IndexError: Too many indices when indexing array with another array

Question

I realize a lot of people have asked about this error, but I have yet to find anything that will help me.

This is my code:

def nanalyze(pupil, pw_sim):
    import numpy as np

    temp_s = abs(pw_sim)**2 * pupil

    vals_x, vals_y = np.where(pupil > 0)
    mask = pupil[vals_x[0]:vals_x[len(vals_x)-1], vals_y[0]:vals_y[len(vals_y)-1]]

    s_i = (np.mean(temp_s[tuple(mask)]**2) / (np.mean(temp_s[tuple(mask)])**2)) - 1
    return s_i

The second-to -last line,s_i = (np.mean(temp_s[tuple(mask)]**2) / (np.mean(temp_s[tuple(mask)])**2)) - 1, is what returns the index error: too many indices.

pupil and pw_sim are each arrays of shape (1024,1024). temp_s is therefore also a (1024,1024) array and mask ends up being a (1023,1023) array. I've tried making mask a (1024,1024) to check and see if I'm getting the index error just because of the difference in shapes, but that doesn't seem to change anything.

It's supposed to index temp_s using mask. I translated this whole thing from Matlab code that had it as temp_s(mask). — songbird813
– songbird813, Commented Jul 31, 2015 at 21:41
if you're creating a tuple - it can't be used as an index (to the best of my knowledge). — Reut Sharabani
– Reut Sharabani, Commented Jul 31, 2015 at 21:41
Hmm. I originally had it as just temp_s[mask] and I got a different error saying I couldn't use mask as an index. I actually found a different question on this site which said that could be rectified by using a tuple — songbird813
– songbird813, Commented Jul 31, 2015 at 21:45

Bi Rico · Accepted Answer · 2015-07-31 22:34:08Z

0

You should probably read about numpy's advanced indexing, sometimes called "fancy indexing".

It's not clear from your question what you're trying to do with temp_s[mask]. If mask is really a mask, you'll need to make sure it has dtype bool and has the same shape as temp_s. If it's not a mask, you'll need to explain better what you're trying to do.

answered Jul 31, 2015 at 22:34

Bi Rico

25.9k3 gold badges57 silver badges75 bronze badges

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

1 Comment

songbird813 Over a year ago

Thanks, I'll look into that. mask is a mask in terms of the module I'm using, but it is not a mask that would have dtype bool. All I know about temp_s[mask] is that it's supposed to index temp_s using mask.

songbird813 · Accepted Answer · 2015-08-03 19:33:34Z

-1

It turns out that simplifying and using different syntax solved the problem. Now, rather than

vals_x, vals_y = np.where(pupil > 0)
mask = pupil[vals_x[0]:vals_x[len(vals_x)-1], vals_y[0]:vals_y[len(vals_y)-1]]

s_i = (np.mean(temp_s[tuple(mask)]**2) / (np.mean(temp_s[tuple(mask)])**2)) - 1

I have

temp_s_nz = np.mean(temp_s[pupil >0])
temp_sq_nz = np.mean(temp_s[pupil >0]**2)

s_i = (temp_sq_nz) / (temp_s_nz)**2 - 1

This change was made by my mentor, so I can't fully explain why it works, but I know that there are no longer any errors and the code is easier to read. I think this works because of the way pupil >0 is defined; in my original code there was room for error in it and now there is not. Thanks for all your help.

edited Aug 3, 2015 at 19:33

answered Aug 3, 2015 at 19:20

songbird813

616 bronze badges

Collectives™ on Stack Overflow

IndexError: Too many indices when indexing array with another array

2 Answers 2

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related