[
https://issues.apache.org/jira/browse/MATH-1153?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14248928#comment-14248928
]
Luc Maisonobe commented on MATH-1153:
-------------------------------------
I have reworked the patch in order to integrate it directly into the
BetaDistribution class.
This involved a few variable renames, and I am clearly not sure I did it
properly.
It seems the goodness of fit test failed in many cases. I finally selected some
seeds for which it succeeded, just to make some progress here. This
is clearly not satisfaying. However, the patch has a side effect of making most
random generator tests fail, as they share a common test based on beta
distribution sampling. They all inherit this test from RandomDataGeneratorTest,
the test is testNextInversionDeviate.
So my current state of mind is that I broke something while updating the patch,
but I do not have the necessary skills to analyze it and even less to fix it.
Could someone look at my attempts. They are available on a MATH-1153 branch in
the git repository.
In the meantime, I propose to postpone this issue after 3.4.
> Sampling from a 'BetaDistribution' is slow
> ------------------------------------------
>
> Key: MATH-1153
> URL: https://issues.apache.org/jira/browse/MATH-1153
> Project: Commons Math
> Issue Type: Improvement
> Reporter: Sergei Lebedev
> Priority: Minor
> Fix For: 3.4
>
> Attachments: ChengBetaSampler.java, ChengBetaSamplerTest.java
>
>
> Currently the `BetaDistribution#sample` uses inverse CDF method, which is
> quite slow for sampling-intensive computations. I've implemented a method
> from the R. C. H. Cheng paper and it seems to work much better. Here's a
> simple microbenchmark:
> {code}
> o.j.b.s.SamplingBenchmark.algorithmBCorBB 1e-3 1000 thrpt 5
> 2592200.015 14391.520 ops/s
> o.j.b.s.SamplingBenchmark.algorithmBCorBB 1000 1000 thrpt 5
> 3210800.292 33330.791 ops/s
> o.j.b.s.SamplingBenchmark.commonsVersion 1e-3 1000 thrpt 5
> 31034.225 438.273 ops/s
> o.j.b.s.SamplingBenchmark.commonsVersion 1000 1000 thrpt 5
> 21834.010 433.324 ops/s
> {code}
> Should I submit a patch?
> R. C. H. Cheng (1978). Generating beta variates with nonintegral shape
> parameters. Communications of the ACM, 21, 317–322.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
