一个充分下降的杂交共轭梯度法

A Hydrid Conjugate Gradient Method with Sufficient Decent Property

对无约束优化问题,基于文献[6]的WYL公式和文献[8]的MFR公式,给出了一个杂交共轭梯度法公式,并建立相应的算法.在不依赖于任何线搜索的条件下,算法的每一步迭代都可以产生一个充分下降方向,且在标准Wolfe线搜索之下可以证明此杂交共轭梯度法是全局收敛的。最后其数值测试结果表明所给的杂交方法是有效的.

In this paper, Based on the WYL formula in [6] and the MFR formula in [8], a hydrid conjugate gradient method is proposed for unconstrained optimization. It can generates sufficient descent condition at each iteration without any line search , and converges globally for nonconvex minimization if the Wolfe line search is used. Some elementary numerical experiments are reported, which show that the proposed method is promising.