一个具有充分下降性的混合共轭梯度法

A Hybrid Conjugate Gradient Method with Sufficient Descent Property

混合共轭梯度法是一个改进的新共轭梯度法,有着比较好的数值表现.在Jia提出的混合共轭梯度法基础上,建立了一个新的具有充分下降性的混合共轭梯度算法;并证明了该算法在强Wolfe型线搜索下具有全局收敛性.数值实验结果表明该算法是有效的.

A hybrid conjugate gradient method is a new and improved conjugate gradient method with better numerical performance.Base on the hybrid conjugate gradient method proposed by Jia,a new hybrid conjugate gradient algorithm with sufficient descent property is presented,and the algorithm possesses the global convergence under the strong Wolfe line search.Numerical results show that the new algorithm was effective.