ValueError: Invalid input for linprog: b_ub must be a 1-D array; b_ub must not have more than one non-singleton dimension and the number of rows in A_ub must equal the number of values in b_ub 仍然报错

非常抱歉，我之前给出的答案有误。在调用linprog函数时，确实需要将b_ub参数传递一个一维数组，且长度与约束矩阵A_ub的行数相等。

这里是修正后的代码：

import numpy as np
from scipy.optimize import linprog


def integer_cutting_plane(c, A, b, bounds):
    relaxed_A = A
    relaxed_b = b
    while True:
        res = linprog(c=c, A_ub=relaxed_A, b_ub=relaxed_b.flatten(), bounds=bounds)
        x = res.x
        if all(int(val) == val for val in x):
            return x.astype(int)

        new_constraint = (relaxed_A @ x <= relaxed_b)
        relaxed_A = np.vstack((relaxed_A, new_constraint))


def get_bounds():
    return [(0, None), (0, None)]


def get_c():
    return np.array([40, 90])


def get_A():
    return np.array([[-9, -7], [-7, -20]])


def get_b():
    return np.array([-56, -70])


if __name__ == '__main__':
    bounds = get_bounds()
    relaxed_A = get_A()
    relaxed_b = get_b()
    integer_cutting_plane(get_c(), relaxed_A, relaxed_b, bounds)

这次我在主函数中添加了两个变量`relaxed_A`和`relaxed_b`，分别用于保存初始的约束矩阵A和约束向量b。然后在调用`integer_cutting_plane`函数时，传递这两个变量而不是直接调用`get_A()`和`get_b()`函数。

这样，通过使用`relaxed_b.flatten()`将`relaxed_b`转换为一维数组，可以解决报错问题。

再次对之前的错误表示歉意，并感谢你的耐心指正。希望这次修改能够解决问题。