d - bound conjecture . in what follows , we would like to study the time evolution of the thermodynamics parameters when the black hole is embedded in an flrw universe driven by phantom energy and accreting this cosmic fluid . in this situation , the scale factor is given by @xmath174 where @xmath175 in ( [ eq : a1 ] ) is the big rip time . in that case , as time increases , the black hole horizon increases monotonically while the cosmic ah decreases monotonically . the ah coincide when @xmath176^{\left\|\frac{3(w+1)}{3w+1}\right\|}$ ] , and after which the black hole singularity will become naked in a finite time , violating the cosmic censorship conjecture @xcite . in figure 2 we show the evolution of entropy and free energy with time for @xmath177 , and different initial mass of the black hole . we observe , that entropy is decreasing and increasing at early times and late times when approaching @xmath34 , respectively , and diverges at the critical ah radius @xmath178 this indicates that at times approaching @xmath34 , the second law of thermodynamics is protected , @xmath179 , while at early times the second lwa of thermodynamics is violated . the important result to note here is that , taking into account the effect of backreaction of the phantom fluid on the black hole in an expanding universe , the second law of thermodynamics becomes valid when approaching the coincidence time , while it is always violated in a purely phantom - energy dominated universe . figure 2 : variation of entropy and free energy as functions of time with @xmath180 from right to left we have @xmath181 it has been advanced that when a black hole is embedded in an expanding universe driven by phantom energy , the gsl is verified under some restrictive conditions @xcite . let us now proceed to discuss the gsl , @xmath182 , where @xmath183 is the geometric entropy associated with the ah and @xmath184 the entropy of the cosmic fluid confined between the bh and the cosmic ah . the entropy of the fluid can be obtained by using the gibbs equation @xcite , @xmath185 where @xmath186 is the temperature of the energy in the vicinity of the ah . using the continuity equation , which is given in our model by @xmath187 and assuming that the cosmic fluid is in thermal equilibrium with the boundary , the evolution of the fluid entropy is then given by @xmath188 using this relation and since the temperature is meaningful only in the near horizon region , we assume that the fluid entropy is the sum of the contributions near the bh and cosmic ah , respectively , @xmath189\right).\ ] ] let us now consider explicitly the case where the hawking - hayward mass is enough small so that the quantities associated with the bh horizon can be neglected . then , we can set @xmath190 and the horizon entropy is due essentially to the contribution near the cosmic ah @xmath191 where @xmath192 is given by ( [ eq : flrw - w ] ) . the time derivative of the phantom fluid entropy becomes @xmath193\right).\label{eq : s - phantom1}\ ] ] now taking the time derivative of @xmath194 , we obtain@xmath195\right)+2\mathcal{f}_{0}\pi r_{c}\dot{r}_{c}.\ ] ] the gsl states that @xmath196 note that in the phantom era @xmath197 , @xmath198 , and since @xmath199 a necessary condition for the gsl to be satisfied is@xmath200\geq0.\ ] ] this condition can be integrated and leads to @xmath201}.\ ] ] using the expression of the cosmic ah , we get the following condition on the derivative of the hawking - hayward quasi - local mass @xmath202}-1\right]^{2}\right].\ ] ] now substituting the scale factor given by ( [ eq : a1 ] ) , and the expression of the hubble parameter we finally obtain @xmath203}}{3\left(1+w\right)}\left(t_{rip}-t\right)^{\frac{-3w}{1 + 3w}}+1\right]^{2}\right].\ ] ] in figure 3 , we plotted the variation of @xmath204 with time for different values of the eos parameter . knowing that the hawking - hayward mass is an increasing function of time in an expanding universe , we observe that the gsl remains protected from the past to the present time if @xmath205 figure 3 : variation of the derivative of the hawking - hayward quasi - local mass as a function of time . using @xmath206 and considering the present time , we get for the mass of the black hole the following constraint@xmath207 assuming the positivity of the mass , we obtain again @xmath208 in order for the gsl to be satisfied . this value of the eos parameter is compatible with the analysis performed on the validity of the d - bound conjecture in the case where the cosmic ah is close to the critical value @xmath209 in this paper we have studied the thermodynamical properties of black holes immersed in an expanding spatially flat flrw universe . we have particularly calculated the entropy and temperature associated with the apparent horizons using the improved thin - layer brick wall method and the dynamical surface gravity , respectively . the temperature and entropy at the apparent horizons ( ah ) display a non trivial behavior as a function of time , and are not scaled by the expected factors @xmath10 and @xmath211 , respectively . the correction terms become relevant when the expansion rate is significant relative to the density energy of the universe . on the other hand , we found that the sum of entropies associated with the ah meets the d - bound conjecture for an eos parameter in the interval , @xmath212 although for @xmath213 the null , strong and dominant energy conditions are violated . we have also discussed the validity of the second law of thermodynamics and the generalized second law for a black hole embedded in phantom energy - dominated flrw universe . the analysis showed that for arbitrary hawking - hayward quasi - local mass , the second law of thermodynamics is protected when approaching the coincidence time , at which the apparent horizons coincide , @xmath214 on the other , in the case of small hawking - hayward quasi - local mass , the gsl is only satisfied if @xmath205 this work was supported by the algerian ministry of high education and scientific research for financial support under the cnepru project : d01720070033 . s. nojiri and s. d. odintsov , phys . b 562 ( 2003 ) ; e. elizalde , s. nojiri and s. d. odintsov , phys . rev . d 70 , 043539 ( 2004 ) ; h. m. sadjadi and m. alimohammadi , phys . rev . d 74 , 103007 ( 2006 ) ; y. cai , h. li and x. zhang [ arxiv : gr - qc/0609039 ] . a. ashtekar and g. j. galloway , adv . * 9 * , 1 ( 2005 ) @xmath215 $ ] ; a. ashtekar and b. krishnan , living rev * 7 * , 10 ( 2004 ) [ arxiv : gr - qc/0407042 ] ; phys . d * 68 * , 104030 ( 2003 ) [ arxiv : gr - qc/0308033 ] ; phys . lett . * 89 * , 261101 ( 2006 ) [ arxiv : gr - qc/0207080 ] ; a. ashtekar and a. corichi , class . quantum grav . * 17 * , 1317 ( 2000 ) [ arxiv : gr - qc/9910068 ] ; a. ashtekar , a. corichi , and d. sudarsky , class . quantum grav . * 20 * , 3413 ( 2003 ) [ arxiv : gr - qc/0305044 ] ; a. ashtekar , [ arxiv : gr - qc/0306115 ] ; a. ashtekar , c. beetle , and j. lewandowski , class . quantum grav . * 19 * , 1195 ( 2002 ) ; phys . d * 64 * , 044016 ( 2001 ) ; a. ashtekar , c. beetle , o. dreyer , s. fairhurst , b. krishnan , j. lewandowski and j. wisniewski , phys . 85 * , 3564 ( 2000 ) ; a. ashtekar , s. fairhurst and b. krishnan , phys . d * 62 * , 104025 ( 2000 ) ; a. ashtekar , c. beetle and s. fairhurst , class . quantum grav . * 17 * , 253 ( 2000 ) ; * 16 * , l1 ( 1999 ) ; j. a. de freitas pacheco and j. e. horvath , class . * 24 * , 5427 ( 2007 ) [ arxiv:0709.1240 ] ; d.c . guariento , j. e. horvath , p. s. custodio , and j. a. de freitas pacheco , gen . . grav . * 40 * , 1593 ( 2008 ) [ arxiv:0711.3641 ] ; p. s. custodio and j. e. horvath , int . d * 14 * , 257 ( 2005 ) s. a. hayward , phys . d * 70 * , 104027 ( 2004 ) [ arxiv : gr - qc/0408008 ] ; phys . lett . * 93 * , 251101 ( 2004 ) [ arxiv : gr - qc/0404077 ] ; phys . . lett . * 81 * , 4557 ( 1998 ) [ arxiv : gr - qc/9807003 ] ; phys . d * 53 * , 1938 ( 1996 ) ; class . quantum grav . * 11 * , 3025 ( 1994 ) ; phys . d * 49 * , 6467 ( 1994 ) ; [ arxiv : gr - qc/9303006 ] ; s. a. hayward , s. mukohyama and m. c. ashworth , phys . lett . * a * 256 , 347 ( 1999 ) [ arxiv : gr - qc/9810006 ] ; mukohyama and s.a . hayward , class . quantum grav . * 17 * , 2153 ( 2000 ) . x. li and z. zhao , phys . d * 62 * , 104001 ( 2000 ) ; f. he , z. zhao and s. w. kim , phys . d * 64 * , 044025 ( 2001 ) ; s. q. wu and m. l. yan , d * 69 * , 044019 ( 2004 ) ; erratum - ibid . d * 73 * , 089902 ( 2006 ) @xmath216}}}}.}$ ] b. wang , y. gong and e. abdalla , phys . d * 74 , * 083520 ( 2006 ) [ arxiv : gr - qc/0511051v1 ] ; y. gong , b. wang and a. wang , jcap * 0701 * , 024 ( 2007 ) [ arxiv : gr - qc/0610151v2 ] ; y. gong , b. wang and a. wang , phys . d * 75 * , 123516 ( 2007 ) [ arxiv : gr - qc/0611155v3 ] ; b. boisseau , g. esposito - farese , d. polarski and a. a. starobinsky , phys . . lett . * 85 * , 2236 ( 200 ) [ arxiv : gr - qc/0001066 ] ; r. r. caldwell , phys . b * 545 * , 23 ( 2002 ) [ arxiv : astrp - ph/9908168 ] ; p. sing , m. sami and n. dadhich , phys . d * 68 * , 023522 ( 2003 ) [ arxiv : heo - th/0305110 ] ; v. b. johri , phys . d * 70 * , 041303 ( 2004 ) [ arxiv : astro - ph/0311293 ] ; p. f. gonzalez - diaz , phys . d * 69 * , 063522 ( 2004 ) [ arxiv : hep - th/0401082];p . f. gonzalez - diaz , phys . lett . * 93 * , 071301 ( 2004 ) [ arxiv : astrp - ph/0404045 ] .	adopting the thin - layer improved brick - wall method , we investigate the thermodynamics of a black hole embedded in a spatially flat friedmann - lemaitre - robertson - walker universe . we calculate the temperature and the entropy at every apparent horizon for arbitrary solution of the scale factor . we show that the temperature and entropy display a non - trivial behavior as a functions of time . in the case of black holes immersed in universe driven by phantom energy , we show that for specific ranges of the equation - of - state parameter and apparent horizons the entropy is compatible with the d - bound conjecture , even the null , dominant and strong energy conditions are violated . in the case of accretion of phantom energy onto black hole with small hawking - hayward quasi - local mass , we obtain an equation - of - state parameter in the range @xmath0 , guaranteeing the validity of the generalized second law . pacs : 04.60.-m ; 98.80.qc ; 95.36.+x ; 98.80.-k key words : dynamical apparent horizons , entropy , generalized second law _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ laboratory of theoretical physics ( lpth ) and department of physics , faculty of sciences , university of jijel , bp 98 ouled aissa , jijel 18000 , algeria _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
as of the durham conference @xcite , the problem of obtaining a goodness of fit in unbinned likelihood fits was an unsolved one . in what follows , we will denote by the vector @xmath0 , the theoretical parameters ( @xmath0 for `` signal '' ) and the vector @xmath1 , the experimentally measured quantities or `` configurations '' . for simplicity , we will illustrate the method where both @xmath0 and @xmath1 are one dimensional , though either or both can be multi - dimensional in practice . we thus define the theoretical model by the conditional probability density @xmath2 . then an unbinned maximum likelihood fit to data is obtained by maximizing the likelihood @xcite , @xmath3 where the likelihood is evaluated at the @xmath4 observed data points @xmath5 . such a fit will determine the maximum likelihood value @xmath6 of the theoretical parameters , but will not tell us how good the fit is . the value of the likelihood @xmath7 at the maximum likelihood point does not furnish a goodness of fit , since the likelihood is not invariant under change of variable . this can be seen by observing that one can transform the variable set @xmath1 to a variable set @xmath8 such that @xmath9 is uniformly distributed between 0 and 1 . such a transformation is known as a hypercube transformation , in multi - dimensions . other datasets will yield different values of likelihood in the variable space @xmath1 when the likelihood is computed with the original function @xmath10 . however , in the original hypercube space , the value of the likelihood is unity regardless of the dataset @xmath11 , thus the likelihood @xmath7 can not furnish a goodness of fit by itself , since neither the likelihood , nor ratios of likelihoods computed using the same distribution @xmath10 is invariant under variable transformations . the fundamental reason for this non - invariance is that only a single distribution , namely , @xmath10 is being used to compute the goodness of fit . in binned likelihood cases , where one is comparing a theoretical distribution @xmath2 with a binned histogram , there are two distributions involved , the theoretical distribution and the data distribution . the @xmath12 of the data is approximated by the bin contents of the histogram normalized to unity . if the data consists of @xmath4 events , the @xmath12 of the data @xmath13 is defined in the frequentist sense as the normalized density distribution in @xmath1 space of @xmath4 events as @xmath14 . in the binned case , we can bin in finer and finer bins as @xmath14 and obtain a smooth function , which we define as the @xmath12 of the data @xmath13 . in practice , one is always limited by statistics and the binned function will be an approximation to the true @xmath12 . we can now define a likelihood ratio @xmath15 such that @xmath16 where we have used the notation @xmath17 to denote the event set @xmath5 . let us now note that @xmath18 is invariant under the variable transformation @xmath19 , since @xmath20 and the jacobian of the transformation @xmath21 cancels in the numerator and denominator in the ratio . this is an extremely important property of the likelihood ratio @xmath18 that qualifies it to be a goodness of fit variable . since the denominator @xmath22 is independent of the theoretical parameters @xmath0 , both the likelihood ratio and the likelihood maximize at the same point @xmath6 . one can also show @xcite that the maximum value of the likelihood ratio occurs when the theoretical likelihood @xmath23 and the data likelihood @xmath24 are equal for all @xmath25 . in the case where the @xmath12 @xmath13 is estimated by binned histograms and the statistics are gaussian , it is readily shown @xcite that the commonly used goodness of fit variable @xmath26 . it is worth emphasizing that the likelihood ratio as defined above is needed and not just the negative log of theoretical likelihood @xmath27 to derive this result . the popular conception that @xmath28 is -2 log @xmath27 is simply incorrect!. it can also be shown that the likelihood ratio defined above can describe the binned cases where the statistics are poissonian @xcite . in order to solve our problem of goodness of fit in unbinned likelihood cases , one needs to arrive at a method of estimating the data @xmath12 @xmath13 without the use of bins . one of the better known methods of estimating the probability density of a distribution in an unbinned case is by the use of probability density estimators @xmath29 , also known as kernel density estimators @xcite @xmath30 . the @xmath12 @xmath13 is approximated by @xmath31 where a kernel function @xmath32 is centered around each data point @xmath25 , is so defined that it normalizes to unity and for large @xmath4 approaches a dirac delta function @xcite . the choice of the kernel function can vary depending on the problem . a popular kernel is the gaussian defined in the multi - dimensional case as @xmath33 where @xmath34 is the error matrix of the data defined as @xmath35 and the @xmath36 implies average over the @xmath4 events , and @xmath37 is the number of dimensions . the hessian matrix @xmath38 is defined as the inverse of @xmath34 and the repeated indices imply summing over . the parameter @xmath39 is a `` smoothing parameter '' , which has@xcite a suggested optimal value @xmath40 , that satisfies the asymptotic condition @xmath41 the parameter @xmath39 will depend on the local number density and will have to be adjusted as a function of the local density to obtain good representation of the data by the @xmath42 . our proposal for the goodness of fit in unbinned likelihood fits is thus the likelihood ratio @xmath43 evaluated at the maximum likelihood point @xmath6 . we consider a simple one - dimensional case where the data is an exponential distribution , say decay times of a radioactive isotope . the theoretical prediction is given by @xmath44 we have chosen an exponential with @xmath45 for this example . the gaussian kernel for the @xmath42 would be given by @xmath46 where the variance @xmath47 of the exponential is numerically equal to @xmath0 . to begin with , we chose a constant value for the smoothing parameter , which for 1000 events generated is calculated to be 0.125 . figure [ genev ] shows the generated events , the theoretical curve @xmath2 and the @xmath42 curve @xmath48 normalized to the number of events . the @xmath42 fails to reproduce the data near the origin due to the boundary effect , whereby the gaussian probabilities for events close to the origin spill over to negative values of @xmath1 . this lost probability would be compensated by events on the exponential distribution with negative @xmath1 if they existed . in our case , this presents a drawback for the @xmath42 method , which we will remedy later in the paper using @xmath42 definitions on the hypercube and periodic boundary conditions . for the time being , we will confine our example to values of @xmath49 to avoid the boundary effect . in order to test the goodness of fit capabilities of the likelihood ratio @xmath18 , we superimpose a gaussian on the exponential and try and fit the data by a simple exponential . and the @xmath42 estimator ( solid ) histogram with no errors . [ genev],width=245 ] figure [ genev1 ] shows the `` data '' with 1000 events generated as an exponential in the fiducial range @xmath50 . superimposed on it is a gaussian of 500 events . more events in the exponential are generated in the interval @xmath51 to avoid the boundary effect at the fiducial boundary at c=1.0 . since the number density varies significantly , we have had to introduce a method of iteratively determining the smoothing factor as a function of @xmath1 as described in @xcite . with this modification in the @xmath42 , one gets a good description of the behavior of the data by the @xmath42 as shown in figure [ genev1 ] . generated as an exponential with decay constant @xmath0=1.0 with a superimposed gaussian of 500 events centered at @xmath1=2.0 and width=0.2 . the @xmath42 estimator is the ( solid ) histogram with no errors . [ genev1],width=245 ] we now vary the number of events in the gaussian and obtain the value of the negative log likelihood ratio @xmath52 as a function of the strength of the gaussian . table [ tab1 ] summarizes the results . the number of standard deviations the unbinned likelihood fit is from what is expected is determined empirically by plotting the value of @xmath52 for a large number of fits where no gaussian is superimposed ( i.e. the null hypothesis ) and determining the mean and @xmath53 of this distribution and using these to estimate the number of @xmath47 s the observed @xmath52 is from the null case . table [ tab1 ] also gives the results of a binned fit on the same `` data '' . it can be seen that the unbinned fit gives a @xmath54 discrimination when the number of gaussian events is 85 , where as the binned fit gives a @xmath55 of 42/39 for the same case . we intend to make these tests more sophisticated in future work . . goodness of fit results from unbinned likelihood and binned likelihood fits for various data samples . the negative values for the number of standard deviations in some of the examples is due to statistical fluctuation . [ tab1 ] [ cols="^,^,^,^ " , ] equation [ pns ] can be used to show that the expectation value of @xmath56 of the parameter @xmath0 is given by @xmath57 where @xmath58 is the average of @xmath0 for individual experiments . equation [ sbar ] states @xmath56 is the weighted average of @xmath58 obtained from individual measurements , the weight for each experiment being the `` data likelihood '' @xmath59 for that experiment . in the absence of experimental bias , @xmath56 would be identical to the true value @xmath60 . it remains to be shown that the weighted average of maximum likelihood values @xmath6 from indiviual experiments also converge to the maximum likelihood point of @xmath61 . also one needs to develop an analytic theory of the goodness of fit for unbinned likelihood fits . finally , one needs to investigate a bit more closely the transformation properties of @xmath61 under change of variable . to conclude , we have proposed a scheme for obtaining the goodness of fit in unbinned likelihood fits . this scheme involves the usage of two @xmath12 s , namely data and theory . in the process of computing the fitted errors , we have demonstrated that the quantity in the joint probability equations that has been interpreted as the `` bayesian prior '' is in reality a number and not a distribution . this number is the value of the @xmath12 of the parameter , which we call the `` unknown concomitant '' at the true value of the parameter . this number is calculated from a combination of data and theory and is seen to be an irrelevant parameter . if this viewpoint is accepted , the controversial practice of guessing distributions for the `` bayesian prior '' can now be abandoned , as can be the terms `` bayesian '' and `` frequentist '' . we show how to use the posterior density to rigorously calculate fitted errors . 9 k. kinoshita , `` evaluating quality of fit in unbinned maximum likelihood fitting '' , proceedings of the conference on advanced statistical techniques in particle physics , durham , march 2002 ippp/02/39 , dcpt/02/78 . + b. yabsley,``statistical practice at the belle experiment , and some questions'',ibid . r. d. cousins,``conference summary '' , ibid . r. a. fisher,``on the mathematical foundations of theoretical statistics '' , _ philos . london ser . a _ * 222 * , 309 - 368(1922 ) ; + r. a. fisher,``theory of statistical estimation '' , _ proc . cambridge philos . * 22 * , 700 - 725 ( 1925 ) . `` a measure of the goodness of fit in unbinned likelihood fits '' , r .raja , long write - up , + http://www-conf.slac.stanford.edu/phystat2003/talks + /raja / raja_bayes_maxlike.pdf `` end of bayesianism ? '' , r.raja , http://www-conf.slac.stanford.edu/phystat2003/talks/raja/raja-end_bayesianism.pdf e. parzen , `` on estimation of a probability density function and mode '' _ ann.math.statis . _ * 32 * , 1065 - 1072 ( 1962 ) . d. scott . _ multivariate density estimation_. john wiley & sons , 1992 . + m. wand and m. jones , _ kernel smoothing_. chapman & hall , 1995 .	maximum likelihood fits to data can be done using binned data ( histograms ) and unbinned data . with binned data , one gets not only the fitted parameters but also a measure of the goodness of fit . with unbinned data , currently , the fitted parameters are obtained but no measure of goodness of fit is available . this remains , to date , an unsolved problem in statistics . using bayes theorem and likelihood ratios , we provide a method by which both the fitted quantities and a measure of the goodness of fit are obtained for unbinned likelihood fits , as well as errors in the fitted quantities . the quantity , conventionally interpreted as a bayesian prior , is seen in this scheme to be a number not a distribution , that is determined from data .
let us assume that a vector boson @xmath13 [ here @xmath13 could be either a fundamental state , like the @xmath14 , or a quark - antiquark bound state like the @xmath15 , or @xmath16 couples to @xmath10 . if it couples also to @xmath17 as all the states above do then by unitarity its exchange contributes also to @xmath18 . let us write the effective coupling between the vector boson @xmath13 and @xmath10 as @xmath19 this coupling , through the diagram of fig . 1 , contributes to the @xmath20 amplitude a term @xmath21 here @xmath22 is the effective coupling of the vector boson @xmath13 to @xmath17 , while @xmath23 . to @xmath17 , which contribute to this process these have not been included in the above , since they do not change our qualitative discussion . these couplings are , however , taken into account in the @xmath24 bounds given in eqs . ( 5 ) and ( 9 ) below . ] as a first approximation , it is sensible to neglect @xmath25 in comparison with @xmath26 . then comparing the above contribution to the @xmath0 process to that of ordinary muon decay , @xmath27 , which proceeds via @xmath28 exchange and ( almost ) identical kinematics , gives the relation : @xmath29_{v-{\rm exch.}}}{\gamma(\mu\to e\nu\bar \nu)}\approx \frac{\tilde g_{v\mu e}^2g_{vee}^2}{m_v^4}/\frac{g_w^4}{m_w^4}~.\ ] ] since @xmath30 and @xmath31 , while @xmath32 , we can rewrite the last expression as @xmath33_{v-{\rm exch . } } \approx \frac{\gamma(v\to e^+e^-)\gamma(v\to \mu^\pm e^\mp)}{\gamma^2(w\to e\nu ) } \left(\frac{m_w}{m_v}\right)^6~.\ ] ] using br@xmath34 and other data pertaining to the @xmath17 widths of the various vector mesons @xmath13 , we find a set of bounds for the two - body lfv branching ratios of these vector bosons . these bounds are : @xmath35 likewise , the generic upper bounds on lfv tau decays br@xmath36 yields @xmath37 with @xmath38 . except for ( 11 ) , these inferred bounds are unlikely to be improved by future experimental data on two - body decays . one can use similar considerations to obtain bounds on the lfv decays of pseudoscalar states . for these purposes , one considers instead of the @xmath0 process the @xmath39 decay , which has a lfv bound br@xmath40 . for this latter process the lfv couplings of the @xmath41 contribute , due to the exchange of these particles in the s - channel . we can again utilize this fact to infer upper bounds on @xmath42 . the @xmath43 vertex , because of gauge invariance , involves two derivatives : @xmath44 where @xmath45 . this derivative coupling , in contrast to the @xmath46 non - derivative coupling encountered earlier , kinematically suppresses the off - shell @xmath47 contribution at @xmath48 relative to what it would be for on - shell @xmath41 decay . consequently the analog to eq . ( 4 ) for the present case , @xmath49_{\pi^o/\eta^o-{\rm exch.}}\approx \frac{\gamma(\pi^o/\eta^o\to\gamma\gamma ) \cdot\gamma(\pi^o/\eta^o\to \mu^\pm e^\mp)}{\gamma^2(w\to e\nu ) } \left(\frac{m_w}{m_{\pi^/\eta}}\right)^6\cdot \left(\frac{\langle s_{\gamma\gamma}\rangle}{m_{\pi/\eta}^2}\right)^2~,\ ] ] contains an extra factor @xmath50 which tends to weaken the bounds one can derive . one finds , for pseudoscalar lfv decays the bounds : @xmath51 in the discussion above we have obtained the quoted bounds purely by concentrating on the contribution of the exchanged state in question to the lfv process . one can imagine , however , additional lfv contributions . for example , for the @xmath39 decay , in addition to @xmath52 exchange in the @xmath25 channel , we have also the contribution of electron exchange in the @xmath53 and @xmath54 channels ( see fig . 2 ) . in this case , however , the stringent bound on the @xmath55 vertex coming from experiment [ br ( @xmath56 strongly suppresses these additional diagrams and causes negligible modifications to the bounds ( 15 ) , ( 16 ) . even in the absence of a strong bound on the @xmath55 coupling , we would like to note that cancellations between @xmath25 and @xmath53 channel contributions are in general expected to be at best rather partial . unless all particles , both external and exchanged , are spinless any specific @xmath25 channel amplitude will have different @xmath57 ( or @xmath58 ) dependence , and will contribute to different combinations of helicity amplitudes than the @xmath53 channel exchange contributions . by the same token , it is clear that cancellations among different angular momentum states exchanged in the @xmath25 channel are also impossible . indeed , for example , the total decay rate for @xmath0 can be expressed as @xmath59^{-1/2 } \nonumber \\ & & { } \times\sum_j ( 2j+1 ) \|a^j_{\alpha\beta,\gamma\delta}(s)\|^2\end{aligned}\ ] ] with @xmath60 the triangular function expressing the initial c.m . momentum in the @xmath25 channel , which here is that of @xmath61 or , equivalently , @xmath62 . the @xmath63 are the partial waves in the jacob - wick expansion of the various @xmath25 channel helicity amplitudes . note that for the @xmath18 case , adding the @xmath53 channel amplitude amounts to enforcing the ( anti ) symmetrization between the @xmath64 and @xmath65 fermions . since fermi statistics does not preclude the vectorial coupling contributing to @xmath18 considered here [ c.f . ( 2 ) ] , no cancellation of @xmath25 and @xmath53 contributions should arise as well . although we have called the bounds we obtained above unitarity bounds , in the strict sense the inferred bounds are not true unitarity bounds as would be the case if the exchanged particle(s ) were on mass shell . to illustrate this point , let us recall a well known example of a true unitarity bound arising in rare kaon decays . this is the lower bound for the br@xmath66 derived from the measured branching ratio of @xmath67 . the @xmath68 process , with an on - shell @xmath69 intermediate state , contributes to i m @xmath70 via unitarity since i m @xmath71 @xmath72 . this contribution provides a strict lower bound to the br@xmath66 , so the apparent violation of this bound in early @xmath73 data was a source for much concern . modern day data , as expected , agrees with this bound . @xcite in the present context , an example of a pure " unitarity bound for a lfv process is provided by the @xmath6 analog " of the @xmath74 process . because the decay @xmath75 is kinematically forbidden , what one should consider instead is @xmath76 ( or @xmath77 ) . the on - shell " @xmath78 emitted in this putative process propagates over a long distance , of order @xmath79 fm , before decaying into @xmath80 in a manner which is completely independent of its production . this will generate a distinct narrow contribution to the corresponding three body processes @xmath81 , @xmath82 , @xmath83 , contributing to the imaginary part of these amplitudes at @xmath84 . hence , for example , there is an attendant lower bound on br@xmath85 $ ] which is simply br@xmath86 . the resulting rigorous upper bound one obtains , @xmath87 unfortunately happens to be rather weak . all the vector ( or pseudoscalars ) used as intermediaries in deriving the bounds in eqs ( 5)-(11 ) and eqs . ( 15 ) and ( 16 ) are not on - shell . thus we must entertain the possibility that their contribution to the three - body decays considered are reduced . this could void , or at least weaken the various strong bounds obtained above . in the rest of this note we will focus on possible mechanisms for such a reduction . the size of the boson exchange contribution to the three - body decay amplitude can be reduced if there are * kinematical * suppressions . these arise when the effective boson couplings are not minimal , involving derivatives ( or momentum factors ) . we already encountered one such case above , when we discussed the @xmath41 contribution to @xmath88 . we want to discuss here whether such kinematical suppressions may not also affect the vector exchange contributions . it is clearly possible to imagine that the lfv @xmath89 vertex , instead of having the form of eq . ( 1 ) , involves an anomalous magnetic moment coupling : @xmath90 in this case the contribution of the virtual @xmath13 to @xmath18 is reduced by @xmath91 this would considerably weaken the bounds in eqs . ( 6)-(8 ) and reduce the bound on @xmath24 decay to only br@xmath92 . it does not seem likely to us , however , that the strong suppression factors appearing in eq . ( 20 ) obtain in practice . indeed , various model calculation involving mixing among heavy neutrinos @xcite lead to @xmath93 and effective @xmath94 flavor violating vertices , which involve non anomalous terms terms like the @xmath95 coupling of eq . ( 1 ) and vectorial couplings like @xmath96 . @xcite hence we believe that the kinematic suppression given by eq . ( 20 ) probably should not be included in our bounds . there is another possible source of suppression which needs to be considered . this is connected to possible `` form factor '' effects due to the * dynamics * which would , for example , reduce the contribution of the various @xmath13 states to @xmath18 compared to the naive expectations . however , the effect of form factors should be minimal or controllable if the lfv is induced by physics at scales much higher than the ew scale or the @xmath24 mass . the effects of dynamics are nicely illustrated in a recent paper by ilana , jack and riemann @xcite . these authors find , in fact , an apparent mild enhancement when the @xmath97 process is induced by relatively light ( @xmath98 gev ) neutrinos . indeed , in this case the on - shell contribution of @xmath99 loops enhances the @xmath24 decay rates relative to the @xmath100 contribution by factors of 10 - 100 . however , such light active neutrinos would contribute to the @xmath14 width and are hence ruled out . thus such an ehancement is not physically expected . for light neutrinos i.e. neutrinos with masses in the ev range , as inferred from the superkamiokande @xcite data are very much below our bounds ] in terms of the dispersive approach adopted here , such a `` form factor '' suppression would result from cancellations in the corresponding partial wave amplitudes . consider , for example , the @xmath101 partial wave amplitude for the @xmath0 process : @xmath102 to get a `` form factor '' suppression , there must be a cancellation between the contributions of the various @xmath103 particles among themselves , or between these contributions and those of the continuum . let us examine these possibilities . for the case of quarkonium intermediate states , besides the lowest energy bound state there are towers of states of the same spin and parity . thus , for example , in eq.(21 ) besides the contribution of the @xmath104 one should also take into acount the exchange of the @xmath105 charmonium bound resonances . is it possible that these additional contributions largely cancel the @xmath106 term in eq . this is unlikely for the following reason . to get @xmath104 exchange to contribute to @xmath18 in the first place , one needs to assume that the lfv physics at a high scale induces an effective four - fermi coupling of the form : @xmath107 such a coupling underlies all the other charmonium contributions . in fact , quark - hadron duality identifies the @xmath108 contributions to @xmath0 as arising from specific portions of the @xmath109 integration region where due to non - perturbative qcd effects @xmath110 @xmath111 bound states dominate , as shown schematically in fig . 3 . both the @xmath112 and @xmath113 couplings appearing in eq . ( 21 ) are proportional to the wave function of @xmath114 at the origin , @xmath115 . thus all the terms in the sum share a common sign fixed by the sign of @xmath116 , with @xmath117 being the charge of the charm quark and cancellation can not occur . similar arguments apply against possible cancellations among the various states in the @xmath118 sector . the above discussion still leaves open the possibility of cancellations in the partial wave amplitude between different quarks - antiquark contributions ( @xmath119 ) or , equivalently , between the various resonant states ( @xmath120 . while possible this seems highly unlikely . for example , even if all the effective couplings @xmath121 were equal due to some universality , and the bubble kinematics were identical , the net contribution would not vanish since the total contribution would be proportional to @xmath122 . furthermore , for the case of light quarks such cancellations can not work even in principle . the @xmath25 dependence for @xmath123 neglected above implies , for example , that a @xmath124 and @xmath78 contribution to the total decay rate cancel only at the level of @xmath125 . in general , lepton flavor violating processes have been analyzed within a specific theoretical framework . in this context , the restrictive role played by the low energy bounds ( @xmath126 conversion , etc . ) has been noted by many authors . @xcite @xcite in this note instead we tried to present in a , relatively model - independent manner , the connections which unitarity implies between some two - body and three - body lfv decays . we have illustrated these connections by focusing on a few processes . clearly , many other bounds can be obtained . indeed , since the particle data group @xcite lists altogether about one hundred lfv processes , many additional results can come from a more comprehensive analysis . we have noted that the bounds that we derived can be avoided if one can kinematically suppress the small @xmath25 contributions ( e.g. by having a purely anomalous magnetic @xmath127 coupling ) , or as a result of some ( rather unlikely ) cancellations . because we can not rule out these possibilities with absolute certainty , we hope that our discussion will not dissuade future efforts to improve the bounds on lfv decays of the @xmath128 . such decays would not only signal new lfv physics but , because of our considerations , this physics must also naturally give cancellations among terms so as to lead to a small @xmath18 branching ratio . thus searching for @xmath129 decays at levels considerably higher than our suggested bounds remains a worthwhile experimental challenge . both s. n. and x. m. z. would like to acknowledge the hospitality of the department of physics and astronomy at ucla , where this work was initiated . s. n. would like to acknowledge the support of usa - israel binational and israeli academy grants . x. m. z. thanks z. h. lin for discussions . the work of r. d. p. was supported in part by the department of energy under contract no . de - fg03 - 91er40662 , task c. 9 particle data group , c. caso _ et al . _ , europ . phys.j . * c3 * , 1 ( 1998 ) . e871 collaboration , d. ambrose _ et al . lett . * 81 * , 4309 ( 1998 ) . j. illana , m. jack , and t. riemann , hep - ph/0001273 , and references therein . `` probing lepton flavor violation in decays of charmonium and bottomonium systems '' , r. d. peccei , j. x. wang and x. m. zhang , may 1998 note ( unpublished ) . superkamiokande collaboration , y fukuda _ et al . lett . * 81 * , 1562 ( 1998 ) . see , for example , p. langacker and d. london , phys . rev . d38 , 886 ( 1988 ) ; d. tommasini _ et al . b444 , 451 ( 1995 ) . 1 . a vector exchange diagram contributing to @xmath18 .	simple unitarity inspired " relations between two- and three - body lepton flavor violating decays are noted and discussed . in the absence of cancellations , the existing strong bounds on @xmath0 and @xmath1 severly constrain two - body lepton flavor violating decays . lepton flavor violating ( lfv ) processes are strongly suppressed in the standard model by powers of ( small ) neutrino masses . such decays signal therefore new physics . at present we have stringent bounds for @xmath2 decays [ br@xmath3 , br@xmath4 and somewhat weaker @xmath5 bounds on lfv @xmath6 decays . @xcite the availability of large samples of decaying vector bosons [ @xmath7 , and @xmath8 or pseudoscalars [ @xmath9 and the clear signature provided by @xmath10 final states suggests searching for lfv two - body decays @xmath11 or @xmath12 . in this note we show that rather simple considerations , based on unitarity , provide rather strong constraints on two - body lfv processes . hence , most three - body @xmath2 and @xmath6 lfv decays are likely to remain more sensitive tests of lepton flavor violation , rather than the corresponding two - body decays .
