Concentration and convergence rates for spectral measures of random matrices (Q1955835): Difference between revisions

The main topic of the paper is the behavior of the spectral measures of large random matrices drawn from several ensembles of interest. Specifically, the authors consider random matrices drawn from Haar measure on the classical Lie groups. In each cases, they estimate the expected Wasserstein distance from the empirical spectral measure to a deterministic references measure, and prove a concentration result for that distance. The bounds obtained are sufficient to obtain almost sure convergence of the empirical spectral measures (with rates in the Wasserstein distance) in all cases.