OPTIMAL CLEANING FOR SINGULAR VALUES OF CROSS-COVARIANCE MATRICES

被引:3
作者
Benaych-Georges, Florent [1 ]
Bouchaud, Jean-Philippe [1 ]
Potters, Marc [1 ]
机构
[1] CFM, Cincinnati, OH 45215 USA
关键词
Random matrices; cross-covariance matrices; rotationally invariant estimator; EIGENVECTORS; ESTIMATOR;
D O I
10.1214/22-AAP1842
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
We give a new algorithm for the estimation of the cross-covariance ma-trix EXY' of two large-dimensional signals X e Rn, Y e Rp in the context where the number T of observations of the pair (X, Y) is large but n/T and p/ T are not supposed to be small. In the asymptotic regime where n, p, T are large, with high probability, this algorithm is optimal for the Frobenius norm among rotationally invariant estimators, that is, estimators derived from the empirical estimator by cleaning the singular values, while letting singular vectors unchanged.
引用
收藏
页码:1095 / 1126
页数:32
相关论文
共 27 条
[1]  
Amsalu S, 2018, Arxiv, DOI arXiv:1804.09472
[2]  
Anderson GW., 2010, Cambridge Studies in Advanced Mathematics, V118
[3]  
BENAYCH-GEORGES F., 2017, ADV TOPICS RANDOM MA, P531
[4]  
Benaych-Georges F, 2022, Arxiv, DOI arXiv:2201.05690
[5]   SPECTRAL ANALYSIS OF THE GRAM MATRIX OF MIXTURE MODELS [J].
Benaych-Georges, Florent ;
Couillet, Romain .
ESAIM-PROBABILITY AND STATISTICS, 2016, 20 :217-237
[6]  
BOSE A., 2019, MONOGRAPHS STAT APPL, V162
[7]  
BOUCHAUD J.-P., 2007, EUR PHYS J B
[8]  
BUN J., 2016, RISK MAGAZINE
[9]   Cleaning large correlation matrices: Tools from Random Matrix Theory [J].
Bun, Joel ;
Bouchaud, Jean-Philippe ;
Potters, Marc .
PHYSICS REPORTS-REVIEW SECTION OF PHYSICS LETTERS, 2017, 666 :1-109
[10]   Additive/Multiplicative Free Subordination Property and Limiting Eigenvectors of Spiked Additive Deformations of Wigner Matrices and Spiked Sample Covariance Matrices [J].
Capitaine, M. .
JOURNAL OF THEORETICAL PROBABILITY, 2013, 26 (03) :595-648