Stochastic Rayleigh-Ritz Ascent: Global Linear Convergence under a Spectral Gap

arXiv:2610.42071 · math.OC, cs.NA, math.NA, math.PR · Submitted 2026-10-30 · Read on arXiv

math.OC, cs.NA, math.NA, math.PR

Submitted: 2026-10-30

Updated: 2026-10-30

Code: https://github.com/JJEWBresch/LinearConvergenceOperatorNormSolver

License: http://creativecommons.org/licenses/by/4.0/

The gist: We consider the adjoint-free random-plane iteration algorithm for finding the top eigenvector of a matrix proposed in (SIAM J.

Terminology

Abstract

We consider the adjoint-free random-plane iteration algorithm for finding the top eigenvector of a matrix proposed in (SIAM J. Matrix Anal. Appl. 47 (2026), 1349-1374). Unlike standard methods such as the power method or gradient descent, the above algorithm lacked linear convergence guarantees in the presence of a spectral gap. In this paper, we establish explicit spectral-gap-dependent global linear convergence rates in expectation, providing quantitative guarantees for the empirically observed convergence.

Sources

Related papers