The local relaxation flow approach to universality of the local statistics for random matrices (Q2428944)

This article presents in a general manner the technique of local relaxation flows for proving universality of the local eigenvalue statistics in the bulk of large random matrices. The introduction contains a concise and pedagogical overview of the problem, of its solution using Dyson's Brownian motion and of its latest refinement, the local relaxation flow. The key ideas and crucial steps are clarified in non-technical terms. A detailed exposition and analysis of the relaxation flow follows, which culminates in the proof of convergence of the local statistics to the equilibrium measure for times \(t\gg N^{-2\epsilon}\). In order to obtain universality of local eigenvalue statistics for random matrix ensembles -- general Wigner and sample covariance matrices -- the authors review the reverse heat flow argument, which allows for the removing of the time restriction. For sample covariance matrices, the article also provides a proof of the local Marchenko-Pastur law on the density of eigenvalues. The article is very much self-contained.