摘要
本文提出了两种求解伪单调变分不等式的定步长的投影算法.这与Solodov&Tseng(1996)和He(1997)的变步长策略不同.我们证明了算法的全局收敛性,并且还在一定条件下证明了算法的Q-线性收敛性.
In this paper, we present two projection-type algorithms for solving pseudo-monotone variational inequality problems. These algorithms use the fixed stepsize strategy, different from the ones presented by Solodov & Tseng (1996) and He (1997) using the variable stepsize strategy. It has been shown that the algorithms are globally convergent and, under some mild conditions, they are convergent Q-linearly.
出处
《运筹学学报》
CSCD
北大核心
2005年第1期58-64,共7页
Operations Research Transactions
基金
Supported by the NSF of China 10001007State Foundations of Ph.D Units 20020141013 research foundation of DUT (2002-03).
关键词
单调变分不等式
求解
全局收敛性
证明
投影算法
线性
变步长
Operations research, variational inequality, pseudo-monotonicity, projection-type algorithm, convergence
