共轭梯度法也就是共轭方向法的一种特殊形式，即初始搜索方向为梯度方向。其中根据β的不同又可细分为FR、PRP等方法，主要参考了文件夹中的四本书的内容对其进行总结。最优化算法本质上都是进行最小值的寻优，FR共轭梯度法也不例外。寻优的问题包括两个重要的方面，一个是方向，一个是步长。(Conjugate gradient method is a special form of conjugate direction method, that is, the initial search direction is gradient direction. Among them, according to the different beta, can be subdivided into FR, PRP and other methods, the main reference to the folder in the contents of the four books to summarize it. In essence, the optimization algorithm is the minimal optimization, and the FR conjugate gradient method is no exception. The problem of optimization involves two important aspects: one is the direction, and the other is the step size.)