[
https://issues.apache.org/jira/browse/MATH-585?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13047176#comment-13047176
]
Darren Wilkinson commented on MATH-585:
---------------------------------------
So, I am not very good at reading patches, and I am not set up right now to be
able to apply it, but if I understand correctly, I think it looks fine. The
idea is that there is just a single instance of the gamma class associated with
each random number stream, and the reason for having it is to keep all of the
constants and variables out of the name space of the random number stream. This
makes sense. It probably could be done with an inner class, which might be
slightly cleaner, but I do not feel strongly about it. Caching stuff is always
a worry for thread safety, but is clearly important for performance. I'm
guessing that the random number streams are not thread safe anyway, and that it
is assumed that threads each have their own random number streams (preferably
seeded differently!). Thanks again for making this a priority. You say it was
quite easy, but I'm sure it was a lot of work!
> Very slow generation of gamma random variates
> ---------------------------------------------
>
> Key: MATH-585
> URL: https://issues.apache.org/jira/browse/MATH-585
> Project: Commons Math
> Issue Type: Improvement
> Affects Versions: 2.2, 3.0
> Environment: All
> Reporter: Darren Wilkinson
> Assignee: Mikkel Meyer Andersen
> Labels: Gamma, Random
> Attachments: MATH585-1.patch
>
> Original Estimate: 6h
> Remaining Estimate: 6h
>
> The current implementation of gamma random variate generation works, but uses
> an inversion method. This is well-known to be a bad idea. Usually a carefully
> constructed rejection procedure is used. To give an idea of the magnitude of
> the problem, the Gamma variate generation in Parallel COLT is roughly 50
> times faster than in Commons Math.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
