一种新的求解拟单调变分不等式的压缩投影算法  

A new projection and contraction algorithm for solving quasimonotone variational inequalities

作　　者：叶明露[1] 邓欢 YE Minglu;DENG Huan(Sichuan Colleges and Universities Key Laboratory of Optimization Theory and Applications,School of Mathematics and Information,China West Normal University,Nanchong,637009,Sichuan,China)

机构地区：[1]西华师范大学数学与信息学院四川高等院校优化理论与应用重点实验室,四川南充637009

出　　处：《运筹学学报》2023年第1期127-137,共11页Operations Research Transactions

基　　金：国家自然科学基金面上项目(No.11871059);西华师范大学培育项目(No.20A024)。

摘　　要：2020年Liu和Yang提出了求解Hilbert空间中拟单调且Lipschitz连续的变分不等式问题的投影算法,简称LYA。本文在欧氏空间中提出了一种新的求解拟单调变分不等式的压缩投影算法,简称NPCA。新算法削弱了LYA中映射的Lipschitz连续性。在映射连续、拟单调且对偶变分不等式解集非空的条件下得到了NPCA所生成点列的聚点是解的结论。当变分不等式的解集还满足一定条件时,得到了NPCA的全局收敛性。数值实验结果表明NPCA所需的迭代步数少于LYA的迭代步数,NPCA在高维拟单调例子中所需的计算机耗时也更少。In 2020,Liu and Yang proposed a projection algorithm(LYA for short)for solving quasi-monotone and Lipschitz continuous variational inequalities problem(VIP for short)in Hilbert space.In this paper,we present a new projection and contraction algorithm(NPCA for short)for solving quasi-monotone VIP in Euclidean space.The new algorithm weakens the Lipschitz continuity of the underlying mapping in LYA.NPCA clusters to the solution of VIP whenever the underlying mapping is continuous,quasi-monotone and the solution set of dual variational inequality is nonempty.The global convergence of NPCA needs an additional assumption about the solution set of VIP.Numerical experiments show that NPCA is more efficient than LYA from the total number of iterative point of view,and the CPU time point of view in high dimensional quasi-monotone VIP.

关 键 词：变分不等式 拟单调 压缩投影算法 连续 

分 类 号：O221.2[理学—运筹学与控制论]