节点文献
两类Armijo-type线搜索下的PRP新算法
Two Classes of the New PRP Methods under the Armijo-Type Line Search
【摘要】 给出了新的PRP算法,该算法具有充分下降性和在无需任何线搜索的情况下自动保持搜索方向在可行域的特性,而且该算法在非凸函数和一致凸函数上具有全局收敛性,最后证明了新算法的线性收敛速度.
【Abstract】 The PRP method is generally believed to be the most efficient conjugate gradient method.A new PRP algorithm is proposed in this paper,which has sufficient descent property and gives the characteristic that the new search direction can automatically keep in a trust region without carrying out any line search rule.What is more,this algorithm possesses superior convergence property for nonconvex function and uniformly convex function.Finally,the linear convergence rate of the new algorithm is proved.
- 【文献出处】 西南大学学报(自然科学版) ,Journal of Southwest University(Natural Science Edition) , 编辑部邮箱 ,2010年07期
- 【分类号】O224
- 【被引频次】3
- 【下载频次】63