[Numpy-discussion] SFMT (faster mersenne twister)

[josef.pktd at gmail.com]

On Sun, Sep 7, 2014 at 4:23 PM, Sturla Molden <sturla.molden at gmail.com>
wrote:

> Benjamin Root <ben.root at ou.edu> wrote:
> > In addition to issues with reproducibility, think of all of the unit
> tests
> > that would break!
>
> That is a reproducibility problem :)
>
> _______________________________________________
> NumPy-Discussion mailing list
> NumPy-Discussion at scipy.org
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>


Related aside about reproducibility of random numbers:

IMO:
scipy.stats.distributions.rvs does not guarantee yet the values of the
random numbers except for those that are directly produced by numpy.
In contrast to numpy.random, scipy's distributions don't have unit tests
for the specific values of the rvs, and the rvs code for specific
distribution could still be improved in some cases, I guess

Josef
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/numpy-discussion/attachments/20140908/6ffd2dbd/attachment.html>