Momentum accelerated power iterations and the restarted Lanczos method

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Barletta, Alessandro, Marshall, Nicholas, Pollock, Sara
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910036331069440
author Barletta, Alessandro
Marshall, Nicholas
Pollock, Sara
author_facet Barletta, Alessandro
Marshall, Nicholas
Pollock, Sara
contents In this paper we compare two methods for finding extremal eigenvalues and eigenvectors: the restarted Lanczos method and momentum accelerated power iterations. The convergence of both methods is based on ratios of Chebyshev polynomials evaluated at subdominant and dominant eigenvalues; however, the convergence is not the same. Here we compare the theoretical convergence properties of both methods, and determine the relative regimes where each is more efficient. We further introduce a preconditioning technique for the restarted Lanczos method using momentum accelerated power iterations, and demonstrate its effectiveness. The theoretical results are backed up by numerical tests on benchmark problems.
format Preprint
id arxiv_https___arxiv_org_abs_2511_05364
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Momentum accelerated power iterations and the restarted Lanczos method
Barletta, Alessandro
Marshall, Nicholas
Pollock, Sara
Numerical Analysis
65F15, 65F08
In this paper we compare two methods for finding extremal eigenvalues and eigenvectors: the restarted Lanczos method and momentum accelerated power iterations. The convergence of both methods is based on ratios of Chebyshev polynomials evaluated at subdominant and dominant eigenvalues; however, the convergence is not the same. Here we compare the theoretical convergence properties of both methods, and determine the relative regimes where each is more efficient. We further introduce a preconditioning technique for the restarted Lanczos method using momentum accelerated power iterations, and demonstrate its effectiveness. The theoretical results are backed up by numerical tests on benchmark problems.
title Momentum accelerated power iterations and the restarted Lanczos method
topic Numerical Analysis
65F15, 65F08
url https://arxiv.org/abs/2511.05364