Python 列表中的值在 for 循环中不会改变

Question

我正在对某种类型的三对角矩阵进行 LU 分解，但我遇到了一个我完全无法理解的 Python 问题。

def trisolve2(b):
    length = len(b)

    # Solve Ly = b
    y = b.copy()
    for i in range(1, length):
        print('y[' + str(i) + '] = ' + str(b[i]) + ' - (-' + str(i) + ' / (' + str(i+1)
            + ')) * ' + str(y[i-1]))
        y[i] = b[i] - (-1 * i / (i+1)) * y[i-1]
    print("y = " + str(y))

    # Solve Ux = y
    x = y.copy()
    for i in range(length - 2, 0, -1):
        print("x = " + str(x))
        x[i] =  (y[i] + x[i+1]) / ((i+2) / (i+1))

    return x

test_b = np.array([[1], [0], [0], [0], [1]])
print("b = " + str(test_b))
test_x = trisolve2(test_b)
print("x = " + str(test_x))

我希望当 i = 1 时，y[i] 被分配到 1/2，它甚至看起来应该与我放在那里的打印语句一样，但由于某种原因我无法包装我的绕着它走永远不会改变。 y 保持 b，x 保持 y 不变。输出是：

b = [[1]
 [0]
 [0]
 [0]
 [1]]
y[1] = [0] - (-1 / (2)) * [1]
y[2] = [0] - (-2 / (3)) * [0]
y[3] = [0] - (-3 / (4)) * [0]
y[4] = [1] - (-4 / (5)) * [0]
y = [[1]
 [0]
 [0]
 [0]
 [1]]
x = [[1]
 [0]
 [0]
 [0]
 [1]]
x = [[1]
 [0]
 [0]
 [0]
 [1]]
x = [[1]
 [0]
 [0]
 [0]
 [1]]
x = [[1]
 [0]
 [0]
 [0]
 [1]]```

Answer 1

您的 numpy 数组默认为 int32，0.5 被四舍五入为最接近的整数：1。

>> b = np.array([[1], [0], [0], [0], [1]])
>> b
array([[1],
       [0],
       [0],
       [0],
       [1]])
>> b.dtype
dtype('int32')

尝试传递dtype 参数：

test_b = np.array([[1], [0], [0], [0], [1]], dtype=float)