[Numpy-discussion] Memory hungry reduce ops in Numpy

Bruce Southey bsouthey@gmail....
Tue Nov 15 16:07:37 CST 2011


On Tue, Nov 15, 2011 at 11:57 AM, Robert Kern <robert.kern@gmail.com> wrote:
> On Tue, Nov 15, 2011 at 16:55, Gael Varoquaux
> <gael.varoquaux@normalesup.org> wrote:
>> On Tue, Nov 15, 2011 at 05:46:52PM +0100, Andreas Müller wrote:
>>>    My question was more along the lines of "why doesn't numpy do the online
>>>    algorithm".
>>
>> It's probably a matter of nobody having had the time and the urge to code
>> it, _and_ do all the extra steps necessary to integrate to the core
>> numpy: working on n dimensions, suporting all the options of the current
>> implementation, doing the tests, doing the docs.
>>
>> I don't think that anybody would frown upon a pull request that
>> implements a better algorithm. It's just work, as you know given your
>> contributions to other project.
>
> Actually, last time I suggested it, it was brought up that the online
> algorithms can be worse numerically. I'll try to find the thread.
>
> --
> Robert Kern
>
> "I have come to believe that the whole world is an enigma, a harmless
> enigma that is made terrible by our own mad attempt to interpret it as
> though it had an underlying truth."
>   -- Umberto Eco
> _______________________________________________
> NumPy-Discussion mailing list
> NumPy-Discussion@scipy.org
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>

There was this one:
http://mail.scipy.org/pipermail/numpy-discussion/2010-November/054007.html

I think part of the problem there was numerical precision as the
difference is about right for double precision.

Bruce


More information about the NumPy-Discussion mailing list