How to Create an Identity Matrix in NumPy Easily

You can create an identity matrix in NumPy using the numpy.eye() function, which returns a 2-D array with ones on the diagonal and zeros elsewhere. Specify the size with the N parameter to get an N x N identity matrix.

Syntax

The basic syntax to create an identity matrix in NumPy is:

• numpy.eye(N, M=None, k=0, dtype=float, order='C')

Where:

• N: Number of rows (and columns if M is not given).
• M: Number of columns (optional, defaults to N).
• k: Index of the diagonal (0 for main diagonal, positive for above, negative for below).
• dtype: Data type of the output array.
• order: Memory layout ('C' for row-major, 'F' for column-major).

python

import numpy as np

# Create a 3x3 identity matrix
identity_matrix = np.eye(3)
print(identity_matrix)

Output

[[1. 0. 0.]
 [0. 1. 0.]
 [0. 0. 1.]]

Example

This example shows how to create a 4x4 identity matrix and a 3x5 matrix with ones on the main diagonal using numpy.eye().

python

import numpy as np

# 4x4 identity matrix
identity_4x4 = np.eye(4)
print('4x4 Identity Matrix:')
print(identity_4x4)

# 3x5 matrix with ones on main diagonal
identity_3x5 = np.eye(3, 5)
print('\n3x5 Matrix with ones on main diagonal:')
print(identity_3x5)

Output

4x4 Identity Matrix:
[[1. 0. 0. 0.]
 [0. 1. 0. 0.]
 [0. 0. 1. 0.]
 [0. 0. 0. 1.]]

3x5 Matrix with ones on main diagonal:
[[1. 0. 0. 0. 0.]
 [0. 1. 0. 0. 0.]
 [0. 0. 1. 0. 0.]]

Common Pitfalls

Common mistakes when creating identity matrices include:

• Confusing numpy.eye() with numpy.identity(). The identity() function only creates square matrices, while eye() can create rectangular matrices.
• Forgetting that the default diagonal is the main diagonal (k=0), so changing k shifts the diagonal of ones.
• Not specifying dtype when you need a specific data type, which defaults to float.

Example of a wrong and right way:

python

import numpy as np

# Wrong: expecting identity() to create non-square matrix (this will error)
try:
    np.identity(3, 5)
except TypeError as e:
    print('Error:', e)

# Right: use eye() for non-square matrix
rect_matrix = np.eye(3, 5)
print('\nCorrect rectangular identity matrix with eye():')
print(rect_matrix)

Output

Error: identity() takes exactly one argument (2 given)

Correct rectangular identity matrix with eye():
[[1. 0. 0. 0. 0.]
 [0. 1. 0. 0. 0.]
 [0. 0. 1. 0. 0.]]

Quick Reference

Summary tips for creating identity matrices in NumPy:

• Use numpy.eye(N) for an N x N identity matrix.
• Use numpy.eye(N, M) for rectangular matrices with ones on the main diagonal.
• Adjust k to move the diagonal of ones up or down.
• Specify dtype if you need integers or other types instead of floats.

Key Takeaways

Use numpy.eye() to create identity matrices with flexible sizes and shapes.

The parameter k controls which diagonal has the ones; 0 means the main diagonal.

numpy.identity() only creates square identity matrices and takes one argument.

Specify dtype in numpy.eye() to control the data type of the matrix elements.

For rectangular identity-like matrices, always prefer numpy.eye() over numpy.identity().