两类无约束优化的充分下降共轭梯度法

Two Sufficient Descent Conjugate Gradient Methods for Unconstrained Optimization

对无约束优化问题提出两类新的充分下降共轭梯度法.在每次迭代过程中,算法均可得到充分下降方向.在适当条件下,证明了算法的全局收敛性.数值结果表明算法可行、有效.

We proposed two sufficient descent conjugate gradient methods for unconstrained optimiza- tion. The sufficient descent direction is always obtained at each iteration. Under some suitable conditions, the global convergence can be induced. Numerical results show that these methods are feasible and effective.