基于进化策略的线性变分不等式求解算法  

Solving linear variational inequality based on evolution strategies

作　　者：张明[1,2] 周永权[2] Ahmed N.Abdalla 

机构地区：[1]大连水产学院理学院,辽宁大连116023 [2]广西民族大学数学与计算机科学学院,广西南宁530006 [3]华中科技大学电气与电子工程学院,湖北武汉430074

出　　处：《计算机工程与设计》2009年第2期378-381,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(60461001);广西自然科学基金项目(0542048)

摘　　要：基于在求解变分不等式过程中存在着传统数字计算机的迭代算法很难满足并行性要求的问题,提出了求解一类线性变分不等式问题的进化策略算法。将进化策略算法用于求解线性变分不等式的数值方法,充分发挥了进化策略算法的全局收敛和并行搜索的特性,满足了工程技术中并行求解变分不等式问题的要求。数值计算结果表明,该算法收敛速度快、精度高、稳定性好,是一种解决线性变分不等式问题的有效方法。A solution to linear variational ine.quality is presented by evolution strategies based on the question that iterative algorithm of traditional numerical computer cannot satisfy parallel in solving variational inequality. The numerical solution that evolution strategies are used to solve linear variational inequality, which sutticiently exerted the advantage ofevolution strategies such as group search and global convergence, anditsatisfiesthequestion of parallel solute the linear variational inequality in engineering technique. The numerical computation results indicate that the algorithm offers an effective way to solve linear variational inequality, high convergence rate, high accuracy and robustness.

关 键 词：变分不等式 进化策略 全局收敛 并行搜索 平均突变算子 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]