The spectrum of heavy tailed random matrices (Q956630)

Let \(X_N\) be an \(N \times N\) random, symmetric matrix with independent, equidistributed entries. If the law \(P\) of the entries has a finite second moment then \textit{E. Wigner} [Ann. Math 2, 325--327 (1958; Zbl 0085.13203)] showed that the empirical distribution of the eigenvalues of \(X_N\) normalized by \(\sqrt{N}\) converges weakly almost surely to the semi-circle distribution. Using classical techniques this paper investigates the situation where \(P\) is in the domain of attraction of an \(\alpha\)-stable law, \(\alpha \in (0,2).\) In this case, if the eigenvalues are re-normalized by a constant \(a_N\) of order \(N^{\frac{1}{\alpha}}\), then their empirical distribution converges to a law \(\mu_{\alpha}\) which depends only on \(\alpha\). The paper discusses three different characterizations of \(\mu_{\alpha}\) and establishes that \(\mu_{\alpha} \) is a heavy tailed probability measure that is symmetric with unbounded support. The measure \(\mu_{\alpha}\) has a smooth density, except possibly on a compact set of capacity zero.