CODEWITHSUNDEEP
pythonpython-3.xnumpy
Mike W
Mike W

Reputation: 1403

Difference in numpy indexing

Why this indexing result in different arrays?

import numpy as np

x = np.array(range(0,2*3*4)).reshape(2,3,4)

print(x[0,:,[2,3]])
print(x[0,:,2:])

the first output is

[[ 2  6 10]
 [ 3  7 11]]

the second one is 

[[ 2  3]
 [ 6  7]
 [10 11]]

in the second case 2: means take from the 2nd value until the end, the last column of that dim is the 3 column, that means that it is taking the 2nd and 3rd dimensions, therefore that is the same as [2,3], so what is the difference between both ways of indexing arrays?

Upvotes: 0

Views: 548

Answers (2)

newkid
newkid

Reputation: 1458

In the first case, x[0,:,[2,3]] implies that numpy will return an array such that x[0,:,2] is the first item followed by x[0,:,3]. In the second case, x[0,:,2:] you are asking numpy for 2nd and 3rd column of the 0th matrix.

The indexing documentation is available here.

Upvotes: 1

Jatentaki
Jatentaki

Reputation: 13103

The rules are different for indexing with an array (or a list) of integers and for slicing. This is explained in-depth in the documentation, in particular in the part on mixing advanced and basic indexing.

Upvotes: 2

Related Questions