摘要
针对等式约束优化问题提出了一个带记忆的等式约束信赖域算法。该算法不同于传统的信赖域方法,此信赖域模型是记忆模型,从全局考虑目标函数的下降性而不完全依赖于当前点信息,采用非单调技术得到了算法的全局收敛性和超线性收敛性。
In the paper a new trust region algorithm with memory model for equality constrained optimization problems is proposed.Different from the tradition trust algorithm,the new algorithm contains the message of the past iteration,which makes the algorithm more farsighted.Moreover,the algorithm is not completely decided by the local nature of the objection function,and numerical results show it is efficient.By adopting non-monotone technique,the global convergence and superlinear convergence of the algorithm are obtained.
出处
《河池学院学报》
2011年第5期33-42,共10页
Journal of Hechi University
基金
国家自然科学基金资助项目(11061011)
广西高校优秀人才资助项目(〔2009〕156)
关键词
记忆模型
信赖域算法
非单调技术
全局收敛性
超线性收敛性
memory model
trust region algorithm
non-monotone technique
global convergence
superlinear convergence
