Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #109029 > unrolled thread

Re: numpy problem

Back to article view | Back to comp.lang.python

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: numpy problem Michael Selik <michael.selik@gmail.com> - 2016-05-23 23:00 +0000

#109029 — Re: numpy problem

On Mon, May 23, 2016 at 9:12 AM <lists@onemanifest.net> wrote:

> > On 23 mei 2016, at 14:19, Peter Otten <__peter__@web.de> wrote:
> > lists@onemanifest.net wrote:
> >
> >> I've got a 2D array
> >> And an array of indexes that for shows which row to keep for each column
> >> of values:
> >>
> >> keep = np.array([2, 3, 1, 9, 2])
> >>
> >> So, the result should be an array like array([ values[2,0], values[3,1],
> >> values[1,2], values[9,3], values[2,4] ]) == np.array([92, 62, 38, 81,
> 44])
> >>
> > values[keep].diagonal()
>
> That seems to do the trick!
>

To clarify, the fancy index is selecting all rows you want to keep, which
gives you a 5x5 array (repeating rows if necessary). Then you are getting
the diagonal of that selection.

If that was already clear, my apologies for the noise.

[toc] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web