diffractive scattering is characterised , in high energy hadron interactions , by final states consisting of two systems well separated in rapidity , which carry the quantum numbers of the initial state hadrons . the process is related through unitarity to inelastic scattering and governs the high energy behaviour of total cross sections . it is described in regge theory @xcite by the exchange of the vacuum singularity , called the pomeron " , and may be interpreted as the differential absorption of the various virtual components of the interacting systems @xcite . it is a challenge for quantum chromodynamics ( qcd ) to explain diffraction in terms of quark and gluon interactions . most diffractive phenomena which include elastic scattering are governed by large distance , soft " processes , which in general are not accessible to perturbative qcd ( pqcd ) calculations . however , for short distance processes , the presence of a hard " scale offers the possibility to use perturbative techniques to calculate diffractive amplitudes . alternatively , at high energy the interaction properties of colour fields are invoked in models which characterize the incident particles as a superposition of colour dipoles with various size to calculate diffractive and total cross sections . ( 6.0,5 ) ( 0.0,0.0 ) an important testing ground for calculations in diffraction is provided by the study of exclusive vector meson ( vm ) production @xmath64 . this process is illustrated in fig . [ fig : vm ] : the intermediate photon of four - momentum @xmath65 converts into a diffractively scattered vm ( , , , , ... ) of mass @xmath66 , while the incoming proton is scattered into a system @xmath67 of mass @xmath68 , which can be a proton ( elastic " scattering ) or a diffractively excited system ( proton dissociation " ) . in vm production , a hard scale can be provided by the photon virtuality @xmath69 , with @xmath70 , the four - momentum transfer @xmath71 from the proton , or by the quark mass ( for heavy vm production ) . the reaction energy is defined by the photon - proton centre of mass energy @xmath6 , with @xmath72 , where @xmath73 is the bjorken scaling variable . the high energy electron - proton collider hera offers access to all these scales , over a wide range of values . the present publication is devoted to the study of the diffractive electroproduction of and mesons , both for elastic and proton dissociative scattering . the data were taken at hera with the h1 detector in the period from 1996 to 2000 . a common analysis of the four channels is performed . measurements of the production cross sections and of the spin density matrix elements , which give access to the helicity amplitudes , are presented as a function of the kinematic variables ( including the dependence of the polarised cross sections ) , @xmath6 , @xmath74 and , for mesons , the dipion mass . the measurement of kinematic dependences and the comparison between different vms provide tests of a large spectrum of predictions . the data cover the interesting transition from the low domain , dominated by soft diffraction , to the higher domain where hard diffraction is expected to be dominant . this offers the opportunity to test the relevance of soft physics features present in the photon and vm wave function , and to study the development of features predicted by pqcd calculations . quantitative tests of pqcd and colour dipole calculations are provided by the comparison with the data of various model predictions . two important aspects of diffraction are tested : the flavour independence of the diffractive process and the factorisation of the process into a hard scattering contribution at the photon vertex and soft diffractive scattering at the proton vertex ( regge factorisation " ) . in addition , valuable information is provided by precise measurements of empirical parameters , in particular the and @xmath74 dependences of the cross sections and the ratio of the proton dissociative to elastic cross sections , as well as the contributions of various backgrounds . the present studies confirm with increased precision previous h1 measurements on @xcite and @xcite electroproduction , mainly in the elastic channel but also in proton dissociation @xcite . the samples analysed here include data taken in 1996 and 1997 , and the present results supersede those presented in @xcite . thanks to the larger statistics , the scope of the investigation is significantly extended . this analysis complements other h1 measurements of exclusive diffractive processes : production of real photons , in photoproduction ( @xmath75 ) at large @xcite and in electroproduction at small ( deeply virtual compton scattering dvcs ) @xcite , production of mesons in photoproduction at low @xcite and large @xcite , of mesons in photo- and electroproduction at low @xcite and large @xcite , of @xmath76 @xcite and @xmath77 @xcite mesons in photoproduction . the zeus collaboration at hera has performed measurements of dvcs @xcite , @xcite , @xmath78 @xcite , @xcite , @xcite and @xmath77 @xcite production . results at lower energy have been published , in particular for electroproduction , by the desy - glasgow @xcite , chio @xcite , nmc @xcite , e665 @xcite and hermes @xcite collaborations . the experimental and theoretical status of diffractive vm production before the high energy fixed target and hera experiments is presented in detail in the review @xcite . the paper is organised as follows . the theoretical context and the models which will be compared to the data are presented in section [ sect : th_context ] . the h1 detector and the event selection criteria are summarised in section [ sect : selection ] , where the kinematic and angular variables are defined . the various signal samples are defined in section [ sect : extraction ] , which also contains a detailed discussion of the backgrounds , a description of the monte carlo ( mc ) simulations used for the analyses and a discussion of the systematic errors affecting the measurements . in section [ sect : cross_sections ] , the measurements of the vm line shapes and of the elastic and proton dissociative cross sections are presented , and vm universality and proton vertex factorisation are discussed . section [ sect : polarisation ] is devoted to the polarisation characteristics of the reactions and their kinematic dependence . a summary of the results and conclusions are given in section [ sect : conclusions ] . since the first observation of high inclusive diffraction @xcite and of vm production at hera , a large number of theoretical studies has been published on diffractive vm production ( see e.g. @xcite ) . reviews of theoretical predictions confronted by the data have been published recently @xcite . calculations are performed following two main approaches , sketched in fig . [ fig : diag ] . the approach based on collinear factorisation , illustrated in fig . [ fig : diag](a ) , describes vm production using the parton content of the proton , in the presence of a hard scale . the colour dipole picture of fig . [ fig : diag](b ) provides a complementary way to describe high energy scattering . ( 14.2,4 . ) ( -.5,0.0 ) ( 6.5,0.0 ) [ [ collinear - factorisation ] ] collinear factorisation + + + + + + + + + + + + + + + + + + + + + + + in a pqcd framework , a collinear factorisation theorem @xcite has been proven for the production of longitudinally polarised vms in the kinematic domain with @xmath79 , @xmath80 and @xmath81 , for leading powers of @xmath69 and for all values of @xmath73 . the longitudinal amplitude , sketched in fig . [ fig : diag](a ) , is given by @xmath82 where @xmath83 is the generalised parton distribution function ( gpd ) for parton @xmath84 in the proton and @xmath85 is the factorisation and renormalisation scale , of the order of @xmath69 . the gpds ( see e.g. @xcite ) , which are an extension of ordinary parton distribution functions ( pdf ) , include correlations between partons with longitudinal momenta @xmath73 and @xmath86 and transverse momenta @xmath74 ; they describe the off - diagonal kinematics ( @xmath87 ) implied by the different squared four - momenta of the incoming photon and outgoing vm . the @xmath88 matrix elements describe the hard scattering from the parton @xmath84 in the proton to the parton @xmath89 in the meson with wave function @xmath90 , where @xmath91 is the fraction of the photon longitudinal momentum carried by one of the quarks . the scale evolution is calculated , in the hera kinematic domain , using the dglap equations , and higher order corrections have been calculated @xcite . collinear factorisation holds for heavy vms @xcite , and its validity is extended to transverse amplitudes at sufficiently high ( see e.g. @xcite ) . [ [ dipole - approach ] ] dipole approach + + + + + + + + + + + + + + + at high energy ( small @xmath73 ) and small , vm production is conveniently studied in the proton rest frame , for all values of . it is described as three factorising contributions , characterised by widely different time scales @xcite , as illustrated in fig . [ fig : diag](b ) : the fluctuation of the virtual photon into a @xmath7 colour dipole , with a coupling depending only on the quark charge , the dipole - proton scattering ( either elastic or proton dissociative scattering ) , and the @xmath7 recombination into the final state vm . the amplitude is @xmath92 where @xmath93 is the transverse distance between the quark and the antiquark , and @xmath94 and @xmath95 are the photon and the vm wave functions , respectively . the diffractive dipole - proton cross section @xmath96 is expected to be flavour independent and to depend only on the dipole transverse size ( the impact parameter between the dipole and the proton is integrated over ) . photons with large virtuality and fluctuations into heavy quarks are dominated by dipoles with small transverse size . in this case , the two quarks tend at large distance to screen each other s colour ( colour transparency " @xcite ) , which explains the small cross section . in several models @xcite , the convolution of the vm wave function with the dipole is expected to play a significant role in vm production , by selecting specific dipoles . it can be noted that the generalised vector meson dominance model @xcite is related to the dipole approach . dipole - proton scattering is modeled at lowest order ( lo ) in pqcd through the exchange of a gluon pair in a colour singlet state @xcite , and in the leading logarithm approximation ( ll @xmath97 ) as the exchange of a bfkl - type gluon ladder . in a ( @xmath98 ) representation , where @xmath99 represents the quark ( or antiquark ) momentum component transverse to the photon direction ( i.e. the fourier transform of the dipole transverse size ) , @xmath100-unintegrated gluon distribution functions are used . the contributions of gluons with small @xmath100 are of a non - perturbative nature , whereas at large @xmath100 they can be obtained from the logarithmic derivative of the usual , integrated , gluon distribution , @xmath101 . in the lo and ll @xmath97 approximations both gluons emitted from the proton carry the same fraction @xmath73 of the proton longitudinal momentum and the cross section is proportional to the square of the gluon density @xcite . calculations beyond the ll @xmath97 approximation take into account the difference between the longitudinal momentum fractions carried by the two gluons ( skewing " effects ) @xcite . at low @xmath73 , vm production can be calculated @xcite , in the absence of a hard scale , using universal dipole - proton cross sections obtained from deep inelastic scattering ( dis ) measurements @xcite . this approach automatically incorporates soft , non - perturbative contributions . such models often involve parton saturation effects , expected from the recombination of high density gluons @xcite as inferred from the observation of geometric scaling @xcite . dglap evolution can also be included , for instance in the model @xcite . the photon - proton cross section can be decomposed into a longitudinal and transverse part , @xmath102 and @xmath103 , respectively . at lo and for @xmath104 , the dependences @xmath105 and @xmath106 are predicted @xcite , and the ratio @xmath107 is predicted to be @xmath108 . modifications to these dependences are expected ( see e.g. @xcite ) , due to the dependence of the gluon density , the quark transverse movement ( fermi motion ) and quark virtuality @xcite , and the dependence of the strong coupling constant @xmath109 . in the dipole approach , the square of the scale @xmath85 of the interaction is of the order of @xmath110 @xmath111 being the current quark mass . it is related to the inverse of the relevant scanning radius " @xcite in the dipole - proton interaction . for longitudinally polarised photons or for heavy quark production , the @xmath7 wave function @xmath94 is concentrated around @xmath112 this suggests that a universal hard scale , @xmath85 , following from the transverse size of the dominant dipoles , can be of the order of @xmath113 . for transverse photons fluctuating into light quarks , in contrast , the wave function is non - zero at the end - points @xmath114 or @xmath115 . these contributions correspond to small @xmath100 values of the quarks forming the dipole , and hence to a large transverse distance between them . the scale @xmath85 is therefore damped to smaller values than for longitudinal photons with the same virtuality , soft contributions may be significant and formal divergences appear in pqcd calculations for @xmath116 @xcite . for moderate @xmath117 values , the @xmath91 distribution of light quark pairs from longitudinal photons can present a non - negligible smearing around the value @xmath118 , which results in a contamination of soft , finite size " effects @xcite . it is estimated that the fully perturbative qcd regime is reached for light vm production by longitudinal photons for above @xmath119 to @xmath120 @xcite . the @xmath6 dependence of vm production is governed by the @xmath121 evolution of the gluon distribution , with @xmath122 increasing from @xmath123 for @xmath124 to @xmath125 for @xmath126 , as measured in the total dis cross sections at hera @xcite . for heavy vms and for longitudinally polarised light vms at sufficiently high , a strong ( hard " ) @xmath6 dependence of the cross section is thus expected , fixed for all vms by the scale . in contrast , the @xmath6 dependence of the transverse cross section is expected to be milder than for longitudinal photons , since the @xmath122 parameter is taken at a smaller value of the effective scale . this may result in a @xmath6 dependence of the cross section ratio @xmath127 . in the framework of regge theory , the existence of two pomerons @xcite is postulated to describe both the soft and hard behaviours of the cross section @xcite . at low @xmath128 ( @xmath129 for elastic scattering ) , the @xmath74 dependence of vm production is well described by an exponentially falling distribution with slope @xmath130 , @xmath131 ( predictions for the dependence are also given e.g. in @xcite ) . in an optical model approach , the slope @xmath130 is given by the sum of the transverse sizes of the scattered system @xmath67 , of the @xmath7 dipole and of the exchanged system , with possibly in addition a vm form factor . neglecting the latter , the @xmath74 slopes for heavy vms and for light vm production by longitudinally polarised photons are expected to take universal values , depending only on , whereas the production of light vms by transverse photons , which is dominated by dipoles with larger transverse size , is expected to exhibit steeper @xmath74 distributions @xcite . this may result in a @xmath74 dependence of @xmath132 . the helicity amplitudes @xmath133 , where @xmath134 and @xmath135 are the vm and photon helicities , respectively , have been calculated in perturbative qcd for the electroproduction of light vms with @xmath136 @xcite . in this domain , the dominant amplitude is the @xmath137-channel helicity conserving ( schc ) amplitude , which describes the transition from a longitudinal photon to a longitudinal vm . other amplitudes are damped by powers of @xmath69 . those leading to the production of a transverse vm , of which the schc amplitude is largest , contain an additional factor @xmath138 . schc violation implies for single helicity flip amplitudes an additional factor @xmath139 , to be squared for the double flip amplitude . this leads , in the kinematic range studied here , to the following hierarchy of amplitude intensities ( assuming natural parity exchange ) : @xmath140 . predictions for vm production are available from a large number of models . quantitative calculations generally imply the choice of pdf or gpd parameterisations or , in colour dipole models , of dipole - proton cross section parameterisations . model calculations also generally imply the choice of vm wave function parameterisations , often taken as following a gaussian shape , with several variants @xcite . in view of the large number of models , no attempt is made in this paper to provide exhaustive comparisons to the data . instead , a few models and parameterisations , representative of recent approaches , are compared to various choices of observables . examples of the uncertainties on the predictions , due to the choice of parton distribution functions and wave function parameterisations , are given for two of the models . * the gpd model of goloskokov and kroll ( gk @xcite ) provides predictions within the handbag factorisation scheme for the longitudinal and transverse amplitudes in the schc approximation . soft physics is described by a gpd parameterisation of the proton structure , constructed from standard pdfs with adequate skewing features and @xmath74 dependences . the end - point singularities are removed with the aid of a specific model for the vm wave function . error bands are provided with the model predictions . * the model of martin , ryskin and teubner ( mrt @xcite ) for meson production is based on parton - hadron duality . open @xmath141 production is calculated in an appropriate spin - angular state and in a specific invariant mass interval , which is then assumed to saturate production , thus neglecting any vm wave function effects . the dependence of the gluon density , described by the anomalous dimension @xmath142 with @xmath143 , is used to calculate the longitudinal and transverse cross sections . skewing effects are parameterised @xcite without explicit use of gpds . predictions using two alternative pdfs are compared with the present data : cteq6.5 m @xcite and mrst-2004-nlo @xcite . * the model presented in the review of ivanov , nikolaev and savin ( ins @xcite ) is framed in the @xmath100-factorisation dipole approach . the helicity amplitudes are calculated perturbatively and then extended into the soft region by constructing parameterisations of the off - forward unintegrated gluon density . the and @xmath6 dependences of the cross sections and the full set of spin density matrix elements are predicted . two wave function models , compact " and large " , are used for mesons , corresponding to two extreme cases for describing the @xmath144 decay width . * the @xmath100-factorisation calculations of ivanov and kirschner ( ik @xcite ) provide predictions for the full set of helicity amplitudes , including helicity flip transitions . similar to the mrt approach , the relevance of pqcd for transverse amplitude calculations is justified by the scale behaviour @xmath145 of the gluon distribution , which avoids divergences for @xmath146 . * the dipole approach of kowalski , motyka and watt ( kmw @xcite ) uses an impact parameter dependent description of the dipole cross section in the non - forward direction @xcite , within the saturation models of golec - biernat and wsthoff ( gw @xcite ) and of iancu _ et al_. ( colour glass condensate cgc @xcite ) . the and @xmath6 dependences of the schc longitudinal and transverse amplitudes are predicted using the dglap evolution equations for @xmath147 . * the dipole approach of marquet , peschanski and soyez ( mps @xcite ) proposes an extension of the saturation model @xcite , geometric scaling being extended to non - forward amplitudes with a linear @xmath74 dependence of the saturation scale . the exponential @xmath74 dependence at the proton vertex is parameterised with a universal slope obtained from previous vm measurements . the diffractive production and decay of and mesons is identified using the following reactions : @xmath148 the events are selected by requiring the detection of the scattered electron and of a pair of oppositely charged particles , and by requiring the absence of additional activity in the detector , except in the region close to the outgoing proton beam , where proton dissociation can contribute . the kinematic domain of the measurements is : @xmath149 the large values of @xmath150 compared to @xmath151 , @xmath152 , @xmath153 and ensure that the process is diffractive , i.e. due to pomeron exchange . the variable @xmath154 , which corresponds to the proton energy loss , is always smaller than @xmath155 . the data studied here were taken with @xmath156 energy electrons or positrons colliding with @xmath157 or @xmath158 protons ( in the rest of this paper the term `` electron '' is used generically to refer to both electrons and positrons ) . the data sets are summarised in table [ table : data_sets ] , where @xmath159 is the @xmath160 centre of mass energy and the lepton beam type is specified . the integrated luminosity of @xmath0 corresponds to running periods with all relevant parts of the detector fully operational . the periods with high prescaling of the triggers relevant for the present analyses are discarded . the published results with @xmath161 @xcite are also presented in table [ table : data_sets ] ( h1 sv " ) . they were obtained in 1995 in a special run of @xmath162 , with the @xmath160 interaction point shifted by @xmath163 in the outgoing @xmath164 beam direction . this data set is not re - analysed in the present publication . data taking & lepton & proton energy & @xmath159 & luminosity + year & beam & ( ) & ( ) & ( @xmath165 ) + 1995 ( sv ) & @xmath166 & @xmath157 & @xmath167 & @xmath168 + 1996 & @xmath166 & @xmath157 & @xmath167 & @xmath169 + 1997 & @xmath166 & @xmath157 & @xmath167 & @xmath170 + 1999 & @xmath171 & @xmath172 & @xmath173 & @xmath174 + 1999 & @xmath166 & @xmath172 & @xmath173 & @xmath175 + 2000 & @xmath166 & @xmath172 & @xmath173 & @xmath176 + a detailed description of the h1 detector can be found in @xcite . only the components essential to the present analysis are described here . the origin of the h1 coordinate system is the nominal @xmath160 interaction point , with the positive @xmath91axis ( forward direction ) along the direction of the proton beam . the polar angles @xmath177 and the particle transverse momenta are defined with respect to this axis , and the pseudorapidity is @xmath178 . a system of two large coaxial cylindrical drift chambers ( cjc ) of @xmath179 length and @xmath180 external radius , with wires parallel to the beam direction , is located in a @xmath181 uniform magnetic field . this provides a measurement of the transverse momentum of charged particles with resolution @xmath182 ( @xmath183 measured in gev ) , for particles emitted from the nominal interaction point with polar angle @xmath184 . drift chambers with wires perpendicular to the beam direction , located inside the inner cjc and between the two cjc chambers , provide measurements of @xmath91 coordinates . track measurements are improved by the use of the central silicon tracker @xcite ( from 1997 onward ) . the interaction vertex is reconstructed from the tracks . the liquid argon ( lar ) calorimeter , located inside the magnet and surrounding the central tracker , covers the angular range @xmath185 . the backward electromagnetic calorimeter spacal ( @xmath186 ) is used to identify scattered electrons . in front of the spacal , the backward drift chamber ( bdc ) provides a precise electron direction measurement . the forward detectors " are sensitive to energy flow close to the outgoing proton beam direction . they consist of the proton remnant tagger ( prt ) , a set of scintillators placed @xmath187 downstream of the interaction point and covering the angles @xmath188 , and the forward muon detector ( fmd ) , a system of drift chambers covering the angular region @xmath189 . the prt and the three layers of the fmd situated closer to the main calorimeter detect secondary particles produced in interactions with the beam collimators or the beam pipe walls of elastically scattered protons at large and of decay products of diffractively excited systems @xmath67 with @xmath190 . for the data collected in 1996 and 1997 , events with @xmath191 were selected by inclusive triggers requesting an electromagnetic energy deposit in the spacal . for the years 1999 and 2000 , diffractive vm events with @xmath192 were registered using several inclusive triggers ; in addition , a special trigger was dedicated to elastic production with @xmath193 . to reduce the data recording rate to an acceptable level , data selected by certain triggers have been dowscaled . in the following , the accepted events are weighted accordingly . for the present analyses , the scattered electron candidate is identified as an electromagnetic cluster with energy larger than @xmath194 reconstructed in the spacal calorimeter . this energy threshold reduces to a negligible level the background of photoproduction events with a wrongly identified electron candidate in the spacal . the electron direction is calculated from the position of the measured interaction vertex and from the bdc signals , when their transverse distance to the cluster barycentre is less than @xmath195 ; if no such bdc signal is registered , the cluster centre is used . the vm candidate selection requires the reconstruction in the central tracking detector of the trajectories of two , and only two , oppositely charged particles . they must originate from a common vertex lying within @xmath196 in @xmath91 of the nominal @xmath160 interaction point , and must have transverse momenta larger than @xmath197 and polar angles within the interval @xmath198 . this ensures a difference in pseudorapidity of at least two units between the most forward track and the most forward cell of the lar calorimeter . the vm momentum is calculated as the vector sum of the two charged particle momenta . the existence of a gap in rapidity between the vm and the forward system @xmath67 is further ensured by two veto conditions : that there is in the central tracker no additional track , except if it is associated to the electron candidate , and that there is in the lar calorimeter no cluster with energy above noise level , @xmath199 , unless it is associated to the vm candidate . these requirements reduce to negligible level the contamination from non - diffractive dis interactions , which are characterised by the absence of a significant gap in rapidity in the fragmentation process . they imply that the mass of the diffractively excited proton system is restricted to @xmath200 . they also contribute to the suppression of backgrounds due to the diffractive production of systems subsequently decaying into a pair of charged particles and additional neutral particles . energy deposits unrelated to the vm event and noise in the calorimeter are monitored from randomly triggered readouts of the detector . the energy threshold of @xmath201 leads to an average loss of @xmath202 of the diffractive vm events . a cut is applied to the difference between the sum of energies and the sum of longitudinal momenta of the scattered electron and vm candidate , @xmath203 . for events where all particles except the forward going system @xmath67 are detected , this quantity is close to twice the incident electron beam energy , @xmath204 . the cut reduces the qed radiation and background contributions in which additional particles remain undetected . to optimise measurements in the selected domain , the kinematic variables are reconstructed from the measured quantities following the algorithms detailed in @xcite . in addition to the nominal beam energies , they make use of well measured quantities in the h1 detector : the electron and vm directions and the vm momentum . the variable is reconstructed from the polar angles of the electron and of the vm ( double angle " method @xcite ) . the modulus of the variable @xmath74 is to very good precision equal to the square of the transverse momentum of the scattered system @xmath67 , which is calculated as the vector sum @xmath205 of the transverse momenta of the vm candidate and of the scattered electron is a measure of @xmath206 , where @xmath207 is the minimum value of @xmath208 kinematically required for the vm and the system @xmath67 to be produced on shell through longitudinal momentum transfer . at hera energies and for the relevant values of @xmath66 and @xmath68 , @xmath207 is negligibly small compared to . in the following the notations @xmath208 is used for @xmath209 . ] . the electron transverse momentum , @xmath210 , is determined using the electron energy obtained from the double angle " method . the variable @xmath6 is reconstructed from the vm energy and longitudinal momentum @xcite . the electron energy measured in the spacal is used only for the calculation of the variable @xmath211 . ( 10.8,7.0 ) ( 0.0,0.0 ) three angles characterise vm electroproduction and two - body decay ( fig . [ fig : dec_ang ] ) . in the helicity frame used for the present measurements , they are chosen as follows . the azimuthal angle @xmath212 is defined in the hadronic centre of mass system as the angle between the electron scattering plane and the vm production plane , which is formed by the directions of the virtual photon and the vm . the two other angles , which describe vm decay , are chosen in the vm rest frame as the polar angle @xmath177 and the azimuthal angle @xmath213 of the positively charged decay particle , @xmath214 , the quantization axis being opposite to the direction of the outgoing system @xmath67 . this section first defines the analysis samples . the backgrounds are then discussed , the monte carlo simulations used to extract the signals are introduced , and the predictions are compared to the distributions of the hadronic invariant mass and of other observables . finally , systematic uncertainties are discussed . four event samples , which correspond approximately to the four processes studied in this paper , are selected following the conditions summarised in tables [ table : signal_vm ] and [ table : signal_diffr ] . these conditions are chosen to minimize background contributions . vector meson & mass range + sample & @xmath215 + sample & @xmath216 + the vm identification relies on the invariant mass of the two particles with trajectories reconstructed in the central tracker ; no decay particle identification is performed . for the sample , the mass calculated under the pion mass hypothesis is required to lie in the range @xmath215 . for the sample , the range @xmath217 is selected , the invariant mass being calculated under the kaon hypothesis . diffractive process & forward detector selection & @xmath74 range + notag sample & no signal above noise & @xmath218 + tag sample & signal detected above noise & @xmath219 + the events in the and samples are further classified in two categories , notag " and tag " , according to the absence or the presence of activity above noise levels in the forward detectors , respectively . elastic production is studied in the notag sample with @xmath218 whereas the tag sample with @xmath5 is used for proton dissociative studies . year & vm & & + 1995 - sv & , & @xmath220 & @xmath221 & & @xmath221 & @xmath222 & @xmath223 & @xmath221 & @xmath6 & @xmath221 & @xmath224 + 1996 - 1997 & , & @xmath222 & @xmath225 & & @xmath226 & @xmath227 & @xmath228 & @xmath221 & @xmath6&@xmath221&@xmath229 + & & @xmath227 & @xmath225 & & @xmath226 & @xmath230 & @xmath223 & @xmath221 & @xmath231&@xmath221&@xmath232 + & & @xmath230 & @xmath225 & & @xmath226 & @xmath233 & @xmath234 & @xmath221 & @xmath231&@xmath221&@xmath224 + & & @xmath233 & @xmath225 & & @xmath226 & @xmath235 & @xmath234 & @xmath221 & @xmath231&@xmath221&@xmath236 + & & @xmath235 & @xmath225 & & @xmath221 & @xmath237 & @xmath238 & @xmath221 & @xmath231&@xmath221&@xmath236 + 1999 - 2000 & notag & @xmath222 & @xmath225 & & @xmath226 & @xmath227 & @xmath228 & @xmath221 & @xmath231&@xmath221&@xmath239 + & , & @xmath227 & @xmath225 & & @xmath226 & @xmath230 & @xmath223 & @xmath221 & @xmath231&@xmath221&@xmath232 + & & @xmath230 & @xmath225 & & @xmath226 & @xmath233 & @xmath234 & @xmath221 & @xmath231&@xmath221&@xmath224 + & & @xmath233 & @xmath221 & & @xmath226 & @xmath235 & @xmath234 & @xmath221 & @xmath231&@xmath221&@xmath240 + & & @xmath235 & @xmath221 & & @xmath221 & @xmath237 & @xmath238 & @xmath221 & @xmath231&@xmath221&@xmath241 + the kinematic domain of the measurements is summarised in table [ table : signal_kin_range ] . it is determined by the detector geometry , the beam energies and the triggers , with the requirement of a reasonably uniform acceptance . the accepted range depends on the data taking period ; for the notag sample in 1999 - 2000 it extends to smaller values than for the tag sample and for the samples , due to the special elastic trigger . for @xmath6 , the regions with good acceptance are determined by the track requirement ; the accepted @xmath6 values increase with and with @xmath242 . the acceptance increases with , mostly because of the non - uniform geometric acceptance of the electron trigger for @xmath243 . monte carlo studies show that the total acceptance increases from @xmath244 ( @xmath245 ) for ( ) elastic production at @xmath246 to about @xmath247 at @xmath248 ( @xmath249 ) and to more than @xmath250 for @xmath251 ( @xmath8 ) , and that they are essentially independent of @xmath6 in the measurement domain . & & + & raw & weighted & raw & weighted + notag sample & @xmath252 & @xmath253 & @xmath254 & @xmath255 + tag sample & @xmath256 & @xmath257 & @xmath258 & @xmath259 + the raw numbers of events selected in the four samples defined by tables [ table : signal_vm]-[table : signal_kin_range ] are given in table [ table : nb - of - events ] , together with the numbers weighted to account for the downscaling applied to certain triggers . several background processes , which affect differently the four data samples and depend on the kinematic domain , are discussed in this section . their contributions are summarised in table [ table : percent - of - bg ] . the non - resonant @xmath260 contribution to the signal , which contributes essentially through interference , is discussed separately in section [ sect : mass - rho ] . the @xmath261 and @xmath262 backgrounds were found , using the grape simulation @xcite , to be completely negligible . & & & & + & & & & + p. diss . events & @xmath263 & @xmath264 & @xmath265 & & @xmath266 & & @xmath267 & @xmath264 & @xmath268 & & @xmath266 & + el . events & & @xmath266 & & @xmath269 & @xmath264 & @xmath270 & & @xmath266 & & @xmath271 & @xmath264 & @xmath272 + @xmath273 & & @xmath266 & & & @xmath266 & & @xmath274 & @xmath264 & @xmath275 & @xmath276 & @xmath264 & @xmath277 + @xmath278 & @xmath279 & @xmath264 & @xmath280 & @xmath281 & @xmath264 & @xmath280 & & @xmath266 & & & @xmath266 & + @xmath78 & @xmath282 & @xmath264 & @xmath280 & @xmath283 & @xmath264 & @xmath280 & @xmath284 & @xmath264 & @xmath265 & @xmath285 & @xmath264 & @xmath268 + & @xmath286 & @xmath264 & @xmath287 & @xmath288 & @xmath264 & @xmath289 & @xmath290 & @xmath264 & @xmath289 & @xmath291 & @xmath264 & @xmath292 + the notag and tag samples correspond roughly to the elastic and proton dissociative processes , respectively . however , cross - contaminations occur , due to the limited acceptance and efficiency of the forward detectors and to the presence of noise . the response of these detectors is modeled using independent measurements , by comparing signals in the various prt and fmd planes . the cross - contaminations are determined for each vm species without _ a priori _ assumptions on the relative production rates of elastic and inelastic events . in a first step , the contaminations are calculated from the numbers of tag and notag events and from the probabilities for elastic and proton dissociative events to deposit a signal in the forward detectors as obtained from the mc simulations . the crossed backgrounds are then determined in an iterative procedure from the simulations , after final tunings to the data . [ [ proton - dissociative - backgrounds - in - the - notag - samples ] ] proton dissociative backgrounds in the notag samples + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + proton dissociative events produce a background to the elastic signals in the notag samples when the mass of the excited baryonic system is too low to give a signal in the forward detectors ( @xmath293 ) or because of inefficiencies of these detectors . the background fraction increases strongly with @xmath128 , because the proton dissociative cross sections have a shallower distribution than the elastic cross sections . in the notag samples with @xmath294 , the proton dissociative background amounts to @xmath295 . [ [ elastic - backgrounds - in - the - tag - samples ] ] elastic backgrounds in the tag samples + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + conversely , elastic background in the proton dissociative samples of tag events is due to unrelated signal or noise in the forward detectors . for @xmath296 , it amounts to @xmath297 , with larger contributions for small values where the elastic to proton dissociative cross section ratio is larger . in addition , when is large enough for the scattered proton to hit the beam pipe walls or adjacent material ( @xmath298 ) , elastic events may give signal in the forward detectors . for production , the contribution from the @xmath299 channel is removed by the requirement @xmath300 , which also suppresses the contribution of the @xmath301 channel ( @xmath302 ) with the @xmath303 meson decaying into a pion pair close to the emission vertex and the @xmath304 being undetected in the calorimeter . the largest background in the selected samples is due to the low mass tail of @xmath273 pair production extending under the peak . it amounts to @xmath305 and depends on . the shape of the @xmath273 distribution corresponding to small values of is discussed in section [ sect : mc ] . a small contamination in the samples is due to the channel @xmath306 ( @xmath307 ) when each photon from the @xmath308 decay remains undetected because it is emitted outside the lar calorimeter acceptance , because the energy deposit in the lar calorimeter does not pass the @xmath201 threshold , or because it is associated with one of the charged pions . this background contributes to the distribution mostly below the selected mass range ; it amounts to @xmath309 of the selected notag sample with @xmath294 and @xmath310 of the tag sample with @xmath296 . the background rate increases with because the non - detection of the @xmath308 decay photons leads in general to an overestimate of the @xmath183 imbalance of the event , @xmath311 , which mimics a large value . the @xmath278 contribution below the @xmath312 signal is negligible . similarly , the diffractive production of @xmath78 mesons decaying in the mode @xmath313 ( @xmath314 ) gives background contributions to the and samples when the @xmath308 decay photons escape detection . in addition , the @xmath315 ( @xmath316 ) channel gives an irreducible background to the signal . the background due to production contributes @xmath317 to the elastic and @xmath318 to the proton dissociative samples , and @xmath284 and @xmath319 for the samples , respectively . the non - detection of photons leads to large reconstructed values for these contributions . note that for the cross sections quoted below , as for results in previous hera papers , the @xmath320 interference is neglected : its contribution is small and cancels when integrated over the mass range . the largest background to the signal and the second largest background to the signal is due to diffractive production . ] . the mesons decay mostly into a meson and a pion pair , leading to final states with four charged pions ( @xmath321 ) or with two charged and two neutral pions ( @xmath322 ) . the @xmath323 events can mimic large or production when the photons from the @xmath308 decays escape detection , which induces a @xmath183 imbalance in the event and a distortion of the @xmath74 distribution , similarly to the @xmath324 and @xmath325 backgrounds . at high , this background affects mostly the notag samples . it is indeed distributed between the notag and tag samples following the elastic to proton dissociative production cross section ratio , whereas genuine high and mesons are essentially produced with proton dissociation and thus contribute mainly to the tag samples . no cross section measurement of diffractive production has been published in the relevant @xmath151 range . the contribution to the signal is thus determined from the data themselves , using a method presented in the h1 analysis of high electroproduction @xcite . the distribution of the variable @xmath326 , which is the cosine of the angle between the transverse components of the candidate momentum , @xmath327 , and of the event missing momentum , @xmath311 , is sensitive to the relative amounts of signal and background . the contribution gives a peak at @xmath328 and a negligible contribution at @xmath329 , since the and the missing @xmath308 s are all emitted roughly in the direction of the . in contrast , the signal gives peaks at @xmath328 and @xmath329 . however , for genuine production , @xmath326 is also correlated to the angle @xmath212 between the production plane and the electron scattering plane , which is distributed according to the a priori unknown value of the combinations of spin density matrix elements and ( eq . ( [ eq : angle_phi ] ) of the appendix ) . an iterative procedure is used to determine simultaneously the amounts of background in the notag and tag samples , the matrix element combinations and ( assumed to be identical for elastic and proton dissociative scattering ) , and the distributions of elastic and proton dissociative production . it is found to converge after a few steps . the results are also used to calculate the background to the signal . the background is estimated to contribute @xmath330 to the notag samples with @xmath294 , and @xmath331 to the tag samples with @xmath296 . monte carlo simulations based on the diffvm program are used to describe , , and vm production and decay , detector response ( acceptances , efficiencies and variable reconstruction ) and radiative effects . the diffvm program @xcite is based on regge theory and vector meson dominance @xcite . the @xmath68 diffractive mass distribution for proton dissociative events contains an explicit simulation of baryonic resonance production for @xmath332 and a dependence @xmath333 for larger masses @xcite , with quark and diquark fragmentation simulated using the jetset programme @xcite . the and mc samples are reweighted according to the measurements of the , @xmath6 and differential cross sections and of the angular vm production and decay distributions : the angle @xmath177 is distributed according to the measurements of the matrix element ( eq . ( [ eq : cosths ] ) ) , the angle to those of the and combinations ( eq . ( [ eq : angle_phi ] ) ) , and the angle to those of the parameter , which in the schc approximation fixes the @xmath334 distribution ( eq . ( [ eq : wcosdelta ] ) ) . for the @xmath78 and backgrounds , the cross section dependences on the kinematic variables , @xmath6 and are taken to be the same as for mesons at the same value . for the two - body @xmath78 decay , the angular distributions are taken as for mesons . for three - body @xmath78 and @xmath212 decays , the angular distributions are chosen to follow @xmath212 and distributions described by the same values of the matrix elements as for two - body decays . for decays decay mode , the two pions do not form a resonance and can be assumed to be in a spin 0 state . the angular decay distribution thus includes the two possible polarisation states of the meson , with the squared amplitude @xmath335 ( @xmath336 ) corresponding to the probability that it is longitudinally ( transversely ) polarised , giving in the schc approximation , with the notations of the appendix : + @xmath337 \right.$ ] + @xmath338 \right \ } $ ] , where @xmath339 . ] , the parameters @xmath340 and @xmath341 describe the angular distributions @xcite . the values @xmath342 are chosen for the present simulations . the ratio of proton dissociative to elastic cross sections is taken from the present analysis and assumed to be the same for all vms . all kinematic and angular distributions are taken to be identical for elastic and proton dissociative scattering , as supported by the present data , except for the dependence of the cross sections . the to cross section ratio is set to that measured in this analysis . the @xmath78 to ratio is taken from zeus measurements @xcite . for production , a to ratio of @xmath343 is used , as a result of the procedure described in the previous section . for , and @xmath78 mesons , the particle mass , width and decay branching ratios are taken from the pdg compilation @xcite . the mass and width of the resonance are taken as @xmath344 and @xmath345 , respectively . for and meson decays into two pseudoscalar mesons , the mass distributions are described by a relativistic breit - wigner function @xmath346 with momentum dependent width , as described in section [ sect : mass_shapes ] . in addition , the mass shape is skewed according to the parameterisation of ross and stodolsky @xcite , @xmath347 with the dependent value of @xmath348 measured in this analysis . the @xmath260 background in the mass region is taken from the skewed breit - wigner distribution for mesons , modified for @xmath349 according to the empirical form @xmath350 , \label{eq : low - mpp}\ ] ] with masses expressed in and the parameter @xmath351 being taken to be @xmath352 . this parameterisation describes the low mass distribution well , as shown in figs . [ fig : mass_distrib - mpp - nt ] , [ fig : mass_distrib - mpp - t ] and [ fig : rh_mass_sod ] , where the cut @xmath353 is applied to suppress genuine production . radiative effects are calculated using the heracles program @xcite . corrections for these effects in the selected kinematic range with @xmath203 are of the order of @xmath354 . all generated events are processed through the full geant @xcite based simulation of the h1 apparatus and are reconstructed using the same program chain as for the data . of particular relevance to the present analysis is the description of the forward detector response ; the activity in these detectors , not related to vm production , is obtained from data taken independently of physics triggers , and is superimposed on generated events in the mc simulations . the and mass distributions are shown in figs . [ fig : mass_distrib - mpp - nt ] to [ fig : mass_distrib_mkk ] , separately for the notag and tag samples . the results of the monte carlo simulations , comprising signal and backgrounds , are also shown . they are reweighted and normalised to the data as described in the previous section . ( 12.0,12.0 ) ( 0.0,0.0 ) ( 12.0,12.0 ) ( 0.0,0.0 ) ( 12.0,12.0 ) ( 0.0,0.0 ) the mass spectra are presented from threshold to masses well above the actual measurement ranges defined in table [ table : signal_vm ] . the spectra in figs . [ fig : mass_distrib - mpp - nt ] and [ fig : mass_distrib - mpp - t ] are presented in four bins in , with the cut @xmath353 . the spectra in fig . [ fig : mass_distrib_mkk ] exhibit the reflection of production and of backgrounds . the mass distributions are well described from the threshold at @xmath355 up to @xmath356 . the backgrounds are small in the mass ranges selected for the physics analyses , shown as the shaded regions in the figures , but their contributions can be distinctly identified outside these domains . in the distributions , they are particularly visible at low mass and , as expected , they contribute mostly at large , especially in the notag sample with @xmath357 of fig . [ fig : mass_distrib - mpp - nt ] . a decrease of the background with increasing for the same ranges in is also observed ( not shown here ) , which is explained by the larger transverse momentum of the virtual photon , resulting in larger @xmath183 values of the decay photons which thus pass the detection threshold and lead to the rejection of the events . in view of the small background in the final selected samples , an analysis of only the mass spectrum , performed in the restricted mass range @xmath215 , is not sufficient to constrain the contribution . controlling this background is crucial for the measurements of the slope and of the matrix element . in the present analysis , the amount of background is obtained from the distribution of the variable ( defined in section [ sect : rhop_bg ] ) . the value determined within the mass range @xmath215 also gives a good description of the mass range @xmath358 , below the actual measurement . this demonstrates the reliability of the background estimate . the mass distribution shown in fig . [ fig : mass_distrib_mkk ] is also very well described . the @xmath260 background under the peak , which contains a contribution obtained from the analysis , is small , the background under the peak amounts to @xmath359 for @xmath246 ( @xmath360 from the @xmath273 low mass tail , @xmath361 from @xmath78 and @xmath362 from production ) , and to @xmath363 for @xmath364 ( @xmath365 , @xmath366 and @xmath365 , respectively ) . an empirical description of the background by zeus , using a simple power law shape , is in agreement with these detailed findings : it amounts to @xmath245 for @xmath246 and @xmath367 for @xmath364 @xcite . ] . figures [ fig : control - plotsa ] and [ fig : control - plotsb ] present several kinematic and angular variable distributions for the samples selected as defined in tables [ table : signal_vm]-[table : signal_kin_range ] . they demonstrate that the simulations , taking into account the detector acceptance and response and the background contributions , correctly describe the data . ( 18.0,18.0 ) ( -1.5,0 . ) ( 18.0,18.0 ) ( -1.5,0.0 ) figure [ fig : control - plotsa ] shows kinematic variable distributions of the and notag samples . the structure observed in the electron polar angle distribution ( a ) results from the different kinematic range selections for the different years . the dip in the distribution ( b ) of the laboratory azimuthal angle @xmath368 of the electron is due to an asymmetric electron trigger acceptance . the @xmath183 distributions of the decay mesons ( e ) , ( i ) reflect the vm mass and the decay angular distributions . the good description of the difference between the azimuthal angles of the decay kaons in the sample ( f ) indicates that the reconstruction of pairs of tracks with small differences in azimuthal angles is under control . a description of similar quality is obtained for the tag samples . figures [ fig : control - plotsb](a)-(c ) present distributions related to the spin density matrix elements . the distributions ( d)-(e ) are sensitive to the values of the matrix element combinations and and to the amount of background especially at high as discussed in section [ sect : rhop_bg ] . the distributions ( f)-(i ) are sensitive to the amount of diffractive backgrounds ( proton dissociation for the notag sample , elastic scattering for the tag sample ) and to the values assumed for the exponential @xmath74 slopes . uncertainties on the detector response and background contributions are listed in table [ table : syst ] . they are estimated by variations in the mc simulations within the indicated limits , which are in most cases determined from the data . global normalisation errors are given separately . + electron polar angle @xmath369 & & all vm & @xmath370 mrad + spacal energy scale & & all vm & @xmath371 + noise threshold in lar & & all vm & @xmath372 + + @xmath373 & @xmath374 & all vm & @xmath375 + @xmath376 & @xmath377 & all vm & @xmath378 + @xmath379 & @xmath380 , @xmath130 in & & el . : @xmath381 + & & & p. diss . : @xmath382 + & & , , @xmath78 & el . : @xmath383 + & & & p. diss . : @xmath384 + + proton dissoc . @xmath385 elastic & & all vm & @xmath386 @xmath387 + shape skewing & @xmath388 & & @xmath375 + vm cross sections & & @xmath389 & @xmath390 @xmath387 + & & @xmath391 & @xmath392 @xmath393 + & & @xmath394 & @xmath395 @xmath396 + decay & @xmath340 and @xmath341 & & see text + + & @xmath397 & , & @xmath398 + , & @xmath399 & , & @xmath400 + & & , & @xmath401 + + luminosity & & all vm & @xmath402 + trigger efficiency & & all vm & @xmath403 + track rec . ( per track ) & & all vm & @xmath404 + width of rel . b .- w . & see text & & @xmath405 + @xmath406 @xmath407 & see @xcite & & @xmath408 + @xmath260 under peak ( @xmath351 param . ) & & & @xmath409 + @xmath410 & @xmath411 & all p. diss . & @xmath375 + the error on the electron polar angle @xmath412 , which affects the measurements and the acceptance calculations , is due to the uncertainty on the absolute positioning of the bdc with respect to the cjc chambers , the uncertainty on the electron beam direction in the interaction region and the error on the @xmath91 position of the interaction vertex . the uncertainty on the energy scale of the spacal calorimeter affects the cross section measurements through the electron energy threshold of @xmath194 and the @xmath211 cut . the uncertainty on losses due to the rejection of events affected by noise in the lar calorimeter or containing energy deposits unrelated to the diffractive event is estimated by varying the energy threshold , both in the data and in the simulation ( where data taken from random triggers are directly superimposed to the simulated events ) . the uncertainties on the simulated cross section dependences on , @xmath6 and affect the bin - to - bin migrations and the extrapolations from the average value of the kinematic variables in a bin to the position where they are presented ( bin centre corrections " ) . an absolute error of @xmath413 is used for the ratio of the proton dissociative ( with @xmath414 ) to elastic cross sections , which corresponds to about @xmath415 relative error . it is estimated by varying by @xmath416 the parameter @xmath348 in the simulated dissociative mass distribution @xmath417 , by varying the slope parameters of the exponential distributions of elastic and proton dissociative events within the experimental limits , and by calculating the cross section ratio using only the prt or only the fmd . the latter covers uncertainties in the inefficiencies of these detectors . for the cross section measurements , the error due to the extraction of the non - resonant @xmath260 background is estimated through the variation of the dependent skewing parameter @xmath348 of the ross - stodolsky parameterisation of eq . ( [ eq : rs ] ) . the errors on the various cross section ratios are taken from the present analysis for the to and to ratios , and from the zeus measurements of the @xmath418 ratio @xcite . the errors due to the uncertainty on the decay angular distribution are estimated by considering the two extreme cases @xmath419 and @xmath420 of the pair of variables defined in @xcite . the uncertainty on the angular distributions are described by varying the values of the matrix element ( for the angle @xmath177 ) , of the combinations and ( for the angle ) and of the parameter ( for the angle @xmath421 ) . the uncertainty on the choice of the momentum dependent width of mesons results in normalisation uncertainties of @xmath422 ( see section [ sect : mass_shapes ] ) . for production , the uncertainty on the @xmath260 background under the signal is estimated by varying the parameter @xmath351 globally from @xmath423 to @xmath424 ( eq . ( [ eq : low - mpp ] ) in section [ sect : mc ] ) , leading to a normalisation error of @xmath425 on the cross section measurements . for the proton dissociative cross sections , the error on the correction for the smearing through the experimental cut @xmath414 is estimated by varying the parameter @xmath348 of the @xmath68 distribution ( @xmath426 , with @xmath375 ) , which leads to an additional normalisation error of @xmath427 on the proton dissociative cross section measurement . the uncertainties on the luminosity measurement , on the triggers and on the track reconstruction efficiency are assumed to affect globally the normalisation only . systematic errors due to limited mc statistics are negligible compared to the statistical precision of the measurements ( the generated samples correspond to at least ten times the data integrated luminosity ) . all systematic errors on the measurements presented in the rest of this paper are calculated from separate quadratic sums of positive and negative effects of the variations listed in table [ table : syst ] . in all figures , measurements are shown with statistical errors ( inner error bars ) and statistical and systematic errors added in quadrature ( full error bars ) . in tables , the errors are given separately : first the statistical , second the systematic errors . overall normalisation errors are not included in the error bars but are quoted in the relevant captions . in this section , measurements of the and line shapes are presented first . the elastic and proton dissociative cross sections are then measured as a function of ( total and polarised cross sections ) , @xmath6 and @xmath74 ( total cross sections ) ; results for different vms are compared . finally , elastic and proton dissociative scatterings are compared , including tests of proton vertex factorisation . model predictions are compared to the data . the cross sections for and production presented in this paper are extracted from the numbers of events in the mass ranges @xmath215 and @xmath217 , respectively . they are corrected for all backgrounds , including for mesons the non - resonant dipion diffractive production ( see section [ sect : mass - rho ] ) . they include all corrections for detector acceptance and response . when quoted at a fixed value of a kinematic variable , the cross sections are evolved from the average value in the bin using dependences measured in this analysis . the cross sections are quoted for the full resonance mass range from the two particle threshold up to the nominal mass plus five times the resonance width : @xmath428 for mesons , the cross sections take into account the branching ratio to the @xmath429 channel . elastic and proton dissociative cross sections are given at the born level ( i.e. they are corrected for qed radiation effects ) in terms of @xmath430 cross sections ( except for the mass shapes , which are given in terms of @xmath160 cross sections ) . the @xmath430 cross sections are extracted from the @xmath160 cross sections in the weizscker - williams equivalent photon approximation @xcite using the definition @xmath431 where the flux @xmath432 of virtual photons @xcite and the inelasticity @xmath433 are given by @xmath434 @xmath435 being the fine structure constant and @xmath164 and @xmath436 the four - momenta of the incident proton and electron , respectively . the distribution of the invariant mass @xmath437 of the vm decay particles is analysed assuming the relativistic breit - wigner distribution @xmath346 with momentum dependent width @xmath438 @xcite : @xmath439 where and @xmath440 are the nominal vm resonance mass and width , @xmath441 is the momentum of the decay particles in the rest frame of the pair with mass @xmath437 , and @xmath442 is the value taken by @xmath441 when @xmath443 . for mesons , the mass extrapolation from the measurement domain @xmath215 to the full range given by eq . ( [ eq : window ] ) , including the correction for skewing effects , implies a correction factor of @xmath444 with a systematic error of @xmath422 due to the theoretical uncertainty on the choice of the momentum dependent width @xcite . for production , a very small extrapolation outside the measurement domain is required , with negligible related error . distributions of the mass in the range @xmath445 , with the cut @xmath353 applied to suppress the signal at low mass , are shown in fig . [ fig : rh_mass_sod ] for elastically produced events in four ranges in , after subtraction of the proton dissociative , , and backgrounds and corrections for detector and qed radiation effects . the mass resolution in the mass range , determined with the mc simulation , is about @xmath446 . ( 12.0,12.0 ) ( 0.0,0.0 ) [ [ skewing ] ] skewing + + + + + + + the mass distributions are skewed towards small masses , especially at low . according to sding s analysis @xcite , this is due to the interference of the meson with background from @xmath164-wave drell - type non - resonant @xmath260 pair production , with positive interference for @xmath447 and negative interference for @xmath448 . following one of the forms of skewing proposed in @xcite , the mass shape is described as @xmath449 where resonant and non - resonant @xmath260 production are supposed to be in phase . the interference is proportional to @xmath450 , which is taken to be independent of the mass ; the very small purely non - resonant contribution is given by @xmath451 . figure [ fig : rh_mass_sod ] shows that the mass shape is well described by eqs . ( [ eq : b_w]-[eq : sod ] ) over the full range @xmath452 , with the skewing parameters fitted in the range @xmath215 . no indication is found for significant additional backgrounds , also outside the mass domain used for the measurements . the skewing effect is also often conveniently parameterised in the form proposed by ross and stodolsky @xcite , given by eq . ( [ eq : rs ] ) . for a fit over the whole range with the parameterisation of ross and stodolsky , the values of the resonance mass and width are @xmath453 and @xmath454 , respectively . the sding parameterisation gives similar values , with larger errors . this is in agreement with the world average values as obtained in photoproduction @xcite : @xmath455 and @xmath456 . within errors , no difference is observed between the elastic and proton dissociative samples . ( 10.0,5.0 ) ( 0.0,0.0 ) ( 6.0,0.0 ) figure [ fig : skew ] presents the dependence of the fitted values of the skewing parameters for elastic production and @xmath348 slightly depend on the fit mass range . at low mass , this is related to the shape uncertainties reflected by the uncertainty in the parameterisation of eq . ( [ eq : low - mpp ] ) . for higher masses , the mass limit dependence may be due to additional interference of mesons with heavier ( ) resonances @xcite . ] , the mass and width of the resonance being fixed to the pdg values @xcite . the skewing effects decrease with increasing , showing that the non - resonant amplitude decreases faster with than the resonant amplitude , as expected on theoretical grounds @xcite . no significant dependence of the skewing parameters is observed as a function of @xmath6 or . ( 6.0,6.0 ) ( 0.0,0.0 ) the mass distribution for elastically produced kaon pairs is shown in fig . [ fig : ph_mass ] , after background subtraction and corrections for detector and qed radiation effects . it is described by the convolution of the breit - wigner function defined by eqs . ( [ eq : b_w]-[eq : gj ] ) with a gaussian function of width @xmath457 describing the mass resolution , as evaluated using the mc simulation . the mass and width of the resonance , fitted over the interval @xmath458 , are @xmath459 and @xmath460 , respectively , reasonably close to the world average values of @xmath461 and @xmath462 @xcite . conversely , when the mass and width are fixed to the nominal values the fitted resolution , which is assumed to be gaussian , is @xmath463 . this value , which is slightly smaller than that obtained from simulations , is interpreted as to come from small systematic effects . as expected @xcite , no indication is found for skewing effects due to interference with non - resonant @xmath464 production . ( 16.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) the measurements of the @xmath465 cross sections for and meson elastic and proton dissociative production are presented in fig . [ fig : sigma_f_qsqplmsq ] as a function of the scaling variable ) . they are quoted for @xmath466 using the @xmath6 dependences parameterised as a function of following the measurements of section [ sect : w - dep_hardening ] . using the fits of the dependence presented below , it is verified that the normalisations of the 1995 ( sv ) cross section measurement @xcite and of the present measurement are in good agreement for mesons ( the ratio is @xmath467 ) . for the data , the 1995 sv measurement is slightly lower than extrapolated from the present result ( the ratio is @xmath468 ) . this difference is attributed to the different treatments of the backgrounds . zeus measurements of and electroproduction are also shown in fig . [ fig : sigma_f_qsqplmsq ] . whereas the measurements agree well , measurements of zeus are a factor @xmath469 above the present data . when an improved estimation of the proton dissociative background , investigated for the latest zeus production study @xcite , is used to subtract this background in their analysis , the cross section ratio of the two experiments is reduced to @xmath470 , which is within experimental errors @xcite . + & el . & + @xmath348 & @xmath471 & @xmath472 + & @xmath473 & @xmath474 + & el . & + @xmath348 & @xmath475 & @xmath476 + & @xmath477 & @xmath478 + + & el . & p. diss . + @xmath479 & @xmath480 & @xmath481 + @xmath482 ( @xmath483 ) & @xmath484 & @xmath485 + & @xmath486 & @xmath487 + & el . & + @xmath479 & @xmath488 & @xmath489 + @xmath482 ( @xmath483 ) & @xmath490 & @xmath491 + & @xmath492 & @xmath493 + the total cross sections roughly follow power laws of the type @xmath494 with values of @xmath348 , fitted over the domain @xmath495 , given in table [ table : qsq_n_fits](a ) . these values are compatible for elastic and proton dissociative scattering . they are also similar for and mesons , which supports the relevance of the scaling variable ) . the generally poor values of for fits with constant values of @xmath348 confirms the observation of @xcite : compared to a simple power law , the cross section dependence is damped for small values of ) and steepens for larger values . an empirical parameterisation @xmath496 provides a significant improvement of the fit and a good description of the data ( table [ table : qsq_n_fits](b ) ) . it is interesting to note that the fitted values of the parameter @xmath479 are close to the value @xmath497 expected in the vector meson dominance model @xcite for the exponent @xmath348 when @xmath498 . predictions of the kmw dipole model @xcite with gw saturation @xcite are compared to the data in fig . [ fig : sigma_f_qsqplmsq ] . the shape of the elastic cross section measurement is well described . the normalisation of the prediction is low by @xmath360 , while the overall normalisation error in the present measurement is of @xmath330 . predictions using cgc saturation @xcite ( not shown ) are nearly indistinguishable , except for the highest bins in where , however , the limited precision of the data does not allow to discriminate . the mrt model @xcite does not provide normalisation predictions , because of the uncertainty on the quark pair invariant mass window corresponding to the meson recombination . for this reason , the predictions for different pdf parameterisations are normalised to the data at @xmath499 . both the cteq6.5 m @xcite and the mrst 2004 nlo pdfs @xcite ( not shown ) lead to predictions which are compatible with the dependence of the data . it should however be noted that the normalisation factors required to fit the data are about @xmath500 for cteq6.5 m but larger than @xmath497 for the mrst04 nlo pdf ( see also @xcite ) . this surprisingly large factor suggests that the gluon contribution in the mrst04 nlo pdfs is underestimated . for elastic production , the kmw predictions describe the shape of the distribution well , but are higher than the data by @xmath501 . the mrt model gives a good description of the dependence of the cross section , with normalisation factors similar to those for mesons . ( 12.0,6.0 ) ( 0.0,0.0 ) ( 6.0,0.0 ) figures [ fig : phi_on_rh_f_qsq](a ) and ( b ) present as a function of and ) , respectively , the ratio of the to elastic cross sections , for which several uncertainties cancel , in particular those related to the subtraction of the proton dissociative background . the ratios are different because the same value of corresponds to different values of ) for and mesons , in view of the mass difference . a slight increase of the ratio with is observed for @xmath502 , whereas the ratio is consistent with being constant when plotted as a function of ) . similar behaviours ( not shown ) are obtained for proton dissociative production . the cross section ratios , computed for the same domains in ) for rho and phi mesons and for @xmath466 , are @xmath503 where the ratio of elastic cross sections includes the 1995 sv measurements ( @xmath161 ) . the measurements are close to the value expected from quark charge counting @xmath504 , but they tend to be slightly lower . qualitatively , the behaviour of the ratio is consistent with the dipole model . at small , the influence of the meson mass on the transverse size of the @xmath7 pair is larger , which implies that colour screening is expected to be larger for mesons than for mesons . in contrast , for @xmath505 , the transverse size of the dipole is given essentially by and symmetry is expected to be restored . ( 16.