A block product preconditioner for saddle point problems

JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS

卷:352

Article

Liao, Li-Dan^1,2  Zhang, Guo-Feng¹  Zhu, Mu-Zheng³ 

[1] Lanzhou Univ, Sch Math & Stat, Lanzhou 730000, Gansu, Peoples R China

[2] Nanchang Univ, Dept Math, Nanchang 330031, Jiangxi, Peoples R China

[3] Hexi Univ, Sch Math & Stat, Zhangye 734000, Peoples R China

关键词: Saddle point problems;    Preconditioner;    Spectral properties;    Optimal parameter;    Krylov subspace method;   

DOI  :  10.1016/j.cam.2018.11.026

来源: Elsevier

【 摘 要 】

In this paper, a block product (BP) preconditioner is established for saddle point problems. Spectral properties of the BP preconditioned matrix are investigated. A strategy for practical choice of quasi-optimal parameter is given. Numerical results on saddle point linear systems arising from Stokes problems and weighted least square problems show that the proposed BP preconditioner is more economic to implement within Krylov subspace acceleration than some extensively studied preconditioners. (C) 2018 Elsevier B.V. All rights reserved.

【 授权许可】

Free