非线性互补问题的并行多分裂松弛迭代算法  

Relaxed Parallel Multi-splitting Iterative Algorithm for the Nonlinear Complementarity Problem

作　　者：段班祥[1] 徐安农[1] 李郴良[1] 

机构地区：[1]桂林电子工业学院计算科学与应用物理系,广西桂林541004

出　　处：《桂林电子工业学院学报》2003年第5期65-68,共4页Journal of Guilin Institute of Electronic Technology

摘　　要：运用矩阵多重分裂理论,同时考虑并行计算与松弛迭代法,得到求解一类非线性互补问题的高效数值算法。当问题的系数矩阵为对角元为正的H-矩阵时,证明了算法的全局收敛性。该算法把大规模问题分解为规模比较小的子问题,再对各子问题并行求解,与已有算法相比较,具有计算量小、计算速度快等特点,因而特别适于求解大规模问题。In this paper, the authors first set up relaxed parallel multisplitting iterative algorithm for solving the nonlinear complementarity problem. And then, they establish the global convergence theory of the algorithm. And they solve subproblems parallel by means of discomposing the largescale problems.The algorithm features quicker velocity and less computational complexity , and is suitable for largescale problems. 

关 键 词：矩阵多重分裂 并行计算 松弛迭代 非线性互补 

分 类 号：O241.6[理学—计算数学]