The factor model problem focuses on the decomposition of a covariance matrix $\Sigma$ to low-rank and diagnonal positive semidefinite matrices. In practice, $\Sigma$ is often not available, and only its empirical counterpart, $\hat{\Sigma}$, is available. To robustify to this approximation error, a common practice is to consider a family of covariance matrices in the vicinity of $\Sigma$.
The linked repository contains MATLAB files corresponding to the numerical results of the paper 'A Saddle Point Algorithm for Robust Data-Driven Factor Model Problems'.
" authors: - family-names: Khodakaramzadeh given-names: Shabnam orcid: "https://orcid.org/0000-0001-6896-3725" title: "Factor Model MATLAB code underlying the publication: A Saddle Point Algorithm for Robust Data-Driven Factor Model Problems" keywords: version: 1 identifiers: - type: doi value: 10.4121/f25270e0-4054-4e9d-85da-5908a07a3874.v1 license: BSD-3-Clause date-released: 2025-06-17