Matrix-free polynomial preconditioning of saddle point systems using the hyper-power method
Abstract
This study explores the integration of the hyper-power sequence, a method commonly employed for approximating the Moore-Penrose inverse, to enhance the effectiveness of an existing preconditioner. The approach is closely related to polynomial preconditioning based on Neumann series. We commence with a state-of-the-art matrix-free preconditioner designed for the saddle point system derived from isogeometric structure-preserving discretization of the Stokes equations. Our results demonstrate that incorporating multiple iterations of the hyper-power method enhances the effectiveness of the preconditioner, leading to a substantial reduction in both iteration counts and overall solution time for simulating Stokes flow within a 3D lid-driven cavity. Through a comprehensive analysis, we assess the stability, accuracy, and numerical cost associated with the proposed scheme.
Cite
@article{arxiv.2311.06926,
title = {Matrix-free polynomial preconditioning of saddle point systems using the hyper-power method},
author = {Michał Łukasz Mika and Marco ten Eikelder and Dominik Schillinger and René Rinke Hiemstra},
journal= {arXiv preprint arXiv:2311.06926},
year = {2023}
}