[Numpy-discussion] SFMT (faster mersenne twister)

James A. Bednar jbednar at inf.ed.ac.uk
Sat Sep 6 13:37:32 EDT 2014


|  Date: Fri, 05 Sep 2014 13:19:57 -0400
|  From: Neal Becker <ndbecker2 at gmail.com>
|  
|  I think it's somewhat debatable whether generating a different
|  sequence of random numbers counts as breaking backward
|  compatibility.

Please don't ever, ever break the sequence of numpy's random numbers!
Please!  We have put a lot of effort into being able to reproduce our
published work exactly, and all of that would be in vain if the
sequence changes.  See e.g.:
 
  http://journal.frontiersin.org/Journal/10.3389/fninf.2013.00044/full

We'd be very happy to see additional number generators appear
*alongside* the existing ones, though, particularly if there are
faster or otherwise better ones!

Jim Bednar

-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.




More information about the NumPy-Discussion mailing list