Covariance matrix estimation in the presence of auxiliary information (Q1895527)

We use empirical likelihood to construct the stochastic processes and give the covariance matrix estimation of \(\Sigma > 0\), and eigenvalue and eigenvector estimation of \(\Sigma\) as efficiently as possible. Many theorems for likelihood are analogous to results of usual stochastic processes, but the asymptotic variances of many estimators are smaller. That is, the estimators in the presence of auxiliary information are asymptotically more efficient than the usual empirical distribution estimator.