We present nonmonotonic trust region algorithm via the conjugate gradient path for solving un-constrained optimization and further prove that the proposed algorithm has global convergence and fastasymptotic convergence.

• 提供了无约束最优化问题的共轭梯度路径非单调信赖域算法。进而获得了共轭梯度路的一些重要性质。基于这些性质和一些合理的假设,证明了算法具有整体收敛性和保持局部超线性收敛速率。