A note on overrelaxation in the Sinkhorn algorithm
Optimization and Control
2021-12-07 v3 Numerical Analysis
Numerical Analysis
Statistics Theory
Statistics Theory
Abstract
We derive an a priori parameter range for overrelaxation of the Sinkhorn algorithm, which guarantees global convergence and a strictly faster asymptotic local convergence. Guided by the spectral analysis of the linearized problem we pursue a zero cost procedure to choose a near optimal relaxation parameter.
Cite
@article{arxiv.2012.12562,
title = {A note on overrelaxation in the Sinkhorn algorithm},
author = {Tobias Lehmann and Max-K. von Renesse and Alexander Sambale and André Uschmajew},
journal= {arXiv preprint arXiv:2012.12562},
year = {2021}
}
Comments
9 pages, 1 figure