摘要
用Schilders分解来推导非对称鞍点问题的约束预条件子,主要讨论了Schilders分解的过程、参数矩阵的选择及预处理矩阵特征值和特征向量的分布,得到了预处理矩阵最小多项式次数的一个上界并给出了约束预处理方法的实现,最后用数值算例加以说明.
In this paper,we consider the use of constraint preconditioning via a special Schilders factorization for nonsymmetric saddle point problems.The choices of the parameter matrcies in the Schilders factorization are discussed.The eigenvalue and eigenvector distribution of the preconditioned matrix are described and an upper bound of the degree of the minimal polynomials for the preconditioned matrix is obtained.Implementation of the preconditioning steps are given.Finally,numerical experiments are presented.
出处
《苏州大学学报(自然科学版)》
CAS
2010年第4期3-8,共6页
Journal of Soochow University(Natural Science Edition)