Local laws for non-Hermitian random matrices and their products

Random Matrices-Theory and Applications. 2020. Vol. 9. No. 4. P. 2150004.

Гётце Ф., Наумов А. А., Тихомиров А. Н.

We consider products of independent \(n \times n\) non-Hermitian random matrices \(\X^{(1)}, \ldots, \X^{(m)}\). Assume that their entries, \(X_{jk}^{(q)},  1 \le j,k \le n, q = 1, \ldots, m\), are independent identically distributed random variables with zero mean, unit variance. G\"otze -- Tikhomirov~\cite{GotTikh2011} and O'Rourke--Sochnikov~\cite{Soshnikov2011} proved that under these assumptions the empirical spectral distribution (ESD) of \(X^{(1)} \cdots X^{(m)}\) converges to the limiting distribution which coincides with the distribution of the \(m\)-th power of random variable uniformly distributed in the unit circle.  In the current paper we provide a local vesion of this result. More precisely, assuming additionally that  \(\E |X_{11}^{(q)}|^{4+\delta} < \infty\) for some \(\delta > 0\), we prove that ESD of \(X^{(1)} \cdots X^{(m)}\) converges to the limiting distribution  on the optimal scale up to \(n^{-1+2a}, 0 < a < 1/2\) (up to some logarithmic factor).  Our results generalize the recent results of Bourgade--Yau--Yin~\cite{Bourgade2014a}, Tao--Vu~\cite{TaoVu2015a} and Nemish~\cite{nemish2017}.  We also give further development of Stein's type approach to estimate the the Stieltjes transform of ESD.

Ключевые слова: logarithmic potential theory, random matrices, products of random matrices, local circle law, Stein's method

ПУБЛИКАЦИЯ ПОДГОТОВЛЕНА ПО РЕЗУЛЬТАТАМ ПРОЕКТА: Анализ неопределенности в моделях большой размерности (2020)