Two-level correlation function of critical random-matrix ensembles

TL;DR

This study numerically analyzes the two-level correlation function at criticality in disordered models with long-range hopping, revealing universal forms dependent on symmetry class and dimensionality.

Contribution

It provides a detailed numerical characterization of the two-level correlation function and spectral properties of critical random-matrix ensembles across different dimensions and symmetries.

Findings

Correlation function follows a universal form involving error function or exponential decay.

Level number variance and spectral compressibility are characterized at criticality.

Results depend on symmetry class and spatial dimensionality.

Abstract

The two-level correlation function $R_{d,\beta}(s)$ of $d$-dimensional disordered models ($d=1$, 2, and 3) with long-range random-hopping amplitudes is investigated numerically at criticality. We focus on models with orthogonal ($\beta=1$) or unitary ($\beta=2$) symmetry in the strong ($b^d \ll 1$) coupling regime, where the parameter $b^{-d}$ plays the role of the coupling constant of the model. It is found that $R_{d,\beta}(s)$ is of the form $R_{d,\beta}(s)=1+\delta(s)-F_{\beta}(s^{\beta}/b^{d\beta})$, where $F_{1}(x)=\text{erfc}(a_{d,\beta} x)$ and $F_{2}(x)=\exp (-a_{d,\beta} x^2)$, with $a_{d,\beta}$ being a numerical coefficient depending on the dimensionality and the universality class. Finally, the level number variance and the spectral compressibility are also considerded.