loss Module

This module provides loss functions frequently encountered in the literature on high dimensional covariance matrix estimation.

Functions

loss_fr(sigma_hat, sigma)	Squared Frobenius norm scaled by 1/p.
loss_mv(sigma_hat, sigma)	The minimum variance loss function of Ledoit and Wolf (2018).
marchenko_pastur(x, c, sigma_sq)	The Marchenko-Pastur distribution.
prial(S_list, sigma_hat_list, sigma[, loss_func])	The percentage relative improvement in average loss (PRIAL) over the sample covariance matrix.