Abstract:Testing large covariance matrices is of fundamental importance in statistical analysis with high-dimensional data. In the past decade, three types of test statistics have been studied in the literature: quadratic form statistics, maximum form statistics, and their weighted combination. It is known that quadratic form statistics would suffer from low power against sparse alternatives and maximum form statistics would suffer from low power against dense alternatives. The weighted combination methods were introduced to enhance the power of quadratic form statistics or maximum form statistics when the weights are appropriately chosen. In this article, we provide a new perspective to exploit the full potential of quadratic form statistics and maximum form statistics for testing high-dimensional covariance matrices. We propose a scale-invariant power-enhanced test based on Fisher's method to combine the p -values of quadratic form statistics and maximum form statistics. After carefully studying the asymptotic joint distribution of quadratic form statistics and maximum form statistics, we first prove that the proposed combination method retains the correct asymptotic size under the Gaussian assumption, and we also derive a new Lyapunov-type bound for the joint distribution and prove the correct asymptotic size of the proposed method without requiring the Gaussian assumption. Moreover, we show that the proposed method boosts the asymptotic power against more general alternatives. Finally, we demonstrate the finite-sample performance in simulation studies and a real application. Supplementary materials for this article are available online.

Testing the Diagonality of A Large Covariance Matrix in A Regression Setting

A hypothesis test for independence of sets of variates in high dimensions

Testing Super-Diagonal Structure in High Dimensional Covariance Matrices

Testing covariates in high-dimensional regression

An Adaptive Test for the Mean Vector in Large-P-small-n Problems

Use of Random Integration to Test Equality of High Dimensional Covariance Matrices

HYPOTHESIS TESTING ON LINEAR STRUCTURES OF HIGH DIMENSIONAL COVARIANCE MATRIX.

HIGH-DIMENSIONAL TWO-SAMPLE COVARIANCE MATRIX TESTING VIA SUPER-DIAGONALS

Nonlinear shrinkage test on a large‐dimensional covariance matrix

Double verification for two‐sample covariance matrices test

Testing High-dimensional Covariance Matrices under the Elliptical Distribution and Beyond

Two Sample Tests for High Dimensional Covariance Matrices

Tests for Covariance Structures with High-Dimensional Repeated Measurements

Testing for patterns and structures in covariance and correlation matrices

Diagonally-Dominant Principal Component Analysis

Limitations on detecting row covariance in the presence of column covariance

Testing and Support Recovery of Multiple High-Dimensional Covariance Matrices with False Discovery Rate Control

Fisher's Combined Probability Test for High-Dimensional Covariance Matrices

Joint testing and false discovery rate control in high-dimensional multivariate regression.

On testing the equality of high dimensional mean vectors with unequal covariance matrices

Comparing large covariance matrices under weak conditions on the dependence structure and its application to gene clustering