The global convergence of the non-quasi-Newton methods with non-monotone line search
The non-quasi-Newton methods for unconstrained optimization was investigated. Non-monotone line search procedure is introduced, which is combined with the non-quasi-Newton family. Under the uniform convexity assumption on objective function, the global convergence of the non-quasi-Newton family was proved.Numerical experiments showed that the non-monotone line search was more effective.
作 者： JIAO Bao-cong LIU Hong-wei 作者单位： JIAO Bao-cong(The Center of Management and Decision Research, Capital Normal University, Beijing 100037, China)LIU Hong-wei(The Institute of Information, Renmin University of China,Beijing 100872,China)
刊 名： 哈尔滨工业大学学报（英文版） EI 英文刊名： JOURNAL OF HARBIN INSTITUTE OF TECHNOLOGY 年，卷(期)： 2006 13(6) 分类号： O174.13 关键词： non-quasi-Newton method non-monotone line search global convergence unconstrained optimization