NumPy Tutorial #5: Array Copy vs View

How can you make a duplicate of a NumPy array? There are two NumPy array methods that you can use to perform that. Those two methods are copy and view. What is the difference between these two?

---

Copy

Let me start with the copy method. The copy method in here can be refered to "deep duplication". Let me show what it means through an example.

Let us create a NumPy array

import numpy as np

list_of_number = [1, 3, 5, 7, 9]

arr = np.array(list_of_number)

Then let us make a duplication of the array called arr2 from the original array using copy method

arr2 = arr.copy()

Let us update an item from arr2 and print arr also with arr2 to see the output

arr2[0] = 90

print(arr)
print(arr2)

You will see the following output

[1 3 5 7 9]
[90  3  5  7  9]

From the output, you can see that even if we update a value of arr2, it will not update the original array. This is what it means by "deep duplication".

---

View

Now let us see what view will do to an array. Let us update our existing code from .copy() to .view()

arr2 = arr.view()

Then update an item from arr2 and print both arr and arr2 to see the output

arr2[0] = 90

print(arr)
print(arr2)

You will see the following output

[90  3  5  7  9]
[90  3  5  7  9]

You will see that any update we perform in the arr2, it also update arr in the process. This can also refered to "shallow copy".

---

There you go, you can use "copy" or "view" to duplicate a NumPy array depending on your situation. Thank you for reading this blog, and see you next time!