Sample canonical correlation coefficients of high-dimensional random vectors: local law and Tracy-Widom limit
arXiv:2002.09643 · doi:10.1142/S2010326322500071
Abstract
Consider two random vectors and , where the entries of and are i.i.d. random variables with mean zero and variance one, and and are and deterministic population covariance matrices. With independent samples of , we study the sample correlation between these two vectors using canonical correlation analysis. We denote by and the sample covariance matrices for and , respectively, and the sample cross-covariance matrix. Then the sample canonical correlation coefficients are the square roots of the eigenvalues of the sample canonical correlation matrix . Under the high-dimensional setting with and as , we prove that the largest eigenvalue of converges to the Tracy-Widom distribution as long as we have . This extends the result in [16], which established the Tracy-Widom limit of the largest eigenvalue of under the assumption that all moments are finite. Our proof is based on a linearization method, which reduces the problem to the study of a random matrix . In particular, we shall prove an optimal local law on its inverse , i.e the resolvent. This local law is the main tool for both the proof of the Tracy-Widom law in this paper, and the study in [22,23] on the canonical correlation coefficients of high-dimensional random vectors with finite rank correlations.
arXiv comment56 pages
References in corpus (15)
- Spectral statistics of Erdős-Rényi graphs I: Local semicircle law
- Universality of covariance matrices
- Multivariate analysis and Jacobi ensembles: largest eigenvalue, Tracy--Widom limits and rates of convergence
- Averaging Fluctuations in Resolvents of Random Band Matrices
- A Necessary and Sufficient Condition for Edge Universality of Wigner matrices
- Delocalization and Diffusion Profile for Random Band Matrices
- A necessary and sufficient condition for edge universality at the largest singular values of covariance matrices
- Independence test for high dimensional data based on regularized canonical correlation coefficients
- Minimax estimation in sparse canonical correlation analysis
- Local law for random Gram matrices
- Edge universality of separable covariance matrices
- Convergence of eigenvector empirical spectral distribution of sample covariance matrices
- Local circular law for the product of a deterministic matrix with a random matrix
- Sample canonical correlation coefficients of high-dimensional random vectors with finite rank correlations
- Limiting distribution of the sample canonical correlation coefficients of high-dimensional random vectors