Eigenvector distributions and optimal shrinkage estimators for large covariance and precision matrices
Preprint
Abstract
This paper focuses on investigating Stein's invariant shrinkage estimators for large sample covariance matrices and precision matrices in high-dimensional settings. We consider models that have nearly arbitrary population covariance matrices, including those with potential spikes. By imposing mild technical assumptions, we establish the asymptotic limits of the shrinkers for a wide range of loss functions. A key contribution of this work, enabling the derivation of the limits of the shrinkers, is a novel result concerning the asymptotic distributions of the non-spiked eigenvectors of the sample covariance matrices, which can be of independent interest.Keywords
All Related Versions
This publication has 55 references indexed in Scilit:
- High Dimensional Normality of Noisy EigenvectorsCommunications in Mathematical Physics, 2022
- Eigenstate Thermalization Hypothesis for Wigner MatricesCommunications in Mathematical Physics, 2021
- Random Band Matrices in the Delocalized Phase I: Quantum Unique Ergodicity and UniversalityCommunications on Pure and Applied Mathematics, 2020
- Convergence of Local Statistics of Dyson Brownian MotionCommunications in Mathematical Physics, 2017
- The Eigenvector Moment Flow and Local Quantum Unique ErgodicityCommunications in Mathematical Physics, 2016
- Edge Universality of Beta EnsemblesCommunications in Mathematical Physics, 2014
- Averaging Fluctuations in Resolvents of Random Band MatricesAnnales Henri Poincare, 2013
- Eigenvector distribution of Wigner matricesProbability Theory and Related Fields, 2011
- Eigenvectors of some large sample covariance matrix ensemblesProbability Theory and Related Fields, 2010
- Lectures on the theory of estimation of many parametersJournal of Mathematical Sciences, 1986