摘要
针对大型稀疏鞍点问题给出了一种新的迭代解法,该方法的构成是基于对系数矩阵进行的一种分裂,A∈Rn×n是对称正定矩阵.利用不完全分解法分解A为LLT+R,通过适当选取预处理矩阵和待定系数,证明该迭代法是收敛的,并且以定理的形式给出了新迭代法收敛的充分必要条件.
A new iterative solution is proposed to solve the problem of large-scale saddle point problem.Based on a splitting for the matrix of coefficients which A∈Rn×n is symmetric positive definite in coefficient matrix,using the incomplete decomposition method which split A to LLT+R,after choosing a pretreated matrix and undermined parameters,the converence of the iteration is proved and sufficient and necessary conditions of the new iteration method proposed becomes convergent in form of theorem.
出处
《怀化学院学报》
2010年第11期29-31,共3页
Journal of Huaihua University
关键词
鞍点
对称正定矩阵
矩阵分裂
收敛性
saddle point problem
symmetric positive definite
matrix splitting
convergence