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) the dipole size effect also explains the strong increase with of the to ratio , scaled according to the quark charge content @xmath506 , as presented in fig . [ fig : su5_sigma_f_qsqplmsq](a ) , and the fact that the ratio is nearly constant and close to unity when studied as a function of ) , as shown in fig . [ fig : su5_sigma_f_qsqplmsq](b ) ( note the different vertical scales ) . although striking , the agreement with su(4 ) universality is however only qualitative , with the scaled to cross section ratios slightly below @xmath115 and the scaled to ratios slightly above @xmath115 . scaling factors obtained from the vm decay widths into electrons @xcite are expected to encompass wave function and soft effects ; the use of the factors given in @xcite modifies the scaled to ratio very little and brings the scaled to ratio slightly below @xmath115 . the separate study of the polarised ( longitudinal and transverse ) cross sections sheds light on the dynamics of the process and on the dependence of the total cross section . soft physics contributions , related to large transverse dipoles , are predicted to play a significant role in transverse cross sections , whereas hard features should be significant in longitudinal amplitudes . at relatively low values of the scale , @xmath507 , soft , finite size " effects are however expected to also affect longitudinal cross sections . the extraction of the polarised cross sections presented in this section implies the use of the measurement of the cross section ratio @xmath508 , which is performed using angular distributions and is discussed in section [ sect : polar_disc_r ] . the total cross section can be expressed as the sum of the contributions of transversely and longitudinally polarised virtual photons : @xmath509 where @xmath510 is the photon polarisation parameter , @xmath511 , with @xmath512 and @xmath513 in the kinematic domain corresponding to the present measurement . ( 16.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) the polarised cross sections , obtained from the measurements of the total cross sections and of @xmath514 , with the value of @xmath510 for the relevant , are presented in fig . [ fig : sigma_pol_f_qsqplmsq_models ] for elastic and production , as a function of ) . + @xmath515 & @xmath516 + @xmath517 & @xmath518 + @xmath519 & @xmath520 + @xmath521 & @xmath522 + results of power law fits with constant exponents are presented in table [ table : qsq_n_fits_polar ] ( the fit quality does not improve with a ) dependent value of @xmath348 ) . the fit values differ from the results @xmath523 for the longitudinal and @xmath524 for the transverse cross sections , obtained from a lo calculation of two gluon exchange @xcite . model predictions for @xmath102 and @xmath103 are compared to the data in fig . [ fig : sigma_pol_f_qsqplmsq_models ] . the gpd predictions of the gk model @xcite are slightly too flat , both for @xmath102 and for @xmath103 , but the global normalisations are within the theoretical and experimental errors , which suggests that higher order effects , not included in the model , are weak . the kmw model @xcite describes well the shapes of the @xmath102 and @xmath103 measurement and the absolute normalisation of @xmath102 , whereas the normalisation is too low for @xmath103 ; this is the reflection of the good description of the shape for @xmath525 and of the prediction for @xmath514 which is systematically too high ( see fig . [ fig : rlt - q2 ] in section [ sect : polar_disc_r ] ) . the mrt @xcite predictions for the polarised cross sections are reasonable , but for production they are too low for @xmath102 and too high for @xmath103 , which reflects the fact that the predictions for @xmath514 are too low ( fig . [ fig : rlt - q2 ] ) . the ins @xmath100-unintegrated model with the compact wave function @xcite gives predictions which are significantly too high both for @xmath102 and for @xmath103 , and too steep for @xmath103 ( not shown ) ; the predictions with the large wave function have better absolute predictions but are too steep for @xmath102 and for @xmath103 . the mps dipole saturation model @xcite describes the data rather well . ( 16.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) the same data and model predictions are presented in fig . [ fig : sigma_pol_f_qsq_scaled ] , where the longitudinal cross sections are divided by and the transverse cross sections by , all being in addition multiplied by the scaling factors @xmath526 to remove trivial kinematic dependences @xcite . the breaking of the formal expectations ( @xmath523 , @xmath524 ) for the @xmath527 dependence of the longitudinal and transverse cross sections is manifest in this presentation . this is expected from the fast increase with of the gluon density at small @xmath73 . note that the cross sections in fig . [ fig : sigma_pol_f_qsq_scaled ] are given for a fixed value of @xmath6 and thus correspond to different values of @xmath73 . the increase with of the scaled longitudinal cross section is slower than that of the scaled transverse cross section . this is reflected in the dependence of the cross section ratio @xmath508 , which is slower than @xmath528 ( see section [ sect : polar_disc_r ] , figs . [ fig : rlt - q2 ] and [ fig : rlt - compil ] ) . figure [ fig : sigma - pol_ratios_f_qsqplmsq ] shows the to and to polarised cross section ratios , scaled according to the quark charge content of the vm ( longitudinal cross sections are affected by very large errors due to the measurement errors on @xmath514 and are not shown ) . ( 16.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) the ratios of the to polarised production cross sections are within uncertainties independent of ) and close to the ratio of the total cross sections ( fig . [ fig : su5_sigma_f_qsqplmsq ] ) , suggesting little effect of the wave functions . in contrast , the ratios of the to transverse cross sections are very different from @xmath115 . this is because the polarisation states for and mesons on the one hand and for mesons on the other hand are very different for the same ) value , in view of the dependence of @xmath514 . the fact that the cross section ratios are consistent with being independent of ) thus indicates that , within the present errors , no large difference is found between the small dipoles involved in transverse production and the dipoles involved in transverse production , for @xmath529 . ( 16.0,16.0 ) ( 0.0,8.0 ) ( 8.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) figure [ fig : sigma_f_w ] displays the @xmath6 dependence of the @xmath465 cross sections for the production of and mesons , for several values of . for the first time , measurements are performed for both the elastic and the proton dissociative channels . the @xmath6 dependence of the cross sections is well described by power laws of the form @xmath530 represented by the straight lines in fig . [ fig : sigma_f_w ] . this parameterisation is inspired by the regge description of hadron interactions at high energy , with @xmath531 in hadron interactions , typical values for the intercept and the slope of the pomeron trajectory are @xmath532 to @xmath533 @xcite and @xmath534 @xcite , respectively . ( 16.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) the @xmath6 dependence of the cross sections is presented in fig . [ fig : alphapom0_f_qsq ] in the form of the intercept of the effective pomeron trajectory , @xmath535 , to allow comparison between different channels with different @xmath74 dependences . the values of @xmath535 are calculated for the present and meson production from the @xmath6 dependences following eqs . ( [ eq : w - fit]-[eq : traj ] ) , using the measured values of @xmath536 and the measurements of for production given in table [ table : shrinkage ] ; the latter are derived from the evolution with @xmath74 of the @xmath6 dependence of the cross section . the measurements of @xmath535 are presented as a function of the scale @xmath537 for , and production , and as a function of @xmath538 for dvcs , as expected for the lo process . up to values of the order of @xmath539 , the @xmath6 dependence of the elastic cross section for both and production is slightly harder than the soft behaviour characteristic of hadron interactions and photoproduction ( fig . [ fig : alphapom0_f_qsq](a ) ) . for the higher range , higher values of @xmath535 are reached , of the order of @xmath540 to @xmath541 , compatible with measurements . this evolution is related to the hardening of the gluon distribution with the scale of the interaction . consistent results are obtained in the proton dissociative channel , but with larger uncertainties ( fig . [ fig : alphapom0_f_qsq](b ) ) . ( 16.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) in principle , the @xmath6 dependence of vm production can put constraints on gluon distributions , including effects like saturation at very low @xmath73 and large @xmath6 values . all models predict a hardening of the @xmath6 distribution with increasing , following from the steepening of the gluon distributions . as examples , predictions are given in fig . [ fig : sigma_f_w_models ] for the gk gpd model @xcite , the ins @xmath100-unintegrated model with the large wave function @xcite and the kmw dipole @xcite with gw saturation @xcite . the mps saturation model @xcite ( not shown ) gives predictions for production nearly identical to those of kmw . in general , relatively small differences are found between the model predictions for the @xmath6 dependence , and the present data do not provide significant discrimination . differences in normalisation between models in fig . [ fig : sigma_f_w_models ] reflect differences in the predicted dependence of the cross sections . ( 16.0,16.0 ) ( 0.0,8.0 ) ( 8.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) the differential cross sections as a function of for and elastic and proton dissociative production are presented in fig . [ fig : sigma_f_t ] for different ranges in . they are well described by empirical exponential laws of the type @xmath131 . ( 16.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) the slope parameters @xmath130 extracted from exponential fits in the range @xmath294 for elastic scattering and @xmath296 for proton dissociation are presented in fig . [ fig : b_f_qsq ] as a function of the scale @xmath537 . the measurements of the proton dissociative slopes are the first precise determination at hera for light vm in electroproduction ; they constitute an important ingredient for the extraction of the elastic @xmath130 slopes . in fig . [ fig : b_f_qsq ] , and measurements by zeus and measurements are also presented as a function of , together with dvcs measurements ( with @xmath538 ) . the present measurements of the @xmath130 slopes for @xmath542 are higher than those of zeus @xcite and also than those of a previous h1 measurement @xcite . two sources of systematic experimental differences are identified . the first is related to the estimation of the proton dissociative background , both in size and in shape . the subtraction of a smaller amount of proton dissociative background and the use of a steeper proton dissociative slope lead to shallower distributions of the elastic cross section and to smaller @xmath130 slope measurements . the use of a central value of @xmath543 for the proton dissociative slope , as assumed in @xcite , compared to the values measured here ( fig . [ fig : b_f_qsq](b ) ) , leads to a decrease of the elastic slope determination by @xmath544 , and a variation by @xmath545 of the amount of proton dissociative background induces a change of the elastic slope measurement by @xmath546 for @xmath547 and @xmath548 for @xmath126 . the second and major source of discrepancy , for both vms , is in the treatment of the , and mostly backgrounds discussed in section [ sect : rhop_bg ] . because of the non - detection of the decay photons , these backgrounds exhibit effective distributions which are much flatter than their genuine distributions and than the signal . neglecting completely the presence of the background would lead in the present analysis to a decrease of the measurement of the elastic @xmath130 slope by @xmath549 for @xmath550 and @xmath551 for @xmath126 . in an optical model inspired approach , the @xmath74 slopes for dvcs and vm production result from the sum of terms describing the form factors due to the transverse sizes of the scattered system @xmath67 ( @xmath552 ) , of the @xmath7 dipole pair ( @xmath553 ) and of the exchange ( @xmath554 ) . an additional form factor reflecting the vm transverse size may also give a contribution , @xmath555 , to the @xmath74 slope for light vm production in models where the wave function plays an important role in the process , while being negligible for dvcs and for . the value of the slope can thus be decomposed as : @xmath556 in elastic scattering , the slope @xmath557 reflects the colour distribution in the proton . for baryonic excited states with size larger than that of the proton , larger slopes ( i.e. steeper @xmath74 distributions ) than for elastic scattering may be expected . in contrast , when the proton is disrupted in the diffractive scattering , no form factor arises from the @xmath67 system and @xmath552 is expected to be @xmath558 . the @xmath559 contribution of the exchange is generally believed to be small and independent of . there is indeed a priori no relation between and the transverse size of the exchange , at least for @xmath136 and for @xmath109 taken to be constant ( ll bfkl ) . it is visible in fig . [ fig : b_f_qsq ] that , already for @xmath560 , the elastic @xmath130 slopes for light vm electroproduction are significantly lower than in photoproduction , showing a departure from purely soft diffraction and a decrease of the relevant @xmath7 dipole transverse size . until the scale reaches values @xmath561 , light vm slopes are however significantly larger than for . this indicates the presence of dipoles with relatively large transverse sizes for light vms in this domain . this is expected in the transverse amplitudes and also in longitudinal amplitudes until the fully hard regime is reached ( finite size " effects ) . light vm and dvcs slopes are compatible when plotted as a function of the scales and , respectively . for large scale values , they are consistent with the data , although they may be slightly higher . all these features confirm that the present domain covers the transition from the regime where soft diffraction dominates light vm production to the regime where hard diffraction dominates . the comparable values of the slopes for , and production in the harder regime suggests that light vm form factors are small . for proton dissociative diffraction , the @xmath74 slopes shown in fig . [ fig : b_f_qsq](b ) have significantly smaller values than for elastic scattering . this is expected for @xmath67 systems above the nucleon resonance region , with vanishing values of @xmath552 . the proton dissociative slopes for and mesons are similar at the same value , but remain larger than for , confirming the presence of large dipoles for @xmath562 or , alternatively , leaving room for a light vm form factor . in figure [ fig : sigma_f_t ] predictions of the mps saturation model @xcite for the @xmath74 dependence of the cross sections are shown , superimposed on the elastic measurements . the data fall faster with than predicted by the model , especially at small . the discrepancy is particularly significant for production . ( 7.0,7.0 ) ( 0.0,0.0 ) a dipole function with a @xmath74 dependent two - gluon form factor has been proposed by frankfurt and strikman ( fs ) @xcite , with @xmath563 which tends to @xmath564 for @xmath565 , with @xmath566 . fits of this parameterisation to the data for and elastic production in several bins in are shown in fig . [ fig : sigma_f_t ] , superimposed on the measurements . the fit quality is good , similar to the exponential fits . figure [ fig : two - gluon ] presents the extracted values of the parameter @xmath567 as a function of for the and elastic channels . the parameter increases with , from about @xmath568 at @xmath569 to about @xmath570 at @xmath571 . a measurement in photoproduction is also shown . the ) dependence of the form factor reflects the dependence of the @xmath74 distributions , as summarised in fig . [ fig : b_f_qsq ] . ( 16.0,8.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) the @xmath6 dependences in four bins in of the cross sections for meson production are presented in fig . [ fig : sigma_w_f_t ] for two values of . the notag ( @xmath294 ) and tag ( @xmath296 ) samples are combined in order to extend the measurement lever arm in . it was checked that , using only the notag events , compatible values of are obtained , although with much larger errors . the combination is also supported by the fact that the values of @xmath535 for the elastic and proton dissociative processes are compatible ( see fig . [ fig : alphapom0_f_qsq ] ) . ( 16.0,4.0 ) ( 0.0,0.0 ) ( 8.0,0.0 ) ( ) & + @xmath572 & @xmath573 + @xmath574 & @xmath575 + the @xmath6 dependences , which are observed to depend on , are parameterised following the power law of eq . ( [ eq : w - fit ] ) . the extracted values of @xmath576 are presented in fig . [ fig : delta_f_t ] . linear fits to the @xmath74 dependence of @xmath577 , following eq . ( [ eq : traj ] ) , give the measurements of the slope of the effective pomeron trajectory reported in table [ table : shrinkage ] . values slightly smaller than @xmath578 and higher than @xmath423 are obtained . in soft diffraction , the non - zero value of the slope of the pomeron trajectory ( @xmath579 ) explains the shrinkage of the forward diffractive peak with increasing @xmath6 : @xmath580 the parameter can thus in principle also be obtained from the evolution with @xmath6 of the exponential slopes for elastic production , but this measurement is affected by the large errors on @xmath130 ( not shown ) . ( 8.0,8.0 ) ( 0.0,0.0 ) figure [ fig : alprim ] summarises measurements by h1 and zeus for dvcs and in photo- and electroproduction of , and mesons . the measurement for photoproduction @xcite , which combines the zeus data at high energy with omega results @xcite at low energy , is @xmath581 , which is lower than the value @xmath582 typical for soft hadronic diffraction and is similar , within errors , to values of in electroproduction . measurements of at large are consistent with @xmath423 , with small errors on the measurements @xcite . in the bfkl description of hard scattering , the value of , which reflects the average transverse momentum @xmath100 of partons along the diffractive ladder , is expected to be small . in regge theory , the reggeon trajectories are fixed by the resonance positions , and slopes do not depend on . evolutions of the effective pomeron trajectories with or are thus an indication of additional effects , e.g. multiple exchanges and rescattering processes . this section presents comparisons of the proton dissociative and elastic channels , for both and meson production . measurements of the @xmath74 integrated cross section ratios are first presented , providing empirical information useful for experimental studies . the factorisation of vm production amplitudes into photon vertex and proton vertex contributions , which can be disentangled by comparing elastic and proton dissociative scatterings , is then discussed : the photon vertex contributions govern the dependence and the relative strength of the various helicity amplitudes , whereas proton vertex form factors govern the @xmath74 dependence . proton vertex factorisation ( regge factorisation " ) has been observed to hold , within experimental uncertainties , for inclusive diffraction @xcite . factorisation is tested here through the study of the independence of the vm production cross section ratios at @xmath104 and through the measurement of the difference @xmath583 between the elastic and the proton dissociative exponential @xmath74 slopes . ( 12.0,6.0 ) ( 0.0,0.0 ) ( 6.0,0.0 ) figure [ fig : pd_on_el_f_qsq ] presents , as a function of , the ratio of the proton dissociative to elastic cross sections , for and mesons . in the ratio , several systematic uncertainties cancel , in particular those related to meson reconstruction . no significant dependence of the ratios on is observed . the average ratios of proton dissociative ( with @xmath414 ) to elastic cross sections , integrated over @xmath74 , are : @xmath584 \frac { \sigma_{tot , { \rm p.~diss.}}^{\rm m_y < 5~{\rm gev } } } { \sigma_{tot , { \rm el . } } } ( \phi ) & = & 0.50 \pm 0.04~{\rm ( stat.)}~^{+ 0.06}_{-0.08}~{\rm ( syst.)}\pm 0.01~{\rm ( norm.)}\ . \label{eq : pd_to_el_ratio}\end{aligned}\ ] ] within uncertainties , the values for the two vms are compatible . using the diffvm model to estimate the contributions of proton dissociative scattering with @xmath585 , the ratio of the proton dissociative cross section for the full @xmath68 mass range to the elastic cross section is found to be close to @xmath115 . this value is used e.g. in @xcite . if the same object ( e.g. a gluon ladder ) is exchanged in proton dissociative and elastic scattering , proton vertex factorisation should be manifest through the independence of the cross section ratio for @xmath104 . ( 12.0,6.0 ) ( 0.0,0.0 ) ( 6.0,0.0 ) for exponentially falling @xmath74 distributions , the cross section ratio at @xmath104 is obtained from the total cross sections and the @xmath130 slopes as @xmath586 figure [ fig : pd_on_el_f_qsq_t0 ] presents , as a function of , the cross section ratios at @xmath104 for and production , as obtained from the total cross section ratios presented in fig . [ fig : pd_on_el_f_qsq ] and the @xmath130 slopes given in fig . [ fig : b_f_qsq ] . the average ratios for both vms are measured as : @xmath587 the ratios are observed to be independent of and consistent for the two vms , which supports proton vertex factorisation . the ratios of the proton dissociative to elastic @xmath130 slopes are also independent of , with average values of @xmath588 this empirical observation is consistent with the independence of the total cross section ratios ( fig . [ fig : pd_on_el_f_qsq ] ) and of the cross section ratios at @xmath104 ( fig . [ fig : pd_on_el_f_qsq_t0 ] ) . in the optical model approach of eq . ( [ eq : optical ] ) , assuming pomeron universality , the difference between the elastic and proton dissociative @xmath130 slopes , @xmath583 , is related only to the proton size and independent of the interaction scale at the photon vertex and of the vm species . ( 6.0,6.0 ) ( 0.0,0.0 ) figure [ fig : b_el - minus - pd ] presents the slope difference @xmath583 for and meson production , as a function of . within errors , independent values for the slope differences are found , with consistent average values of @xmath589 these observations support proton vertex factorisation , with a proton form factor contribution of about @xmath590 . measurements of photo- and electroproduction are also presented in fig . [ fig : b_el - minus - pd ] . they are consistent with independence , with @xmath591 , a value significantly smaller than for and production ; for dvcs @xcite , the measurement is @xmath592 . the difference observed between light and heavy vms is difficult to understand in the optical model , since the contributions to the slopes of the @xmath141 dipole form factors and of possible vm form factors should cancel in the difference . it may indicate that the hard regime is not reached for and mesons in the present kinematic domain . information on the spin and parity properties of the exchange and on the contribution of the various polarisation amplitudes are accessed in diffractive vm production through the distributions of the angles @xmath177 , @xmath213 and @xmath212 defined in fig . [ fig : dec_ang ] . the present section presents , successively , the measurements of the spin density matrix elements , a discussion of the nature of the exchange , measurements of the longitudinal over transverse cross section ratio @xmath514 , and measurements of the ratios and relative phases of the helicity amplitudes . the results are compared with qcd models . in the formalism of schilling and wolf @xcite , summarised in the appendix , the angular distributions allow the measurement of spin density matrix elements given in the form @xmath593 , which are normalised bilinear combinations of the complex helicity amplitudes @xmath594 , @xmath595 and @xmath596 being the helicities of the virtual photon and of the vm , respectively , and @xmath597 and @xmath598 those of the incoming proton and of the outgoing baryonic system @xmath67 . at hera , the proton beam is not polarised and the helicity of the outgoing baryonic system @xmath67 is not measured ; the helicities @xmath597 and @xmath598 are thus integrated over . for the electron beam , transverse polarisation builds up progressively over the running period through the sokolov - ternov effect but the related matrix elements are measurable only for @xmath599 , where @xmath600 is the electron mass , and are not accessible in electroproduction . the electron beam is thus treated here as unpolarised . in these conditions , a total of 15 independent components of the spin density matrix remain accessible to measurement . under natural parity exchange ( npe ) in the @xmath74 channel poles with @xmath601 , @xmath602 and @xmath603 being the particle parity and spin , respectively . ] , five @xmath133 amplitudes are independent : two helicity conserving amplitudes ( @xmath604 and @xmath605 ) , two single helicity flip amplitudes ( @xmath606 and @xmath607 ) and one double flip amplitude ( @xmath608 ) . the 15 matrix elements enter the normalised angular distribution @xmath609 which is given in eq . ( [ eq : w ] ) of the appendix . they are measured as projections of the @xmath609 distribution onto 15 orthogonal functions of the @xmath177 , and angles , listed in appendix c of @xcite . in practice , each matrix element is given by the average value of the corresponding ( @xmath177 , , ) function , calculated over the relevant data sample . for production , the @xmath78 , and background contributions to the angular distributions are subtracted following the results of the monte carlo simulations ; no correction is performed for the interfering non - resonant @xmath260 channel but this is expected to have a small effect since the interference contribution is small , decreases with and changes sign at the resonance mass value , so that it largely cancels when integrated over the selected mass range ( see fig . [ fig : rh_mass_sod ] ) . for production , the @xmath78 , and dipion backgrounds are subtracted . kinematic and angular distributions are corrected for detector acceptance and migration effects . the systematic errors on the measurements are estimated by varying the mc simulations according to the list given in table [ table : syst ] . in addition , a systematic error related to the binning is assigned to the acceptance correction used for determining the average value of the projection functions ; it is quantified by varying the number of bins in the @xmath177 , and angular variables . for both and mesons , the matrix element measurements for the elastic and proton dissociative channels are found to be compatible within experimental errors . in order to improve the statistical significance of the measurements and to reach higher values , the notag and tag samples with @xmath294 and @xmath296 , respectively , are combined . the large notag sample is not used because of the large background , as shown in figs . [ fig : mass_distrib - mpp - nt](c)-(d ) . ( 16.0,16.0 ) ( 0.0,0.0 ) ( 16.0,16.0 ) ( 0.0,0.0 ) ( 16.0,16.0 ) ( 0.0,0.0 ) ( 16.0,16.0 ) ( 0.0,0.0 ) the matrix element measurements are presented as a function of for and production in figs . [ fig : matelem_f_qsq_rho ] and [ fig : matelem_f_qsq_phi ] , and as a function of and the mass for production in two intervals of , in figs . [ fig : matelem_f_t_two_qsq_rho ] and [ fig : matelem_f_m_two_qsq_rho ] . ( 16.5,4.5 ) ( 0.0,0.0 ) ( 4.0,0.0 ) ( 8.0,0.0 ) ( 12.,0.0 ) the present measurements as a function of and confirm with increased precision the previous h1 results @xcite and they are globally compatible with zeus measurements as a function of @xcite . no significant dependence of the matrix elements with @xmath6 is observed within the present data . measurements ( not shown ) of the matrix elements and , obtained from fits to the @xmath610 and distributions as given by eqs . ( [ eq : cosths]-[eq : angle_varphi ] ) of the appendix , are in agreement with those presented in figs . [ fig : matelem_f_qsq_rho ] to [ fig : matelem_f_m_two_qsq_rho ] . for the combinations and for mesons , measurements from fits of eq . ( [ eq : angle_phi ] ) to the @xmath212 distribution , which give smaller errors than the projection method , are presented in fig . [ fig : rfive - one - comb ] . figures [ fig : matelem_f_qsq_rho ] and [ fig : matelem_f_qsq_phi ] present , superimposed on the and measurements , predictions of the gk gpd model @xcite and of the ins @xmath100-unintegrated model @xcite for two different wave functions ; for the gk model , the schc approximation is used and only non - zero elements are shown . for production ( fig . [ fig : matelem_f_qsq_rho ] ) , taking into account the experimental and theoretical uncertainties and the use of the schc approximation , the gk model @xcite gives a description of the data which is reasonable in shape but does not describe the normalisation well . the ins model @xcite reproduces the gross features of the evolution but there are problems in the details . the model with the compact wave function describes the matrix element evolution , but it fails for the other elements which are non - zero under schc ( , i m , re , i m ) ; on the other hand , the model with the large wave function gives a rather good description of these four elements , but fails badly for . in addition , both wave functions predict too low values for , also in the regime with @xmath611 . for mesons ( fig . [ fig : matelem_f_qsq_phi ] ) with less statistics , the picture is slightly different for the ins model @xcite , where the use of a large wave function gives a better description of all matrix elements , including , than the compact wave function . ( 16.5,4.5 ) ( 0.0,0.0 ) ( 4.0,0.0 ) ( 8.0,0.0 ) ( 12.,0.0 ) the observation at low energy @xcite of dominant natural parity exchange ( npe ) supports the attribution of the vacuum quantum numbers ( @xmath612 ) to the pomeron ; the recent observation by the hermes collaboration @xcite of the presence at low energy of a small contribution ( about @xmath305 ) of unnatural parity exchange is attributed to quark exchange ( @xmath613 , @xmath614 or @xmath615 exchange ) . at high energy , the modeling of diffraction as two gluon exchange implies a npe character , in particular in the gk gpd model @xcite . with unpolarised beams and for a single value of the beam energies , the only accessible information about the parity of the exchange is the asymmetry @xmath616 between natural ( @xmath617 ) and unnatural ( @xmath618 ) parity exchange for transverse photons , using eq . ( [ eq : npe - asym ] ) of the appendix . measurements of @xmath619 as a function of and for and mesons are presented in fig . [ fig : npe - asym ] . they are globally compatible with @xmath115 , which supports npe for transverse photons . natural parity exchange is assumed in the following . inspection of figs . [ fig : matelem_f_qsq_rho ] and [ fig : matelem_f_qsq_phi ] shows that , for both and meson electroproduction , the five matrix elements listed in eq . ( [ eq : non-0-schc ] ) of the appendix ( , , i m , re , i m ) , which contain products of the two helicity conserving amplitudes , @xmath620 and @xmath605 , are significantly different from zero , with the schc relations of eq . ( [ eq : schc_pairing ] ) being approximately satisfied . in addition , with the significant exception of , the other matrix elements are small or consistent with @xmath423 . in the present kinematic domain , schc is thus a reasonable approximation , which can be used to obtain information on the transition amplitudes . in order to decrease the sensitivity to the schc violating amplitudes , which increase with ( see sections [ sect : polar_disc_hel_flip ] and [ sect : polar_disc_ampl_ratios ] ) , only events with @xmath294 are used in the rest of this section . [ [ psi - distributions - phase - delta - between - the - schc - amplitudes ] ] @xmath621 distributions ; phase @xmath622 between the schc amplitudes + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + ( 4.5,4.5 ) ( 0.0,0.0 ) under schc , the angular distribution @xmath609 reduces to a function of the angles @xmath177 and @xmath334 , eq . ( [ eq : wcosdelta ] ) , which allows the extraction in this approximation of the cross section ratio @xmath508 and of the phase @xmath622 between the @xmath620 and @xmath605 amplitudes . measurements of @xmath623 obtained from two - dimensional fits of eq . ( [ eq : wcosdelta ] ) with @xmath514 left free are presented in fig . [ fig : cosdelta ] as a function of for and production ( @xmath294 ) . they are in agreement with the measurements obtained with @xmath514 fixed to the values measured in the schc approximation using the matrix element and eq . ( [ eq : r_schc ] ) . the measurements of @xmath624 are close to @xmath115 , indicating that the transverse and longitudinal amplitudes are nearly in phase . for production with @xmath625 , @xmath624 differs however significantly from @xmath115 , as is also observed for around @xmath626 in the low energy measurement by hermes @xcite . an indication of an increase of @xmath624 toward @xmath115 at high may be present in the data . an interpretation of a value of @xmath623 different from @xmath115 at high energy in terms of a @xmath6 dependence of @xmath132 will be given in section [ sect : polar_disc_ampl_phases ] . a significant violation of schc is observed in figs . [ fig : matelem_f_qsq_rho ] and [ fig : matelem_f_qsq_phi ] through the non - zero value of the matrix element , for and for mesons ( see also fig . [ fig : rfive - one - comb ] for the combination measurement for mesons ) . the matrix element is proportional to the product @xmath627 of @xmath620 , the leading schc amplitude , and @xmath606 , the helicity flip amplitude describing the transition from a transverse photon to a longitudinal vm . in figs . [ fig : matelem_f_t_two_qsq_rho ] and [ fig : matelem_f_m_two_qsq_rho ] , non - zero values , with dependent strengths , are also observed in production for the matrix elements re , re and i m , which contain the product @xmath628 of @xmath606 and the