摘要
基于PRP方法和HS方法在算法参数结构,算法性质和数值表现方面的共性、构造了一种求解无约束优化问题的三项共轭梯度法。该算法所确定的搜索方向不依赖于线搜索条件,恒为充分下降方向,并在Wolfe线搜索的条件下证明了该算法全局收敛性。最后选用部分测试函数进行了数值试验,试验结果表明,该算法不仅能保证全局收敛性,而且还具有较快的收敛速度。
Based on PRP method and HS method parameter in the algorithm structure,the common nature of the algorithm and numerical simulation,a three conjugate gradient method is built for solving unconstrained optimization problems.Determining the search direction of the algorithm is independent of line search condition,it is constant for a full down direction,and in Wolfe line search condition,the algorithm is global convergence.Finally test function is carried on the numerical experiments,the results show that the algorithm not only can guarantee the global convergence,but also has fast convergence speed.
出处
《长江大学学报（自科版）（上旬）》
2014年第8期1-3,131,共4页
JOURNAL OF YANGTZE UNIVERSITY (NATURAL SCIENCE EDITION) SCI ＆ ENG
基金
湖北省教育科学"十二五"规划课题(2012B310)
长江大学工程技术学院基金项目(13J0802)
关键词
无约束优化
三项共轭梯度法
WOLFE线搜索
全局收敛性
unconstrained optimization
three-term conjugate gradient method
Wolfe line search
global convergent
