[Numpy-discussion] Faster np.triu_indices

Stéfan van der Walt stefan at sun.ac.za
Mon Sep 2 03:38:41 EDT 2013

On Mon, Sep 2, 2013 at 12:38 AM, Daniel Smith <malorian at me.com> wrote:
> This is just a first run at the function, and unfortunately does not work for k<0. However, it does return the correct results for k>=0 and is between 2-8 faster on my machine then `np.triu_indices`. Any thoughts on this?

That looks like a good start to me.  Let's get a pull request up that
fixes it for k < 0 as well, and then continue the discussion there.


More information about the NumPy-Discussion mailing list