共轭梯度法（Conjugate Gradient）是介于最速下降法与牛顿法之间的一个方法，它仅需利用一阶导数信息，但克服了最速下降法收敛慢的缺点，又避免了牛顿法需要存储和计算Hesse矩阵并求逆的缺点，共轭梯度法不仅是解决大型线性方程组最有用的方法之一，也是解大型非线性最优化最有效的算法之一。-Conjugate gradient method (Conjugate Gradient) between the steepest descent between law and Newton' s method is a method, it is only the first derivative information, but to overcome the steepest descent method of slow convergence shortcomings, but also avoid the Newton method needs to be stored and calculate the Hesse matrix and the inverse of the shortcomings of the conjugate gradient method is not only the most useful way to solve the large linear equations, one is also the solution of large-scale nonlinear optimization one of the most effective algorithm.