np. transpose ＆ np.expand_ dims

One 、numpy.transpose(a, axes=None)

   function ：
     Transform the high-dimensional array on the coordinate axis , That is, perform transpose operation .
   Parameters ：
    a： Numpy Input array
    axes： If specified , It must contain [0,1,…,N-1] Arranged tuples or lists , among N yes a The number of shaft ( It can be understood as dimension ). Returns the second row of the array i The axes will correspond to axes[i] Entered number axis . If not specified , The default is range(a.ndim)[::-1], This will reverse the order of the shafts
   return ：
    dst： Transposed array

  
   If expressed in linear algebra, it is the same as the following figure ：

   Put it simply transpose() The function of Change the position of different dimensions in the array .
   for instance ：

import numpy as np

x = np.ones((1,2,3))  # We first create a three-dimensional array 

We use transpose() Function exchange x Positions of different dimensions of the array ：

x_temp = np.transpose(x, (1, 0, 2))

Obviously , We can see x The first of an array of 0 Uighur 2 Dimension exchange position , It was (1,2,3) The format array becomes (2,1,3) Format .

Two 、numpy.expand_dims(a, axes=None)

   function ：
     Add a new dimension to the array
   Parameters ：
    a： Numpy Input array
    axes： Place new dimensions （ Or multiple dimensions ） The location of .
   return ：
    dst： The array after adding the new dimension

Example ：

x = np.ones((1,2,3)) 
x_temp2 = np.expand_dims(x,0)  # stay 0 Dimension insert a new dimension 

x = np.ones((1,2,3)) 
x_temp2 = np.expand_dims(x,axis=(0, 1))  #axis It could also be a tuple