- np.transpose(a,axes)
- By deafault,reverse the dimensions,otherwise permute the axes according to the values given.
- a : input array
- axes : list of ints,optional
Transpose is a normal operation in matrix,means change the order of matrix dimention,such as a tensor with dimention of (2,3),if we perform a transpose on it ,it's shape will be changed into (3,2).And the argument axes is an value to control the transpose dimentions.if we have a tensor with shape of (3,4,3) and axes=(1,0,2) ,the transform fucntion will change the shape of (3,4,3) into (4,3,3).There are two example to explain it.
Example 1:
import numpy as np
a=np.array([[1,2],
[3,4]])
b=np.transpose(a)
print(b)
Result:
[[1 3]
[2 4]]
Example 2:
import numpy as np
a=np.ones((3,4,3))
b=np.transpose(a,(1,0,2))
print(b.shape)
Result:
(4, 3, 3)