English

Gaussian Regularization of the Pseudospectrum and Davies' Conjecture

Functional Analysis 2020-04-23 v4 Numerical Analysis Numerical Analysis Probability Spectral Theory

Abstract

A matrix ACn×nA\in\mathbb{C}^{n\times n} is diagonalizable if it has a basis of linearly independent eigenvectors. Since the set of nondiagonalizable matrices has measure zero, every ACn×nA\in \mathbb{C}^{n\times n} is the limit of diagonalizable matrices. We prove a quantitative version of this fact conjectured by E.B. Davies: for each δ(0,1)\delta\in (0,1), every matrix ACn×nA\in \mathbb{C}^{n\times n} is at least δA\delta\|A\|-close to one whose eigenvectors have condition number at worst cn/δc_n/\delta, for some constants cnc_n dependent only on nn. Our proof uses tools from random matrix theory to show that the pseudospectrum of AA can be regularized with the addition of a complex Gaussian perturbation. Along the way, we explain how a variant of a theorem of \'Sniady implies a conjecture of Sankar, Spielman and Teng on the optimal constant for smoothed analysis of condition numbers.

Keywords

Cite

@article{arxiv.1906.11819,
  title  = {Gaussian Regularization of the Pseudospectrum and Davies' Conjecture},
  author = {Jess Banks and Archit Kulkarni and Satyaki Mukherjee and Nikhil Srivastava},
  journal= {arXiv preprint arXiv:1906.11819},
  year   = {2020}
}

Comments

17pp. Fixed a mistake in the appendix, all results are unchanged. Accepted version, to appear in CPAM

R2 v1 2026-06-23T10:05:47.862Z