How do I reference a specific value in a nested NumPy array?

Question

I have a function from an external package that returns a nested group of NumPy arrays. For example, from IPython:

In [28]: out = Function()

In [29]: out
Out[29]: 
[[array([ 1.,  1.,  0.,  1.,  0.])],
 [array([ 4.,  4.,  5.,  4.,  5.])],
 [array([ 3.,  1.,  0.,  0.,  1.])],
 [array([ 3.,  6.,  1.,  6.,  4.])],
 [array([ 3.,  0.,  1.,  1.,  1.])],
 [array([  3.,  17.,  10.,  25.,  23.])],
 [array([ 0.,  0.,  0.,  0.,  0.])],
 [array([ 0.,  0.,  0.,  0.,  0.])],
 [array([ 0.,  4.,  2.,  5.,  3.])],
 [array([ 0.,  0.,  2.,  2.,  2.])],
 [array([ 0.,  0.,  0.,  0.,  0.])],
 [array([ 0.,  0.,  0.,  0.,  0.])],
 [array([ 0.,  0.,  0.,  0.,  0.])],
 [array([  0.,   1.,   6.,  11.,  15.])]]

I can assign something like:

In [30]: a = out[9]

In [31]: a
Out[31]: [array([ 0.,  0.,  2.,  2.,  2.])]

And then finally:

In [32]: b = a[0]

In [33]: b[-1]
Out[33]: 2.0

To get a specific value that comes out of that array - namely, the last value of the 9th array in the function's output. But I'd really not prefer to make a torrent of variables every time I want to reference something - is there a clean way of referencing a specific interior part of a nested array like this?

Are you sure this strange list of list of numpy arrays is worth it? You should be able to treat the data as a 2D numpy array for many purposes. — Daniel
– Daniel, Commented Dec 5, 2013 at 21:32
@Ophion It's part of an external package. If I was implementing it myself? No. But since I don't want to reimplement the thing from scratch, I'm making due for now. — Fomite
– Fomite, Commented Dec 6, 2013 at 5:44

DSM · Accepted Answer · 2013-12-05 20:02:09Z

2

That's a strangely shaped object: it's a list of 1-element lists whose element is a 1D numpy array. So we need three indices:

>>> out[9]
[array([ 0.,  0.,  2.,  2.,  2.])]
>>> out[9][0]
array([ 0.,  0.,  2.,  2.,  2.])
>>> out[9][0][-1]
2.0

which will also work if we want to modify it:

>>> out[9][0][-1] *= 100
>>> out[9][0][-1]
200.0
>>> out[9]
[array([   0.,    0.,    2.,    2.,  200.])]

edited Dec 5, 2013 at 20:02

answered Dec 5, 2013 at 19:53

DSM

355k67 gold badges606 silver badges504 bronze badges

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

1 Comment

Fomite Over a year ago

Yeah - it's an oddly shaped object indeed, and I was having trouble unpacking it in my mind. But the multitude of indices works. Thanks for your help.

Collectives™ on Stack Overflow

How do I reference a specific value in a nested NumPy array?

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related