Copying vs Viewing Arrays in NumPy

When working with NumPy arrays, it’s crucial to understand whether a new operation creates a copy or a view of the array. This distinction determines whether modifying one affects the other.

Copy vs View: Key Differences

Creating a Copy

A copy creates a completely new array with its own data. Changes made to the copy do not affect the original array.

Example: Using copy()

import numpy as np
 
# Original array
array = np.array([1, 2, 3, 4, 5])
 
# Create a copy
array_copy = array.copy()
 
# Modify the copy
array_copy[0] = 99
 
print("Original array:", array)  # Output: [1 2 3 4 5]
print("Modified copy:", array_copy)  # Output: [99 2 3 4 5]

Output:

Original array: [1 2 3 4 5]
Modified copy: [99 2 3 4 5]

Creating a View

A view is a reference to the original array. Changes made to the view also reflect in the original array.

Example: Using Slicing (View)

# Original array
array = np.array([1, 2, 3, 4, 5])
 
# Create a view
array_view = array[1:4]
 
# Modify the view
array_view[0] = 99
 
print("Original array:", array)  # Output: [ 1 99  3  4  5]
print("Modified view:", array_view)  # Output: [99  3  4]

Output:

Original array: [ 1 99  3  4  5]
Modified view: [99  3  4]

Identifying Copies and Views

Use the base attribute to determine if an array is a view. If base is None, the array is a copy.

Example:

# Original array
array = np.array([1, 2, 3, 4, 5])
 
# Create a copy and a view
array_copy = array.copy()
array_view = array[1:4]
 
# Check base attribute
print("Copy base:", array_copy.base)  # Output: None
print("View base:", array_view.base)  # Output: [1 2 3 4 5]

Output:

Copy base: None
View base: [1 2 3 4 5]

Practical Implications

When to Use a Copy:

• If the original array must remain unchanged.
• For independent data manipulations.

When to Use a View:

• To save memory when working with large datasets.
• For temporary or localized changes.

Try It Yourself

Problem 1: Identifying Copy vs View

Create an array and a sliced view of it. Modify the view and observe the changes in the original array. Use the base attribute to verify if it’s a copy or a view.

import numpy as np
 
# Create an array
array = np.array([10, 20, 30, 40, 50])
 
# Create a view
array_view = array[2:4]
 
# Modify the view
array_view[0] = 99
 
# Check base attribute
print("Original array:", array)  # Output: [10 20 99 40 50]
print("View base:", array_view.base)  # Output: [10 20 99 40 50]

Problem 2: Working with Copies

Create an array and make a copy of it. Modify the copy and verify that the original array remains unchanged.

import numpy as np
 
# Create an array
array = np.array([5, 10, 15, 20, 25])
 
# Create a copy
array_copy = array.copy()
 
# Modify the copy
array_copy[2] = 99
 
# Check arrays
print("Original array:", array)  # Output: [ 5 10 15 20 25]
print("Modified copy:", array_copy)  # Output: [ 5 10 99 20 25]