second schc amplitude @xmath605 . the data tend to support the relation @xmath629 of eq . ( [ eq : t01-non - schc_pairing ] ) . other matrix elements are , within errors , consistent with @xmath423 when integrated over @xmath74 . these findings confirm the previous h1 observation @xcite in production that the helicity flip amplitude is significantly different from @xmath423 in the present domain and is dominant among the schc violating amplitudes , supporting the hierarchy ( see for instance @xcite ) @xmath630 note that helicity violation as such is not a signature for hard processes . when integrated over , the amplitude in the present kinematic domain is larger for low than for large , as shown by the matrix element measurement in fig . [ fig : matelem_f_m_two_qsq_rho ] . at low energy and for @xmath631 around @xmath632 , the amplitude is non - zero , with @xmath633 to @xmath634 for @xmath6 about @xmath635 @xcite and @xmath636 to @xmath637 for @xmath638 @xcite . the matrix element increases with @xmath208 , as observed in fig . [ fig : matelem_f_t_two_qsq_rho ] ( see also fig . [ fig : rfive - one - comb ] ) . this is expected on quite general grounds for helicity flip amplitudes , as will be discussed in section [ sect : polar_disc_ampl_ratios ] . the cross section ratio @xmath508 is one of the most important observables in the study of light vm production since it is sensitive to the interaction dynamics , including effects related to the interacting dipole size or depending on the vm wave function . in the schc approximation , @xmath514 can be calculated from the matrix element : @xmath639 in view of the observed violation of schc , a better approximation takes into account the dominant helicity flip amplitude @xmath606 and uses in addition the measurement of : @xmath640 where npe is assumed and the amplitudes are taken to be in phase . as expected , the effect of this improved approximation is mostly significant at large values , in view of the increase with of the helicity flip amplitudes : the corresponding measurement of @xmath514 is lower than that obtained in the schc approximation by about @xmath641 for @xmath642 and about @xmath643 for @xmath644 , independently of . integrated over @xmath74 , this makes a @xmath645 difference . measurements of @xmath514 are presented in the following using the improved approximation of eq . ( [ eq : r_t01 ] ) . the general features of the kinematic variable dependences discussed below are similar when the schc approximation of eq . ( [ eq : r_schc ] ) is used . ( 14.0,7.0 ) ( 0.0,0.0 ) ( 7.0,0.0 ) the measurements of @xmath514 presented in fig . [ fig : rlt - q2 ] show a strong increase with , which is tamed at large , a feature already noted in previous h1 @xcite and zeus @xcite publications . for production , the gk gpd model @xcite , the mrt model @xcite and the ins model @xcite with the compact wave function give a good description of the measurements , whereas the kmw @xcite predictions are too high and the ins model with the large wave function is ruled out . the predictions of the mps model @xcite ( not shown ) are very similar to those of kmw up to @xmath646 , and then slightly lower . the dependence of the ik @xcite model ( not shown ) is similar to that of the mrt model , since it is derived in a similar way . for production , the kmw model gives a good description while the mrt predictions are too low ; within the quoted uncertainty , the gk model describes the data ; for the ins model , the large wave function gives a slightly better description than the compact wave function ; the predictions of the mps model ( not shown ) are again similar to those of kmw , although slightly higher at low . ( 10.0,10.0 ) ( 0.0,0.0 ) @xmath514 measurements for , and mesons are presented as a function of the scaling variable in fig . [ fig : rlt - compil ] . the improved approximation , eq . ( [ eq : r_t01 ] ) , is used for the present data whereas the schc approximation is used for the other data , which makes little difference for the @xmath74 integrated measurements . a smooth and common behaviour is observed for the three vms over the full @xmath647 range and the full energy range , from the fixed target experiments to the hera collider measurements . the data are close to a law @xmath648 , represented by the dotted line , but they lie systematically below the line , with a slower increase of @xmath514 with increasing . these features are easily understood in the mrt @xcite and ik @xcite models where the formal @xmath649 evolution is damped by a factor @xmath650 and the taming of the @xmath514 evolution results from the decrease of @xmath142 with increasing . ( 14.0,14.0 ) ( 0.0,7.0 ) ( 7.0,7.0 ) ( 3.5,0.0 ) the @xmath6 dependence of @xmath514 is presented for meson production in fig . [ fig : rlt - wtm](a ) for three intervals in . because of the strong correlation in detector acceptance between @xmath6 and , the lever arm in @xmath6 for each domain in is rather limited . as discussed in section [ sect : th_context ] , the onset of hard diffraction , characterised by a strong @xmath6 dependence , is expected to be delayed for transverse amplitudes compared to longitudinal amplitudes . a harder @xmath6 dependence is thus expected for @xmath102 than for @xmath103 , resulting in an increase of @xmath514 with @xmath6 . in view of the limited precision , no significant conclusion can be drawn from the present measurements . figure [ fig : rlt - wtm](b ) presents the measurement of @xmath514 as a function of for mesons , in two bins in . for exponentially falling @xmath74 distributions , this can be translated into a measurement of the difference between the longitudinal and transverse @xmath74 slopes , through the relation @xmath651 . measurements of the slope difference @xmath652 extracted from a fit of the @xmath74 dependence of @xmath514 are given in table [ tab : bl - bt ] ( for completeness , the result for production in one bin in @xmath74 is also given in spite of the large errors ) . the errors are dominated by the systematic uncertainty on the background subtraction . a slight indication ( @xmath653 ) is found for a negative value of @xmath652 in the higher bin in . the use of the schc approximation of eq . ( [ eq : r_schc ] ) instead of the improved approximation of eq . ( [ eq : r_t01 ] ) for the measurement of @xmath514 does not affect the measurements of @xmath652 . ( ) & @xmath652 ( ) + + @xmath572 & @xmath654 + @xmath574 & @xmath655 + + @xmath656 & @xmath657 + a difference between the @xmath130 slopes is expected to indicate a difference between the transverse size of the dominant dipoles for longitudinal and transverse amplitudes ( see e.g. @xcite ) . the indication for a negative value of @xmath652 in the higher bin in is consistent with the expectation that @xmath102 reaches a harder qcd regime than @xmath103 . conversely , the absence of a dependence of @xmath514 in the lower range is consistent with the interpretation of @xmath130 slope measurements in section [ sect : t_slopes_elast ] , suggesting that large dipoles may be present in longitudinal amplitudes ( finite size " effects ) for moderate values of the scale @xmath658 . ( 6.0,6.0 ) ( 0.0,0.0 ) a striking decrease of the cross section ratio @xmath514 with the increase of the mass , which was also reported by zeus @xcite , is observed in fig . [ fig : rlt - wtm](c ) . this strong effect is not expected in calculations where the meson is treated as a particle with well defined mass and wave function . a simple interpretation of the dependence follows from the formal @xmath659 dependence of the cross section ratio , if the mass @xmath660 is understood as the dipion mass rather than the nominal resonance mass . such an interpretation is in line with the open quark approach of the mrt parton - hadron duality model @xcite , and is qualitatively supported by the calculations superimposed to the data in fig . [ fig : rlt - wtm](c ) @xcite . the mass dependence of @xmath514 expected from the interference of resonant and non - resonant @xmath260 production , discussed in @xcite , is small compared to that observed here and should decrease with . the @xmath130 slopes of the distributions do not show any significant dependence on the mass ( see fig . [ fig : bfmq2 ] ) , which indicates that the dependence of @xmath514 can not be explained by an hypothetic kinematic selection of dipoles with specific size , related either to transverse or longitudinal amplitudes . all this suggests that the vm wave function plays a limited role in the description of vm diffractive production . the measurements of the spin density matrix elements presented in figs . [ fig : matelem_f_qsq_rho ] to [ fig : matelem_f_m_two_qsq_rho ] give access to the ratios and relative phases of the helicity amplitudes . following the ik analysis @xcite , four amplitude ratios , taken relative to the dominant @xmath620 amplitude , are measured from global fits to the 15 matrix element measurements , assuming npe and taking all amplitudes as purely imaginary ; negative values correspond to opposite phases . the measurements are presented in the following sections for and mesons as a function of and , and additionally for mesons as a function of the invariant mass . the relative phases are then discussed . ( 16.5,9.0 ) ( 0.0,4.5 ) ( 4.0,4.5 ) ( 8.0,4.5 ) ( 12.,4.5 ) ( 0.0,0.0 ) ( 4.0,0.0 ) ( 8.0,0.0 ) ( 12.,0.0 ) the dependence of the four amplitude ratios for and meson production are presented in fig . [ fig : ampl - ratios - f - qsq ] . the strong decrease with of the amplitude ratio for both vms , which is consistent with a linear increase with @xmath661 , is related to the increase of the cross section ratio @xmath514 through the dominance of the schc amplitudes . for the first time , a dependence of the amplitude ratio is also observed , for meson production . this dependence is also visible in the comparison of the two ranges in figs . [ fig : ampl - ratios - f - t ] and [ fig : ampl - ratios - f - m ] . no significant dependence is observed for the amplitude ratios and . in the ik @xcite model , the amplitude ratio is given by @xmath662 where the decrease with of the anomalous dimension @xmath142 slows down the evolution , and the amplitude ratio is given by @xmath663 the model describes the evolution well for values of @xmath664 and @xmath665 , or @xmath666 and @xmath667 ( not shown ) . the latter is preferred for the description of , though the physical interpretation of this high value for the parameter @xmath142 is unclear . the @xmath74 dependence of the amplitudes , empirically parameterised as exponentially falling , is mainly determined by the proton and vm form factors . it is a reasonable assumption that these form factors affect in a similar way all amplitudes , and that their effects cancel in matrix elements and in amplitude ratios @xcite . the study of the @xmath74 dependence of the amplitude ratios thus gives access , in the reaction dynamics , to features specific to the different amplitudes . note , however , that this line of reasoning neglects the different @xmath74 dependences for transverse and longitudinal amplitudes , related to different dipole sizes . ( 16.5,9.0 ) ( 0.0,4.5 ) ( 4.0,4.5 ) ( 8.0,4.5 ) ( 12.,4.5 ) ( 0.0,0.0 ) ( 4.0,0.0 ) ( 8.0,0.0 ) ( 12.,0.0 ) figure [ fig : ampl - ratios - f - t ] shows , for both vms , the dependences of the four amplitude ratios . for the first time , a decrease with of the ratio of amplitudes is observed , both for and for production ( figs . [ fig : ampl - ratios - f - t](a ) and ( e ) ) . the increase with of the normalised helicity flip amplitudes , which could be deduced from the behaviour of the matrix element , is confirmed in figs . [ fig : ampl - ratios - f - t](b ) and ( f ) . for the second single flip amplitude , , negative values with increased strength relative to are observed in fig . [ fig : ampl - ratios - f - t](c ) at large . finally , non - zero values are found in fig . [ fig : ampl - ratios - f - t](d ) for the ratio of the double flip to the amplitude , with negative values of the ratio and intensity increasing with for both bins in . the dependence of the to amplitude ratio , which is not predicted in the ik model , eq . [ eq : ik : alpha ] , may be understood as an indication of different transverse dipole sizes in transverse and longitudinal photon scattering , as discussed in section [ sect : polar_disc_r - t ] for the @xmath74 dependence of the cross section ratio @xmath514 . this is substantiated by the calculation of the cross section ratio using the helicity amplitude ratios , the cross section ratio @xmath508 being given by : @xmath668 following the procedure of section [ sect : polar_disc_r - t ] , the difference between the longitudinal and transverse slopes are extracted from the @xmath74 dependence of @xmath514 . the results are given in table [ tab : bl - bt_ampl ] . for production , the same effect is observed as in table [ tab : bl - bt ] , where the value of @xmath514 was obtained only from the measurements of the and matrix elements using eq . ( [ eq : r_t01 ] ) : a value of @xmath669 consistent with @xmath423 for @xmath670 , and a negative value for @xmath671 . errors are reduced due to the use of all amplitude ratios in the global fits , and the value of @xmath669 in the range with @xmath671 is @xmath672 away from @xmath423 . for production , the limited statistics do not allow to measure separately the slope difference in two bins in . ( ) & @xmath652 ( ) + + @xmath572 & @xmath673 + @xmath574 & @xmath674 + + @xmath656 & @xmath675 + the @xmath74 dependence of the helicity flip amplitudes for light quarks can be explained as follows . in the case of the @xmath606 amplitude , the virtual photon with transverse polarisation fluctuates into a quark and an antiquark which , given their opposite helicities , must be in an orbital momentum state with projection @xmath115 onto the photon direction . during the hard interaction , the dipole size and the quark and antiquark helicities are unchanged , but a transverse momentum @xmath676 is transferred to the dipole , which modifies its line of flight and thus allows a change of the orbital momentum projection . the @xmath606 amplitude , which describes the production of a longitudinal meson from a transverse photon , is thus proportional to @xmath677 . similar reasons explain the @xmath74 dependence of the @xmath607 amplitude . note that , at variance with the case of light vms , for heavy vms with a non - relativistic wave function ( @xmath112 ) , the exchange of orbital momentum can not take place , thus implying schc . in the ik model the dependence of the single - flip to no - flip amplitude ratio is given by eq . ( [ eq : ik : beta ] ) , and that of by @xmath678 respectively , where the negative value of the ratio is consistent with the data in the higher domain , fig . [ fig : ampl - ratios - f - t](c ) . in the two - gluon exchange picture of diffraction for the double flip @xmath679 amplitude , the change by two units from the photon to the vm helicities requires in addition spin transfer by the exchanged gluons . the observation of a non - zero value for this amplitude may thus provide important information concerning gluon polarisation in the proton @xcite . the prediction of the ik model for is @xmath680 with a dependence proportional to . the model describes the dependence of the data , but the negative sign of , both for @xmath670 and @xmath671 , is at variance with the model expectation ; this is attributed to the strong approximations involved in the parameterisations @xcite . ( 16.5,4 . ) ( 0.0,0.0 ) ( 4.0,0.0 ) ( 8.0,0.0 ) ( 12.,0.0 ) no significant @xmath6 dependence of the amplitude ratios is observed ( not shown ) , which follows from the absence of a @xmath6 dependence of the matrix elements . the strong dependence of the @xmath132 cross section ratio observed in fig . [ fig : rlt - wtm](c ) is confirmed in the ratio of the dominant schc amplitudes , as seen in fig . [ fig : ampl - ratios - f - m ] , with a similar hint for . as suggested in section [ sect : polar_disc_r - m ] , these features may be related to the @xmath681 dependences in eqs . ( [ eq : ik : alpha ] ) and ( [ eq : ik : beta ] ) . in an extension of the fits performed in the previous sections , the phases between the amplitudes can be left free . to ensure proper convergence , the number of fitted quantities has to be reduced . in view of their small values , the approximation is made to put to @xmath423 the amplitudes @xmath607 and @xmath679 . when the phase difference @xmath682 is left free , it is pushed to the bound @xmath115 ; it is therefore fixed to this value in the fit is close to @xmath115 is at variance with calculations in @xcite , where an attempt was made in a gpd approach to estimate the size of the @xmath606 amplitude within the handbag approach . the prediction in @xcite that the amplitudes should be out of phase depends in fact strongly on a number of assumptions @xcite . ] . the fitted phase difference @xmath683 is found to be systematically lower than 1 , with the amplitude ratios and being compatible with those presented in the previous section . the average value of the phase difference for mesons is @xmath684 which confirms the result of section [ sect : polar_disc_schc ] under the schc approximation , that the dominant longitudinal and transverse amplitudes are nearly but not completely in phase . this paper reports on the measurement of diffractive and meson electroproduction at high energy , both in the elastic and proton dissociative channels . the data were taken in the years 1996 to 2000 with the h1 detector at the @xmath160 collider hera , in the kinematic domain @xmath685 , @xmath4 , @xmath5 and @xmath414 . the total , longitudinal and transverse cross sections are measured as a function of the scaling variable @xmath686 . they roughly follow power laws , and are well described by empirical parameterisations allowing the power to linearly depend on @xmath686 . the to total cross section ratios are found to be independent of @xmath686 and consistent for elastic and proton dissociative scattering , with a value close to but slightly lower than the ratio expected from quark charge counting , @xmath687 . the measurements significantly differ from the formal predictions @xmath523 and @xmath524 for the @xmath527 dependence of the longitudinal and transverse cross sections , respectively , which is attributed mainly to the increase with of the gluon density at small @xmath73 . the cross sections increase with the photon proton centre of mass energy @xmath6 , which is parameterised in the regge inspired form @xmath688 , where @xmath622 increases significantly with . this hardening " of the @xmath6 distribution is described in terms of the intercept @xmath535 of the effective regge trajectory . for values of the scale @xmath537 up to about @xmath539 , the @xmath6 dependence of and production is slightly harder than the soft behaviour characteristic of hadron interactions and photoproduction , @xmath532 to @xmath533 . for the higher range , values of @xmath535 of the order of @xmath540 to @xmath541 are reached , compatible with measurements . dvcs measurements show a similar behaviour as a function of the scale @xmath538 . the @xmath74 dependences of the cross sections are well described as exponentially falling distributions @xmath689 , up to values of @xmath632 for elastic production and @xmath539 for proton dissociation . the @xmath74 slopes are measured for all four channels , providing the first precise determination at hera of the proton dissociative slopes for light vm electroproduction . the values of the @xmath74 slopes are lower than those in photoproduction and they decrease with increasing scale , in a way which is common to light vms and dvcs . values of the @xmath74 slopes comparable to those for production , or slightly larger , are reached for a scale @xmath561 , which suggests that light vm form factors are small and confirms that the dominant longitudinal amplitudes approach a perturbative behaviour for around @xmath424 to @xmath569 . the correlation between the @xmath6 and @xmath74 dependences of the cross sections is parameterised in the form of the slope of the effective pomeron trajectory . for meson production , this slope is smaller than that in soft hadron - hadron interactions , albeit with large errors . the ratio of the proton dissociative to elastic cross sections for @xmath690 and the difference between the elastic and proton dissociative slopes are measured to be independent of . these observations support the relevance of the factorisation of the process into a hard scattering contribution at the photon vertex and a soft diffractive scattering at the proton vertex ( regge factorisation " ) . the value measured for and production for the slope difference , @xmath691 , however , is larger than for production . polarisation effects are studied through the measurement of @xmath692 spin density matrix elements , which are normalised bilinear combinations of the complex helicity amplitudes @xmath693 . the dependence on the kinematic variables and , for mesons , on the dipion mass is measured . the main feature in the present domain is the dominance of the @xmath137-channel helicity conserving ( schc ) amplitudes , @xmath620 and @xmath605 , with @xmath694 . in addition , a significant breaking of schc is manifest through the non - zero value of the matrix element , especially at large values . the ratio @xmath508 of the longitudinal to transverse cross sections increases strongly with , as predicted in pqcd , with a scaling behaviour as a function of @xmath647 for the different vms . the linear dependence @xmath695 predicted at lo , however , is damped for large values of . no @xmath6 dependence of @xmath514 is observed within errors . for @xmath74 , an indication of the dependence of @xmath514 is found for meson production with @xmath671 . this can be interpreted as a difference between the longitudinal and transverse @xmath74 slopes , @xmath652 , which differs from zero by @xmath653 , with dominant systematic errors . a strong dependence of @xmath514 is observed for meson production , both for smaller and larger than @xmath696 . this behaviour may be interpreted as following from the general @xmath659 dependence of vm production , if the mass @xmath660 is understood as the dipion mass rather than the nominal resonance mass . the ratio of the helicity amplitudes is measured from global fits to the 15 matrix elements . several features expected in pqcd are observed for the first time . a decrease with increasing is found for the amplitude ratio , which supports the higher twist nature of the helicity flip amplitudes . the amplitude ratio is observed to decrease with increasing , which may be related to different transverse sizes of transverse and longitudinal dipoles . this is substantiated by the non - zero value of the slope difference @xmath652 obtained from the measurement of @xmath514 from global fits of the helicity amplitudes , with a @xmath672 significance . at large , the amplitude ratio which involves the second single flip amplitude is found to exhibit a dependence . finally , a non - zero value at large is found for the ratio which involves the double flip amplitude , an observation which may provide information on gluon polarisation in the proton . the phase between the @xmath620 and @xmath605 amplitudes is measured to be non - zero , which may suggest different @xmath6 dependences of the transverse and longitudinal amplitudes . the general features of the kinematic dependences of the cross sections and of the spin density matrix elements are understood qualitatively in qcd . in particular , the @xmath6 and @xmath74 dependences indicate that hard " , perturbative qcd features become dominant in the longitudinal cross section in the present kinematic domain , for @xmath697 . the measurements are globally described by models using gpds or a dipole approach , which differ in detail but agree on the gross features . the study of vm production at hera thus provides new insights for the understanding of qcd and the interplay of soft and hard diffraction . we are grateful to the hera machine group whose outstanding efforts have made this experiment possible . we thank the engineers and technicians for their work in constructing and maintaining the h1 detector , our funding agencies for financial support , the desy technical staff for continual assistance , and the desy directorate for the hospitality which they extend to the non - desy members of the collaboration . we thank a. bruni , j .- cudell , m. diehl , s.v . goloskokov , d.yu . ivanov , i. ivanov , r. kirschner , p. kroll , g. soyez , m. strikman , t. teubner and g. watt for useful discussions and for providing us with the predictions of their calculations . [ [ sect : polar : ang_distri ] ] matrix elements + + + + + + + + + + + + + + + in the formalism of @xcite , the spin density matrix elements are normalised sums of products of two helicity amplitudes @xmath698 . they are given in the form @xmath593 , where the notation @xmath699 of the upper index @xmath700 denotes the combination of unpolarised transverse and longitudinal photons and @xmath701 components is only possible through measurements with different polarisation parameters @xmath510 , i.e. with different beam energies in the same detector configuration . in this case , 18 matrix elements in total can be measured . ] , the notations @xmath702 and @xmath703 are used for vm production by transverse photons with orthogonal linear polarisations , and @xmath704 and @xmath705 for the interference between vm production by transverse and longitudinal photons . the lower indices @xmath706 refer to the vm helicities @xmath596 of the pair of amplitudes . in the absence of longitudinal beam polarisation , 15 independent components of the spin density matrix can be measured ( 8 additional matrix elements are accessible with a longitudinally polarised lepton beam ) . they enter in the normalised angular distribution @xmath609 : @xmath707 the matrix elements are measured as projections of the normalised angular distribution , eq . ( [ eq : w ] ) , onto orthogonal functions of the @xmath177 , and angles , with one specific function corresponding to each matrix element ( see appendix c of @xcite ) . in practice , each matrix element is measured as the average value of the corresponding function , taken over all events in the data sample . alternatively , fits to the projections of the angular distribution @xmath609 onto each of the three angles provide measurements of the matrix elements and and of the combinations ( ) and ( ): @xmath708 natural parity exchange ( npe ) in the @xmath74 channel implies the following relations between amplitudes ) implies that , for @xmath709 , the trajectory exchanged in the @xmath74 channel has natural parity . ] : @xmath710 for unnatural parity exchange , an additional factor @xmath711 appears in the right hand side of eq . ( [ eq : npe ] ) . under npe and integrating over the nucleon polarisations , the number of independent @xmath693 amplitudes is reduced from 9 to 5 : two helicity conserving amplitudes ( @xmath604 and @xmath712 ) , two single helicity flip amplitudes ( @xmath713 and @xmath714 ) and one double flip amplitude ( @xmath715 ) . in general , longitudinally polarised lepton beams are required to separate natural and unnatural parity exchange process . however , unpolarised beams allow the measurement of the asymmetry @xmath716 between natural ( @xmath617 ) and unnatural ( @xmath618 ) parity exchange for transverse photons : @xmath717 the measurement of the corresponding asymmetry for longitudinal photons requires different values of @xmath510 , i.e. different beam energies . in the approximation of @xmath137-channel helicity conservation ( schc ) @xcite , the helicity of the virtual photon ( measured in the helicity frame defined in section [ sect : variables ] ) is retained by the final state vm ( with the nucleon helicity also remaining unchanged ) . single and double helicity flip amplitudes thus vanish ( @xmath718 ) and only five matrix elements are non - zero : @xmath719 under schc and npe , the following relations hold between these elements : @xmath720 in the case of schc , only two independent parameters are left , conveniently chosen as the cross section ratio @xmath508 and the phase @xmath622 between the @xmath620 and @xmath605 amplitudes , with @xmath721 the angular distribution @xmath609 then reduces to a function of @xmath177 and @xmath334 , the angle between the electron scattering plane and the meson decay plane , in the frame : @xmath722 the precision of measurements performed in the schc approximation , especially at large , can be improved by retaining the dominant helicity flip amplitude @xmath606 . five additional matrix elements are then non - zero , supplementing the five elements given in eq . ( [ eq : non-0-schc ] ) : @xmath723 under npe , the following relations hold in addition to the schc relations ( [ eq : schc_pairing ] ) : @xmath724 f.d . et al . _ , [ h1 collaboration ] , phys * b659 * ( 2008 ) 796 [ arxiv:0709.4114 ] ; + f.d . et al . _ , [ h1 collaboration ] , desy-09 - 109 , accepted for publication by phys . lett . * b * , arxiv:0907.5289 . j. breitweg _ [ zeus collaboration ] , eur . j. * c14 * ( 2000 ) 213 [ hep - ex/9910038 ] ; + _ idem _ , eur . j. * c2 * ( 1998 ) 247 [ hep - ex/9712020 ] ; + m. derrick _ et al . _ [ zeus collaboration ] , z. phys . * c73 * ( 1997 ) 253 [ hep - ex/9609003 ] ; + _ idem _ , z. phys . * c69 * ( 1995 ) 39 [ hep - ex/9507011 ] . a. airapetian _ [ hermes collaboration ] , eur . j. * c17 * ( 2000 ) 389 ; + k. ackerstaff _ [ hermes collaboration ] , eur . j. * c18 * ( 2000 ) 303 ; + a. airapetian _ [ hermes collaboration ] , eur . j. * c62 * ( 2009 ) 659 [ arxiv:0901.0701 ] . kopeliovich and b.g . zakharov , phys . * d44 * ( 1991 ) , 3466 ; + b.z . kopeliovich , j. nemchik , n.n . nikolaev and b.g . zakharov , phys . * b309 * ( 1993 ) 179 ; + _ idem _ , phys . * b324 * ( 1994 ) 469 . d.yu . ivanov and r. kirschner , phys . rev . * d58 * ( 1998 ) 114026 . i. royen and j .- r . cudell , nucl . b545 * ( 1999 ) 505 ; + i. royen , phys . lett . * b513 * ( 2001 ) 337 . d. schildknecht , g.a . schuler and b. surrow , b449 * ( 1999 ) 328 ; + g. cvetic , d. schildknecht , b. surrow and m. tentyukov , eur . j. * c20 * ( 2001 ) 77 . j.r . forshaw and g. shaw , jhep * 0412 * ( 2004 ) 052 . forshaw , r. sandapen and g. shaw , phys . rev * d69 * ( 2004 ) 094013 ; + _ idem _ , jhep * 0611 * ( 2006 ) 025 . ivanov , a. schaefer , l. szymanowski and g. krasnikov , eur . j. * c34 * ( 2004 ) 297 ; + d. yu . ivanov , l. szymanowski and g. krasnikov , jetp lett . * 80 * ( 2004 ) 226 . ivanov , m.i . kotsky and a. papa , eur . j. * c38 * ( 2004 ) 195 . h. kowalski , l. motyka and g. watt , phys . * d74 * ( 2006 ) 074016 . dosch and e. ferreira , eur . j. * c51 * ( 2007 ) 83 . goloskokov and p. kroll , eur . j. * c53 * ( 2008 ) 367 . c. marquet , r. peschanski and g. soyez , phys * d76 * ( 2007 ) 034011 . k. goulianos , phys . * 101 * ( 1983 ) 169 . t. sjstrand , comput . * 82 * ( 1994 ) 74 . c. collard , phd thesis , universit libre de bruxelles ( ulb ) ( 2002 ) , appendix a , + http://www-h1.desy.de/psfiles/theses/h1th-296.ps . ryskin and yu.m . shabelski , phys . * 62 * ( 1999 ) 980 [ yad . fiz . * 62 * ( 1999 ) 1047 ] . ryskin and yu.m . shabelski , phys . * 61 * ( 1998 ) 81 [ yad . fiz . * 61 * ( 1998 ) 89 ] ; + fig . 6 is corrected in : m.g . ryskin and yu.m . shabelski , hep - ph/9704279 ( 1997 ) . jaroszkiewicz and p.v . landshoff , phys . rev . * d10 * ( 1974 ) 170 ; + a. donnachie and p.v . landshoff , nucl . b231 * ( 1984 ) 189 . [ omega collaboration ] , nucl . b209 * ( 1982 ) 56 . 