摘要: |
CF-PCG算法是牛顿法和预优共轭梯度法结合起来求解牛顿方程的一种非精确牛顿法。笔者将自动微分技术应用到该算法中,并证明应用自动微分技术的CF-PCG方法具有更高的效率。 |
关键词: 牛顿法 预优共轭梯度法 自动微分 牛顿方程 |
DOI: |
修订日期:2000-07-17 |
基金项目:国家自然科学基金资助项目 |
|
CF-PCG Method Using Automatic Differentiation and Its Efficience Analysis |
|
|
Abstract: |
Newton PCG method is an inexact Newton like method. It is an organic combination of Newton's method and preconditioned conjugate gradient method. Automatic differentiation method are applied in this algorithm, it has been shown that CF PCG method appling automatic differentiation techniques is more efficient than the original CF PCG method using symbolic differentiation method. |
Key words: Newton's method,preconditioned conjugate gradient method,automatic differentiation |