8
0

Covariance estimation with direction dependence accuracy

Abstract

We construct an estimator Σ^\widehat{\Sigma} for covariance matrices of unknown, centred random vectors X, with the given data consisting of N independent measurements X1,...,XNX_1,...,X_N of X and the wanted confidence level. We show under minimal assumptions on X, the estimator performs with the optimal accuracy with respect to the operator norm. In addition, the estimator is also optimal with respect to direction dependence accuracy: Σ^u,u\langle \widehat{\Sigma}u,u\rangle is an optimal estimator for σ2(u)=EX,u2\sigma^2(u)=\mathbb{E}\langle X,u\rangle^2 when σ2(u)\sigma^2(u) is ``large".

View on arXiv
Comments on this paper