6533b862fe1ef96bd12c68f7

RESEARCH PRODUCT

Spectral properties of random non-self-adjoint operators

subject

description

In this thesis we are interested in the spectral properties of random non-self-adjoint operators. Weare going to consider primarily the case of small random perturbations of the following two types of operators: 1. a class of non-self-adjoint h-differential operators Ph, introduced by M. Hager [32], in the semiclassical limit (h→0); 2. large Jordan block matrices as the dimension of the matrix gets large (N→∞). In case 1 we are going to consider the operator Ph subject to small Gaussian random perturbations. We let the perturbation coupling constant δ be e (-1/Ch) ≤ δ ⩽ h(k), for constants C, k > 0 suitably large. Let ∑ be the closure of the range of the principal symbol. Previous results on the same model by M. Hager [32], W. Bordeaux-Montrieux [4] and J. Sjöstrand [67] show that if δ ⪢ e(-1/Ch) there is, with a probability close to 1, a Weyl law for the eigenvalues in the interior of the pseudospectrumup to a distance ⪢ (-h ln δ h) 2/3 to the boundary of ∑. We will study the one- and two-point intensity measure of the random point process of eigenvalues of the randomly perturbed operator and prove h-asymptotic formulae for the respective Lebesgue densities describing the one- and two-point behavior of the eigenvalues in ∑. Using the density of the one-point intensity measure, we will give a complete description of the average eigenvalue density in ∑ describing as well the behavior of the eigenvalues at the pseudospectral boundary. We will show that there are three distinct regions of different spectral behavior in ∑. The interior of the of the pseudospectrum is solely governed by a Weyl law, close to its boundary there is a strong spectral accumulation given by a tunneling effect followed by a region where the density decays rapidly. Using the h-asymptotic formula for density of the two-point intensity measure we will show that two eigenvalues of randomly perturbed operator in the interior of ∑ exhibit close range repulsion and long range decoupling. In case 2 we will consider large Jordan block matrices subject to small Gaussian random perturbations. A result by E.B. Davies and M. Hager [16] shows that as the dimension of the matrix gets large, with probability close to 1, most of the eigenvalues are close to a circle. They, however, only state a logarithmic upper bound on the number of eigenvalues in the interior of that circle. We study the expected eigenvalue density of the perturbed Jordan block in the interior of thatcircle and give a precise asymptotic description. Furthermore, we show that the leading contribution of the density is given by the Lebesgue density of the volume form induced by the Poincarémetric on the disc D(0, 1).