Speaker: Sourav Chatterjee
Affiliation: University of California, Berkeley, Dept. of Statistics
Title: A measure concentration technique, with an application to random matrices
Time and Place: Friday, 11/11, 12:35-1:30pm, Milner 317.
Abstract: This talk is about a novel approach to measure concentration that uses couplings and rates of convergence of Markov chains to obtain concentration inequalities. A concentration result for the empirical distribution of the eigenvalues of sums of random hermitian matrices will be derived as an application of the method. This also gives an example of concentration for discontinuous functions, which can be a significant area of application for the new technique.