Matrix-free stochastic calculation of operator norms without using adjoints

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Bresch, Jonas, Lorenz, Dirk A., Schneppe, Felix, Winkler, Maximilian
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917120376307712
author Bresch, Jonas
Lorenz, Dirk A.
Schneppe, Felix
Winkler, Maximilian
author_facet Bresch, Jonas
Lorenz, Dirk A.
Schneppe, Felix
Winkler, Maximilian
contents This paper considers the problem of computing the operator norm of a linear map between finite dimensional Hilbert spaces when only evaluations of the linear map are available and under restrictive storage assumptions. We propose a stochastic method of random search type to maximize the Rayleigh quotient and employ an exact line search in the random search directions. Moreover, we show that the proposed algorithm converges to the global maximum (the operator norm) almost surely and a sublinear convergence behavior for the corresponding eigenvector and eigenvalue equation. Finally, we illustrate the performance of the method with numerical experiments.
format Preprint
id arxiv_https___arxiv_org_abs_2410_08297
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Matrix-free stochastic calculation of operator norms without using adjoints
Bresch, Jonas
Lorenz, Dirk A.
Schneppe, Felix
Winkler, Maximilian
Numerical Analysis
65F35, 15A60, 68W20
This paper considers the problem of computing the operator norm of a linear map between finite dimensional Hilbert spaces when only evaluations of the linear map are available and under restrictive storage assumptions. We propose a stochastic method of random search type to maximize the Rayleigh quotient and employ an exact line search in the random search directions. Moreover, we show that the proposed algorithm converges to the global maximum (the operator norm) almost surely and a sublinear convergence behavior for the corresponding eigenvector and eigenvalue equation. Finally, we illustrate the performance of the method with numerical experiments.
title Matrix-free stochastic calculation of operator norms without using adjoints
topic Numerical Analysis
65F35, 15A60, 68W20
url https://arxiv.org/abs/2410.08297