Accessing object data slicing numpy array

Question

So i'm working on optimizing some python code by using numpy arrays instead of for each loops. Is there any way to access fields from a class during slicing of a numpy array?

class foo:

    the_int = 0

    def __init__(self, i):
        self.bar = i

one, two = foo(5), foo(10)
ar = array([one, two])

int_array = ar[0:2].the_int

#I want int_array = [5, 10]

If that isn't possible in that manner, how would I efficiently generate my "int_array" variable without using a for each loop to loop through "ar", gather "the_int" from each entry?

Thanks, Kyle

score 2 · Accepted Answer · edited May 23 '17 at 12:24

2

Why are you using a numpy array to store PyObjects? You won't get the performance improvement you think. See here.

Using a list, you can use a list comprehension instead:

class foo:

    the_int = 0

    def __init__(self, i):
        self.bar = i

one, two = foo(5), foo(10)

ar = [one, two]

int_array = [i.bar for i in ar[0:2]]

edited May 23 '17 at 12:24

Community

1
1

answered Apr 15 '16 at 14:07

user3684792

2,542
2
18
23

Ahh, that makes sense. Didn't realize that numpy arrays shouldn't store objects. So there's no way with objects to avoid the for each loop? – Kyle Hunter Apr 15 '16 at 14:09
1

even though the syntax is similar, the list comprehension I have used is not actually a for loop – user3684792 Apr 15 '16 at 14:14
Yeah, just noticed that. It's significantly faster that my original for loop. Thanks a lot for your help! – Kyle Hunter Apr 15 '16 at 14:16
1

Something, `for loop` or otherwise, has to query each object (or the list or array) for its `bar` attribute. Your object has been coded in Python, so that query is a Python operation. – hpaulj Apr 15 '16 at 15:49

Accessing object data slicing numpy array

1 Answers1