Why does out of range numpy list slicing not raise IndexError

Asked May 07 '20 at 19:22

Active May 07 '20 at 20:03

Viewed 27 times

This is the shape of the array I'm using and I'm wondering why this doesn't result in an IndexError. The array result is just the same array.

import numpy as np
example = np.zeros((240, 320, 3))
example = example[-999:999,-999:999,:]

This holds for an example like this too:

x = np.array([[2,3,4],[3,4,5]])
a = x[-10:10,-10:10]

Since I'm relying on this raising an exception, it is confusing for me. Does numpy just index using the in bounds array and not the values that lay outside the range?

edited May 07 '20 at 20:03

asked May 07 '20 at 19:22

mkohler

1

My best guess is that they try to keep the slice behaviour consistent with lists and other sequences, as shown in the dupe – roganjosh May 07 '20 at 19:29
you can use `np.arange(-999,1000,1)`. Slicing is "forgiving", indexing is not. – FBruzzesi May 07 '20 at 19:29
See also https://stackoverflow.com/questions/54632256/numpy-slicing-with-bound-checks – roganjosh May 07 '20 at 19:39

Why does out of range numpy list slicing not raise IndexError

0 Answers0