摘要
本文考虑线性约束非线性规划问题，提出了一类共轭投影梯度法，证明了算法的全局收敛性，并对算法的二次终止性，超线性收敛特征进行了分析．算法的优点是（１）采用计算机上容易实现的Ａｒｍｉｊｏ线性搜索规则，（２）初始点不要求一定是可行点，可以不满足线性等式约束。
In this paper, a class of conjugate projection gradient algorithms for optimization problem with linear constraints is presented, and its global convergence, quadratic termination, super linear convergence rate are analyzed. The advantage of these algorithms is (1) the Armijo’s line search which is easily realized by computer is used in algorithms; (2) don’t demand that initial point must be the feasible point of problem; (3) the convergence rate of algorithms is expected to be accelerated.
出处
《高校应用数学学报（A辑）》
CSCD
北大核心
1997年第2期209-218,共10页
Applied Mathematics A Journal of Chinese Universities(Ser.A)
基金
山东省自然科学基金
青年基金
关键词
收敛性
共轭投影梯度法
非线性规划
超线性收敛
Conjugate Projection, Armijo’s Line Search, Convergence, Quadratic Termination, Superlinear Convergence Rate.
