How to Use np.expand_dims in NumPy: Syntax and Examples

Use np.expand_dims to add a new axis to a NumPy array at a specified position, increasing its dimensions by one. This is useful to reshape arrays for broadcasting or aligning shapes without changing data.

Syntax

The syntax of np.expand_dims is:

• np.expand_dims(a, axis)

Where:

• a is the input array.
• axis is the position where the new axis is inserted. It can be a positive or negative integer.

python

np.expand_dims(a, axis)

Example

This example shows how to add a new axis to a 1D array to make it 2D by inserting the axis at position 0 and 1.

python

import numpy as np

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

# Add new axis at position 0 (makes shape (1, 3))
expanded_0 = np.expand_dims(arr, axis=0)

# Add new axis at position 1 (makes shape (3, 1))
expanded_1 = np.expand_dims(arr, axis=1)

print('Original shape:', arr.shape)
print('Expanded shape axis=0:', expanded_0.shape)
print('Expanded array axis=0:\n', expanded_0)
print('Expanded shape axis=1:', expanded_1.shape)
print('Expanded array axis=1:\n', expanded_1)

Output

Original shape: (3,)
Expanded shape axis=0: (1, 3)
Expanded array axis=0:
 [[1 2 3]]
Expanded shape axis=1: (3, 1)
Expanded array axis=1:
 [[1]
 [2]
 [3]]

Common Pitfalls

Common mistakes when using np.expand_dims include:

• Using an axis value outside the valid range. The axis must be between -a.ndim-1 and a.ndim.
• Confusing np.expand_dims with reshape or newaxis. While related, expand_dims is clearer for adding a single axis.

python

import numpy as np

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

# Wrong: axis out of range
# np.expand_dims(arr, axis=3)  # This will raise an error

# Correct: axis within range
expanded = np.expand_dims(arr, axis=2)
print('Expanded shape:', expanded.shape)

Output

Expanded shape: (3, 1)

Quick Reference

Parameter	Description
a	Input array to expand
axis	Position to insert new axis (int, can be negative)
Returns	Array with one more dimension

Key Takeaways

Use np.expand_dims to add a new axis to an array without changing data.

The axis parameter controls where the new dimension is inserted.

Axis must be within the range [-a.ndim-1, a.ndim].

It is useful for preparing arrays for broadcasting or model inputs.

np.expand_dims is clearer and safer than manual reshaping for adding axes.