Newton法与共轭梯度法的组合方法

    Combination of Newton's Method and Conjugate Gradient Method

    • 摘要: 本文在Newton法与最速下降法的组合方法的基础上提出了Newton法与共轭梯度法的组合方法,并证明了该方法的全局收敛性。该方法能有效地应用于目标函数的Hessian矩阵不正定或病态等实际问题。计算HK的过程简化,大大减少了算法的计算量。

       

      Abstract: This paper presents a combined method of Newton's method and Conjugate cradient method based on the combined method of Newton's method and the Gradient Descent method.The global convergence of the algorithm is proved.This method could be effectively applied to practical problems where Hessian matrix of objective function is not positive definite or is ill-conditioned.The process of computing HK is simplified resulting in great reduction in the amount of algorithm computation.

       

    /

    返回文章
    返回