“Minimizing Frobenius Norm in Covariance Cleaning for Heavy-Tailed Distributions is Asymptotically Equivalent to Minimizing Information Loss in Random Matrix Theory”

https://pubmed.ncbi.nlm.nih.gov/38115530

This study reveals that minimizing the Frobenius norm in covariance cleaning for heavy-tailed distributions, such as Student’s t distributions, is asymptotically equivalent to minimizing information loss, expanding the applicability of random matrix theory.