Searcharxiv⌕ Search

arXiv subjects

Gordon Blower

Publications and source records attributed to Gordon Blower.

29 records · Page 2Linked to original sources

Concentration of the invariant measures for the periodic Zakharov, KdV, NLS and Gross--Piatevskii equations in 1D and 2D

This paper concerns Gibbs measures $ν$ for some nonlinear PDE over the $D$-torus ${\bf T}^D$. The Hamiltonian $H=\int_{{\bf T}^D} \Vert\nabla u\Vert^2 - \int_{{\bf T}^D} \vert u\vert^p$ has canonical equations with solutions in $Ω_N=\{ u\in L^2({\bf T}^D) :\int \vert u\vert^2\leq N\}$. For $D=1$ and $2\leq p<6$, $Ω_N$ supports the Gibbs measure $ν(du)=Z^{-1}e^{-H(u)}\prod_{x\in {\bf T}} du(x)$ which is normalized and formally invariant under the flow generated by the PDE. The paper proves that $(Ω_N, \Vert\cdot\Vert_{L^2}, ν)$ is a metric probability space of finite diameter that satisfies the logarithmic Sobolev inequalities for the periodic $KdV$, the focussing cubic nonlinear Schrödinger equation and the periodic Zakharov system. For suitable subset of $Ω_N$, a logarithmic Sobolev inequality also holds in the critical case $p=6$. For $D=2$, the Gross--Piatevskii equation has $H=\int_{{\bf T}^2} \Vert\nabla u\Vert^2-\int_{{\bf T}^2} (V\ast \vert u\vert^2 ) \vert u\vert^2$, for a suitable bounded interaction potential $V$ and the Gibbs measure $ν$ lies on a metric probability space $(Ω, \Vert\cdot\Vert_{H^{-s}}, ν)$ which satisfies $LSI$. In the above cases, $(Ω, d, ν)$ is the limit in $L^2$ transportation distance of finite-dimensional $(Ω_n, \Vert \cdot \Vert,ν_n)$ given by Fourier sums.

math.PR↗

Hill's Spectral Curves and the Invariant Measure of the Periodic KdV Equation

This paper analyses the periodic spectrum of Schrödinger's equation $-f''+qf=λf$ when the potential is real, periodic, random and subject to the invariant measure $ν_N^β$ of the periodic KdV equation. This $ν_N^β$ is the modified canonical ensemble, as given by Bourgain ({Comm. Math. Phys.} {166} (1994), 1--26), and $ν_N^β$ satisfies a logarithmic Sobolev inequality. Associated concentration inequalities control the fluctuations of the periodic eigenvalues $(λ_n)$. For $β, N>0$ small, there exists a set of positive $ν_N^β$ measure such that $(\pm \sqrt{2(λ_{2n}+λ_{2n-1})})_{n=0}^\infty$ gives a sampling sequence for Paley--Wiener space $PW(π)$ and the reproducing kernels give a Riesz basis. Let $(μ_j)_{j=1}^\infty$ be the tied spectrum; then $(2\sqrt{μ_j}-j)$ belongs to a Hilbert cube in $\ell^2$ and is distributed according to a measure that satisfies Gaussian concentration for Lipschitz functions. The sampling sequence $(\sqrt{μ_j})_{j=1}^\infty$ arises from a divisor on the spectral curve, which is hyperelliptic of infinite genus. The linear statistics $\sum_j g(\sqrt{λ_{2j}})$ with test function $g\in PW(π)$ satisfy Gaussian concentration inequalities.

math.SP↗

Logarithmic Sobolev inequalities and spectral concentration for the cubic Schrödinger equation

The nonlinear Schrödinger equation NLSE(p, β), -iu_t=-u_{xx}+β| u|^{p-2} u=0, arises from a Hamiltonian on infinite-dimensional phase space \Lp^2(\mT). For p\leq 6, Bourgain (Comm. Math. Phys. 166 (1994), 1--26) has shown that there exists a Gibbs measure μ^β_N on balls Ω_N= {ϕ\in \Lp^2(\mT) : | ϕ|^2_{\Lp^2} \leq N} in phase space such that the Cauchy problem for NLSE(p,β) is well posed on the support of μ^β_N, and that μ^β_N is invariant under the flow. This paper shows that μ^β_N satisfies a logarithmic Sobolev inequality for the focussing case β<0 and 2\leq p\leq 4 on Ω_N for all N>0; also μ^β satisfies a restricted LSI for 4\leq p\leq 6 on compact subsets of Ω_N determined by Hölder norms. Hence for p=4, the spectral data of the periodic Dirac operator in \Lp^2(\mT; \mC^2) with random potential ϕsubject to μ^β_N are concentrated near to their mean values. The paper concludes with a similar result for the spectral data of Hill's equation when the potential is random and subject to the Gibbs measure of KdV.

math.SP↗

On tau functions for orthogonal polynomials and matrix models

Let v be a real polynomial of even degree, and let ρbe the equilibrium probability measure for v with support S; so that v(x)\geq 2\int \log |x-y| ρ(dy)+C_v for some constant C_v with support S. Then S is the union of finitely many bounded intervals with endpoints delta_j, and ρis given by an algebrais weight w(x) on S. The system of orthogonal polynomials for w gives rise to the Magnus--Schlesinger differential equations. This paper identifies the tau function of this system with the Hankel determinant det[\in x^{j+k}ρ(dx)] of ρ. The solutions of the Magnus--Schlesinger equations are realised by a linear system, which is used to compute the tau function in terms of a Gelfand--Levitan equaiton. The tau function is associated with a potential q and a scattering problem for the Schrodinger operator with potential q. For some algebro-geometric potentials, the paper solves the scattering problem in terms of linear systems. The theory extends naturally to elliptic curves and resolves the case where S has exactly two intervals.

