Cleaning large-dimensional covariance matrices for correlated samples

07/03/2021

∙

by   Zdzislaw Burda, et al.

∙

0

∙

share

A non-linear shrinkage estimator of large-dimensional covariance matrices is derived in a setting of auto-correlated samples, thus generalizing the recent formula by Ledoit-Péché. The calculation is facilitated by random matrix theory. The result is turned into an efficient algorithm, and an associated Python library, shrinkage, with help of Ledoit-Wolf kernel estimation technique. An example of exponentially-decaying auto-correlations is presented.