Edge universality of separable covariance matrices
Abstract
In this paper, we prove the edge universality of largest eigenvalues for separable covariance matrices of the form . Here is an random matrix with , where are random variables with zero mean and unit variance, and and are respectively and deterministic non-negative definite symmetric (or Hermitian) matrices. We consider the high-dimensional case, i.e. as . Assuming and some mild conditions on and , we prove that the limiting distribution of the largest eigenvalue of coincide with that of the corresponding Gaussian ensemble (i.e. the with being an Gaussian matrix) as long as we have , which is a sharp moment condition for edge universality. If we take , then becomes the normal sample covariance matrix and the edge universality holds true without the vanishing third moment condition. So far, this is the strongest edge universality result for sample covariance matrices with correlated data (i.e. non-diagonal ) and heavy tails, which improves the previous results in \cite{BPZ1,LS} (assuming high moments and diagonal ), \cite{Anisotropic} (assuming high moments) and \cite{DY} (assuming diagonal ).
Keywords
Cite
@article{arxiv.1809.04572,
title = {Edge universality of separable covariance matrices},
author = {Fan Yang},
journal= {arXiv preprint arXiv:1809.04572},
year = {2019}
}
Comments
58 pages, 1 figure