摘要
本文提出了二类解约束优化问题的广义既约型梯度法,从统一角度研究了投影梯度法和既约梯度法的结构及其全局收敛性.本文结果统一、推广了常见的可行方向法.
This paper presents two classes of generalized reduced-type gradient methods for constrained optimization, and studies the constructure and global convergence of projection gradient method and reduced gradient method from the point of unified approach. The results of the paper unify and extend common feasible direction methods.
出处
《应用数学》
CSCD
北大核心
1993年第2期145-150,共6页
Mathematica Applicata
基金
校青年科学基金资助课题~~
关键词
约束优化
既约梯度法
广义
Constrained optimization
Projection gradient method
Reduced gradient method
Unified approach
Generalized reduced-type gradient method
Global convergence