NumPy transpose()

The transpose() method swaps the axes of the given array similar to the transpose of a matrix in mathematics. In the case of arrays with more than two dimensions, transpose() permutes the axes based on the given argument.

Example

import numpy as np

originalArray = np.array([[1, 2], [3, 4]])

# swaps the rows and columns of an array transposedArray = np.transpose(originalArray)
print(transposedArray) ''' Output: [[1 3] [2 4]] '''

transpose() Syntax

The syntax of transpose() is:

numpy.transpose(array, axes = None)

transpose() Arguments

The transpose() method takes two arguments:

  • array - the array to be transposed
  • axes(optional) - the axes of the transposed matrix ( tuple or list of integers )

transpose() Return Value

The transpose() method returns an array with axes permuted.


Example 1: Transposing an Array Without Axes Argument

import numpy as np

originalArray = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

# if axes argument is skipped, the shape of array is reversed transposedArray = np.transpose(originalArray)
print(transposedArray)

Output

[[1 4 7]
 [2 5 8]
 [3 6 9]]

Transposing a two-dimensional NumPy array is the same as transposing a matrix.


Example 2: Transposing a 1D Array

If we use the transpose() method on a one-dimensional array, the method returns the original array.

import numpy as np

originalArray = np.array([1, 2, 3, 4])

# try to swap the rows and columns of the given array transposedArray = np.transpose(originalArray)
print(transposedArray)

Output

 [1 2 3 4]

Note: The transpose() method cannot increase the dimension of an array. Hence, the method cannot convert a unidirectional row array to a column array.


Example 3: Transposing a 3D Array

The axes argument (second argument) defines which axes are swapped.

import numpy as np

originalArray = np.array([[[0, 1], [2, 3]], [[4, 5], [6, 7]]])

# swap the rows and columns of the array based on given axes print('When axes are (2, 0, 1):') transposedArray = np.transpose(originalArray, (2, 0, 1))
print(transposedArray)
# swap the rows and columns of the array based on given axes print('When axes are (0, 2, 1):') transposedArray = np.transpose(originalArray, (0, 2, 1))
print(transposedArray)

Output

When axes are (2, 0, 1):
[[[0 2]
  [4 6]]

 [[1 3]
  [5 7]]]

When axes are (0, 2, 1):
[[[0 2]
  [1 3]]

 [[4 6]
  [5 7]]]

Note: NumPy uses indices to represent dimensions, unlike traditional concepts like x, y, and z. The dimensions are represented as dimension 0, dimension 1, and so on.

Thus, the axes argument cannot have repeated elements since a dimension can only appear once.

1. Using (2, 0, 1) as the second argument to transpose()

In the original array, (x, y, z) are represented by (0, 1, 2).

If we use (2, 0, 1) as the second argument to the transpose() method, it means

  • z-axis (index 2) becomes the new x-axis
  • x-axis (index 0) becomes the new y-axis
  • y-axis (index 1) becomes the new z-axis

2. Using (0, 2, 1) as the second argument to transpose()

If we use (0, 2, 1) as the second argument to the transpose() method, it means

  • x-axis (index 0) will remain unchanged
  • z-axis (index 2) becomes the new y-axis
  • y-axis (index 1) becomes the new z-axis

Note: It is hard to visualize the working of transpose() for arrays having more than two dimensions. To learn more, visit working of transpose() on a 3D array.