一类新的求解无约束最优化问题下降算法及收敛性  

A New Class of Efficient Descent Method and Its Congenvence

作　　者：温坤文[1] 

机构地区：[1]嘉应学院,广东梅州514015

出　　处：《广东技术师范学院学报》2012年第3期7-10,60,共5页Journal of Guangdong Polytechnic Normal University

基　　金：嘉应学院自然科学基金项目(No.2011KJY08)

摘　　要：对求解无约束优化问题提出了一类新的下降算法,并且给出了HS算法与其相结合的两类混合算法.在Wolfe线搜索下,不需要给定下降条件,即证明了它们的全局收敛性.此外,数值实验表明:此文的算法是有效的.In this paper, we propose a new class of descent method and also give two combinations methods based on Hestenes-Stiefel and this new method. We proved their global congenvence in Wolfe line search without descent condition. In addition, two numerical tests have been done and the results show that the new method is effective.

关 键 词：无约束最优化 下降算法 混合算法 WOLFE线搜索 全局收敛性 

分 类 号：O175[理学—数学]