English

Central limit theorem for eigenvectors of heavy tailed matrices

Probability 2014-06-02 v3

Abstract

We consider the eigenvectors of symmetric matrices with independent heavy tailed entries, such as matrices with entries in the domain of attraction of α\alpha-stable laws, or adjacencymatrices of Erdos-Renyi graphs. We denote by U=[uij]U=[u_{ij}] the eigenvectors matrix (corresponding to increasing eigenvalues) and prove that the bivariate process Bs,tn:=n1/21ins,1jnt(uij2n1),B^n_{s,t}:=n^{-1/2}\sum_{1\le i\le ns, 1\le j\le nt}(|u_{ij}|^2 -n^{-1}), indexed by s,t[0,1]s,t\in [0,1], converges in law to a non trivial Gaussian process. An interesting part of this result is the n1/2n^{-1/2} rescaling, proving that from this point of view, the eigenvectors matrix UU behaves more like a permutation matrix (as it was proved by Chapuy that for UU a permutation matrix, n1/2n^{-1/2} is the right scaling) than like a Haar-distributed orthogonal or unitary matrix (as it was proved by Rouault and Donati-Martin that for UU such a matrix, the right scaling is 11).

Keywords

Cite

@article{arxiv.1310.7435,
  title  = {Central limit theorem for eigenvectors of heavy tailed matrices},
  author = {Florent Benaych-Georges and Alice Guionnet},
  journal= {arXiv preprint arXiv:1310.7435},
  year   = {2014}
}

Comments

31 pages. In this version, we added some details to several proofs

R2 v1 2026-06-22T01:55:26.778Z