NumPy - Insert Axes to an Array

---

---

Insert Axes to an Array in NumPy

Inserting axes into a NumPy array refers to adding new dimensions to the existing array. This is particularly useful when you need to align data for broadcasting, reshape arrays for specific operations, or add new dimensions to facilitate operations such as stacking or concatenation.

The primary function used for this purpose is np.expand_dims(), which adds a new axis at a specified position.

Insert Axes Using expand_dims() Function

The np.expand_dims() function in NumPy is used to insert a new axis (dimension) into an existing array, thereby increasing its dimensionality. Following is the syntax −

numpy.expand_dims(a, axis)

Where,

• a: It is the input array.
• axis: It is the position in the dimensions where the new axis is to be inserted.

Example

In the following example, we start with creating a 1D array. Using np.expand_dims() function, we add a new axis at position 1, transforming the array into a 2D column vector −

import numpy as np

# Creating a 1D NumPy array
arr = np.array([1, 2, 3, 4])

# Adding a new axis to create a 2D column vector
expanded_arr = np.expand_dims(arr, axis=1)

print("Original Array:\n", arr)
print("Array with New Axis:\n", expanded_arr)

Following is the output obtained −

Original Array:
[1 2 3 4]
Array with New Axis:
[[1]
 [2]
 [3]
 [4]]

Insert Axes to Create Higher-dimensional Arrays

Inserting axes to create higher-dimensional arrays means adding new dimensions to an existing array, which changes its shape. This process allows you to expand the array from, for example, 1D to 2D or from 2D to 3D.

This is useful for operations like broadcasting or reshaping data to fit specific requirements.

Example

In this example, we are expanding a 1D NumPy array to a 3D array by adding two new axes using the np.expand_dims() function −

import numpy as np

# Creating a 1D NumPy array
arr = np.array([1, 2, 3])

# Adding two new axes to create a 3D array
expanded_arr = np.expand_dims(arr, axis=(0, 2))

print("Original Array:\n", arr)
print("3D Array with New Axes:\n", expanded_arr)
print("Shape of 3D Array:", expanded_arr.shape)

This will produce the following result −

Original Array:
[1 2 3]
3D Array with New Axes:
[[[1]
  [2]
  [3]]]
Shape of 3D Array: (1, 3, 1)

Insert Axes Using "None" or "np.newaxis" Indexing

An alternative way to insert axes is by using None or np.newaxis indexing. This approach is generally used for its simplicity and readability in code.

By indexing with "None" or "np.newaxis", you can expand a 1D array to 2D or 3D, or adjust the shape as needed.

Example

In the example below, we are transforming a 1D NumPy array into a 2D row vector by adding a new axis with np.newaxis indexing −

import numpy as np

# Creating a 1D NumPy array
arr = np.array([1, 2, 3, 4])

# Adding a new axis to create a 2D row vector
row_vector = arr[:, np.newaxis]

print("Original Array:\n", arr)
print("2D Row Vector:\n", row_vector)

Following is the output of the above code −

Original Array:
[1 2 3 4]
2D Row Vector:
 [[1]
 [2]
 [3]
[4]]

Combining Multiple Arrays with Inserted Axes

Combining multiple arrays with inserted axes involves adding new dimensions to each array so they align properly for concatenation or stacking.

By inserting axes, you ensure that arrays of different shapes can be joined together. This technique allows for flexible data manipulation and performing various operations.

Example

In the following example, we are adding a new axis to each 1D array to convert them into 2D row vectors. We then concatenate these row vectors along axis 0, resulting in a 2D array where each original array is now a separate row −

import numpy as np

# Creating two 1D arrays
arr1 = np.array([1, 2, 3])
arr2 = np.array([4, 5, 6])

# Adding new axes to create 2D row vectors
arr1_expanded = arr1[np.newaxis, :]
arr2_expanded = arr2[np.newaxis, :]

# Concatenating the row vectors along axis 0
combined_arr = np.concatenate([arr1_expanded, arr2_expanded], axis=0)

print("Array 1 with New Axis:\n", arr1_expanded)
print("Array 2 with New Axis:\n", arr2_expanded)
print("Combined Array:\n", combined_arr)

The output obtained is as shown below −

Array 1 with New Axis:
[[1 2 3]]
Array 2 with New Axis:
 [[4 5 6]]
Combined Array:
[[1 2 3]
 [4 5 6]]