【np.power()参数为矩阵】

形状相同

用例代码

a1 = np.array([[1,2,3],
              [4,5,6]])

a2 = np.array([[1,2,3],
               [4,5,6]])
print(np.power(a1,a2))               

输出

[[    1     4    27]
 [  256  3125 46656]]

结论：两相同形状矩阵，按元素一一对应计算。

行数不同

用例代码

a1 = np.array([[1,2,3],
              [4,5,6]])

a3 = np.array([1,2,3])

a4 = np.array([[1,2,3],
              [1,2,3]])
print(np.power(a1,a3))
print(np.power(a1,a4))
print(np.power(a3,a1))
print(np.power(a4,a1))

输出

[[  1   4  27]
 [  4  25 216]]
[[  1   4  27]
 [  4  25 216]]
[[  1   4  27]
 [  1  32 729]]
[[  1   4  27]
 [  1  32 729]]

结论：触发广播机制，小矩阵扩展成和大矩阵相同形状再计算。

列数不同

用例代码

a1 = np.array([[1,2,3],
              [4,5,6]])
a5 = np.array([[1,2],
              [4,5]])
print(np.power(a1,a5))

输出

Traceback (most recent call last):
  File "D:/Users/**/PycharmProjects/pythonProject/main.py", line 38, in <module>
    print(np.power(a1,a5))
ValueError: operands could not be broadcast together with shapes (2,3) (2,2) 

结论：报错，不让这么干。