numpy: Indexing

import numpy as np

A = np.array([[1, 4, 9], [2, 8, 18]])
print(A)

[[ 1  4  9]
 [ 2  8 18]]

A[1,2]

18

What's the result of this?

A[:,1]

array([4, 8])

And this?

A[1:,:1]

array([[2]])

One more:

A[:,[0,2]]

Iclicker questions

#clear Select the correct output: A) B) C) D)

#clear Select the correct output: A) [[ 8 3 18] [ 5 1 2]] B) [[ 3 18] [ 1 2]] C) [[3 18]] D) [[8 3 18]]

For higher-dimensional arrays we can use ... like:

a = np.random.rand(3,4,2)
a.shape

a[...,1].shape

---

Indexing into numpy arrays usually results in a so-called view.

a = np.zeros((4,4))

Let's call b the top-left $2\times 2$ submatrix.

b = a[:2,:2]

What happens if we change b?

b[1,0] = 5

a

To decouple b from a, use .copy().

b = b.copy()
b[1,1] = 7
print(b)
print(a)

iclicker question

#clear What is the output of print(A)? A) [[ 1 4 9 3] [ 2 8 3 18] [ 4 5 2 2] [ 6 4 6 3]] B) [[ 1 4 9 3] [ 2 8 3 18] [ 4 5 1 2] [ 6 4 6 3]] C) none of the above

---

You can also index with boolean arrays:

a = np.random.rand(4,4)

a

a_big = a>0.5
a_big

a[a_big]

Also each index individually:

a_row_sel = [True, True, False, True]

a[a_row_sel,:]

---

And with index arrays:

a

x,y = np.nonzero(a > 0.5)

x

y

a[(x,y)]