基于修正割线条件的NDL方法

A modified DL method with global convergence

基于带有自调比参数的修正DL方法,提出一个带有修正割线条件的修正共轭梯度法NDL方法.在满足Wolfe线搜索的条件下,证明了该方法具有充分下降性;最后分析并且证明,该方法在满足强Wolfe条件下,对一般函数都会具有全局收敛性.

Based on the modified DL method with self scaling parameters, a modified conjugate gradient method （NDL） with modified secant conditions has been proposed,The step length satisfies the Wolfe line search, it is proved that the meth od has sufficient descending property. Finally, the method is proved to have a general function under the strong Wolfe condition. There is a global convergence.