commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alex D Herbert (JIRA)" <j...@apache.org>
Subject [jira] [Created] (RNG-93) SmallMeanPoissonSampler
Date Mon, 15 Apr 2019 15:04:00 GMT
Alex D Herbert created RNG-93:
---------------------------------

             Summary: SmallMeanPoissonSampler 
                 Key: RNG-93
                 URL: https://issues.apache.org/jira/browse/RNG-93
             Project: Commons RNG
          Issue Type: Bug
          Components: sampling
    Affects Versions: 1.3
            Reporter: Alex D Herbert
            Assignee: Alex D Herbert


The SmallMeanPoissonSampler accepts a mean up to 0.5 * Integer.MAX_VALUE. However at a value
above approximately mean=744.44 the probability p0 is zero:

{code:java}
p0 = Math.exp(-mean);
{code}

This causes an impossible to reach condition within the sample loop and the sampler will always
return a sample of the upper limit. This is:

{code:java}
limit = (int) Math.ceil(Math.min(1000 * mean, Integer.MAX_VALUE));
{code}

If the sampler is updated to reject any mean where p0 = 0 then this will serve as a warning
to users that the sampler is invalid and they should switch to the LargeMeanPoissonSampler.




--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message