56
15
v1v2 (latest)

Limiting distributions for eigenvalues of sample correlation matrices from heavy-tailed populations

Abstract

Consider a pp-dimensional population xRp{\mathbf x} \in\mathbb{R}^p with iid coordinates in the domain of attraction of a stable distribution with index α(0,2)\alpha\in (0,2). Since the variance of x{\mathbf x} is infinite, the sample covariance matrix Sn=n1i=1nxixi{\mathbf S}_n=n^{-1}\sum_{i=1}^n {{\mathbf x}_i}{\mathbf x}'_i based on a sample x1,,xn{\mathbf x}_1,\ldots,{\mathbf x}_n from the population is not well behaved and it is of interest to use instead the sample correlation matrix Rn={diag(Sn)}1/2Sn{diag(Sn)}1/2{\mathbf R}_n= \{\operatorname{diag}({\mathbf S}_n)\}^{-1/2}\, {\mathbf S}_n \{\operatorname{diag}({\mathbf S}_n)\}^{-1/2}. This paper finds the limiting distributions of the eigenvalues of Rn{\mathbf R}_n when both the dimension pp and the sample size nn grow to infinity such that p/nγ(0,)p/n\to \gamma \in (0,\infty). The family of limiting distributions {Hα,γ}\{H_{\alpha,\gamma}\} is new and depends on the two parameters α\alpha and γ\gamma. The moments of Hα,γH_{\alpha,\gamma} are fully identified as sum of two contributions: the first from the classical Mar\v{c}enko-Pastur law and a second due to heavy tails. Moreover, the family {Hα,γ}\{H_{\alpha,\gamma}\} has continuous extensions at the boundaries α=2\alpha=2 and α=0\alpha=0 leading to the Mar\v{c}enko-Pastur law and a modified Poisson distribution, respectively. Our proofs use the method of moments, the path-shortening algorithm developed in [18] and some novel graph counting combinatorics. As a consequence, the moments of Hα,γH_{\alpha,\gamma} are expressed in terms of combinatorial objects such as Stirling numbers of the second kind. A simulation study on these limiting distributions Hα,γH_{\alpha,\gamma} is also provided for comparison with the Mar\v{c}enko-Pastur law.

View on arXiv
Comments on this paper