3.3 & @xmath725 & @xmath726 & @xmath727 & @xmath728 & @xmath729 & @xmath730 + 6.5 & @xmath731 & @xmath732 & @xmath733 & @xmath734 & @xmath735 & @xmath736 + 11.9 & @xmath737 & @xmath738 & @xmath739 & @xmath740 & @xmath741 & @xmath742 + 23.0 & @xmath743 & @xmath744 & @xmath745 & @xmath746 & @xmath747 & @xmath748 + 2.65 & @xmath749 & @xmath750 & @xmath751 + 2.95 & @xmath752 & @xmath753 & @xmath754 + 3.30 & @xmath755 & @xmath756 & @xmath757 + 3.70 & @xmath758 & @xmath759 & @xmath760 + 4.15 & @xmath761 & @xmath762 & @xmath763 + 4.65 & @xmath764 & @xmath765 & @xmath766 + 5.20 & @xmath767 & @xmath768 & @xmath769 + 5.85 & @xmath770 & @xmath768 & @xmath771 + 6.55 & @xmath772 & @xmath773 & @xmath774 + 7.35 & @xmath775 & @xmath776 & @xmath777 + 8.20 & @xmath778 & @xmath779 & @xmath780 + 9.20 & @xmath781 & @xmath782 & @xmath783 + 10.3 & @xmath784 & @xmath782 & @xmath785 + 11.5 & @xmath786 & @xmath787 & @xmath788 + 12.9 & @xmath789 & @xmath790 & @xmath791 + 14.5 & @xmath792 & @xmath793 & @xmath794 + 16.5 & @xmath795 & @xmath796 & @xmath797 + 18.8 & @xmath798 & @xmath799 & @xmath800 + 21.7 & @xmath801 & @xmath802 & @xmath803 + 25.0 & @xmath804 & @xmath805 & @xmath806 + 29.3 & @xmath807 & @xmath808 & @xmath809 + 35.0 & @xmath810 & @xmath811 & @xmath812 + 46.0 & @xmath813 & @xmath814 & @xmath815 + 2.71 & @xmath816 & @xmath817 & @xmath818 + 3.21 & @xmath819 & @xmath820 & @xmath821 + 3.82 & @xmath822 & @xmath823 & @xmath824 + 4.52 & @xmath770 & @xmath825 & @xmath826 + 5.36 & @xmath827 & @xmath828 & @xmath829 + 6.35 & @xmath830 & @xmath831 & @xmath832 + 7.60 & @xmath833 & @xmath834 & @xmath835 + 9.30 & @xmath836 & @xmath837 & @xmath838 + 12.00 & @xmath839 & @xmath840 & @xmath841 + 14.85 & @xmath842 & @xmath843 & @xmath844 + 19.20 & @xmath845 & @xmath846 & @xmath847 + 32.15 & @xmath848 & @xmath849 & @xmath850 + 2.71 & @xmath851 & @xmath852 & @xmath853 + 3.21 & @xmath854 & @xmath828 & @xmath855 + 3.82 & @xmath856 & @xmath857 & @xmath858 + 4.52 & @xmath859 & @xmath860 & @xmath861 + 5.36 & @xmath862 & @xmath863 & @xmath864 + 6.35 & @xmath865 & @xmath866 & @xmath867 + 7.60 & @xmath868 & @xmath869 & @xmath870 + 9.30 & @xmath871 & @xmath872 & @xmath873 + 12.00 & @xmath874 & @xmath875 & @xmath876 + 14.85 & @xmath877 & @xmath878 & @xmath879 + 19.20 & @xmath880 & @xmath735 & @xmath881 + 32.15 & @xmath882 & @xmath883 & @xmath884 + 3.3 & @xmath885 & @xmath886 & @xmath887 + 6.6 & @xmath888 & @xmath889 & @xmath890 + 11.9 & @xmath891 & @xmath892 & @xmath893 + 18.6 & @xmath894 & @xmath895 & @xmath896 + 31.3 & @xmath897 & @xmath744 & @xmath898 + 2.9 & @xmath899 & @xmath900 & @xmath901 & 3.94 & @xmath902 & @xmath903 & @xmath904 + 4.1 & @xmath905 & @xmath906 & @xmath907 & 5.14 & @xmath908 & @xmath909 & @xmath910 + 6.6 & @xmath911 & @xmath912 & @xmath913 & 7.64 & @xmath914 & @xmath915 & @xmath916 + 11.9 & @xmath905 & @xmath917 & @xmath913 & 12.94 & @xmath918 & @xmath919 & @xmath913 + 18.6 & @xmath920 & @xmath921 & @xmath922 & 19.64 & @xmath923 & @xmath924 & @xmath925 + 31.3 & @xmath926 & @xmath927 & @xmath928 & 32.34 & @xmath929 & @xmath726 & @xmath928 + 2.9 & @xmath930 & @xmath931 & @xmath932 & @xmath933 & @xmath934 & @xmath935 + 4.1 & @xmath936 & @xmath937 & @xmath938 & @xmath939 & @xmath940 & @xmath941 + 6.6 & @xmath942 & @xmath943 & @xmath944 & @xmath945 & @xmath946 & @xmath947 + 11.9 & @xmath948 & @xmath949 & @xmath950 & @xmath951 & @xmath952 & @xmath953 + 18.6 & @xmath954 & @xmath955 & @xmath956 & @xmath957 & @xmath840 & @xmath958 + 31.3 & @xmath959 & @xmath960 & @xmath961 & @xmath962 & @xmath963 & @xmath964 + 3.3 & @xmath965 & @xmath966 & @xmath967 & @xmath968 & @xmath969 & @xmath970 + 6.6 & @xmath971 & @xmath972 & @xmath973 & @xmath974 & @xmath975 & @xmath976 + 15.8 & @xmath977 & @xmath978 & @xmath979 & @xmath980 & @xmath981 & @xmath982 + 3.3 & @xmath983 & @xmath984 & @xmath985 & @xmath986 + 3.3 & @xmath987 & @xmath988 & @xmath989 & @xmath824 + 3.3 & @xmath990 & @xmath991 & @xmath992 & @xmath993 + 3.3 & @xmath994 & @xmath995 & @xmath996 & @xmath997 + 3.3 & @xmath998 & @xmath999 & @xmath1000 & @xmath1001 + 6.6 & @xmath1002 & @xmath1003 & @xmath1004 & @xmath1005 + 6.6 & @xmath1006 & @xmath1007 & @xmath1008 & @xmath1009 + 6.6 & @xmath1010 & @xmath1011 & @xmath828 & @xmath1012 + 6.6 & @xmath1013 & @xmath1014 & @xmath1015 & @xmath1016 + 6.6 & @xmath1017 & @xmath1018 & @xmath1019 & @xmath1016 + 11.9 & @xmath1020 & @xmath1021 & @xmath975 & @xmath1022 + 11.9 & @xmath1023 & @xmath1024 & @xmath863 & @xmath1025 + 11.9 & @xmath1026 & @xmath1027 & @xmath1028 & @xmath1029 + 11.9 & @xmath1030 & @xmath1031 & @xmath1032 & @xmath1033 + 11.9 & @xmath1034 & @xmath1035 & @xmath1036 & @xmath1037 + 19.5 & @xmath1038 & @xmath1039 & @xmath949 & @xmath1040 + 19.5 & @xmath1041 & @xmath1042 & @xmath1043 & @xmath950 + 19.5 & @xmath1044 & @xmath1045 & @xmath981 & @xmath1046 + 19.5 & @xmath1047 & @xmath1048 & @xmath1049 & @xmath1050 + 19.5 & @xmath1051 & @xmath1052 & @xmath840 & @xmath1053 + 35.6 & @xmath1054 & @xmath540 & @xmath1055 & @xmath1056 + 35.6 & @xmath1057 & @xmath1058 & @xmath1059 & @xmath1060 + 35.6 & @xmath1061 & @xmath1062 & @xmath972 & @xmath1060 + 35.6 & @xmath1063 & @xmath1064 & @xmath949 & @xmath982 + 35.6 & @xmath1065 & @xmath1066 & @xmath1067 & @xmath982 + 3.3 & @xmath1068 & @xmath1069 & @xmath1070 & @xmath1071 + 3.3 & @xmath1072 & @xmath1073 & @xmath1074 & @xmath932 + 3.3 & @xmath1075 & @xmath1076 & @xmath1077 & @xmath1078 + 3.3 & @xmath1010 & @xmath1079 & @xmath1080 & @xmath1081 + 3.3 & @xmath1082 & @xmath1083 & @xmath1084 & @xmath1085 + 7.5 & @xmath1002 & @xmath1086 & @xmath1087 & @xmath1088 + 7.5 & @xmath1006 & @xmath1089 & @xmath1090 & @xmath1091 + 7.5 & @xmath1010 & @xmath1035 & @xmath1092 & @xmath1093 + 7.5 & @xmath1013 & @xmath1094 & @xmath1004 & @xmath1095 + 7.5 & @xmath1017 & @xmath1096 & @xmath1097 & @xmath1098 + 22.5 & @xmath1054 & @xmath1099 & @xmath1059 & @xmath1100 + 22.5 & @xmath1057 & @xmath1101 & @xmath972 & @xmath1102 + 22.5 & @xmath1061 & @xmath290 & @xmath1067 & @xmath1103 + 22.5 & @xmath1063 & @xmath1104 & @xmath1105 & @xmath1106 + 22.5 & @xmath1065 & @xmath1107 & @xmath981 & @xmath1108 + 3.3 & @xmath1068 & @xmath1109 & @xmath1004 & @xmath1110 + 3.3 & @xmath1072 & @xmath1111 & @xmath831 & @xmath1112 + 3.3 & @xmath1075 & @xmath1113 & @xmath946 & @xmath1114 + 3.3 & @xmath1010 & @xmath1115 & @xmath1116 & @xmath1117 + 3.3 & @xmath1082 & @xmath1118 & @xmath1119 & @xmath1120 + 6.6 & @xmath1002 & @xmath1121 & @xmath869 & @xmath1122 + 6.6 & @xmath1006 & @xmath1123 & @xmath1124 & @xmath1125 + 6.6 & @xmath1010 & @xmath1126 & @xmath837 & @xmath1127 + 6.6 & @xmath1013 & @xmath1128 & @xmath966 & @xmath1129 + 6.6 & @xmath1017 & @xmath1130 & @xmath860 & @xmath1131 + 15.8 & @xmath1054 & @xmath1132 & @xmath1133 & @xmath1060 + 15.8 & @xmath1057 & @xmath1134 & @xmath1059 & @xmath1135 + 15.8 & @xmath1061 & @xmath1136 & @xmath949 & @xmath982 + 15.8 & @xmath1063 & @xmath1137 & @xmath1105 & @xmath1138 + 15.8 & @xmath1065 & @xmath1139 & @xmath1036 & @xmath1040 + 5.0 & @xmath1140 & @xmath1045 & @xmath943 & @xmath1141 + 5.0 & @xmath1142 & @xmath1143 & @xmath837 & @xmath1144 + 5.0 & @xmath1145 & @xmath1146 & @xmath1147 & @xmath1148 + 5.0 & @xmath1149 & @xmath1150 & @xmath1151 & @xmath1152 + 5.0 & @xmath1153 & @xmath1154 & @xmath1155 & @xmath1156 + 5.0 & @xmath236 & @xmath1157 & @xmath1158 & @xmath1159 + + 3.3 & @xmath1160 & @xmath1161 & @xmath1162 & @xmath1163 & @xmath1164 & @xmath1165 + 6.6 & @xmath1166 & @xmath978 & @xmath1167 & @xmath1168 & @xmath1164 & @xmath1169 + 11.9 & @xmath1170 & @xmath1171 & @xmath961 & @xmath1172 & @xmath1173 & @xmath1174 + 19.5 & @xmath1175 & @xmath1171 & @xmath961 & @xmath1176 & @xmath1173 & @xmath1174 + 35.6 & @xmath1177 & @xmath1178 & @xmath1179 & @xmath1180 & @xmath1181 & @xmath1174 + + 3.3 & @xmath1182 & @xmath1183 & @xmath1184 & @xmath1185 & @xmath1173 & @xmath1165 + 6.6 & @xmath1186 & @xmath729 & @xmath1187 & @xmath1185 & @xmath1188 & @xmath1165 + 15.8 & @xmath1189 & @xmath1190 & @xmath1191 & @xmath740 & @xmath1161 & @xmath1165 + + 3.3 & @xmath1192 & @xmath1183 & @xmath1193 & @xmath1177 & @xmath1173 & @xmath1179 + 7.5 & @xmath1194 & @xmath1195 & @xmath1196 & @xmath343 & @xmath1173 & @xmath730 + 22.5 & @xmath1197 & @xmath1198 & @xmath1199 & @xmath1200 & @xmath1181 & @xmath730 + + 5.0 & @xmath1201 & @xmath735 & @xmath1202 & @xmath469 & @xmath960 & @xmath1187 + 3.3 & @xmath1203 & @xmath1204 & @xmath1205 & @xmath1206 + 3.3 & @xmath1207 & @xmath1208 & @xmath1209 & @xmath1210 + 3.3 & @xmath168 & @xmath1211 & @xmath1212 & @xmath1213 + 3.3 & @xmath1214 & @xmath1215 & @xmath1216 & @xmath1217 + 3.3 & @xmath1218 & @xmath1219 & @xmath1220 & @xmath1221 + 3.3 & @xmath1222 & @xmath1223 & @xmath1224 & @xmath1225 + 3.3 & @xmath1226 & @xmath1227 & @xmath1228 & @xmath1229 + 6.6 & @xmath1203 & @xmath1230 & @xmath1231 & @xmath1232 + 6.6 & @xmath1207 & @xmath1233 & @xmath753 & @xmath1234 + 6.6 & @xmath168 & @xmath1235 & @xmath1236 & @xmath1237 + 6.6 & @xmath1214 & @xmath1238 & @xmath762 & @xmath1239 + 6.6 & @xmath1218 & @xmath1240 & @xmath1241 & @xmath1242 + 6.6 & @xmath1222 & @xmath1243 & @xmath1244 & @xmath1245 + 6.6 & @xmath1226 & @xmath1246 & @xmath1247 & @xmath1248 + 11.5 & @xmath1203 & @xmath1249 & @xmath1236 & @xmath1250 + 11.5 & @xmath1207 & @xmath1251 & @xmath1252 & @xmath1253 + 11.5 & @xmath168 & @xmath1254 & @xmath1255 & @xmath1256 + 11.5 & @xmath1214 & @xmath1257 & @xmath1258 & @xmath1259 + 11.5 & @xmath1218 & @xmath1260 & @xmath1261 & @xmath1262 + 11.5 & @xmath1222 & @xmath1263 & @xmath1264 & @xmath1265 + 11.5 & @xmath1226 & @xmath1266 & @xmath1267 & @xmath1268 + 17.4 & @xmath1203 & @xmath1269 & @xmath1270 & @xmath1271 + 17.4 & @xmath1207 & @xmath1272 & @xmath1273 & @xmath1274 + 17.4 & @xmath168 & @xmath1275 & @xmath1276 & @xmath1277 + 17.4 & @xmath1214 & @xmath1278 & @xmath1279 & @xmath1280 + 17.4 & @xmath1218 & @xmath1281 & @xmath1282 & @xmath1283 + 17.4 & @xmath1222 & @xmath1284 & @xmath1285 & @xmath1286 + 17.4 & @xmath1226 & @xmath1287 & @xmath1288 & @xmath1289 + 33.0 & @xmath1203 & @xmath1290 & @xmath1291 & @xmath1292 + 33.0 & @xmath1207 & @xmath1293 & @xmath1294 & @xmath1295 + 33.0 & @xmath168 & @xmath1296 & @xmath1297 & @xmath1298 + 33.0 & @xmath1214 & @xmath1299 & @xmath1300 & @xmath1301 + 33.0 & @xmath1218 & @xmath1302 & @xmath1303 & @xmath1304 + 33.0 & @xmath1222 & @xmath1305 & @xmath1306 & @xmath1307 + 33.0 & @xmath1226 & @xmath1308 & @xmath1309 & @xmath1310 + 3.3 & @xmath1311 & @xmath1312 & @xmath1313 & @xmath1314 + 3.3 & @xmath1315 & @xmath1235 & @xmath1316 & @xmath1317 + 3.3 & @xmath1318 & @xmath1319 & @xmath1320 & @xmath1321 + 3.3 & @xmath1322 & @xmath1323 & @xmath1324 & @xmath1325 + 3.3 & @xmath1326 & @xmath1327 & @xmath1328 & @xmath1329 + 3.3 & @xmath1330 & @xmath1331 & @xmath1332 & @xmath1333 + 3.3 & @xmath1334 & @xmath1335 & @xmath1336 & @xmath1337 + 3.3 & @xmath1338 & @xmath1339 & @xmath1340 & @xmath1341 + 3.3 & @xmath1342 & @xmath1343 & @xmath1344 & @xmath1345 + 3.3 & @xmath1346 & @xmath1347 & @xmath1348 & @xmath1349 + 3.3 & @xmath1350 & @xmath1351 & @xmath1352 & @xmath1353 + 6.6 & @xmath1311 & @xmath1354 & @xmath1355 & @xmath1356 + 6.6 & @xmath1315 & @xmath1357 & @xmath1355 & @xmath1358 + 6.6 & @xmath1318 & @xmath1359 & @xmath1273 & @xmath1360 + 6.6 & @xmath1322 & @xmath1361 & @xmath1362 & @xmath1363 + 6.6 & @xmath1326 & @xmath1364 & @xmath1365 & @xmath1366 + 6.6 & @xmath1330 & @xmath1367 & @xmath1368 & @xmath1369 + 6.6 & @xmath1334 & @xmath1370 & @xmath1371 & @xmath1372 + 6.6 & @xmath1338 & @xmath1373 & @xmath1374 & @xmath1375 + 6.6 & @xmath1342 & @xmath1376 & @xmath1377 & @xmath1378 + 6.6 & @xmath1346 & @xmath1379 & @xmath1380 & @xmath1381 + 6.6 & @xmath1350 & @xmath1382 & @xmath1383 & @xmath1384 + 15.8 & @xmath1311 & @xmath1385 & @xmath13863 & @xmath1387 + 15.8 & @xmath1315 & @xmath1388 & @xmath13894 & @xmath1390 + 15.8 & @xmath1318 & @xmath1391 & @xmath13928 & @xmath1393 + 15.8 & @xmath1322 & @xmath1394 & @xmath13955 & @xmath1396 + 15.8 & @xmath1326 & @xmath1397 & @xmath13987 & @xmath1399 + 15.8 & @xmath1330 & @xmath1400 & @xmath14013 & @xmath1402 + 15.8 & @xmath1334 & @xmath1403 & @xmath14046 & @xmath1405 + 15.8 & @xmath1338 & @xmath1406 & @xmath14073 & @xmath1408 + 15.8 & @xmath1342 & @xmath1409 & @xmath14108 & @xmath1411 + 15.8 & @xmath1346 & @xmath1412 & @xmath14139 & @xmath1414 + 15.8 & @xmath1350 & @xmath1415 & @xmath14165 & @xmath1417 + 3.3 & @xmath1203 & @xmath1418 & @xmath1419 & @xmath1420 + 3.3 & @xmath1207 & @xmath1421 & @xmath753 & @xmath1422 + 3.3 & @xmath168 & @xmath1423 & @xmath1424 & @xmath1425 + 3.3 & @xmath1214 & @xmath1426 & @xmath1427 & @xmath1428 + 3.3 & @xmath1218 & @xmath1429 & @xmath1430 & @xmath1431 + 3.3 & @xmath1222 & @xmath1031 & @xmath1432 & @xmath1433 + 3.3 & @xmath1226 & @xmath1434 & @xmath1435 & @xmath1436 + 6.6 & @xmath1203 & @xmath1437 & @xmath1438 & @xmath1439 + 6.6 & @xmath1207 & @xmath1440 & @xmath1441 & @xmath1442 + 6.6 & @xmath168 & @xmath1443 & @xmath1444 & @xmath1445 + 6.6 & @xmath1214 & @xmath1446 & @xmath1447 & @xmath1448 + 6.6 & @xmath1218 & @xmath1449 & @xmath1270 & @xmath1450 + 6.6 & @xmath1222 & @xmath1451 & @xmath1452 & @xmath1453 + 6.6 & @xmath1226 & @xmath1454 & @xmath1455 & @xmath1456 + 15.8 & @xmath1203 & @xmath1457 & @xmath1458 & @xmath1459 + 15.8 & @xmath1207 & @xmath1460 & @xmath1282 & @xmath1461 + 15.8 & @xmath168 & @xmath1462 & @xmath1463 & @xmath1464 + 15.8 & @xmath1214 & @xmath1465 & @xmath1466 & @xmath1467 + 15.8 & @xmath1218 & @xmath1394 & @xmath1468 & @xmath1469 + 15.8 & @xmath1222 & @xmath1470 & @xmath1471 & @xmath1472 + 15.8 & @xmath1226 & @xmath1473 & @xmath1383 & @xmath1474 + 5.0 & @xmath1475 & @xmath1476 & @xmath1477 & @xmath1478 + 5.0 & @xmath1318 & @xmath1479 & @xmath1480 & @xmath1481 + 5.0 & @xmath1330 & @xmath1482 & @xmath1483 & @xmath1484 + 5.0 & @xmath1346 & @xmath1485 & @xmath1486 & @xmath1487 + + 3.3 & @xmath1488 & @xmath1489 & @xmath1490 + 6.6 & @xmath1491 & @xmath1492 & @xmath1493 + 11.5 & @xmath1494 & @xmath1495 & @xmath1496 + 17.4 & @xmath1497 & @xmath1498 & @xmath1499 + 33.0 & @xmath1500 & @xmath1501 & @xmath1502 + + 3.3 & @xmath1503 & @xmath1504 & @xmath1505 + 6.6 & @xmath1506 & @xmath1507 & @xmath1508 + 15.8 & @xmath1509 & @xmath1510 & @xmath1511 + + 3.3 & @xmath1512 & @xmath849 & @xmath1513 + 6.6 & @xmath1514 & @xmath1178 & @xmath1515 + 15.8 & @xmath1516 & @xmath1171 & @xmath1517 + + 5.0 & 2.21 & @xmath264 0.37 & @xmath1518 + + 3.3 & @xmath1519 & @xmath1520 & @xmath1174 + 6.6 & @xmath1521 & @xmath1164 & @xmath1174 + 11.5 & @xmath1522 & @xmath1523 & @xmath1174 + 17.4 & @xmath1524 & @xmath1523 & @xmath1165 + 33.0 & @xmath1525 & @xmath960 & @xmath1165 + + 3.3 & @xmath1166 & @xmath1523 & @xmath1165 + 6.6 & @xmath1526 & @xmath1173 & @xmath1165 + 15.8 & @xmath1527 & @xmath1173 & @xmath1165 + + 0.08 & @xmath1528 & @xmath1529 & @xmath1530 & @xmath1531 + 0.08 & @xmath1532 & @xmath1533 & @xmath1534 & @xmath1535 + 0.08 & @xmath1536 & @xmath1537 & @xmath1538 & @xmath1539 + 0.32 & @xmath1540 & @xmath1541 & @xmath1542 & @xmath1543 + 0.32 & @xmath1544 & @xmath1545 & @xmath1546 & @xmath1547 + 0.32 & @xmath1548 & @xmath1549 & @xmath1084 & @xmath1550 + 0.69 & @xmath1528 & @xmath1551 & @xmath1552 & @xmath1553 + 0.69 & @xmath1532 & @xmath1554 & @xmath1555 & @xmath1556 + 0.69 & @xmath1536 & @xmath1557 & @xmath1558 & @xmath1559 + 1.45 & @xmath1528 & @xmath1560 & @xmath1147 & @xmath1561 + 1.45 & @xmath1532 & @xmath1562 & @xmath840 & @xmath1563 + 1.45 & @xmath1536 & @xmath1564 & @xmath1565 & @xmath1566 + + 0.08 & @xmath1567 & @xmath1568 & @xmath985 & @xmath1569 + 0.08 & @xmath1570 & @xmath1568 & @xmath1571 & @xmath1572 + 0.08 & @xmath1573 & @xmath173 & @xmath985 & @xmath1574 + 0.32 & @xmath1567 & @xmath1575 & @xmath1008 & @xmath1576 + 0.32 & @xmath1570 & @xmath1577 & @xmath1578 & @xmath1579 + 0.32 & @xmath1573 & @xmath1580 & @xmath1581 & @xmath1582 + 0.69 & @xmath1567 & @xmath1583 & @xmath1124 & @xmath1584 + 0.69 & @xmath1570 & @xmath1585 & @xmath943 & @xmath1586 + 0.69 & @xmath1573 & @xmath1587 & @xmath1124 & @xmath1588 + 1.47 & @xmath1567 & @xmath1589 & @xmath972 & @xmath1590 + 1.47 & @xmath1570 & @xmath1591 & @xmath1133 & @xmath1592 + 1.47 & @xmath1573 & @xmath1132 & @xmath1059 & @xmath1593 + + -0.08 & @xmath343 & @xmath1594 & @xmath1174 + -0.32 & @xmath1595 & @xmath1596 & @xmath1165 + -0.69 & @xmath1163 & @xmath1597 & @xmath1598 + -1.45 & @xmath1599 & @xmath1600 & @xmath1601 + + -0.08 & @xmath343 & @xmath1596 & @xmath1174 + -0.32 & @xmath343 & @xmath1597 & @xmath1174 + -0.69 & @xmath1602 & @xmath1603 & @xmath1179 + -1.48 & @xmath1604 & @xmath1605 & @xmath1606 + 2.9 & @xmath1197 & @xmath1173 & @xmath1607 & @xmath1608 & @xmath909 & @xmath1609 + 4.1 & @xmath1522 & @xmath1188 & @xmath979 & @xmath1610 & @xmath919 & @xmath1611 + 6.6 & @xmath1182 & @xmath1523 & @xmath1612 & @xmath1613 & @xmath1614 & @xmath1615 + 11.9 & @xmath1197 & @xmath1188 & @xmath742 & @xmath1616 & @xmath1617 & @xmath1618 + 18.6 & @xmath1619 & @xmath1188 & @xmath1620 & @xmath1621 & @xmath1614 & @xmath1622 + 31.3 & @xmath1166 & @xmath1623 & @xmath1624 & @xmath1625 & @xmath1626 & @xmath1627 + 3.3 & @xmath1197 & @xmath1628 & @xmath1629 & @xmath1630 & @xmath1631 & @xmath1632 + 6.6 & @xmath1633 & @xmath1597 & @xmath1634 & @xmath1635 & @xmath1636 & @xmath1637 + 15.8 & @xmath1638 & @xmath1413 & @xmath1639 & @xmath1640 & @xmath1641 & @xmath1637 + @xmath1653 ( ) & & & & & + & @xmath1654 @xmath1614 @xmath1655 & @xmath1656 @xmath909 @xmath1657 & @xmath1658 @xmath1659 @xmath1660 & @xmath1661 @xmath1662 @xmath1663 & @xmath1664 @xmath1665 @xmath1666 + re & @xmath1667 @xmath1668 @xmath1669 & @xmath1670 @xmath900 @xmath1671 & @xmath1672 @xmath1668 @xmath1673 & @xmath1674 @xmath1659 @xmath1675 & @xmath1676 @xmath1636 @xmath1677 + & @xmath1678 @xmath919 @xmath1679 & @xmath1680 @xmath915 @xmath1681 & @xmath1682 @xmath919 @xmath1683 & @xmath1684 @xmath1685 @xmath1686 & @xmath1687 @xmath1688 @xmath1689 + & @xmath1690 @xmath1691 @xmath1692 & @xmath1693 @xmath1694 @xmath1695 & @xmath1696 @xmath1694 @xmath1697 & @xmath1698 @xmath1699 @xmath1700 & @xmath1701 @xmath1702 @xmath1703 + & @xmath1704 @xmath1705 @xmath1706 & @xmath1690 @xmath1707 @xmath1708 & @xmath1709 @xmath1688 @xmath1710 & @xmath1711 @xmath1712 @xmath1713 & @xmath1714 @xmath1715 @xmath1716 + re & @xmath1717 @xmath1659 @xmath1718 & @xmath1719 @xmath909 @xmath1720 & @xmath1721 @xmath1617 @xmath1722 & @xmath1723 @xmath1724 @xmath1689 & @xmath1725 @xmath1726 @xmath1727 + & @xmath1728 @xmath1729 @xmath1730 & @xmath1731 @xmath919 @xmath1683 & @xmath1732 @xmath1733 @xmath1734 & @xmath1735 @xmath1736 @xmath1737 & @xmath1738 @xmath1739 @xmath1740 + & @xmath1741 @xmath1617 @xmath1742 & @xmath1743 @xmath1668 @xmath1675 & @xmath1744 @xmath1659 @xmath1745 & @xmath1746 @xmath1724 @xmath1747 & @xmath1748 @xmath927 @xmath1749 + i m & @xmath1750 @xmath924 @xmath1751 & @xmath1752 @xmath909 @xmath1753 & @xmath1754 @xmath924 @xmath1755 & @xmath1756 @xmath1688 @xmath1757 & @xmath1758 @xmath1759 @xmath1760 + & @xmath1761 @xmath1724 @xmath1762 & @xmath1763 @xmath1764 @xmath1765 & @xmath1766 @xmath1724 @xmath1767 & @xmath1768 @xmath726 @xmath1769 & @xmath1770 @xmath1771 @xmath1772 + & @xmath1773 @xmath921 @xmath1774 & @xmath1775 @xmath909 @xmath1776 & @xmath1777 @xmath921 @xmath1778 & @xmath1779 @xmath1662 @xmath1780 & @xmath1781 @xmath1726 @xmath1782 + re & @xmath1783 @xmath912 @xmath1784 & @xmath1785 @xmath1786 @xmath1787 & @xmath1788 @xmath912 @xmath1789 & @xmath1790 @xmath909 @xmath1791 & @xmath1696 @xmath921 @xmath1792 + & @xmath1793 @xmath1668 @xmath1794 & @xmath1795 @xmath1796 @xmath1681 & @xmath1687 @xmath906 @xmath1797 & @xmath1798 @xmath921 @xmath1755 & @xmath1773 @xmath1685 @xmath1799 + & @xmath1800 @xmath1801 @xmath1802 & @xmath1803 @xmath1786 @xmath1804 & @xmath1805 @xmath1801 @xmath1806 & @xmath1803 @xmath1668 @xmath1807 & @xmath1808 @xmath1614 @xmath1809 + & @xmath1779 @xmath915 @xmath1810 & @xmath1811 @xmath1796 @xmath1812 & @xmath1813 @xmath915 @xmath1812 & @xmath1773 @xmath1814 @xmath1787 & @xmath1680 @xmath1815 @xmath1816 + @xmath1653 ( ) & & & & & + & @xmath1817 @xmath900 @xmath1818 & @xmath1819 @xmath900 @xmath1820 & @xmath1821 @xmath1668 @xmath1822 & @xmath1823 @xmath924 @xmath1824 & @xmath1825 @xmath1707 @xmath1826 + re & @xmath1678 @xmath915 @xmath1827 & @xmath1828 @xmath1796 @xmath1730 & @xmath1670 @xmath900 @xmath1802 & @xmath1829 @xmath917 @xmath1830 & @xmath1831 @xmath1764 @xmath1832 + & @xmath1833 @xmath906 @xmath1834 & @xmath1773 @xmath1796 @xmath1835 & @xmath1672 @xmath915 @xmath1812 & @xmath1836 @xmath1617 @xmath1837 & @xmath1838 @xmath1815 @xmath1839 + & @xmath1840 @xmath1841 @xmath1842 & @xmath1843 @xmath921 @xmath1844 & @xmath1845 @xmath1662 @xmath1846 & @xmath1847 @xmath1848 @xmath1849 & @xmath1850 @xmath1851 @xmath1852 + & @xmath1853 @xmath1854 @xmath1855 & @xmath1813 @xmath1617 @xmath1856 & @xmath1857 @xmath1729 @xmath1858 & @xmath1859 @xmath1860 @xmath1861 & @xmath1798 @xmath1862 @xmath1863 + re & @xmath1684 @xmath1814 @xmath1864 & @xmath1865 @xmath900 @xmath1837 & @xmath1781 @xmath1668 @xmath1837 & @xmath1866 @xmath921 @xmath1867 & @xmath1868 @xmath1707 @xmath1869 + & @xmath1870 @xmath1814 @xmath1871 & @xmath1872 @xmath915 @xmath1873 & @xmath1874 @xmath917 @xmath1875 & @xmath1876 @xmath1685 @xmath1877 & @xmath1878 @xmath1631 @xmath1879 + & @xmath1880 @xmath909 @xmath1881 & @xmath1882 @xmath900 @xmath1837 & @xmath1883 @xmath909 @xmath1784 & @xmath1741 @xmath921 @xmath1884 & @xmath1885 @xmath1685 @xmath1886 + i m & @xmath1882 @xmath917 @xmath1887 & @xmath1777 @xmath900 @xmath1888 & @xmath1889 @xmath917 @xmath1890 & @xmath1891 @xmath1892 @xmath1893 & @xmath1894 @xmath1631 @xmath1895 + & @xmath1896 @xmath1897 @xmath1898 & @xmath1899 @xmath919 @xmath1900 & @xmath1901 @xmath1733 @xmath1902 & @xmath1903 @xmath1631 @xmath1904 & @xmath1905 @xmath927 @xmath1906 + & @xmath1678 @xmath1814 @xmath1907 & @xmath1775 @xmath915 @xmath1673 & @xmath1682 @xmath917 @xmath1908 & @xmath1909 @xmath1897 @xmath1910 & @xmath1911 @xmath1707 @xmath1912 + re & @xmath1913 @xmath1801 @xmath1871 & @xmath1914 @xmath1915 @xmath1888 & @xmath1763 @xmath1916 @xmath1917 & @xmath1738 @xmath906 @xmath1918 & @xmath1919 @xmath903 @xmath1920 + & @xmath1773 @xmath912 @xmath1921 & @xmath1922 @xmath1786 @xmath1923 & @xmath1924 @xmath1796 @xmath1923 & @xmath1925 @xmath917 @xmath1926 & @xmath1927 @xmath1659 @xmath1928 + & @xmath1929 @xmath1916 @xmath1742 & @xmath1930 @xmath1915 @xmath1931 & @xmath1932 @xmath1916 @xmath1933 & @xmath1934 @xmath912 @xmath1935 & @xmath1936 @xmath906 @xmath1742 + & @xmath1798 @xmath1796 @xmath1923 & @xmath1682 @xmath1915 @xmath1835 & @xmath1937 @xmath1916 @xmath1812 & @xmath1938 @xmath915 @xmath1939 & @xmath1940 @xmath1814 @xmath1941 + 2.9 & @xmath1942 & @xmath1943 & @xmath1944 & @xmath1748 & @xmath1945 & @xmath1946 + 4.1 & @xmath1947 & @xmath1948 & @xmath1949 & @xmath1950 & @xmath1951 & @xmath1952 + 6.6 & @xmath1953 & @xmath1954 & @xmath1955 & @xmath1956 & @xmath1957 & @xmath1958 + 11.9 & @xmath1959 & @xmath1960 & @xmath1961 & @xmath1962 & @xmath1963 & @xmath1964 + 18.5 & @xmath1965 & @xmath1966 & @xmath1967 & @xmath1968 & @xmath1969 & @xmath1970 + 31.3 & @xmath1942 & @xmath1971 & @xmath1972 & @xmath1973 & @xmath1969 & @xmath1974 + & & + 0.03 & @xmath1975 & @xmath1976 & @xmath1977 & @xmath1687 & @xmath1963 & @xmath1978 + 0.13 & @xmath1979 & @xmath1976 & @xmath1980 & @xmath1878 & @xmath1981 & @xmath1982 + 0.31 & @xmath1983 & @xmath1971 & @xmath1984 & @xmath1985 & @xmath1986 & @xmath1987 + 0.70 & @xmath1988 & @xmath1989 & @xmath1990 & @xmath1991 & @xmath1992 & @xmath1993 + 1.44 & @xmath1994 & @xmath1995 & @xmath1996 & @xmath1997 & @xmath1998 & @xmath1999 + + 3.1 & @xmath1185 & @xmath1181 & @xmath2000 & 0.03 & @xmath1163 & @xmath960 & @xmath2001 + 4.8 & @xmath2002 & @xmath1623 & @xmath2003 & 0.13 & @xmath1602 & @xmath849 & @xmath2004 + 7.5 & @xmath2005 & @xmath978 & @xmath2006 & 0.31 & @xmath2005 & @xmath849 & @xmath2007 + 12.2 & @xmath2008 & @xmath2009 & @xmath961 & 0.70 & @xmath2010 & @xmath741 & @xmath2011 + 23.1 & @xmath1172 & @xmath2012 & @xmath2006 & 1.44 & @xmath2013 & @xmath849 & @xmath2014 + + 3.3 & @xmath2015 & @xmath2016 & @xmath1620 & 0.08 & @xmath2017 & @xmath1183 & @xmath1165 + 6.6 & @xmath1185 & @xmath2009 & @xmath1620 & 0.34 & @xmath2018 & @xmath1198 & @xmath2006 + 15.8 & @xmath2019 & @xmath978 & @xmath2006 & 1.05 & @xmath1604 & @xmath2020 & @xmath2021 + + 3.1 & @xmath2034 & @xmath2035 & @xmath2036 + 4.8 & @xmath2037 & @xmath2038 & @xmath2039 + 7.5 & @xmath2040 & @xmath2041 & @xmath2042 + 12.2 & @xmath2043 & @xmath2044 & @xmath2045 + 23.1 & @xmath2046 & @xmath2047 & @xmath2048 + + 3.3 & @xmath2049 & @xmath2050 & @xmath2051 + 6.6 & @xmath2052 & @xmath2053 & @xmath2054 + 15.8 & @xmath2055 & @xmath2056 & @xmath2057 + + @xmath2058 & @xmath2059 & @xmath2060 & @xmath2061 + @xmath2062 & @xmath2063 & @xmath2050 & @xmath2064 + @xmath2065 & @xmath2066 & @xmath2038 & @xmath2067 + @xmath2068 & @xmath740 & @xmath2069 & @xmath2070 + + @xmath2071 & @xmath2072 & @xmath1399 & @xmath2073 + @xmath2074 & @xmath2075 & @xmath2076 & @xmath2077 + @xmath1082 & @xmath2078 & @xmath2079 & @xmath2080 + @xmath2081 & @xmath2082 & @xmath2083 & @xmath2084 + + @xmath2085 & @xmath2086 & @xmath2087 & @xmath2088 + @xmath1057 & @xmath2089 & @xmath2090 & @xmath2091 + @xmath2092 & @xmath2093 & @xmath2094 & @xmath2095 + @xmath2096 & @xmath2055 & @xmath2097 & @xmath2098 + + @xmath2099 & @xmath2100 & @xmath2051 & @xmath2035 + @xmath977 & @xmath2101 & @xmath2102 & @xmath2103 + @xmath1595 & @xmath2104 & @xmath2105 & @xmath2106 + + @xmath2099 & @xmath2107 & @xmath2038 & @xmath2108 + @xmath977 & @xmath2109 & @xmath2110 & @xmath2111 + @xmath1595 & @xmath2112 & @xmath2113 & @xmath2114 + + @xmath1522 & @xmath2115 & @xmath2116 & @xmath2117 + @xmath2118 & @xmath2119 & @xmath2102 & @xmath2036 + @xmath2120 & @xmath2120 & @xmath2121 & @xmath2122 + @xmath2008 & @xmath2123 & @xmath2121 & @xmath2124 + @xmath1595 & @xmath1166 & @xmath2125 & @xmath2126 + + @xmath1522 & @xmath2127 & @xmath1378 & @xmath2128 + @xmath2118 & @xmath2129 & @xmath2054 & @xmath2130 + @xmath2120 & @xmath2082 & @xmath2131 & @xmath2132 + @xmath2008 & @xmath2133 & @xmath2069 & @xmath2134 + @xmath1595 & @xmath2017 & @xmath2135 & @xmath2136 + + @xmath2137 & @xmath2138 & @xmath892 & @xmath2139 + @xmath2140 & @xmath2141 & @xmath1498 & @xmath2142 + @xmath1172 & @xmath2143 & @xmath892 & @xmath2144 + + @xmath2137 & @xmath2145 & @xmath1648 & @xmath2146 + @xmath2140 & @xmath2147 & @xmath2148 & @xmath2149 + @xmath1172 & @xmath2150 & @xmath846 & @xmath2151 + + 3.1 & @xmath2152 @xmath919 @xmath2153 & @xmath2154 @xmath919 @xmath2155 & @xmath1678 @xmath912 @xmath2156 & @xmath1853 @xmath906 @xmath1810 & @xmath2157 @xmath1764 @xmath1780 + 4.8 & @xmath2158 @xmath1614 @xmath2159 & @xmath2160 @xmath1617 @xmath2161 & @xmath2162 @xmath906 @xmath1791 & @xmath2163 @xmath909 @xmath1810 & @xmath2164 @xmath1707 @xmath2165 + 7.5 & @xmath2166 @xmath903 @xmath2167 & @xmath2168 @xmath917 @xmath2169 & @xmath1793 @xmath1801 @xmath2170 & @xmath1885 @xmath906 @xmath1888 & @xmath2171 @xmath1641 @xmath2172 + 12.2 & @xmath2173 @xmath917 @xmath2174 & @xmath1896 @xmath1617 @xmath2175 & @xmath2176 @xmath915 @xmath1806 & @xmath2177 @xmath909 @xmath1742 & @xmath2178 @xmath726 @xmath2179 + 23.1 & @xmath2180 @xmath1814 @xmath2170 & @xmath2181 @xmath921 @xmath2182 & @xmath1847 @xmath1668 @xmath2183 & @xmath2184 @xmath1814 @xmath2185 & @xmath2186 @xmath2187 @xmath2188 + + 3.4 & @xmath2189 @xmath2190 @xmath2191 & @xmath2192 @xmath1815 @xmath2193 & @xmath1781 @xmath1814 @xmath2194 & @xmath2195 @xmath924 @xmath1755 & @xmath2196 @xmath2197 @xmath2198 + 6.6 & @xmath2199 @xmath1897 @xmath1689 & @xmath2200 @xmath1707 @xmath2201 & @xmath1924 @xmath1814 @xmath2202 & @xmath2203 @xmath1733 @xmath2204 & @xmath2205 @xmath2206 @xmath2207 + 15.7 & @xmath2208 @xmath1707 @xmath2209 & @xmath2210 @xmath2211 @xmath2212 & @xmath1865 @xmath924 @xmath2213 & @xmath2214 @xmath924 @xmath2215 & @xmath2216 @xmath2217 @xmath2218 + + + @xmath2219 & @xmath2220 @xmath1814 @xmath2221 & @xmath1914 @xmath1814 @xmath2222 & @xmath2223 @xmath900 @xmath2224 & @xmath1857 @xmath903 @xmath1787 & @xmath2225 @xmath1733 @xmath2226 + @xmath2227 & @xmath2228 @xmath1815 @xmath2229 & @xmath2230 @xmath1636 @xmath2231 & @xmath1711 @xmath919 @xmath1837 & @xmath1991 @xmath1733 @xmath2232 & @xmath2233 @xmath2234 @xmath2235 + @xmath2236 & @xmath2237 @xmath2238 @xmath2239 & @xmath2240 @xmath1694 @xmath2241 & @xmath2242 @xmath2243 @xmath2244 & @xmath2245 @xmath1726 @xmath2246 & @xmath2247 @xmath2187 @xmath2248 + + @xmath2219 & @xmath1658 @xmath915 @xmath2249 & @xmath2250 @xmath906 @xmath2251 & @xmath1940 @xmath1796 @xmath1921 & @xmath1813 @xmath912 @xmath1835 & @xmath2252 @xmath1733 @xmath2253 + @xmath2227 & @xmath2254 @xmath1668 @xmath2255 & @xmath2256 @xmath1814 @xmath2257 & @xmath2223 @xmath900 @xmath1926 & @xmath2258 @xmath1668 @xmath2259 & @xmath2260 @xmath1626 @xmath2261 + @xmath2262 & @xmath2263 @xmath1614 @xmath2264 & @xmath1738 @xmath2190 @xmath2265 & @xmath2266 @xmath919 @xmath2267 & @xmath2268 @xmath1659 @xmath2269 & @xmath2270 @xmath883 @xmath2271 + + @xmath2219 & @xmath2272 @xmath1764 @xmath2273 & @xmath1674 @xmath921 @xmath2274 & @xmath2223 @xmath1668 @xmath1834 & @xmath1704 @xmath1614 @xmath1679 & @xmath2275 @xmath1688 @xmath2276 + @xmath2227 & @xmath2277 @xmath1724 @xmath2278 & @xmath1870 @xmath1631 @xmath2279 & @xmath1690 @xmath921 @xmath2280 & @xmath2281 @xmath1815 @xmath2282 & @xmath2283 @xmath2206 @xmath2284 + @xmath2262 & @xmath2285 @xmath2286 @xmath2287 & @xmath2288 @xmath2289 @xmath2290 & @xmath2291 @xmath1626 @xmath2292 & @xmath1991 @xmath726 @xmath2293 & @xmath2233 @xmath2294 @xmath2295 + & & & & & + + @xmath2296 & @xmath2297 @xmath1897 @xmath2298 & @xmath2242 @xmath1636 @xmath2299 & @xmath1717 @xmath1614 @xmath2300 & @xmath2301 @xmath921 @xmath2302 & @xmath2303 @xmath927 @xmath2304 + @xmath2305 & @xmath2306 @xmath1659 @xmath2307 & @xmath2308 @xmath1764 @xmath1760 & @xmath2309 @xmath906 @xmath1884 & @xmath2223 @xmath917 @xmath2310 & @xmath2311 @xmath2190 @xmath1792 + @xmath2312 & @xmath2313 @xmath1724 @xmath2314 & @xmath2315 @xmath1707 @xmath2316 & @xmath1709 @xmath1814 @xmath2317 & @xmath1711 @xmath1733 @xmath2318 & @xmath2319 @xmath1688 @xmath2320 + @xmath2321 & @xmath2322 @xmath2323 @xmath2324 & @xmath2325 @xmath1759 @xmath2326 & @xmath2327 @xmath1707 @xmath2328 & @xmath2329 @xmath2197 @xmath2330 & @xmath2331 @xmath2289 @xmath2332 + @xmath2262 & @xmath2333 @xmath2334 @xmath2335 & @xmath2336 @xmath744 @xmath2337 & @xmath2338 @xmath2206 @xmath2339 & @xmath2340 @xmath2341 @xmath2342 & @xmath2343 @xmath1691 @xmath2344 + + @xmath2296 & @xmath2345 @xmath903 @xmath2346 & @xmath2340 @xmath1617 @xmath2169 & @xmath2309 @xmath915 @xmath2202 & @xmath2245 @xmath909 @xmath2347 & @xmath2171 @xmath2348 @xmath2349 + @xmath2305 & @xmath2350 @xmath915 @xmath1941 & @xmath2351 @xmath903 @xmath2352 & @xmath2163 @xmath1796 @xmath2353 & @xmath1811 @xmath900 @xmath1794 & @xmath2354 @xmath1636 @xmath2355 + @xmath2312 & @xmath2356 @xmath919 @xmath2357 & @xmath2358 @xmath1659 @xmath2169 & @xmath1922 @xmath915 @xmath2359 & @xmath1670 @xmath909 @xmath2360 & @xmath2361 @xmath2211 @xmath2362 + @xmath2321 & @xmath2363 @xmath2211 @xmath2364 & @xmath2365 @xmath2211 @xmath2366 & @xmath2367 @xmath924 @xmath1689 & @xmath2368 @xmath2190 @xmath1755 & @xmath2369 @xmath726 @xmath2370 + @xmath2262 & @xmath2371 @xmath2348 @xmath2372 & @xmath1711 @xmath1736 @xmath2373 & @xmath2374 @xmath1641 @xmath2375 & @xmath2376 @xmath2211 @xmath2226 & @xmath2377 @xmath732 @xmath2378 +	diffractive electroproduction of and mesons is measured at hera with the h1 detector in the elastic and proton dissociative channels . the data correspond to an integrated luminosity of @xmath0 . about @xmath1 and @xmath2 events are analysed in the kinematic range of squared photon virtuality @xmath3 , photon - proton centre of mass energy @xmath4 and squared four - momentum transfer to the proton @xmath5 . the total , longitudinal and transverse cross sections are measured as a function of , @xmath6 and . the measurements show a transition to a dominantly hard " behaviour , typical of high gluon densities and small @xmath7 dipoles , for larger than @xmath8 to @xmath9 . they support flavour independence of the diffractive exchange , expressed in terms of the scaling variable , and proton vertex factorisation . the spin density matrix elements are measured as a function of kinematic variables . the ratio of the longitudinal to transverse cross sections , the ratio of the helicity amplitudes and their relative phases are extracted . several of these measurements have not been performed before and bring new information on the dynamics of diffraction in a qcd framework . the measurements are discussed in the context of models using generalised parton distributions or universal dipole cross sections . # 1 # 1#2#3 _ ann . rev . nucl . part . sci . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ eur . phys . j. _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ erratum _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ ibid . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ int . j. mod . phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ jetp lett . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ mod . phys . lett . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ nucl . instr . meth . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ nuovo cim . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ nucl . phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ phys . lett . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ phys . rep . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ phys . rev . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ phys . rev . lett . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ prog . th . phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ rev . mod . phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ rep . prog . phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ sov . j. nucl . phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ sov . phys . jept _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ zeit . phys . _ * # 1 * ( # 2 ) # 3 desy 09 - 093 issn 0418 - 9833 + june 2009 + * at hera * h1 collaboration accepted by _ jhep _ + f.d . aaron@xmath10 , m. aldaya martin@xmath11 , c. alexa@xmath12 , v. andreev@xmath13 , b. antunovic@xmath11 , a. asmone@xmath14 , s. backovic@xmath15 , a. baghdasaryan@xmath16 , e. barrelet@xmath17 , w. bartel@xmath11 , k. begzsuren@xmath18 , a. belousov@xmath13 , j.c . bizot@xmath19 , v. boudry@xmath20 , i. bozovic - jelisavcic@xmath21 , j. bracinik@xmath22 , g. brandt@xmath11 , m. brinkmann@xmath23 , v. brisson@xmath19 , d. bruncko@xmath24 , a. bunyatyan@xmath25 , g. buschhorn@xmath26 , l. bystritskaya@xmath27 , a.j . campbell@xmath11 , k.b . cantun avila@xmath28 , f. cassol - brunner@xmath29 , k. cerny@xmath30 , v. cerny@xmath31 , v. chekelian@xmath26 , a. cholewa@xmath11 , j.g . contreras@xmath28 , j.a . coughlan@xmath32 , g. cozzika@xmath33 , j. cvach@xmath34 , j.b . dainton@xmath35 , k. daum@xmath36 , m. dek@xmath11 , y. de boer@xmath11 , b. delcourt@xmath19 , m. del degan@xmath37 , j. delvax@xmath38 , e.a . de wolf@xmath38 , c. diaconu@xmath29 , v. dodonov@xmath39 , a. dossanov@xmath26 , a. dubak@xmath40 , g. eckerlin@xmath11 , v. efremenko@xmath27 , s. egli@xmath41 , a. eliseev@xmath13 , e. elsen@xmath11 , a. falkiewicz@xmath42 , l. favart@xmath38 , a. fedotov@xmath27 , r. felst@xmath11 , j. feltesse@xmath43 , j. ferencei@xmath24 , d .- j . fischer@xmath11 , m. fleischer@xmath11 , a. fomenko@xmath13 , e. gabathuler@xmath35 , j. gayler@xmath11 , s. ghazaryan@xmath16 , a. glazov@xmath11 , i. glushkov@xmath44 , l. goerlich@xmath42 , n. gogitidze@xmath13 , m. gouzevitch@xmath11 , c. grab@xmath37 , t. greenshaw@xmath35 , b.r . grell@xmath11 , g. grindhammer@xmath26 , s. habib@xmath45 , d. haidt@xmath11 , c. helebrant@xmath11 , r.c.w . henderson@xmath46 , e. hennekemper@xmath47 , h. henschel@xmath44 , m. herbst@xmath47 , g. herrera@xmath48 , m. hildebrandt@xmath41 , k.h . hiller@xmath44 , d. hoffmann@xmath29 , r. horisberger@xmath41 , t. hreus@xmath49 , m. jacquet@xmath19 , m.e . janssen@xmath11 , x. janssen@xmath38 , l. jnsson@xmath50 , a.w . jung@xmath47 , h. jung@xmath11 , m. kapichine@xmath51 , j. katzy@xmath11 , i.r . kenyon@xmath22 , c. kiesling@xmath26 , m. klein@xmath35 , c. kleinwort@xmath11 , t. kluge@xmath35 , a. knutsson@xmath11 , r. kogler@xmath26 , p. kostka@xmath44 , m. kraemer@xmath11 , k. krastev@xmath11 , j. kretzschmar@xmath35 , a. kropivnitskaya@xmath27 , k. krger@xmath47 , k. kutak@xmath11 , m.p.j . landon@xmath52 , w. lange@xmath44 , g. latovika - medin@xmath15 , p. laycock@xmath35 , a. lebedev@xmath13 , g. leibenguth@xmath37 , v. lendermann@xmath47 , s. levonian@xmath11 , g. li@xmath19 , k. lipka@xmath11 , a. liptaj@xmath26 , b. list@xmath23 , j. list@xmath11 , n. loktionova@xmath13 , r. lopez - fernandez@xmath48 , v. lubimov@xmath27 , l. lytkin@xmath39 , a. makankine@xmath51 , e. malinovski@xmath13 , p. marage@xmath38 , ll . marti@xmath11 , h .- u . martyn@xmath53 , s.j . maxfield@xmath35 , a. mehta@xmath35 , a.b . meyer@xmath11 , h. meyer@xmath11 , h. meyer@xmath54 , j. meyer@xmath11 , v. michels@xmath11 , s. mikocki@xmath42 , i. milcewicz - mika@xmath42 , f. moreau@xmath20 , a. morozov@xmath51 , j.v . morris@xmath32 , m.u . mozer@xmath38 , m. mudrinic@xmath21 , k. mller@xmath55 , p. murn@xmath56 , th . naumann@xmath44 , p.r . newman@xmath22 , c. niebuhr@xmath11 , a. nikiforov@xmath11 , g. nowak@xmath42 , k. nowak@xmath55 , m. nozicka@xmath11 , b. olivier@xmath26 , j.e . olsson@xmath11 , s. osman@xmath50 , d. ozerov@xmath27 , v. palichik@xmath51 , i. panagoulias@xmath57@xmath58 , m. pandurovic@xmath21 , th . papadopoulou@xmath57@xmath58 , c. pascaud@xmath19 , g.d . patel@xmath35 , o. pejchal@xmath30 , e. perez@xmath59 , a. petrukhin@xmath27 , i. picuric@xmath15 , s. piec@xmath44 , d. pitzl@xmath11 , r. plaakyt@xmath11 , b. pokorny@xmath23 , r. polifka@xmath30 , b. povh@xmath39 , t. preda@xmath12 , v. radescu@xmath11 , a.j . rahmat@xmath35 , n. raicevic@xmath15 , a. raspiareza@xmath26 , t. ravdandorj@xmath18 , p. reimer@xmath34 , e. rizvi@xmath52 , p. robmann@xmath55 , b. roland@xmath38 , r. roosen@xmath38 , a. rostovtsev@xmath27 , m. rotaru@xmath12 , j.e . ruiz tabasco@xmath28 , z. rurikova@xmath11 , s. rusakov@xmath13 , d. lek@xmath30 , d.p.c . sankey@xmath32 , m. sauter@xmath37 , e. sauvan@xmath29 , s. schmitt@xmath11 , l. schoeffel@xmath33 , a. schning@xmath60 , h .- c . schultz - coulon@xmath47 , f. sefkow@xmath11 , r.n . shaw - west@xmath22 , l.n . shtarkov@xmath13 , s. shushkevich@xmath26 , t. sloan@xmath46 , i. smiljanic@xmath21 , y. soloviev@xmath13 , p. sopicki@xmath42 , d. south@xmath61 , v. spaskov@xmath51 , a. specka@xmath20 , z. staykova@xmath11 , m. steder@xmath11 , b. stella@xmath14 , g. stoicea@xmath12 , u. straumann@xmath55 , d. sunar@xmath38 , t. sykora@xmath38 , v. tchoulakov@xmath51 , g. thompson@xmath52 , p.d . thompson@xmath22 , t. toll@xmath23 , f. tomasz@xmath24 , t.h . tran@xmath19 , d. traynor@xmath52 , t.n . trinh@xmath29 , p. trul@xmath55 , i. tsakov@xmath62 , b. tseepeldorj@xmath63 , j. turnau@xmath42 , k. urban@xmath47 , a. valkrov@xmath30 , c. valle@xmath29 , p. van mechelen@xmath38 , a. vargas trevino@xmath11 , y. vazdik@xmath13 , s. vinokurova@xmath11 , v. volchinski@xmath16 , m. von den driesch@xmath11 , d. wegener@xmath61 , ch . wissing@xmath11 , e. wnsch@xmath11 , j. ek@xmath30 , j. zlek@xmath34 , z. zhang@xmath19 , a. zhokin@xmath27 , t. zimmermann@xmath37 , h. zohrabyan@xmath16 , f. zomer@xmath19 , and r. zus@xmath12
