摘要
在目标函数一致凸的条件下,探讨了非拟牛顿法在Wolfel线性搜索和Goldstein线性搜索下的全局收敛性予以证明,得出在求解无约束优化问题当δk充分小时,用B k+1替代2f(x k),使非拟牛顿方程不仅利用了函数梯度值信息,还利用了函数值的信息。
Under the condition of objective function is uniformly convex,this paper discusses the quasi-newton method in Wolfe-Powell linear search and global convergence under Goldstein linear search to be proved,it is concluded that in solving unconstrained optimization problem when full hours,Bk+1with alternative,make the quasi-newton equation using the function gradient value information,not only has also used the information function value.
作者
孟红军
Meng Hongjun(Education Department of Chuzhou City Vocational College,Chuzhou,Anhui 239000)
出处
《呼伦贝尔学院学报》
2020年第2期86-91,共6页
Journal of Hulunbuir University
基金
2018年安徽省教育厅重点项目“小教专业全科教师培养模式改革研究”项目编号(2018jyxm1327)
2018年滁州城市职业学院院级重点项目“高职院校创新创业课程体系构建与实践”项目编号(2018jyxm06)。
