[Paper Review] Universality in Complex Wishart ensembles: The 2 cut case

M. Y. Mo|ArXiv.org|Sep 22, 2008

Random Matrices and Applications32 references3 citations

TL;DR

This paper studies the universality of eigenvalue distributions in complex Wishart ensembles with two distinct eigenvalues in the covariance matrix, deriving conditions under which the limiting spectral density splits into two disjoint intervals. Using Riemann-Hilbert analysis, it establishes that the local eigenvalue statistics converge to the sine kernel in the bulk and the Tracy-Widom distribution at the edge, confirming universal behavior in the 2-cut case.

ABSTRACT

We studied the universality of Wishart ensembles whose covariance matrix has 2 distinct eigenvalues. We studied the asymptotic limit when the number of both eigenvalues goes to infinity and obtained universality results. In this case, the limiting eigenvalue distribution can be supported on 1 or 2 disjoint intervals. We obtained a necessary and sufficient condition on the parameters such that the limiting distribution is supported on 2 disjoint intervals and have computed the eigenvalue density in the limit. Furthermore, by using Riemann-Hilbert analysis, we have shown that under proper rescaling of the eigenvalues, the limiting correlation kernel is given by the sine kernel and the Airy kernel in the bulk and the edge of the spectrum respectively. As a consequence, the behavior of the largest eigenvalue in this model is described by the Tracy-Widom distribution.

Motivation & Objective

• To determine the conditions under which the limiting eigenvalue distribution of complex Wishart matrices with two distinct covariance eigenvalues splits into two disjoint intervals.
• To derive an explicit formula for the limiting spectral density in the 2-cut regime.
• To establish universality of local eigenvalue statistics—specifically, convergence to the sine kernel in the bulk and the Tracy-Widom distribution at the edge—under appropriate scaling.
• To rigorously analyze the asymptotic behavior of eigenvalue correlation kernels using Riemann-Hilbert techniques.

Proposed method

• Derives the joint probability density function of eigenvalues using the Harish-Chandra/Itzykson-Zuber formula for unitary integrals.
• Applies asymptotic analysis to the eigenvalue density under the scaling limits $ N/M \to c $, $ N_1/N \to \beta $, with $ 0 < c, \beta < 1 $, and $ N, M \to \infty $.
• Uses an algebraic equation involving parameters $ a, \beta, c $ to characterize the support of the limiting spectral measure.
• Applies Riemann-Hilbert analysis to the orthogonal polynomial system associated with the eigenvalue density, enabling asymptotic evaluation of correlation kernels.
• Establishes convergence of the rescaled correlation kernel to the sine kernel in the bulk and the Airy kernel at the edge via detailed asymptotic expansions.
• Uses the Tracy-Widom distribution to describe the limiting behavior of the largest eigenvalue in the 2-cut case.

Experimental results

Research questions

• RQ1 Under what conditions on $ a, \beta, c $ does the limiting eigenvalue distribution of the complex Wishart ensemble split into two disjoint intervals?
• RQ2 What is the explicit form of the limiting spectral density when the support consists of two intervals?
• RQ3 Does the local eigenvalue statistics in the bulk and at the edge of the spectrum converge to universal forms—specifically, the sine kernel and the Tracy-Widom distribution—under the given scaling?
• RQ4 How does the Riemann-Hilbert method enable the derivation of universal correlation kernels in the 2-cut case?
• RQ5 What is the role of the discriminant of the quartic equation (1.8) in determining the spectral support structure?

Key findings

• The limiting spectral measure is supported on two disjoint intervals if and only if the discriminant $ \Delta $ of the quartic polynomial (1.8) is positive.
• When $ \Delta > 0 $, the limiting density $ \rho(z) $ is given explicitly by a formula involving cube roots and square roots of a quartic discriminant expression (1.9).
• The local eigenvalue correlation kernel in the bulk converges to the sine kernel under appropriate scaling, confirming bulk universality.
• At the edge of the spectrum, the correlation kernel converges to the Airy kernel, implying that the largest eigenvalue fluctuates according to the Tracy-Widom distribution.
• The edge locations $ \lambda_k $ satisfy $ \lambda_k = \lambda_k^N + O(M^{-1}) $, ensuring stability of edge locations under finite-size corrections.
• The convergence of the correlation kernel to universal forms is established via Riemann-Hilbert analysis, with error terms controlled up to $ O(M^{-1/3}) $.