Comparison between two types of large sample covariance matrices (Q2451115)

The author proves that central limit theorems of eigenvalue statistics of \(\mathcal S\) and \textbf{S} are different as \(n\rightarrow \infty\) with \(p\over n\) approaching a positive constant, where \(\mathcal S\) and \textbf{S} are sample covariance matrices (with/without empirical centering). Moreover, it is also proved that such a different behavior is not observed in the average behavior of eigenvectors.