| 摘要: |
| 利用矩阵条件数的求解方法，求解基于MBFGS割线条件的修正DL共轭梯度法中的参数t，提出带有优选参数的修正DL共轭梯度法；假设搜索方向有下降性，并通过强Wolfe线搜索求解步长，证明了新的共轭梯度法对一般函数有全局收敛性；最后比较了新的共轭梯度法的数值有效性。 |
| 关键词: 共轭梯度法 强Wolfe线搜索 条件数 全局收敛性 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| The Modified DL Conjugate Gradient Method with Optimal Parameter Choices |
|
WU Shuang jiang
|
| Abstract: |
| Making use of the solution of matrix condition number，based on the MBFGS secant condition，the parameter t of modified DL conjugate gradient method is for solution and modified DL conjugate gradient method with optimal parameter choices is presented.If line search direction satisfies descent condition and step size is obtained by strong Wolfe line search，globally convergence for general functions is proved.Finally，the value effectiveness of the new conjugate gradient method is compared. |
| Key words: conjugate gradient method strong Wolfe line search condition number global convergence |
