The spectrum of some nonlinear random matrices

Modern data science often requires one to consider “nonlinear random matrices,” a broad term for random-matrix models whose construction involves a nonlinear function applied entrywise. Such models are typically far from classical random matrix theory, and in principle entrywise nonlinearities can affect the eigenvalues in a complicated way. However, recent years have seen a number of results on nonlinear models whose spectrum is surprisingly simple. We give one such result, emphasizing general random-matrix techniques like free probability and orthogonal polynomials. Joint work with Sofiia Dubova, Yue M. Lu, and Horng-Tzer Yau.

Friday, Feb. 2nd at 12pm, with lunch, lounge at CMSA (20 Garden Street). Also by Zoom: https://harvard.zoom.us/j/92410768363