Sparse Pseudospectral Shattering

TL;DR

This paper demonstrates that sparse random perturbations can regularize the eigenvalues and eigenvectors of nonnormal matrices, similar to dense perturbations, with implications for numerical stability in eigenvalue problems.

Contribution

It shows that adding sparse Gaussian noise to matrices achieves eigenvalue and eigenvector regularization comparable to dense noise, improving stability analysis techniques.

Findings

Sparse perturbations of size O(n log^2 n) regularize eigenvalues and eigenvectors.

Sparse perturbations of size O(n^{1+α}) achieve logarithmic condition numbers.

Reduction of the problem to least singular value estimates of shifted matrices.

Abstract

The eigenvalues and eigenvectors of nonnormal matrices can be unstable under perturbations of their entries. This renders an obstacle to the analysis of numerical algorithms for non-Hermitian eigenvalue problems. A recent technique to handle this issue is pseudospectral shattering [BGVKS23], showing that adding a random perturbation to any matrix has a regularizing effect on the stability of the eigenvalues and eigenvectors. Prior work has analyzed the regularizing effect of dense Gaussian perturbations, where independent noise is added to every entry of a given matrix [BVKS20, BGVKS23, BKMS21, JSS21]. We show that the same effect can be achieved by adding a sparse random perturbation. In particular, we show that given any $n\times n$ matrix $M$ of polynomially bounded norm: (a) perturbing $O(n\log^2(n))$ random entries of $M$ by adding i.i.d. complex Gaussians yields $\log\kappa_V(A)=O(\text{poly}\log(n))$ and $\log (1/\eta(A))=O(\text{poly}\log(n))$ with high probability; (b) perturbing $O(n^{1+\alpha})$ random entries of $M$ for any constant $\alpha>0$ yields $\log\kappa_V(A)=O_\alpha(\log(n))$ and $\log(1/\eta(A))=O_\alpha(\log(n))$ with high probability. Here, $\kappa_V(A)$ denotes the condition number of the eigenvectors of the perturbed matrix $A$ and $\eta(A)$ denotes its minimum eigenvalue gap. A key mechanism of the proof is to reduce the study of $\kappa_V(A)$ to control of the pseudospectral area and minimum eigenvalue gap of $A$, which are further reduced to estimates on the least two singular values of shifts of $A$. We obtain the required least singular value estimates via a streamlining of an argument of Tao and Vu [TV07] specialized to the case of sparse complex Gaussian perturbations. [Rest of abstract in pdf].