Id: arxiv/2603.19703 Url: https://arxiv.org/abs/2603.19703 Tags: math.ST, cs.LG Score: 0 Title: Minimax and Adaptive Covariance Matrix Estimation under Differential Privacy Domain: math.ST Source: arxiv Authors: T. Tony Cai, Yicheng Li Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-03-20 Page: https://delphi.science/r/arxiv/2603.19703 Estimating covariance matrices is fundamental to a wide range of statistical applications. This paper studies minimax and adaptive estimation of high-dimensional covariance matrices under $ρ$-zero-concentrated differential privacy ($ρ$-zCDP) over three nested classes: the pointwise-decay class $\mathcal{H}_α$, the row-tail class $\mathcal{G}_α$, and the separated-block class $\mathcal{F}_α$. We consider both squared operator norm loss and normalized squared Frobenius norm loss. For $\mathcal{H}_α$ and $\mathcal{G}_α$, we develop center--outer dyadic estimators tailored to the refined…