Moment Estimates for the Spectral Norm of Random Matrices with Dependent Entries
Abstract
This paper studies the moments for the spectral norm of random matrices with dependent entries. In particular, we consider a random matrix $BA$, where $A$ is a random matrix with independent mean zero subexponential entries, and $B$ is a deterministic matrix. We show a sharp moment bound for the spectral norm of an $m\times n$ matrix $BA$ based on a comparison theorem due to Latała, van Handel and Youssef. Applying this result, we prove an estimate of the smallest singular value of an $N\times n$ random subexponential matrix.
- Publication:
-
arXiv e-prints
- Pub Date:
- July 2023
- DOI:
- 10.48550/arXiv.2307.03069
- arXiv:
- arXiv:2307.03069
- Bibcode:
- 2023arXiv230703069D
- Keywords:
-
- Mathematics - Probability;
- 60F05;
- 60F17
- E-Print:
- arXiv admin note: text overlap with arXiv:0812.2432 by other authors