关键词: 无约束优化, 混合型共轭梯度法, 充分下降条件, 数值分析

Abstract: We considered several hybrid conjugate gradient methods, determined their step length by weak Wolfe line search, and generated the descent direction satisfying the sufficient descent condition by orthogonalization strategy. The effectiveness of these algorithms was verified by CUTEst test problems, and the numerical performance of these algorithms was analyzed.

Key words: unconstrained optimization, hybrid conjugate gradient method, sufficient descent condition, numerical analysis