高斯赛德尔迭代程序

高斯赛德尔迭代非常常用，看到网上很多例子写的不够简洁，这里我写了一个，供参考

import numpy as np

def gauss_seidel(A,b,x1,eps=1.e-6):
    n = len(A)
    max_iter = 200
    iters = 0
    while   abs(np.dot(A[0,:],x1) - b)[0] > eps and iters < max_iter:#对二维矩阵，只判断了一行 
        #判断，为了减少计算量，只判断第一个点和最后一个点
        for i in range(n):
            median_value1 = 0.
            for j in range(0,n):
                if j != i:      #只需要挖掉i==j的点就可以了
                    median_value1 = median_value1+A[i,j]*x1[j]
            x1[i] = (b[i] - median_value1 )/A[i,i]
        iters += 1
    return x1,iters


A = np.array([
[10.,-1.,-2],
[-1.,10.,-2.],
[-1.,-1.,5. ]
]) 
b = [72.,83.,42.]
initial = np.array([1.,1.,1.])

x1,iters = gauss_seidel(A,b,initial)
print("iters = ",iters)
print("solution = ",x1)
print(np.dot(A,x1))