Test for high-dimensional correlation matrices (Q2328063)

A general framework for testing correlation structures is developed for the one, two and the multiple testing problems. It is considered under a high-dimensional setting when both the sample size and the data dimension go to infinity. The test statistics are designed to deal with both the dense and the sparse alternatives, since they are the sum (or the maximum) of two terms (one for dense alternative and the other for the sparse alternative). The asymptotic null distribution, power function and unbiasedness of each test statistics is established. A simulation study is presented, together with an application to the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset.