Using itertools to index an array

Question

Here is a section of code which is giving me a different answer to what I would expect. The line: print list(x) does what I expect. I would expect the line: print random_array[list(x)] to return the value of that element in the array but it returns three arrays. If for example list(x) returns [9, 8, 7] then random_array[9, :, :], random_array[8, :, :], random_array[7, :, :] will be printed. Can someone please explain to me why this is? And how I can get the expected answer?

import numpy as np
import itertools

random_array = np.random.randint(0, 9, (10, 10, 10))
my_iterator = itertools.product(range(10),range(10),range(10))

for x in my_iterator:
    print list(x)
    print random_array[list(x)]

I have closed my answer. Numpy does support random_array[(2,3,3)] — ninjagecko
– ninjagecko, Commented Feb 24, 2012 at 14:44

ninjagecko · Accepted Answer · 2012-02-24 14:47:35Z

3

You are passing in a list rather than a tuple:

# What you are doing
random_array[[2, 3, 3]]  # semantics: [arr[2], arr[3], arr[3]]

# What you want to be doing
random_array[(2, 3, 3)]  # semantics: arr[2][3][3], same as arr[2,3,3]

In short: do not cast your tuples to lists using list(...).

edited Feb 24, 2012 at 14:47

answered Feb 24, 2012 at 14:38

ninjagecko

91.5k24 gold badges144 silver badges153 bronze badges

Sign up to request clarification or add additional context in comments.

7 Comments

nicholaschris Over a year ago

I see that I am passing in a list rather than I slice. So: random_array[[2, 3, 3]]

Marcin Over a year ago

Can you expand the tuple like: random_array[*(2, 3, 3)]

ninjagecko Over a year ago

@Marcin: that's exactly what I was in the process of testing

ninjagecko Over a year ago

I believe this answer explains more than the chosen answer.

Bi Rico Over a year ago

Here are some docs if anyone wants to know more about numpy indexing. docs.scipy.org/doc/numpy/reference/arrays.indexing.html

|

Bogdan · Accepted Answer · 2012-02-24 14:45:27Z

1

I think what you want is:

print random_array[x[0], x[1], x[2]]

If you pass a list as index to numpy it will iterate trough the index list and get you that slice of elements. For example:

>>> test = numpy.array(range(10))
>>> idx = [1, 2, 3]
>>> test[idx]
array([1, 2, 3])

edited Feb 24, 2012 at 14:45

answered Feb 24, 2012 at 14:39

Bogdan

8,3166 gold badges51 silver badges66 bronze badges

Comments

georg · Accepted Answer · 2012-02-24 14:54:40Z

1

How about

print random_array[x]

When you pass a list, advanced indexing is taking place, which is not what you want.

edited Feb 24, 2012 at 14:54

answered Feb 24, 2012 at 14:46

georg

216k57 gold badges324 silver badges401 bronze badges

Comments

Marcin · Accepted Answer · 2012-02-24 14:32:34Z

0

You say:

I would expect the line: print random_array[list[x]) to return the value of that element in the array

But your code does not contain such a line. I expect this is the cause of your problem.

answered Feb 24, 2012 at 14:32

Marcin

50.1k18 gold badges137 silver badges207 bronze badges

4 Comments

kͩeͣmͮpͥ ͩ Over a year ago

Stop being such a pendant. Clearly he's referring to the last line of his extract.

nicholaschris Over a year ago

Yes that was a typo. I meant "I would expect the line: print random_array[list(x)]" and I have edited the question accordingly.

Marcin Over a year ago

@DavidKemp: Tell that to your python interpreter.

nicholaschris Over a year ago

Marcin has a point I guess. Precision is important in all aspects.

Collectives™ on Stack Overflow

Using itertools to index an array

4 Answers 4

7 Comments

Comments

Comments

4 Comments

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

7 Comments

Comments

Comments

4 Comments

Related