English

Random matrix representations of critical statistics

Disordered Systems and Neural Networks 2009-11-04 v1 Statistical Mechanics

Abstract

We consider two random matrix ensembles which are relevant for describing critical spectral statistics in systems with multifractal eigenfunction statistics. One of them is the Gaussian non-invariant ensemble which eigenfunction statistics is multifractal, while the other is the invariant random matrix ensemble with a shallow, log-square confinement potential. We demonstrate a close correspondence between the spectral as well as eigenfuncton statistics of these random matrix ensembles and those of the random tight-binding Hamiltonian in the point of the Anderson localization transition in three dimensions. Finally we present a simple field theory in 1+1 dimensions which reproduces level statistics of both of these random matrix models and the classical Wigner-Dyson spectral statistics in the framework of the unified formalism of Luttinger liquid. We show that the (equal-time) density correlations in both random matrix models correspond to the finite-temperature density correlations of the Luttinger liquid. We show that spectral correlations in the invariant ensemble with log-square confinement correspond to a Luttinger liquid in the 1+1 curved space-time with the event horizon, similar to the phonon density correlations in the sonic analogy of Hawking radiation in black holes.

Keywords

Cite

@article{arxiv.0911.0615,
  title  = {Random matrix representations of critical statistics},
  author = {V. E. Kravtsov},
  journal= {arXiv preprint arXiv:0911.0615},
  year   = {2009}
}

Comments

21 pages, 4 figures, Chapter in the Handbook on Random Matrix Theory

R2 v1 2026-06-21T14:07:01.949Z