摘要
针对无约束优化问题,将Goldstein非精确线搜索技术引入伪Newton-B族算法.在假设目标函数f(x)二阶连续可微有下界,水平集L={x|f(x)≤f(x(1))}有界的条件下,证明该算法对一般目标函数的全局收敛性,得到一个条件更弱的结论.
The Pesudo-Newton-B class methods with inexact line searches (Goldstein rule) for unconstrained optimization. These methods are globally convergent when applied to a general objective function under the weak condition.
出处
《湖南文理学院学报(自然科学版)》
CAS
2007年第3期16-17,共2页
Journal of Hunan University of Arts and Science(Science and Technology)