English

How many eigenvalues of a Gaussian random matrix are positive?

Statistical Mechanics 2015-03-17 v1 Mathematical Physics math.MP

Abstract

We study the probability distribution of the index N+{\mathcal N}_+, i.e., the number of positive eigenvalues of an N×NN\times N Gaussian random matrix. We show analytically that, for large NN and large N+\mathcal{N}_+ with the fraction 0c=N+/N10\le c=\mathcal{N}_+/N\le 1 of positive eigenvalues fixed, the index distribution P(N+=cN,N)exp[βN2Φ(c)]\mathcal{P}({\mathcal N}_+=cN,N)\sim\exp[-\beta N^2 \Phi(c)] where β\beta is the Dyson index characterizing the Gaussian ensemble. The associated large deviation rate function Φ(c)\Phi(c) is computed explicitly for all 0c10\leq c \leq 1. It is independent of β\beta and displays a quadratic form modulated by a logarithmic singularity around c=1/2c=1/2. As a consequence, the distribution of the index has a Gaussian form near the peak, but with a variance Δ(N)\Delta(N) of index fluctuations growing as Δ(N)logN/βπ2\Delta(N)\sim \log N/\beta\pi^2 for large NN. For β=2\beta=2, this result is independently confirmed against an exact finite NN formula, yielding Δ(N)=logN/2π2+C+O(N1)\Delta(N)= \log N/2\pi^2 +C+\mathcal{O}(N^{-1}) for large NN, where the constant CC has the nontrivial value C=(γ+1+3log2)/2π20.185248...C=(\gamma+1+3\log 2)/2\pi^2\simeq 0.185248... and γ=0.5772...\gamma=0.5772... is the Euler constant. We also determine for large NN the probability that the interval [ζ1,ζ2][\zeta_1,\zeta_2] is free of eigenvalues. Part of these results have been announced in a recent letter [\textit{Phys. Rev. Lett.} {\bf 103}, 220603 (2009)].

Keywords

Cite

@article{arxiv.1012.1107,
  title  = {How many eigenvalues of a Gaussian random matrix are positive?},
  author = {Satya N. Majumdar and Céline Nadal and Antonello Scardicchio and Pierpaolo Vivo},
  journal= {arXiv preprint arXiv:1012.1107},
  year   = {2015}
}

Comments

25 pages, 6 figures

R2 v1 2026-06-21T16:53:54.983Z