Spectrum of non-Hermitian heavy tailed random matrices (Q644762)

Let \((X_{jk} )_{1\leq{}j,k\leq{}n}\) be a matrix of i.i.d. complex random variables with cumulative distribution function \(F\). It is already known that, if \(F\) has finite positive variance \(\sigma^2\), then, for every \(z\in\mathbb C\), there exists a probability measure \({\mathcal Q}_{\sigma,z}\) on \([0,\infty)\) depending only on \(\sigma\) and \(z\), with explicit Cauchy-Stieltjes transform, such that a.s. (almost surely) \[ \nu_{{1\over\sqrt{n}}X-zI}\;\mathop{\rightsquigarrow}\limits_{n\to\infty}\;{\mathcal Q}_{\sigma,z} \] where \(\rightsquigarrow\) denotes the weak convergence of probability measures. The paper formulates generalization to the case when \(F\) does not have a finite second moment. Let \((X_{jk} )_{j,k\geq{}1}\) be i.i.d. complex random variables such that \(\left| X_{jk} \right|\) is in the domain of attraction of an \(\alpha\)-stable law, with \(0 < \alpha < 2\). The main result presented in the paper is a heavy tailed counterpart of Girko's circular law. Namely, under some additional smoothness assumptions on the law of \(X_{jk}\), the authors show that there exist a deterministic sequence \(a_n \sim n^{1/\alpha}\) and a probability measure \(\mu_\alpha\) on \(\mathbb C\) depending only on \(\alpha\) such that, with probability one, the empirical distribution of the eigenvalues of the rescaled matrix \(\left(\alpha_n^{-1} X_{jk}\right )_{1\leq{}j,k\leq{}n}\) converges weakly to \(\mu_\alpha\) as \(n\to\infty\). The presented approach combines Aldous \& Steele's objective method with Girko's Hermitization using logarithmic potentials. The underlying limiting object is defined on a bipartized version of Aldous's Poisson weighted infinite tree. Recursive relations on the tree provide some properties of \(\mu_\alpha\). In contrast with the Hermitian case, the authors find that \(\mu_\alpha\) is not heavy tailed.