Edge universality of separable covariance matrices (Q2279318)

Let \(X=(x_{ij})\) be a \(n\times N\) random matrix with real entries \(x_{ij}=N^{-1/2} q_{ij}\), \(i=\overline{1,n}\), \(j=\overline{1,N}\), where \(q_{ij}\) are {i.i.d.} random variables satisfying \[ \mathbb{E}q_{11}=0 \text{ and } \mathbb{E}|q_{11}|^2=1. \] Let \(A\in\mathcal{M}_{n\times n}\) and \(B\in\mathcal{M}_{N\times N}\) be two deterministic non-negative definite symmetric matrices. The main goal of the paper (see Theorem 2.7 and Remark 2.9) is to prove the edge universality of largest eigenvalues for covariance matrices of the form \[ Q=A^{1/2}XBX^*A^{1/2}. \]