Randomized average block iterative methods for solving factorised linear systems

被引:0
|
作者
Zhao, Jing [1 ]
Wang, Xiang [1 ]
Zhang, Jianhua [2 ]
机构
[1] Nanchang Univ, Dept Math, Nanchang 330031, Jiangxi, Peoples R China
[2] East China Univ Technol, Sch Sci, Nanchang 330013, Jiangxi, Peoples R China
基金
中国国家自然科学基金;
关键词
Factorised linear systems; Randomized average block Kaczmarz method; Randomized extended average block Kacz-marz method; RECOMMENDER SYSTEMS; KACZMARZ METHOD; RATES;
D O I
10.2298/FIL2314603Z
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
Recently, some randomized iterative methods are proposed to solve large-scale factorised linear systems. In this paper, we present two randomized average block iterative methods which still take advantage of the factored form and need not perform the entire matrix. The new methods are pseudoinverse-free and can be implemented for parallel computation. Furthermore, we analyze their convergence behaviors and obtain the exponential convergence rate. Finally, some numerical examples are carried out to show the effectiveness of our new methods.
引用
收藏
页码:4603 / 4620
页数:18
相关论文
共 50 条