[Numpy-discussion] Request for enhancement to numpy.random.shuffle

[John Zwinck]

On Sun, Oct 12, 2014 at 6:51 AM, Warren Weckesser
<warren.weckesser at gmail.com> wrote:
> I created an issue on github for an enhancement
> to numpy.random.shuffle:
>     https://github.com/numpy/numpy/issues/5173

I like this idea.  I was a bit surprised there wasn't something like
this already.

> A small wart in this API is the meaning of
>
>   shuffle(a, independent=False, axis=None)
>
> It could be argued that the correct behavior is to leave the
> array unchanged. (The current behavior can be interpreted as
> shuffling a 1-d sequence of monolithic blobs; the axis argument
> specifies which axis of the array corresponds to the
> sequence index.  Then `axis=None` means the argument is
> a single monolithic blob, so there is nothing to shuffle.)
> Or an error could be raised.

Let's think about it from the other direction: if a user wants to
shuffle all the elements as if it were 1-d, as you point out they
could do this:

  shuffle(a, axis=None, independent=True)

But that's a lot of typing.  Maybe we should just let this do the same thing:

  shuffle(a, axis=None)

That seems to be in keeping with the other APIs taking axis as you
mentioned.  To me, "independent" has no relevance when the array is
1-d, it can simply be ignored.

John Zwinck