Re: [Boost-bugs] [Boost C++ Libraries] #12066: Accuracy of modified Bessel functions

Subject: Re: [Boost-bugs] [Boost C++ Libraries] #12066: Accuracy of modified Bessel functions
From: Boost C++ Libraries (noreply_at_[hidden])
Date: 2016-03-14 12:48:24

#12066: Accuracy of modified Bessel functions
------------------------------------------+--------------------------
  Reporter:  Pavel Holoborodko <pavel@…>  |      Owner:  johnmaddock
      Type:  Library Submissions          |     Status:  new
 Milestone:  To Be Determined             |  Component:  math
   Version:  Boost 1.61.0                 |   Severity:  Optimization
Resolution:                               |   Keywords:
------------------------------------------+--------------------------

Comment (by johnmaddock):

 >I think usage of extended precision is vital for measuring actual error
 of float/double precision routines, especially on magnitudes comparable to
 machine epsilon.

 Yes of course.

 >I will try to prepare PR in near future (anyone more experienced with
 boost is very welcome to do this, as coefficients are publicly available).

 I'm tied up at present, but yes, I saw that your coefficients were
 available.

 Ultimately we would need at least 80-bit precision as an option as well,
 so the general form of the new approximations is as useful as the
 coefficient values (we have the code needed to optimise/generate new
 coefficients, figuring out the form of the approximation is the hard
 part).  Might be interesting to see if they'll stretch to 128-bit
 precision as well.

-- 
Ticket URL: <https://svn.boost.org/trac/boost/ticket/12066#comment:6>
Boost C++ Libraries <http://www.boost.org/>
Boost provides free peer-reviewed portable C++ source libraries.