English

A large deviation principle for Wigner matrices without Gaussian tails

Probability 2014-10-29 v2

Abstract

We consider n×nn\times n Hermitian matrices with i.i.d. entries XijX_{ij} whose tail probabilities P(Xijt)\mathbb {P}(|X_{ij}|\geq t) behave like eatαe^{-at^{\alpha}} for some a>0a>0 and α(0,2)\alpha \in(0,2). We establish a large deviation principle for the empirical spectral measure of X/nX/\sqrt{n} with speed n1+α/2n^{1+\alpha /2} with a good rate function J(μ)J(\mu) that is finite only if μ\mu is of the form μ=μscν\mu=\mu_{\mathrm{sc}}\boxplus\nu for some probability measure ν\nu on R\mathbb {R}, where \boxplus denotes the free convolution and μsc\mu_{\mathrm{sc}} is Wigner's semicircle law. We obtain explicit expressions for J(μscν)J(\mu_{\mathrm{sc}}\boxplus\nu) in terms of the α\alphath moment of ν\nu. The proof is based on the analysis of large deviations for the empirical distribution of very sparse random rooted networks.

Keywords

Cite

@article{arxiv.1207.5570,
  title  = {A large deviation principle for Wigner matrices without Gaussian tails},
  author = {Charles Bordenave and Pietro Caputo},
  journal= {arXiv preprint arXiv:1207.5570},
  year   = {2014}
}

Comments

Published in at http://dx.doi.org/10.1214/13-AOP866 the Annals of Probability (http://www.imstat.org/aop/) by the Institute of Mathematical Statistics (http://www.imstat.org)

R2 v1 2026-06-21T21:40:24.641Z