Proof methods in random matrix theory (Q6158178)

This article is about the two well-known methods to find limit laws from the so-called empirical spectral distributions (EMD) coming from random matrices. Let \(X\) be a random matrix with dimension \(n\) where all the entries \(X(i,j)\) are independent random variables. One important question is to study the measures concentrated on the eigenvalues (EMD) of models such as \(A=X+X^{T}\) or \(B=XX^{T}\), where \(T\) represents the transpose operation. After multiplying \(X\) by \(\frac{1}{\sqrt{n}}\), and under general assumption on the random variables \(X(i,j)\), the EMDs of \(A\) and \(B\) converge to the so-called semicircle law and the Marchenko-Pastur law, respectively. To obtain these results, one can use the method of moments or the Stieltjes transform method. The first one is to identify the moments of the limit distribution, and the second consists in finding the Stieltjes transform (ST) of the distribution, with also helps to identify the limit distribution. The moment method has a flavor more in combinatorics, and the ST method finds differential equations that characterize the ST. In this paper the authors give a detailed exposition of both methods applied to both models, \(A\) and \(B\).