for each @xmath5 , let @xmath6 be a bounded domain in @xmath7 with a smooth boundary @xmath8 . assume , for brevity , that @xmath6 is homeomorphic to a ball in @xmath7 and contains @xmath9 . consider the domain @xmath10 . assume that the boundary @xmath11 of @xmath12 is smooth enough and denote by @xmath13 the inward unit normal to @xmath11 . assume that @xmath13 is not parallel to the subspace @xmath14 for any @xmath15 . denote by @xmath16 the domain in @xmath17 obtained from @xmath12 by contraction : @xmath18 . if @xmath19 , @xmath20 is a narrow tube ( or a strip for @xmath21 ) for @xmath22 . if @xmath23 , then @xmath20 is a thin layer . consider the problem : @xmath24 where @xmath25 is the inward unit normal to @xmath26 . the functions @xmath27 and @xmath28 are sufficiently regular and bounded ; @xmath27 is assumed to be nonnegative . our goal in this paper is to study the behavior of solution of problem ( [ initialequationwith_e ] ) as @xmath29 . using probabilistic methods , we will prove that @xmath30 converges as @xmath29 to the solution @xmath31 of the problem : @xmath32 here @xmath33 is the volume of @xmath6 and @xmath34 is the surface area of @xmath8 . one can expect that , under certain assumptions on the nonlinear term @xmath35 in ( [ initialequationwith_e ] ) , the solution @xmath30 can be approximated by a running - wave - type solution . corresponding results on the standard reaction diffusion equation ( [ initialequationwithout_e ] ) ( see chapter @xmath36 and @xmath37 in @xcite ) allow to describe the asymptotic wavefront motion for ( [ initialequationwith_e ] ) . we will see how the motion of the interface ( wavefront ) depends on the behavior of the cross - sections @xmath6 of the domain @xmath12 . in particular , using the results of @xcite ( chapter 6 ) we will see that in the case of the nonlinear term of k - p - p type the wavefront can have jumps . consider the wiener process @xmath38 in @xmath20 with instantaneous normal reflection on the boundary of @xmath20 . its trajectories can be described by the stochastic differential equations : @xmath39 here @xmath40 and @xmath41 are independent wiener process in @xmath42 and @xmath7 respectively and @xmath43 is a point inside @xmath20 . moreover @xmath44 and @xmath45 are projections of the unit inward normal vector to @xmath46 on @xmath42 and @xmath7 respectively . it is easy to see that @xmath47 and @xmath48 , where @xmath49 denotes euclidean length . furthermore @xmath50 is the local time for the process @xmath38 on @xmath46 , i.e. it is a continuous , non - decreasing process that increases only when @xmath51 such that the lebesque measure @xmath52 ( see for instance @xcite ) . if @xmath38 is defined by ( [ stochasticprocesswithreflection1 ] ) , then as it can be derived from theorem 2.5.1 in @xcite , @xmath30 satisfies the following integral equation in the functional space : @xmath53,\label{feynmankacformulabeforelimit1}\ ] ] where @xmath54 denotes expectation and the subscript @xmath43 denotes the initial point of @xmath55 . equation ( [ feynmankacformulabeforelimit1 ] ) has a unique solution if , say , @xmath56 has a bounded derivative in @xmath57 . let @xmath58 be the solution of the stochastic differential equation @xmath59 then the solution @xmath31 of equation ( [ initialequationwithout_e ] ) satisfies the equality : @xmath60.\label{feynmankacformulaafterlimit1}\ ] ] we prove that the component @xmath61 of the process @xmath38 converges in a certain sense to @xmath58 . this together with uniform in @xmath62 bounds for @xmath30 and its derivatives allow to prove that the solution of ( [ feynmankacformulabeforelimit1 ] ) converges to the solution of ( [ feynmankacformulaafterlimit1 ] ) as @xmath29 uniformly on each compact subset of @xmath63 . in the next section we consider averaging of integrals in local time . this result allows in section 3 to prove convergence of the integral in the right side of the first of equations in ( [ stochasticprocesswithreflection1 ] ) to the integral term in ( [ limitingstochasticprocess1 ] ) and convergence of exponents in ( [ feynmankacformulabeforelimit1 ] ) and ( [ feynmankacformulaafterlimit1 ] ) . together with a - priori bounds obtained in section 3 , this implies convergence of @xmath30 to @xmath31 . some results concerning wavefront propagation are presented in section 4 . let @xmath64 be a smooth and bounded function . we want to consider the limiting behavior as @xmath29 of expressions like @xmath65 ( see lemma 2.1 below ) . we will assume that the unit inward normal @xmath13 to @xmath11 and the function @xmath64 are both three times differentiable in @xmath66 and @xmath67 . * lemma 2.1 . * define @xmath68 , where @xmath69 is the surface element on @xmath8 . then for every @xmath70 and small enough @xmath71 , there exists a constant @xmath72 independent of @xmath71 such that : 1 . @xmath73 . 2 . for every @xmath74 we have + @xmath75 . the proof of lemma 2.1 relies on the following lemma , which we prove first . * lemma 2.2 . * for every @xmath70 and small enough @xmath71 , there exists a constant @xmath76 independent of @xmath71 such that : @xmath77[lemmarde22 ] * proof . * consider the auxiliary problem @xmath78 where @xmath79 and @xmath80 is a parameter . let @xmath81 where @xmath34 is the surface area of @xmath6 and @xmath33 is the volume of @xmath6 . as it can be derived from @xcite , a smooth in @xmath66 and @xmath67 solution @xmath82 of problem ( [ specificneymmanproblem ] ) exists and is bounded together with its first and second derivatives . so we can apply it formula to the function @xmath83 , and get : @xmath84 recalling now that @xmath47 and @xmath48 and that @xmath85 satisfies ( [ specificneymmanproblem ] ) one easily concludes that there is an @xmath86 such that for every @xmath87 : @xmath88,\nonumber\end{aligned}\ ] ] where for any function @xmath89 , @xmath90 . here , we also used the fact that the local time is increasing function of @xmath91 . this proves the statement of the lemma . @xmath92 * proof of lemma 2.1 . * we consider the auxiliary problem @xmath93 where @xmath79 and @xmath80 is a parameter . the necessary and sufficient condition for the existence of a solution for ( [ generalneymmanproblem1 ] ) is that @xmath94 where @xmath69 is the surface element on @xmath8 and @xmath95 . applying it formula to the function @xmath83 and using the bounds obtained in lemma 2.2 we get the following inequalities : @xmath96 which proves statement ( i ) of the lemma . for part ( ii ) one makes use of the doob maximal inequalities ( see @xcite , page 14 ) : @xmath97^{2}&\leq & 4 \parallel \|\nabla_{x}v\|^{2}\parallel t\nonumber\\ e[\sup_{0\leq t \leq t}\|\int_{0}^{t}(\nabla_{y}v(x_{s}^{\epsilon},y_{s}^{\epsilon}/\epsilon),dw_{s}^{2})\|]^{2}&\leq & 4 \parallel \|\nabla_{y}v\|^{2}\parallel t\nonumber\end{aligned}\ ] ] then , following the procedure that proved part @xmath98 we get that there is an @xmath99 such that for every @xmath100 : @xmath101 which together with chebyshev inequality proves statement ( ii ) of the lemma . in this section we consider the limit as @xmath103 of the solution @xmath102 to problem ( [ initialequationwith_e ] ) . the result is given in theorem 3.4 . the proof will proceed as follows . first ( in proposition 3.2 . ) we write down an integral equation in the space of trajectories for the solution of ( [ initialequationwith_e ] ) . then in lemma 3.3 we consider the mean square limit as @xmath103 of the underlying stochastic process with instantaneous normal reflection on the boundary of @xmath20 ( see ( [ stochasticprocesswithreflection1 ] ) ) . lastly an important ingredient to the proof are the a - priori bounds for @xmath102 and its derivatives . these a - priori bounds are independent of @xmath71 , their derivation is standard and are given for completeness in proposition 3.7 . we assume that the initial function @xmath104 of problem ( [ initialequationwith_e ] ) is bounded , non - negative and can have finite number of simple discontinuities . the function @xmath56 is assumed to be uniformly bounded in all arguments , continuous in x , y , lipschitz continuous in u and that there exist constants @xmath105 such that @xmath106 for @xmath107 . in addition we assume that the boundary of @xmath108 satisfies @xmath109 , where @xmath110 . * remark 3.1 . * for the existence of a classical solution to ( [ initialequationwith_e ] ) one actually needs only to assume @xmath111 . the assumption @xmath112 is being done solely for the purpose of lemma 3.2 and theorem 3.3 . let @xmath113 in @xmath114 satisfy ( [ stochasticprocesswithreflection1 ] ) . then we have : * proposition 3.2 . * problem ( [ initialequationwith_e ] ) has a unique classical solution in @xmath115 which satisfies : @xmath53.\label{feynmankacformulabeforelimit}\ ] ] * proof . * under our assumptions , the uniqueness and existence of a classical solution to ( [ initialequationwith_e ] ) follows from theorem 7.5.13 of @xcite ) . the equation ( [ feynmankacformulabeforelimit ] ) follows from theorem 2.5.1 of @xcite . @xmath92 in order now to consider the limit as @xmath103 of ( [ feynmankacformulabeforelimit ] ) , we need first to examine the asymptotic behavior of @xmath61 as @xmath103 . we will prove that @xmath61 converges as @xmath29 to @xmath58 , where @xmath58 is the solution to @xmath116 where @xmath95 . hence , we see that as @xmath29 , the effect of the reflection on the boundary is an extra drift term . a sketch of the proof for the above result is given in chapter 7 of @xcite . more details are given here . * lemma 3.3 . * for any @xmath70 we have @xmath117 * proof . * it is not difficult to see that @xmath118 and that @xmath119 then lemma 2.1 with @xmath120 and @xmath121 implies that for small enough @xmath71 there exists a constant @xmath72 independent of @xmath71 such that @xmath122 now we write @xmath123\nonumber\\ & + & [ \int_{0}^{t}\frac{1}{2}\nabla \log ( v(x^{\epsilon}_{s}))ds-\int_{0}^{t}\frac{1}{2}\nabla \log ( v(x_{s}))ds]\label{beforegronwallinequalityforstochasticprocess}\end{aligned}\ ] ] then gronwall lemma and ( [ limitlocaltimeintegral2 ] ) give : @xmath124 which is the statement of the lemma . @xmath92 consider now the solution , @xmath57 , to the equation @xmath125,\label{feynmankacformulaafterlimit}\ ] ] where @xmath126 for notational convenience we will also denote @xmath127 . since @xmath128 is lipschitz continuous in @xmath57 , the solution of ( [ feynmankacformulaafterlimit ] ) exists and is unique . our assumptions on the functions @xmath129 and the boundary @xmath8 , imply that the solution @xmath57 to ( [ feynmankacformulaafterlimit ] ) is actually the classical solution of the following parabolic problem : @xmath130 * theorem 3.4 . * under our assumptions , we have that @xmath131 where @xmath30 , @xmath31 are the solutions to ( [ initialequationwith_e ] ) and ( [ finalequationwithout_e ] ) respectively . * by proposition 3.6 and the well known theorem of ascoli - arzela we get that there exists a subsequence of @xmath132 ( which for convenience we will denote again by @xmath132 ) and a function @xmath57 , such that : @xmath133 we will prove that @xmath57 actually satisfies ( [ feynmankacformulaafterlimit ] ) which then implies that @xmath57 satisfies ( [ finalequationwithout_e ] ) . fix @xmath91 and @xmath66 and consider the solution @xmath134 to the elliptic boundary value problem : @xmath135 problem ( [ neymmanproblemforparabolicproblemandu ] ) is solvable if @xmath136 proceeding similarly now to lemma 2.1 and recalling that @xmath85 satisfies ( [ neymmanproblemforparabolicproblemandu ] ) , we see that there is a constant @xmath137 such that : @xmath138^{2})\label{limitintegralwithlocaltimeandfunction1}\end{aligned}\ ] ] we observe that @xmath139 depends on @xmath71 only through functions that are uniformly bounded in @xmath71 ( proposition 3.7 ) . this observation and lemma 2.2 imply that as @xmath140 : @xmath141 moreover the lebesque dominated convergence theorem , lemma 3.3 , the compactness of the family @xmath132 and ( [ newnonlinearitybeforelimit ] ) , imply that as @xmath103 : @xmath142 where @xmath143 and @xmath58 are given by ( [ newnonlinearitybeforelimit ] ) , ( [ newnonlinearity ] ) and ( [ limitingstochasticprocess ] ) respectively . now let @xmath144 be the solutions to ( [ feynmankacformulabeforelimit ] ) and ( [ feynmankacformulaafterlimit ] ) respectively . taking into account relations ( [ limitintegralwithlocaltimeandfunction2 ] ) , ( [ limitintegralwithlocaltimeandfunction3 ] ) , the weak convergence of @xmath61 to @xmath58 as @xmath103 ( which is implied by lemma 3.3 ) and proposition 3.2 we get the statement of the theorem . @xmath92 we conclude this section with the a - priori bounds for the hlder norm of the solution and for the sup - norm of the solution , the first and the second derivatives of the solution of ( [ initialequationwith_e ] ) . these bounds will be uniform in @xmath71 . the method follows closely @xcite . let us first introduce some notation . we write @xmath145 , @xmath146 , @xmath147 and @xmath148 , where @xmath149 . for @xmath150 , @xmath70 and for any function @xmath89 we write : @xmath151 moreover for notational convenience we will write @xmath152 . * lemma 3.5 . * under our assumptions there exists a constant @xmath153 , independent of @xmath154 , such that @xmath155 in @xmath156 . * * lemma 3.5 can be proven using equation ( [ feynmankacformulabeforelimit ] ) . here we give an analytic proof of the claim . for any fixed @xmath157 we define the function @xmath158 it is easy to show that @xmath159 in the weak sense . let us choose now @xmath160 , where @xmath161 is such that if @xmath107 then @xmath106 for some @xmath162 . then @xmath163 let us now assume that @xmath164 attains a maximum positive value on the boundary @xmath165 at the point @xmath166 . since @xmath102 is continuous up to the boundary , there exists a connected set @xmath167 such that @xmath168 , @xmath169 and @xmath170 on @xmath167 , i.e. @xmath171 on @xmath167 . since @xmath166 is a maximum for @xmath164 and @xmath25 is the inward normal derivative we get that @xmath172 at @xmath166 . but on @xmath167 we have that @xmath173 . taking into account the particular choice of @xmath174 and that @xmath106 for @xmath107 , we get that @xmath175 at @xmath166 . thus we have a contradiction and so maximum principle implies that @xmath176 lastly maximum principle again implies that @xmath177 . @xmath92 let us consider the following linear parabolic pde : @xmath178 where @xmath129 are bounded smooth functions . under the standard hypotheses problem ( [ initiallinearequationwith_e ] ) has a unique classical solution ( theorem 5.3.2 in @xcite ) . * lemma 3.6 . * there is a constant @xmath179 , independent of @xmath71 , and an open set @xmath180 such that for any @xmath181 : @xmath182 * proof . * we will give just a sketch of the proof , since the analysis follows @xcite , @xcite , @xcite and @xcite . the calculations are lengthy but standard . we solve the second initial - boundary value problem ( [ initiallinearequationwith_e ] ) by reducing it to an integral equation , i.e. we write : @xmath183 where @xmath184 $ ] is the fundamental solution to the heat equation and @xmath185 is the solution to a voltera type integral equation : @xmath186\phi^{\epsilon}(\tau,\xi)d\partial d^{\epsilon}_{\xi}d\tau \label{volteraforphi}\\ & + & 2[\int_{d^{\epsilon}}\frac{\partial \gamma^{\epsilon}(t , z,0,\xi)}{\partial \gamma^{\epsilon}}f(\xi)d\xi+ \epsilon c(z)\int_{d^{\epsilon}}\gamma^{\epsilon}(t , z,0,\xi)f(\xi)d\xi]\nonumber\end{aligned}\ ] ] let us now define @xmath187 it can be shown ( see @xcite ) that there is a hlder continuous ( in space variables ) and bounded ( with bound and hlder coefficient independent of @xmath71 ) solution @xmath188 for ( [ volteraforphi ] ) , expressed in the form : @xmath189 using the boundedness and the hlder continuity of ( [ finalphi ] ) and ( [ represenataion1 ] ) , one can show ( see @xcite , @xcite , @xcite and @xcite ) that there is a constant @xmath179 , independent of @xmath71 , such that @xmath190 @xmath92 now we are ready to prove the result for the a - priori bounds : * proposition 3.7 . * there is a constant @xmath179 , independent of @xmath71 , and an open set @xmath180 such that for any @xmath191 ( @xmath192 is the constant from lemma 3.6 . ) : @xmath193 where @xmath102 is a classical solution to ( [ initialequationwith_e ] ) . * we will use schauder s fixed point theorem . let us first define for convenience @xmath194 . let @xmath195 be the banach space of all functions @xmath196 that are continuous in @xmath156 with norm @xmath197 . for any @xmath198 , let @xmath199 be the set @xmath200 . for every @xmath201 define @xmath202 to be the solution to the following problem : @xmath203 then similarly as in lemma 3.6 , one can write : @xmath204 where @xmath205 satisfies : @xmath206\phi^{\epsilon}(\tau,\xi)d\partial d^{\epsilon}_{\xi}d\tau \nonumber\\ & + & 2[\int_{d^{\epsilon}}\frac{\partial \gamma^{\epsilon}(t , z,0,\xi)}{\partial \gamma^{\epsilon}}f(\xi)d\xi+ \epsilon c(z , u^{\epsilon})\int_{d^{\epsilon}}\gamma^{\epsilon}(t , z,0,\xi)f(\xi)d\xi]\label{schaudervolteraforphi}\end{aligned}\ ] ] we shall prove that @xmath207 has a fixed point . since @xmath102 and @xmath28 are bounded functions , one can show , in the same way as in the proof of lemma 3.6 , that the function @xmath205 that satisfies ( [ schaudervolteraforphi ] ) is bounded and hlder continuous ( in space variables ) with bound and hlder constant independent of @xmath71 . using this result and representation ( [ schauderinitialsolutionwith_e ] ) one can conclude ( lemma 3.6 ) that there is a constant @xmath179 such that @xmath208 so @xmath207 maps @xmath199 into itself for an appropriately chosen constant @xmath179 . now let @xmath209 be a sequence of functions that belong to @xmath199 and @xmath210 be defined by ( [ schauderinitialsolutionwith_e ] ) and ( [ schaudervolteraforphi ] ) when @xmath211 . assume that @xmath212 as @xmath213 . we need to show that @xmath214 as @xmath213 . the continuity of the function @xmath215 in @xmath57-variables imply that @xmath216 as @xmath213 . this and ( [ schauderinitialsolutionwith_e ] ) give us @xmath214 . therefore @xmath207 is a continuous map . next we need to show that @xmath207 maps @xmath199 into a compact subset of @xmath199 . this is an easy consequence of theorem 7.1.1 of @xcite , which states that for @xmath217 , the bounded subsets of @xmath218 are pre - compact subsets of @xmath195 . lastly @xmath219 is a closed convex set of the banach space @xmath218 . therefore by schauder s fixed point theorem we get that @xmath207 has a fixed point , i.e. there exists a @xmath102 such that @xmath220 and actually @xmath221 @xmath92 in this section we will see some applications of theorem 3.4 to the question of wave front propagation in narrow domains . as we mentioned in the introduction , corresponding results on the standard reaction diffusion equation ( [ initialequationwithout_e ] ) ( see chapter @xmath36 and @xmath37 in @xcite , @xcite and @xcite ) allow to describe the asymptotic wavefront motion for ( [ initialequationwith_e ] ) . we will focus on two different cases . in subsection 4.1 we consider the case where the functions @xmath222 , @xmath223 , @xmath224 and @xmath225 change slowly in @xmath66 , i.e. @xmath226 , @xmath227 , @xmath228 and @xmath229 for @xmath230 . we first assume that the nonlinear boundary term in ( [ initialequationwith_e ] ) , @xmath56 , is of k - p - p type for @xmath231 , i.e. @xmath232 is positive for @xmath233 , negative for @xmath234 and @xmath235 . we will see how the motion of the wavefront depends on the behavior of the cross - sections @xmath6 of the domain @xmath12 . in particular , using the results of @xcite ( chapter 6 ) we will see that in the case of the nonlinear term of k - p - p type and for @xmath236 the wavefront can have jumps . actually , the jumps of the wavefront appear at positions where the tube becomes thinner . the results are given in theorem 4.1.2 , theorem 4.1.5 and theorem 4.1.7 . then we briefly discuss the bistable case , i.e. when @xmath237 for @xmath238 and @xmath239 for @xmath240 , where @xmath241 . in this case we consider a specific example and we will see how the asymptotic speed of the wavefront depends on the surface area to volume ratio @xmath242 . in subsection 4.2 , we return to the k - p - p case , but now we consider front propagation when @xmath243 and the boundary @xmath244 of @xmath108 is determined by stationary random processes on @xmath245 on some probability space @xmath246 . the conclusion is in theorem 4.2.7 . we will denote by @xmath247 the nonlinear term in ( [ initialequationwithout_e ] ) . obviously the type of @xmath128 ( k - p - p or bistable ) is determined by @xmath232 and vice - versa . let us assume that the functions @xmath222 , @xmath223 , @xmath224 and @xmath225 change slowly in @xmath66 , i.e. @xmath248 , @xmath227 , @xmath228 and @xmath229 for @xmath230 . we start with the case where the nonlinear term @xmath128 of ( [ initialequationwithout_e ] ) is of k - p - p type . we additionally assume that the closure of the support of @xmath27 , @xmath249 , coincides with the closure of its interior . lastly we take for brevity @xmath250 and @xmath251 ( recall that @xmath252 ) to be an increasing function . let @xmath253\rightarrow \mathbb{r}^{1}$ ] and introduce the functional @xmath254ds,\label{r_functional } & $ \phi$ is absolutely continuous \cr + \infty , & for the rest of $ \mathcal{c}_{0,t}.$\cr } \ ] ] put @xmath255 we say that condition ( n ) is satisfied if for any @xmath256 and @xmath257 @xmath258 as it is mentioned in chapter 10 of @xcite , condition ( n ) is fulfilled for the smooth and increasing function @xmath259 . moreover as we shall see in theorem 4.1.2 , @xmath260 determines the motion of the wave front for @xmath102 for small enough @xmath154 . let us consider @xmath31 , the solution to equation ( [ initialequationwithout_e ] ) , for @xmath19 . if we set @xmath261 , then @xmath262 is the solution to the following parabolic problem : @xmath263 under the assumptions above , the following theorem , which is a reformulation of theorem 6.2.1 of @xcite , states that @xmath260 determines the motion of the wave front for @xmath264 under condition ( n ) : * theorem 4.1.1 . * let @xmath264 be the solution to ( [ finalequationwith_d ] ) . then under condition ( n ) we have : @xmath265 let us consider now equation ( [ initialequationwith_e ] ) for @xmath19 , @xmath248 , @xmath229 in a slowly changing in @xmath66 narrow domain @xmath266 , so that @xmath227 , @xmath228 . let us define @xmath267 . under the assumptions above , theorems 3.4 and 4.1.1 imply that @xmath260 will determine the motion of the wave front in this case too , as follows : * theorem 4.1.2 . * the following statement holds : @xmath268 so the equation @xmath269 defines the position of the interface ( wavefront ) between areas where @xmath270 ( for @xmath154 and @xmath74 small enough ) is close to @xmath271 and to @xmath272 . actually , as we shall see below the wavefront may have jumps . it is known ( see chapter 6 in @xcite ) , that because of the dependance of @xmath259 on @xmath66 , the wave front of @xmath262 may have jumps and new sources may be `` igniting '' ahead of the front . we will give sufficient conditions that guarantee such jumps for a class of smooth and increasing functions @xmath273 . hence theorem 4.1.2 implies that one can predict appearances of new sources and jumps of the wave front of @xmath270 for @xmath154 and @xmath74 small enough . reaction - diffusion eqautions ( rde s ) with nonlinear boundary conditions in narrow domains let @xmath274 be such that @xmath275 . such a @xmath276 is defined in a unique way . to @xmath277 at time @xmath278.,width=8,height=4 ] we have the following proposition ( see chapter 6 in @xcite for more details ) : * proposition 4.1.3 . * let @xmath279 be as in figure 1 and @xmath280 . then the wavefront jumps from @xmath281 to @xmath277 at time @xmath282 ( see figure 1 ) , i.e. : 1 . if @xmath283 then @xmath284 for a connected set : + @xmath285 . if @xmath286 then the set where @xmath284 consists of two connected components : + @xmath287 . + the set @xmath288 is at a positive distance from the set @xmath289 for @xmath286 . 3 . if @xmath290 then @xmath284 for a connected set : + @xmath291 . based now on comparison results ( lemma 4.1.4 ) we will give sufficient conditions that guarantee jumps of the wavefront . in particular we will prove ( theorem 4.1.5 ) that if @xmath259 is a rapidly increasing smooth function , then @xmath274 such that @xmath275 is as in figure 1 . the functional @xmath292 defined in ( [ r_functional ] ) and the function @xmath260 defined in ( [ definitionofw ] ) depend also on @xmath293 . hence we will write sometimes @xmath294 and @xmath295 in order to emphasize this dependence . we have the following comparison result : * lemma 4.1.4 . * 1 . let @xmath296 be a positive number . then @xmath297 . 2 . let @xmath192 be a positive number and define @xmath298 . then @xmath299 . 3 . let @xmath300 be two functions such that @xmath301 for every @xmath250 . then @xmath302 . * * let us write @xmath303 and let @xmath304 be the extremal so that @xmath305 . such an extremal satisfies the following euler - lagrance equation : @xmath306 let us define now the function @xmath307 . we claim that the function @xmath308 is the extremal so that @xmath309 . indeed it is easy to see that the definition of @xmath310 and the fact that @xmath311 imply that @xmath312 . moreover @xmath310 satisfy an euler - lagrange equation of the form ( [ equationfortheextremal1 ] ) with @xmath313 and @xmath314 replaced by @xmath315 and @xmath316 respectively . this proves the claim , which implies part ( i ) of the lemma . part ( ii ) of the lemma can be proven in a similar way . we define @xmath317 and let @xmath318 to be the extremal so that @xmath319 . then similarly as it is done in part ( i ) , one should consider the function @xmath308 that is defined by @xmath320 . we prove now part ( iii ) of the lemma . let us define @xmath321 and @xmath322 . moreover let @xmath323 be the extremal so that @xmath324 . since @xmath301 we have @xmath325 furthermore , it is easy to see that @xmath260 is an increasing function of @xmath91 . let us assume now that @xmath326 . this assumption and the fact that @xmath327 imply that @xmath328 . by recalling the definition of function @xmath329 , one easily concludes that : @xmath330 however inequality ( [ comparisonlemma2 ] ) contradicts ( [ comparisonlemma1 ] ) . therefore @xmath302 . @xmath92 in section 6.2 of @xcite , it is proven that if @xmath273 , instead of the smooth function @xmath331 , is a piecewise constant function , denoted by @xmath332 , such that @xmath333 with @xmath334 , then the function @xmath335 such that @xmath336 is not monotone , as in figure 1 . more specifically the curves connecting the point @xmath337 with @xmath338 and @xmath338 with @xmath339 are line segments and for @xmath340 , @xmath335 is the solution to @xmath341 moreover in this case @xmath342 we will write @xmath343 and @xmath344 to emphasize the dependence of @xmath278 and @xmath345 on the function @xmath332 . with the help of the result above and lemma 4.1.4 we will give sufficient conditions that guarantee jumps of the wavefront of @xmath264 ( and by theorem 4.1.2 of @xmath30 for @xmath154 and @xmath74 small enough ) for a class of smooth and increasing functions @xmath293 . let us define the set @xmath346 it is easy to see that @xmath167 is a non - empty set . * theorem 4.1.5 . * let @xmath332 be the step function defined in ( [ stepfunction ] ) such that @xmath347 . consider real numbers @xmath296 and @xmath192 such that 1 . 2 . @xmath349 $ ] . then for any smoothly increasing function @xmath273 such that @xmath350 the wavefront corresponding to @xmath293 has jumps . in particular the excitation reaches the region @xmath351 before it reaches the point @xmath352 , where @xmath353 is a small enough positive number and @xmath352 is as in figures 2 and 3 . * proof . * let us define @xmath354 . since @xmath348 , the function d(x ) is shifted vertically upwards and horizontally to the left . so we get that @xmath355 ( see figure 2 ) . parts ( i ) and ( ii ) of lemma 4.1.4 imply that @xmath356 . this and part ( iii ) of lemma 4.1.4 give that if @xmath293 satisfies ( [ sufficientconditionforajump1 ] ) , then @xmath276 will satisfy ( see figure 3 ) : @xmath357 , width=6,height=5 ] , width=6,height=5 ] we know that @xmath358 and @xmath359 are not monotone ( recall that @xmath360 and @xmath361 are piecewise constant functions ) . we will show that @xmath276 is also not monotone ( i.e it is as in figure 1 ) . let us assume that @xmath362 where @xmath363 is as in ( [ timesofjump0 ] ) and @xmath364 is defined similarly to @xmath365 in ( [ timesofjump1 ] ) with @xmath366 replaced by @xmath367 respectively . in particular ( [ sufficientconditionforajump2 ] ) holds if condition ( ii ) above holds , i.e. if @xmath349 $ ] . moreover the condition @xmath368 in the definition of @xmath167 in ( [ admissibleset ] ) implies that @xmath369>1 $ ] , which has to be true since @xmath348 . inequality ( [ sufficientconditionforajump2 ] ) can be equivalently written as @xmath370 . by this and ( [ relationforwavefrontposition1 ] ) we immediately get that @xmath371 which , since @xmath372 , implies that @xmath276 is as in figure 1 and so new sources are igniting ahead of the wavefront . in figures 2 and 3 we see an illustration of the construction . @xmath92 * example . * an example of a function @xmath273 that satisfies the requirements of theorem 4.1.5 is @xmath373 where @xmath347 , @xmath374 satisfy assumptions ( i ) and ( ii ) of theorem 4.1.5 , @xmath375 and the constants @xmath376 and @xmath377 are chosen so that @xmath378 and @xmath379 . in particular now if @xmath380 , i.e. @xmath381 , is an increasing smooth function that satisfies the requirements of theorem 4.1.5 , then the jump of the wavefront of @xmath382 , for @xmath154 and @xmath74 small enough , occurs when @xmath242 increases rapidly . this implies , at least when the tube @xmath108 retains its shape as @xmath66 increases , that the jumps of the wave front occur at places where the tube @xmath108 becomes thinner , i.e. when @xmath33 decreases significantly . * remark 4.1.6 . * similar results hold for layers as well , i.e. for @xmath5 with @xmath23 . using the results in @xcite one can consider the limiting behavior as @xmath383 of @xmath382 when condition ( n ) is not fulfilled . we will briefly discuss the result for the general case @xmath5 . instead now of function @xmath260 defined by ( [ definitionofw ] ) , we consider the function @xmath384 one can prove that @xmath385 is lipschitz continuous and that @xmath386 . then theorem 2.1 in @xcite and theorem 3.4 imply that @xmath385 determines the motion of the wave front as follows : * theorem 4.1.7 . * the following statements hold : 1 . for any compact subset @xmath387 of the interior of @xmath388 , @xmath389 2 . for any compact subset @xmath390 of @xmath391 , @xmath392 we conclude subsection 4.1 with the case that the nonlinear term @xmath128 of ( [ initialequationwithout_e ] ) is of bistable type , i.e. @xmath393 for @xmath238 , @xmath394 for @xmath395 , where @xmath241 . this problem was considered in @xcite and it was also presented in section 6.4 of @xcite . here we restrict the analysis to a concrete example that allows to give an exact formula for the asymptotic speed of the wavefront of @xmath270 for @xmath154 and @xmath74 small enough . as we will see the asymptotic speed of the wavefront is proportional to the square root of the surface area to volume ratio @xmath396 . to be specific let @xmath5 , @xmath397 , @xmath398 and assume that the function @xmath264 ( compare with ( [ finalequationwith_d ] ) ) is the solution to @xmath399 consider a point @xmath5 to be excited at time @xmath91 , if @xmath264 ( the solution to ( [ finalequationwith_d1 ] ) ) is close to @xmath272 and non - excited if @xmath264 is close to @xmath271 . then the corollary of theorem 4.1 of @xcite gives us that for small @xmath74 the region @xmath400 becomes excited and the region @xmath401 becomes non - excited after a short starting phase . now let @xmath267 , where @xmath30 is the solution to ( [ initialequationwith_e ] ) . theorem 3.4 implies that the same conclusions hold for @xmath382 for @xmath154 and @xmath74 small enough . to compute the asymptotic propagation speed of excitation at @xmath5 , let us consider the equation for the wave profile : @xmath402 as it can be verified by direct substitution , equation ( [ runningwavebistablehomogeneouscaseexample ] ) is solvable if @xmath403 is given by the formula @xmath404 moreover , in our case , ( [ wavefrontspeed ] ) is also the asymptotic propagation speed of excitation at @xmath405 and it is independent of direction . lastly , it is known that as the size of @xmath6 increases ( without changing shape ) , the surface area to volume ratio @xmath242 decreases . in the case @xmath243 , this fact , equation ( [ wavefrontspeed ] ) and theorem 3.4 imply that the wavefront of @xmath270 ( for @xmath154 and @xmath74 small enough ) slows down when the tube becomes thicker . a similar result holds for layers . in this subsection we consider wave front propagation for the solution of ( [ initialequationwith_e ] ) for small @xmath154 , when @xmath243 , the boundary @xmath244 of @xmath108 is determined by stationary and ergodic random processes on @xmath245 and the nonlinear boundary term in ( [ initialequationwith_e ] ) ( for @xmath231 , i.e. @xmath232 ) is of k - p - p type . as we did in subsections 4.1 , we will first consider ( theorem 4.2.6 ) wavefront propagation for the solution of ( [ initialequationwithout_e ] ) and then with the aid of theorem 3.4 we will consider ( theorem 4.2.7 ) wavefront propagation for the solution of ( [ initialequationwith_e ] ) for small enough @xmath154 . as we will see the cross sections @xmath6 of @xmath12 affect the speed of the wavefront through the surface to volume ratio @xmath242 . in sections @xmath406 of @xcite wave front propagation for equations like ( [ initialequationwithout_e ] ) is considered in the case where there is no drift term and the randomness comes only from the nonlinear part of the equation . moreover in @xcite the authors considered the case of reaction - diffusion equations of type ( [ initialequationwithout_e ] ) with a random drift and homogeneous in @xmath66 nonlinear term . in the case considered here , both the drift and and the nonlinear term are random . in @xcite , 524 - 525 , the author remarks that one could use the procedure developed in sections @xmath406 of @xcite to study wavefronts in one - dimensional uniformly bounded random drift with random nonlinear term . we will see that one can prove theorem 4.2.6 , which is analogous to theorem 7.6.1 in @xcite , by following the proof of theorem 7.6.1 in @xcite . we make use of the results in @xcite and of the fact that the operator of the equation ( [ initialequationwithout_e ] ) is self adjoint with respect to an appropriate inner product ( it has the form @xmath407 ) . actually the latter simplifies the analysis significantly . instead of repeating the proof of @xcite here , we will only outline the differences . let us first list our assumptions . consider a probability space @xmath246 . we assume that the random field @xmath408 ( namely the volume ) is three times continuously differentiable , i.e. @xmath409 , with @xmath410 probability one . suppose that @xmath411 is a random vector function on @xmath246 and that it is measurable , stationary in @xmath66 and translation in @xmath66 generates an ergodic transformation of the space @xmath412 . moreover the function @xmath413 is assumed bounded , with zero mean ( i.e. @xmath414=0 $ ] ) . we additionally assume ( for the purposes of lemma 4.2.1 and 4.2.3 ) that there is a set of nonzero @xmath410 probability on which @xmath415^{-1}dx=+\infty.