Re: [Boost-bugs] [Boost C++ Libraries] #9393: quantile for hypergeometric distribution - precision/rounding issue

Subject: Re: [Boost-bugs] [Boost C++ Libraries] #9393: quantile for hypergeometric distribution - precision/rounding issue
From: Boost C++ Libraries (noreply_at_[hidden])
Date: 2013-11-26 00:06:32

#9393: quantile for hypergeometric distribution - precision/rounding issue
--------------------------------------+-------------------------
  Reporter:  anonymous                |      Owner:  johnmaddock
      Type:  Feature Requests         |     Status:  new
 Milestone:  To Be Determined         |  Component:  math
   Version:  Boost Development Trunk  |   Severity:  Showstopper
Resolution:                           |   Keywords:
--------------------------------------+-------------------------

Comment (by Paul McClellan <paulm@…>):

 Hi John,

 When computing the lower tail probabilities (inverse cdf) I set the
 rounding policy to integer_round_down, and when computing the upper tail
 probabilities (inverse survival function) I set the rounding policy to
 integer_round_up. This seems most consistent in the context of lower or
 upper tail tests.

 In my very limited testing, when I use these rounding policies for
 hypergeometric quantiles I find I get the most accurate, or at least the
 most consistent, results in boost::math if I set the fudge_factor defined
 in hypergeometric_quantile_imp() an passed to round_x_from_p() to 1. That
 is, I remove the fudge_factor from the computations.

 Perhaps this can help suggest an approach to this issue.

 Best regards,
 Paul

-- 
Ticket URL: <https://svn.boost.org/trac/boost/ticket/9393#comment:3>
Boost C++ Libraries <http://www.boost.org/>
Boost provides free peer-reviewed portable C++ source libraries.