numpy中的broadcast，以maximum和加法为例

用numpy对array进行maximum或加法操作，当两个参数的维度不同时，首先以加法为例。

>>>a = np.arange(5).reshape(-1,1)
>>>a                                                           # a shape is (3,1)
array([[0],
       [1],
       [2],
       [3],
       [4]])
>>>b = np.array([10,20,30])
b                                                          # b shape is (3,)
>>>array([10, 20, 30])
>>>a+b
array([[10, 20, 30],
       [11, 21, 31],
       [12, 22, 32],
       [13, 23, 33],
       [14, 24, 34]])                           #  a+b shape is (3,3)  用低维度的array与高维度中维度不同的axis依次相加

同理，以maximum为例，维度再加一级

>>>a = np.arange(6).reshape(-1,1,2)
>>>a
array([[[0, 1]],
       [[2, 3]],
       [[4, 5]]])                           # a shape is (3,1,2)

>>>b = np.array([[