基于二阶导数的优化算法

本文针对梯度法收敛速度慢，易于陷入局部极值的缺点。提出了一种新的基于二阶导数的下降方向，由此导出了新的优化算法即ND算法。该算法计算简便，收敛速度较梯度法快。实例验证效果较好。

A Study on the Optimization Algorithm with the Introduction of Second Derivative

In this paper, the Gradient Method has defects: it has a slow convergencerate and goes to pratial value easily. A new descending direction with second derivative isproposed, it result a new optimization algorithm is ND algorithm. The computationalalgorithm computing is easy, The convergence rate of ND algorithm is faster than that ofgradient method. The real eXamples show that computing result are satisfactory.