摘要
为了在高性能计算机上求解广义鞍点问题,对于合适的系数矩阵,本文提出混合并行迭代法及其加速形式.并详细讨论了新方法的收敛性.
A class of parallel hybrid iterative methods and its accelerated overrelaxation variant are established for solving the generalized saddle-point problems. The new method converges to the so- lution under suitable restrictions on coefficient matrices.
出处
《应用数学》
CSCD
北大核心
2012年第2期282-287,共6页
Mathematica Applicata
基金
Supported by the NSF of China(11071184)
the NSF of Shanxi Province(2010011006)
the Shanxi Scholarship Council of China Research Grant(2010087)
关键词
并行迭代
广义鞍点问题
AOR迭代法
收敛性
Parallel iteration
Generalized saddle-point problem
AOR method
Convergence
