摘要
It is known that stepsize’s choice plays a key role in convergence and efficiency of the extragradient method, which is a special projection-type method, for solving monotone variational inequality problems. In this paper, by analyzing the existing stepsize rules, a predictor stepsize rule without the bounded restriction is proposed, and a corrector stepsize rule with (approximate) optimality is also presented. The corresponding convergence properties and numerical examples are shown.
It is known that stepsize's choice plays a key role in convergence and efficiency of the extragradient method, which is a special projection-type method, for solving monotone variational inequality problems. In this paper, by analyzing the existing stepsize rules, a predictor stepsize rule without the bounded restriction is proposed, and a corrector stepsize rule with (approximate) optimality is also presented. The corresponding convergence properties and numerical examples are shown.
出处
《计算数学》
CSCD
北大核心
2000年第2期197-208,共12页
Mathematica Numerica Sinica
基金
国家自然科学基金!(19971002
19871049)