math.CA↗

A logarithmic Sobolev inequality for the invariant measure of the periodic Korteweg--de Vries equation

The periodic KdV equation u_t=u_{xxx}+βuu_x arises from a Hamiltonian system with infinite-dimensional phase space L^2(T). Bourgain has shown that there exists a Gibbs measure νon balls \{ϕ:\VertΦ\Vert^2_{L^2}\leq N\} in the phase space such that the Cauchy problem for KdV is well posed on the support of ν, and νis invariant under the KdV flow. This paper shows that νsatisfies a logarithmic Sobolev inequality. The stationary points of the Hamiltonian on spheres are found in terms of elliptic functions, and they are shown to be linearly stable. The paper also presents logarithmic Sobolev inequalities for the modified periodic KdV equation and the cubic nonlinear Schrödinger equation, for small values of N.

math.AP↗

On linear systems and tau functions associated with Lame's equation and Painleve's equation VI

Painleve's transcendental differential equation P_{VI} may be expressed as the consistency condition for a pair of linear differential equations with 2 by 2 matrix coefficients with rational entries. By a construction due to Tracy and Widom, this linear system is associated with certain kernels which give trace class operators on Hilbert space. This paper expresses such operators in terms of the Hankel operators Γ_ϕof linear systems which are realised in terms of the Laurent coefficients of the solutions of the differential equations. Let P_{(t infty)}:L^2(0, \infty)\to L^2(t, \infty) be the orthogonal projection. For such, the Fredholm determinant τ(t)=det (I-P_{(t, \infty)}Γ_ϕ) defines the tau function, which is here expressed in terms of the solutions of a matrix Gelfand--Levitan equation. For suitable paramters, solutions of the hypergeometric equation give a linear system with similar properties. For meromorphic transfer functions \hatϕthat have poles on an arithmetic progression, the corresponding Hankel operator has a simple form with respect to an exponential basis in L^2(0, \infty); so τ(t) can be expressed in terms of finite determinants. This applies to elliptic functions of the second kind, such as satisfy Lame's equation with \ell=1.

math.FA↗

Linear systems and determinantal random point fields

Tracy and Widom showed that fundamentally important kernels in random matrix theory arise from differential equations with rational coefficients. More generally, this paper considers symmetric Hamiltonian systems abd determines the properties of kernels that arise from them. The inverse spectral problem for self-adjoint Hankel operators gives a sufficient condition for a self-adjoint operator to be the Hankel operator on $L^2(0, \infty)$ from a linear system in continuous time; thus this paper expresses certain kernels as squares of Hankel operators. For a suitable linear system $(-A,B,C)$ with one dimensional input and output spaces, there exists a Hankel operator $Γ$ with kernel $ϕ_{(x)}(s+t)=Ce^{-(2x+s+t)A}B$ such that $\det (I+(z-1)ΓΓ^\dagger)$ is the generating function of a determinantal random point field.

math.FA↗

Hankel operators that commute with second-order differential operators

Suppose that $Γ$ is a continuous and self-adjoint Hankel operator on $L^2(0, \infty)$ and that $Lf=-(d/dx(a(x)df/dx))+b(x)f(x)$ with $a(0)=0$. If $a$ and $b$ are both quadratic, hyperbolic or trigonometric functions, and $ϕ$ satisfies a suitable form of Gauss's hypergeometric equation, or the confluent hypergeometric equation, then $LΓ=ΓL$. The paper catalogues the commuting pairs $Γ$ and $L$, including important cases in random matrix theory. There are also results proving rapid decay of the singular numbers of Hankel integral operators with kernels that are analytic and of exponential decay in the right half plane.

math.FA↗

Integrable operators and the squares of Hankel operators

Integrable operators arise in random matrix theory, where they describe the asymptotic eigenvalue distributions of large self-adjoint random matrices from the generalized unitary ensembles. This paper gives sufficient conditions for an integrable operator to be the square of a Hankel operator, and applies the condition to the Airy, associated Laguerre, modified Besses and Whittaker functions.

math.FA↗

Operators associated with the soft and hard spectral edges of unitary ensembles

Using Hankel operators and shift-invariant subspaces on Hilbert space, this paper develops the theory of the operators associated with soft and hard edges of eigenvalue distributions of random matrices. Tracy and Widom introduced a projection operator $W$ to describe the soft edge of the spectrum of the Gaussian unitary ensemble. The subspace $WL^2$ is simply invariant under the translation semigroup $e^{itD}$ $(t\geq 0)$ and invariant under the Schrödinger semigroup $e^{it(D^2+x)}$ $(t\geq 0)$; these properties characterize $WL^2$ via Beurling's theorem. The Jacobi ensemble of random matrices has positive eigenvalues which tend to accumulate near to the hard edge at zero. This paper identifies a pair of unitary groups that satisfy the von Neumann--Weyl anti-commutation relations and leave invariant certain subspaces of $L^2(0,\infty)$ which are invariant for operators with Jacobi kernels. Such Tracy--Widom operators are reproducing kernels for weighted Hardy spaces, known as Sonine spaces. Periodic solutions of Hill's equation give a new family of Tracy--Widom type operators.

math.FA↗

Concentration inequalities on product spaces with applications to Markov processes

For a stochastic process with state space some Polish space, this paper gives sufficient conditions on the initial and conditional distributions for the joint law to satisfy Gaussian concentration inequalities, transportation inequalities and also logarithmic Sobolev inequalities in the case of the Euclidean space. In several cases, the obtained constants are of optimal order of growth with respect to the number of variables, or are independent of this number. These results extend results known for mutually independent variables to weakly dependent variables under Dobrushin-Shlosman type conditions.

math.PR↗