Gaussian fluctuations for sample covariance matrices with dependent data (Q1931868): Difference between revisions

The authors derive the central limit theorem for the traces of powers of the large-dimensional sample covariance matrix when the entries of the data matrix are assumed to be dependent without specifying the dependence structure. The result holds under the assumption that the number of dependent entries for each element of the data matrix does not grow too fast as the sample size grows up. Several important results about cumulants are presented as well.