\label{conditionsfordrift}\ ] ] if condition ( [ conditionsfordrift ] ) holds on a set of nonzero measure then , by the ergodicity assumption , it must hold with @xmath410 probability one . as far as the non - linear term @xmath416 is concerned , in addition to the stationarity and ergodicity assumptions , we also make the following assumptions . for all @xmath236 , @xmath28 is of k.p.p type , i.e. @xmath232 is positive for @xmath233 , negative for @xmath234 , continuous in @xmath57 for @xmath417 and @xmath418 . moreover with @xmath410 probability one , the function @xmath419 satisfies a lipschitz condition of the form @xmath420 such that for all @xmath421 and @xmath236 , @xmath422 where @xmath423 is a diffusion process with random generator @xmath424 . the initial function @xmath104 is assumed to be nonnegative , bounded from above and non - random . let now @xmath425 be the function defined by the equality @xmath426ds\ } ] , \hspace{0.2 cm } z\in \mathbb{r},\label{mufunction}\ ] ] where @xmath427 and @xmath428 is the first hitting time of the process @xmath58 to the point @xmath271 . for @xmath428 one has the following lemma : * lemma 4.2.1 . * condition ( [ conditionsfordrift ] ) and @xmath429=0 $ ] imply that @xmath430 . * proof . * it follows directly from the proof of lemma 4.4 of @xcite if one notes that the drift term is @xmath431 . @xmath92 * lemma 4.2.2 . * under the assumptions imposed above , function @xmath425 has the following properties : 1 . for all @xmath432 , @xmath433ds\}$ ] . 2 . function @xmath425 is convex , lower semicontinuous and monotonically non - decreasing in @xmath434 . moreover @xmath425 is continuously differentiable and the derivative @xmath435 is positive and monotonically increasing for @xmath436 , where @xmath437 is a non - positive number ( which actually is the discontinuity point of @xmath425 , as property ( iii ) below shows ) . 3 . @xmath438 for @xmath439 and @xmath440 for @xmath441 where @xmath442 . * proof . * property ( i ) can be proven as proposition 2.1 of @xcite . property ( ii ) follows similarly as theorem 7.5.1(ii ) of @xcite . property ( iii ) follows analogously to theorem 7.5.1(iii ) of @xcite . here one uses the fact that the operator of ( [ initialequationwithout_e ] ) has the form @xmath407 ) , i.e. it is self adjoint . @xmath92 we also observe that @xmath443 where @xmath444 $ ] . as it has been proven in lemma 2.2 of @xcite , function @xmath445 has properties ( i)-(iii ) of lemma 4.2.2 as well ( for @xmath446 ) . in addition the following lemma holds , which is a restatement of proposition 4.1 of @xcite . * lemma 4.2.3 . * condition ( [ conditionsfordrift ] ) and @xmath429=0 $ ] imply that the discontinuity point of @xmath445 is @xmath447 . we will assume that @xmath448 ( by lemma 4.2.2(iii ) or lemma 4.2.3 we already know that @xmath449 ) and we define @xmath450 $ ] for @xmath451 . lemma 4.2.2 and the fact that @xmath443 imply that the arguments in the beginning of section @xmath452 of @xcite carry out here as well . therefore we conclude that there is a unique @xmath453 such that @xmath454 and @xmath455 . * remark 4.2.4 . * we would like to emphasize that the existence and uniqueness of a positive @xmath456 follows mainly from properties ( i)-(iii ) of @xmath425 ( lemma 4.2.2 ) . in particular property ( iii ) holds because the operator of ( [ initialequationwithout_e ] ) is self adjoint . similarly as theorem 7.6.1 in @xcite was proven , one can prove theorem 4.2.6 below . note that by following the proof of theorem 7.6.1 in @xcite , one needs to estimate certain probabilities for @xmath428 and @xmath58 . for this purpose we have the following lemma : * lemma 4.2.5 . * let @xmath353 be a positive number and @xmath457 . then 1 . @xmath458 , @xmath410-a.s . 2 . @xmath459 } p_{x}\lbrace x_{s}\in u_{\delta}(0)\rbrace>0 $ ] , @xmath410-a.s . 3 . for @xmath460 and @xmath461 we have @xmath462 * proof . * the proof of all statements follows from the corresponding statements for @xmath40 in place of @xmath58 ( see for example section 7.5 of @xcite ) and by the girsanov s theorem on the absolute continuous change of measures in the space of trajectories . @xmath92 therefore we have the following theorem : * theorem 4.2.6 . * let @xmath243 and @xmath31 satisfy equation ( [ initialequationwithout_e ] ) . under our assumptions we have : 1 . for all @xmath463 , @xmath464 2 . let us define @xmath465 and assume that there is a constant @xmath466 such that for any @xmath467 and @xmath243 , @xmath468 then for all @xmath469 , @xmath470 finally theorem 3.4 and theorem 4.2.6 imply : * theorem 4.2.7 . * let @xmath471 and @xmath30 satisfy equation ( [ initialequationwith_e ] ) . under our assumptions we have : 1 . for all @xmath463 , @xmath472 2 . let us define @xmath465 and assume that there is a constant @xmath466 such that for any @xmath467 and @xmath243 , @xmath468 then for all @xmath469 , @xmath473 * remark 4.2.8 . * theorem 4.2.6 was proven in ( @xcite ) with the assumption in part ( ii ) replaced by the assumption that for any @xmath467 and @xmath474 , @xmath475 which is however difficult to verify . obviously the assumption made in part ( ii ) of theorems 4.2.6 and 4.2.7 implies ( [ assumption1 ] ) . m. freidlin was partially supported by the nsf . the authors would like to thank prof . k. trivisa and prof . m. grillakis for helpfull discussions . a. kolmogorov , i. petrovskii , n. piskunov , @xmath477tude de l@xmath478quation de la diffusion avec croissence de la mati@xmath479re et son application a un probl@xmath479me biologique , moscov university bull . , vol . 1 ( 1937 ) , pp . 1 - 25 .	second initial boundary problem in narrow domains of width @xmath0 for linear second order differential equations with nonlinear boundary conditions is considered in this paper . using probabilistic methods we show that the solution of such a problem converges as @xmath1 to the solution of a standard reaction - diffusion equation in a domain of reduced dimension . this reduction allows to obtain some results concerning wave front propagation in narrow domains . in particular , we describe conditions leading to jumps of the wave front . keywords : reaction diffusion equations , narrow domains , wave front propagation , instantaneous reflection @xmath2__department of mathematics , university of maryland , college park , 20742 , _ _ @xmath3 _ maryland , usa , e - mail : [email protected]_ @xmath4__department of mathematics , university of maryland , college park , 20742 , _ _ @xmath3 _ maryland , usa , [email protected]_
the hilbert space of the system can be represented as the graph shown in fig . [ fgr:1 ] . the vertices of this graph represent the many - body eigenstates @xmath27 which are classified by the number of particles @xmath0 and the energy @xmath90 . directed edges connect these vertices according to the transitions between the eigenstates generated by the coupling to the vacuum reservoir . the dynamics of the system can be considered as a walk through the graph . transitions should satisfy two constraints : ( i ) @xmath29 , ( ii ) @xmath30 . by possible transitions we mean all the transitions satisfying the first constraint . we divide them into allowed transitions which also satisfy the second constraint and forbidden transitions which do not satisfy the second constraint . we also introduce two categories for the eigenstates of the system . for a given initial state we define an achievable eigenstate as an eigenstate which can be reached from the initial one through the allowed transitions . we define stable eigenstates as the eigenstates from which no further transitions are allowed . only achievable stable eigenstates are present in the steady state of the system . we note that the dynamics of the system clearly does not satisfy the detailed balance principle since all possible transitions are irreversible . sites is considered . bars representing many - body eigenstates @xmath27 are placed into the coordinate system number of particles - energy. initial maximally occupied pure state is indicated by the green colour . allowed transitions and some forbidden transitions for @xmath40 are shown with solid and dashed arrows correspondingly . the achievable eigenstates are indicated with the blue colour . the stable eigenstates are represented with empty bars . after the loss of particles the systems in the ensemble can be in either the lowest energy @xmath91-particle eigenstate or the lowest energy @xmath2-particle eigenstate . ] the binding energy @xmath23 determines the allowed transitions and the eigenstates which are present in the steady state of the system . the dependence of the number of these eigenstates on @xmath23 for the initial maximally occupied pure state is shown in fig . [ fgr:2 ] . for the extreme values of the binding energy there is only one achievable stable eigenstate : ( i ) the empty eigenstate for @xmath92 ( all possible transitions are allowed ) , ( ii ) the maximally occupied eigenstate for @xmath93 ( all possible transitions are forbidden ) . for calculations we set @xmath94 which corresponds to the maximal value of the number of eigenstates present in the steady state of the system . -site lattice is prepared in the maximally occupied pure state . the plot shows the number of achievable stable eigenstates as the function of the binding energy . ] particles and different energies : ( a ) @xmath95 , ( b ) @xmath96 , ( c ) @xmath97 , ( d ) @xmath98 . the stationary distributions are of the boltzmann form in each @xmath0-particle sector for all these initial conditions ( temperatures are presented in tables).,title="fig : " ] particles and different energies : ( a ) @xmath95 , ( b ) @xmath96 , ( c ) @xmath97 , ( d ) @xmath98 . the stationary distributions are of the boltzmann form in each @xmath0-particle sector for all these initial conditions ( temperatures are presented in tables).,title="fig : " ] particles and different energies : ( a ) @xmath95 , ( b ) @xmath96 , ( c ) @xmath97 , ( d ) @xmath98 . the stationary distributions are of the boltzmann form in each @xmath0-particle sector for all these initial conditions ( temperatures are presented in tables).,title="fig : " ] particles and different energies : ( a ) @xmath95 , ( b ) @xmath96 , ( c ) @xmath97 , ( d ) @xmath98 . the stationary distributions are of the boltzmann form in each @xmath0-particle sector for all these initial conditions ( temperatures are presented in tables).,title="fig : " ] as stated in the paper , the steady state of the system is characterized by the boltzmann distribution in each sector . here we demonstrate that the variation of the initial state of the system only changes parameters of the stationary distribution but not its shape . [ fgr:3 ] shows stationary distributions for the initial pure states with @xmath99 particles and the different energies @xmath100 . we observe that ( i ) temperatures @xmath4 increase as @xmath100 increases , ( ii ) for the initial states with close energies stationary distributions are similar . the amplitude of the state - to - state fluctuations of the transition rate as the function of the transition energy depends on the size of the system . [ fgr:6 ] shows that the standard deviation of the state - to - state fluctuations becomes smaller as the size of the system increases . lattices of the different sizes are obtained by taking out sites from the 14-site lattice .	relaxation dynamics of complex quantum systems with strong interactions towards the steady state is a fundamental problem in statistical mechanics . the steady state of subsystems weakly interacting with their environment is described by the canonical ensemble which assumes the probability distribution for energy to be of the boltzmann form . the emergence of this probability distribution is ensured by the detailed balance of the transitions induced by the interaction with the environment . here we consider relaxation of an open correlated quantum system brought into contact with a reservoir in the vacuum state . we refer to such a system as emissive since particles irreversibly evaporate into the vacuum . the steady state of the system is a statistical mixture of the stable eigenstates arising due to the binding energy . we found that , despite the absence of the detailed balance , the stationary probability distribution over these eigenstates is of the boltzmann form in each @xmath0-particle sector . a quantum statistical ensemble corresponding to the steady state is characterized by different temperatures in the different sectors , in a contrast to the gibbs ensemble . we investigate the transition rates between the eigenstates to understand the emergence of the boltzmann distribution and find their exponential dependence on the transition energy . we argue that this property of transition rates is generic for a wide class of emissive quantum many - body systems . whether and how a quantum system brought out of equilibrium reaches its steady state is a fundamental question which has recently attracted much attention @xcite . expectation values of local observables after relaxation are typically determined by integrals of motion while the memory of microscopic details of an initial state is lost . in quantum systems the mechanism of thermalization is rooted in the properties of individual eigenstates as stated in the eigenstate thermalization hypothesis ( eth ) @xcite . the basic statement of eth for nonintegrable systems is the smoothness of eigenstate expectation values of local observables as functions of eigenenergies . for systems with integrals of motion this statement remains true if eigenstates are taken from the same subspace @xcite . in the steady state the density matrix of any subsystem is diagonal and its elements are the same for the typical initial eigenstates of the full system @xcite . eth may break down for rare eigenstates in the low - energy part of the spectrum @xcite . a subsystem weakly coupled to the rest of the system can be viewed as an open system with its surroundings acting as a reservoir . transitions induced by the coupling to the reservoir govern relaxation dynamics of the open system to the steady state which is described by the ( generalized ) gibbs ensemble @xcite . these transitions satisfy the detailed balance principle in the steady state and the probability distribution over many - body eigenstates of the open system is of the boltzmann form . however , this general scenario does not describe a special case of an open correlated quantum system brought into contact with a reservoir in the vacuum state . in this case particles evaporating from the system never return , and therefore the detailed balance principle does not hold . in the following we refer to such a system as emissive . although particles can only leave the system , its steady state may be populated if there is a binding energy for escaping particles . such a populated steady state is a statistical mixture of multiple eigenstates whose probability distribution is not a priori known . the quantum statistical ensemble for emissive systems defined in this way has never been considered so far . -site lattice ( geometry is shown in the inset ) is shown . the initial state of the system is the maximally occupied pure state with @xmath1 . in sectors with at least @xmath2 stable states the data is fitted with the boltzmann distribution @xmath3 . the temperatures @xmath4 are presented in the table . ] open systems coupled to a vacuum reservoir appear ubiquitously in various fields of natural sciences including surface science @xcite , quantum optics @xcite , nuclear physics @xcite and astrophysics @xcite . in the field of cold atoms an established experimental example is a trapped atomic or molecular gas in a vacuum chamber @xcite . collisions of trapped particles result in internal equilibrium of the gas which alters as particles escape . the loss of particles can be accompanied by cooling . this process of evaporative cooling is a key technological development used to achieve bose - einstein condensation of cold atoms in magneto - optical traps @xcite . the systems studied in experiments with ultracold atoms typically contain from @xmath5 to @xmath6 particles . statistical properties of these systems are usually probed by measuring their local observables . recent experimental progress has made it possible to study smaller systems @xcite . for these systems a direct measurement of the probability distribution over many - body states can be realized . in particular , it would allow to probe statistical properties of the steady state of emissive quantum systems . in this paper we study an emissive quantum system of hard - core bosons on a lattice . we demonstrate that the probability distribution in the steady state is of the boltzmann form in each @xmath0-particle sector . we characterize this steady state by a quantum statistical ensemble and discuss its application to calculating expectation values of observables . we show that the physical mechanism behind the emergence of the boltzmann distribution is rooted in the behaviour of the transition rate . this rate appears to be a regular function of the transition energy though there is no detailed balance with the reservoir . we connect this behaviour with statistical properties of off - diagonal matrix elements of local annihilation operators entering expressions for transition rates . we study a system of hard - core bosons on a two - dimensional lattice coupled to the vacuum reservoir with the hamiltonian @xmath7 . the lattice is described by the hamiltonian @xmath8 with a constraint that each site can be occupied with no more than one particle . here @xmath9 ( @xmath10 ) is a creation ( annihilation ) operator on the site @xmath11 , @xmath12 are hopping amplitudes and the sum runs over the pairs of the nearest - neighbour sites . for calculations we use a lattice of @xmath13 sites ( see inset of figure 1 ) . to ensure non - integrablility of the lattice , the different values are assigned to the hopping amplitudes @xmath12 . these values are taken from the interval between @xmath14 and @xmath15 ( in arbitrary units of energy ) . the vacuum reservoir is described by the hamiltonian @xmath16 with operators @xmath17 ( @xmath18 ) creating ( annihilating ) particles in reservoir modes @xmath19 with energy @xmath20 . the interaction between the lattice and the reservoir is introduced through the hamiltonian @xmath21 , where the unit step function @xmath22 accounts for the binding energy @xmath23 . this binding energy is a tunable parameter in our calculations . we consider the weak coupling limit @xmath24 and assume that escaping particles immediately lose the coherence with the lattice and do not return back to it . we describe the reduced dynamics of the lattice by a master equation @xcite for its density matrix @xmath25 . the density matrix is represented in the basis of eigenstates of @xmath26 . these eigenstates @xmath27 are divided into sectors according to the number of particles @xmath0 in the lattice . corresponding eigenenergies are denoted by @xmath28 . we perform leading order calculations of the transitions between the eigenstates generated by the coupling to the reservoir @xcite . in the absence of special symmetries of the lattice there are two constraints on these transitions : ( i ) @xmath29 which is due to the type of coupling , and ( ii ) @xmath30 which is due to the binding energy for escaping particles . we consider the dynamics of the system at the coarse - grained time scale @xmath31 , where @xmath32 is the characteristic time of the particle loss process . in the weak coupling regime @xmath33 is much larger than the dephasing time , so that only the diagonal elements of the density matrix contribute to dynamics . these elements @xmath34 may be viewed as the probability distribution of the lattice over its eigenstates . the master equation is reduced to the system of the rate equations for this probability distribution @xmath35 where @xmath36 is the transition rate from the state @xmath27 into the state @xmath37 . for transitions allowed by the selection rules the rates are estimated using fermi s golden rule as @xmath38 here the step - like density of states of the reservoir @xmath39 has been used that corresponds to the two - dimensional motion of escaping particles . the particle loss process can be viewed as a sequence of transitions which brings the system from the initial to one of the stable states from which no further transitions can occur ( see supplementary ) . there is the trivial stable state with no particles and occupied stable states which appear due to the constraint on the transition energy . for calculations we use the maximally occupied pure state with @xmath1 as the initial one and choose @xmath40 to maximize the number of the stable states which can be achieved ( see supplementary ) . for the @xmath13-site lattice this number is @xmath41 and the maximal occupation of the stable states is @xmath42 . the probability distribution of the system in a steady state is shown in figure 1 . in each @xmath0-particle sector of the hilbert space ( having at least @xmath2 stable states ) this stationary distribution is found to be of the boltzmann form . parameters of the exponential fit of the data @xmath43 define a quantum statistical ensemble . this ensemble is characterized by the set of temperatures @xmath4 and the probabilities @xmath44 for the system to have @xmath0 particles in the steady state . the partial partition functions @xmath45 are summed over achievable stable states . we assign zero probabilities to the unstable states which are not present in the ensemble . stationary expectation values of the observables are given by a standard expression @xmath46 . we note that the steady state can be characterized by this statistical ensemble for an arbitrary initial state of the system ( see supplementary ) . variation of the initial state only changes parameters @xmath44 and @xmath4 of the distribution . [ [ emergence - of - the - distribution . ] ] emergence of the distribution . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the emergence of the boltzmann distribution in the emissive system can be understood from statistics of the transitions which the system undergoes as particles evaporate . for each sector we calculate the occurrence probabilities of the intermediate states , i.e. the states through which the system evolves from the initial to one of the stable states . these distributions in sectors ( referred to as intermediate ) are shown in figure 2 as functions of the energy of the states . in sectors with @xmath47 the dependence is smooth and fluctuations of the data are less than the size of points . in each sector we consider the part of the spectrum accumulating the probability of @xmath48 and approximate corresponding points by the boltzmann distribution ( [ eqn : ensemble ] ) . for @xmath49 the data includes stable states and the temperatures are close to those in figure 1 . we note that measuring the occurence probabilities of the intermediate states will require collecting statistics from an ensemble of emissive systems . assuming that the spectrum of the system is discrete and not degenerate , a sequence of the intermediate states can be directly determined by measuring the energies of escaping particles . -particle sector is shown . the initial state of the system is the maximally occupied pure state with @xmath1 . for illustrative purposes the rates are multiplied by the factor @xmath50 . the stable states are indicated with empty circles . in each sector the points accumulating @xmath51 of the total probability are fitted with the exponential functions @xmath52 . the temperatures @xmath4 are presented in the table . ] the smooth energy dependence of intermediate probability distributions can be understood from the statistical properties of the transition rates in the system . figure 3 shows @xmath36 plotted as a function of the transition energy @xmath53 . the data for @xmath54 in each sector can be approximated by a regular dependence with state - to - state fluctuations . for an arbitrary initial @xmath0-particle ensemble the probability distribution in @xmath55-particle sector after the loss of a particle is determined by @xmath56 possible state - to - state fluctuations of @xmath57 are convolved with the regular function of the transition energy and almost do not translate into the probability distribution in the following sector . the probability distribution becomes a smooth function of energy after the loss of several particles if transition rates in neighboring sectors are uncorrelated . the analysis of lattices of different sizes ( see supplementary ) shows that the transition rate is always a regular function of the transition energy with state - to - state fluctuations . the fluctuations decrease with the increase of the size of the lattice . we expect that they vanish in the thermodynamic limit , though it is not evident from the available range of lattice sizes . this assumption is supported by the analysis of the transition rates in the thermodynamic limit presented below . [ [ thermodynamic - limit . ] ] thermodynamic limit . + + + + + + + + + + + + + + + + + + + + let us show that the exponential dependence of the transition rate on the transition energy is a universal feature of quantum many - body systems coupled to the vacuum reservoir . for this purpose we consider evaporation in a generic system of @xmath58 particles with well defined energy @xmath59 . temperature @xmath60 and chemical potential @xmath61 of the system are related to the density of many - body states @xmath62 via well known thermodynamic relations @xmath63 , @xmath64 . it is reasonable to assume that the energy @xmath65 of a particle escaping to the vacuum obeys the boltzmann distribution with some temperature @xmath66 ( not necessarily equal to @xmath60 , e.g. because of the adiabatic cooling during evaporation ) . the probability density of the energy of the leaving particle is @xmath67 , where @xmath68 is the density of single particle states in the vacuum reservoir . on the other hand , the same probability can be estimated with the fermi s golden rule as @xmath69 . one can use the first order taylor series expansion of @xmath70 and the thermodynamic relations above to show that @xmath71 . comparing two expressions for @xmath72 , we conclude that @xmath73 . for our specific case of two - dimensional reservoir with @xmath74 the transition rate is a regular exponential function of the transition energy , in accordance with results shown in figure [ fgr : transitions ] . the difference between the inverse temperatures of the gas and the leaving particle is determined by the slope angle of @xmath75 function . we conclude that an exponential dependence of the transition rate on the transition energy is a property of the wide class of emissive quantum many - body systems where escaping particles satisfy the boltzmann distribution . from the states in the @xmath0-particle sector to the states in the @xmath55-particle sector as a function of the transition energy @xmath76 for @xmath77 . the allowed transitions correspond to @xmath78 . for illustrative purposes @xmath79 points are shown for each @xmath0 and the rates are multipled by the factor @xmath50 . ] the regular dependence of the transition rate on the transition energy relates to statistical properties of the off - diagonal matrix elements @xmath80 which enter the expression ( [ eqn : rates ] ) . it is important to note that @xmath10 are non - hermitian annihilation operators and the matrix elements are calculated between states in the different sectors , in contrast to known studies @xcite . we argue that the smooth dependence of these matrix elements on the transition energy is similar to that of eigenstate expectation values in eth . fluctuations of the expectation values vanish as the size of the system increases @xcite . from our scaling analysis we expect that fluctuations of @xmath80 have the same property . we note that an experimental verification of our findings will require measuring the probability distribution of a system under study over its many - body states @xcite . this type of measurements would extend the method of determining temperature from average values of ( local ) observables . while the latter is available both for finite - size and macroscopic systems , the former is only realizable for the systems of finite size . [ [ coolingheating - effect . ] ] cooling / heating effect . + + + + + + + + + + + + + + + + + + + + + + + emissive systems are typically cooled down by the particle loss process since leaving particles accumulate on average more energy than the remaining ones . preparation of the system in a gibbs state with @xmath81 allows us to study the cooling / heating effect in the system and its dependence on the initial temperature @xmath82 . fig . [ fgr:4 ] shows the inverse temperature @xmath83 as a function of @xmath0 for the gibbs states with @xmath84 and different initial temperatures . the energy spectrum of the system is bounded both from below and above , so the initial states with negative @xmath82 can also be considered . in the case of a negative initial temperature the system always cools down as particles escape . in the case of a small positive initial temperature the loss of the first particle leads to the abrupt heating of the system . this can be explained by the shift between the spectra of the neighbouring sectors which leads to the possibility of transitions from the lowest energy state in the @xmath0-particle sector to the excited @xmath55-particle states . for the system under study this condition can be satisfied only for @xmath85 so that no heating is observed below half - filling . on the number of particles for different initial conditions which are defined by the set of initial inverse temperatures @xmath86 ( shown in the inset table ) . the plot demonstrates that system may undergo both heating and cooling process during the evaporation depending on the initial temperature of the system and the filling of the system . ] [ [ conclusions . ] ] conclusions . + + + + + + + + + + + + to summarize , we demonstrate an emergence of the boltzmann distribution in each @xmath0-particle sector of an emissive system of hard - core bosons on a lattice . the steady state of the system is described by a quantum statistical ensemble characterized by a set of temperatures . smooth probability distributions are the consequence of the statistical properties of the transition rates in the system . the transition rates are expressed in terms of the off - diagonal matrix elements of local annihilation operators and depend smoothly on the transition energy . this allowed us to draw the analogy with eth which asserts the smooth dependence of the eigenstate expectation values on the energy . we expect that this feature of transition rates is a generic property of the correlated emissive quantum systems and the formation of the smooth boltzmann distribution can be observed in contemporary experiments with ultracold atoms . the physics demonstrated in this paper is strongly determined by a particular choice of the reservoir which can only absorb particles . this excludes thermalization within the same sector due to virtual processes and makes low - energy populated states stable . authors are greatful to a. e. antipov , p. v. elyutin , p. griins , b. krippa , j. p. f. leblanc , and p. ribeiro for useful discussions . the authors acknowledge a financial support from the rfbr grant 14 - 02 - 01219 and the dynasty foundation . 10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , & _ _ * * , ( ) . , , & _ _ * * , ( ) . _ et al . _ _ _ * * , ( ) . _ et al . _ _ _ * * , ( ) . , , , & _ _ * * , ( ) . , & _ _ * * , ( ) . _ _ * * , ( ) . _ _ * * , ( ) . , & _ _ * * , ( ) . , , & _ _ * * , ( ) . , & _ _ * * , ( ) . , & _ _ * * , ( ) . , & _ _ * * , ( ) . , , & _ _ * * , ( ) . , & _ _ * * , ( ) . , & _ _ * * , ( ) . , & _ _ * * , ( ) . _ _ * * , ( ) . , , & _ _ * * , ( ) . & _ _ * * , ( ) . , , , & _ _ * * , ( ) . , & _ _ * * , ( ) . , , & _ _ * * , ( ) . & _ _ ( , , ) . & _ _ ( , , ) . & _ _ ( , ) . _ et al . _ _ _ * * , ( ) . _ _ ( , , ) . _ et al . _ _ _ * * , ( ) . _ _ * * , ( ) . , & _ _ * * ( ) . _ _ * * , ( ) . , , , & _ _ * * , ( ) . & _ _ * * , ( ) . & _ _ * * ( ) . , , , & _ _ * * , ( ) . _ et al . _ _ _ * * , ( ) . , , , & _ _ * * , ( ) . , , , & _ _ * * , ( ) . & _ _ ( , , ) . & _ _ ( , ) . , , , & _ _ * * , ( ) . _ _ * * , ( ) . , & _ _ * * , ( ) . , & _ _ * * , ( ) . , & _ _ * * , ( ) . , & _ _ * * , ( ) . , , , & _ _ * * , ( ) . * quantum statistical ensemble for emissive correlated systems + ( supplementary materials ) * alexey m. shakirov,@xmath87 yulia e. shchadilova,@xmath88 and alexey n. rubtsov@xmath87 _ @xmath88 russian quantum center , novaya 100 , 143025 skolkovo , moscow region , russia + _ @xmath89 department of physics , lomonosov moscow state university , leninskie gory 1 , 119991 moscow , russia _

Subsets and Splits

No saved queries yet

Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.