Large deviation function for the number of eigenvalues of sparse random graphs inside an interval

TL;DR

This paper introduces a method to compute the large deviation rate function for the number of eigenvalues in an interval of sparse random matrices, revealing asymmetries related to eigenstate localization.

Contribution

The paper develops a general approach to exactly determine the large deviation rate function for eigenvalue counts in sparse matrices, applied to Erd"os-Rényi graphs and Anderson models on regular graphs.

Findings

Rate function is asymmetric, reflecting eigenstate localization.

Number variance scales linearly with system size for any disorder.

Good agreement between theory and numerical simulations.

Abstract

We present a general method to obtain the exact rate function $\Psi_{[a,b]}(k)$ controlling the large deviation probability $\text{Prob}[\mathcal{I}_N[a,b]=kN] \asymp e^{-N\Psi_{[a,b]}(k)}$ that a $N \times N$ sparse random matrix has $\mathcal{I}_N[a,b]=kN$ eigenvalues inside the interval $[a,b]$. The method is applied to study the eigenvalue statistics in two distinct examples: (i) the shifted index number of eigenvalues for an ensemble of Erd\"os-R\'enyi graphs and (ii) the number of eigenvalues within a bounded region of the spectrum for the Anderson model on regular random graphs. A salient feature of the rate function in both cases is that, unlike rotationally invariant random matrices, it is asymmetric with respect to its minimum. The asymmetric character depends on the disorder in a way that is compatible with the distinct eigenvalue statistics corresponding to localized and delocalized eigenstates. The results also show that the number variance $\sigma_{N}^{2}$ for the Anderson model on a regular graph scales as $\sigma_{N}^{2} \propto N$ ($N \gg 1$) for any nonzero disorder, which is consistent with the absence of level-repulsion in the extended phase. Our theoretical findings are thoroughly compared to numerical diagonalization in both cases, showing a reasonable good agreement.