NumPy copy()

Summary: in this tutorial, you’ll learn how to use the NumPy copy() method to create a copy of an array rather than a view.

Introduction to the NumPy copy() method

When you slice an array, you get a subarray. The subarray is a view of the original array. In other words, if you change elements in the subarray, the change will be reflected in the original array. For example:

import numpy as np

a = np.array([
    [1, 2, 3],
    [4, 5, 6]
])

b = a[0:, 0:2]
print(b)

b[0, 0] = 0
print(b)
print(a)Code language: Python (python)

How it works.

First, create a 2D array:

a = np.array([
    [1, 2, 3],
    [4, 5, 6]
])Code language: Python (python)

Second, slice the array a and assign the subarray to the variable b:

b = a[0:, 0:2]Code language: Python (python)

The variable b is:

[[1 2]
 [4 5]]Code language: Python (python)

Third, change the element at index [0,0] in the subarray b to zero and display the variable b:

b[0, 0] = 0
print(b)Code language: Python (python)

[[0 2]
 [4 5]]Code language: Python (python)

Since b is a view of array a, the change is also reflected in array a:

print(a)Code language: Python (python)

[[0 2 3]
 [4 5 6]]Code language: Python (python)

The reason numpy creates a view instead of a new array is that it doesn’t have to copy data therefore improving performance.

However, if you want a copy of an array rather than a view, you can use copy() method. For example:

import numpy as np

a = np.array([
    [1, 2, 3],
    [4, 5, 6]
])

# make a copy
b = a[0:, 0:2].copy()
print(b)

b[0, 0] = 0
print(b)

print(a)Code language: Python (python)

In this example:

First, call the copy() method of array a to make a copy of a subarray and assign it to the variable b.

Second, change the element at index [0,0] of the array b, because both arrays are independent, the change doesn’t affect array a.

Summary

• When you slice an array, you’ll get a view of the array.
• Use the copy() method to make a copy of an array rather than a view.