摘要:共軛梯度法是求解大規模約束問題的有效算法,的選取不同構成不同的共軛梯度法,由此通過修正提出了求解無約束優化問題的一種改進的共軛梯度法,并在wolf線搜索下證明了它的全局收斂性。
關鍵詞:無約束優化問題;共軛梯度法;wolf線搜索;全局收斂性
中圖分類號O1-0文獻標志碼:A文章編號:1673-291X(2008)13-0227-02
參考文獻:
[1] Zoutendijk.G.NonlinearProgramming,ComputationalMethods[M].Amsterdam:North-Holland,1970:37-86.
[2] 戴或虹,袁亞湘.非線性共扼梯度法[M].上海:上海科技出版社,2000.
[3] 潘翠英,陳蘭平.求解無約束優化問題的一類新的下降算法[J].應用數學學報,2007,(1):88-98.
[4] 陳蘭平,焦寶聰.一般無約束優化問題的廣義擬牛頓法[J].數學進展,2007,(1):81-85.
[5] 陳寶林.最優化理論與算法[M].北京:清華大學出版社,1989.
A New Conjugate Gradient Method for Unconstrained Optimization Problems
JING Chun-xia1, CHEN Zhong1,HE Yong-qiang2
(1.School of Information and Mathematics, Yangtze University, Jinzhou 434023,China;
2.Sinopec co. Jiangsu oilfield Branch,Yangzhou 211600 ,China)
Abstract: In the ordinary circumstances, conjugate gradient method is the effective algorithm which solves the large-scale restraint question, different Selection of constructs different conjugate gradient method. We propose a new conjugate gradient method for unconstrained optimization problems byupdate and prove that method with wolf line search converges globally.
Key words: unconstrained optimization problem; conjugate gradient method; wolf line searc; global convergence
[責任編輯王曉燕]