Subject: Re: [Boost-bugs] [Boost C++ Libraries] #8621: erf function gives wrong results with pgcpp - PGI 10.4
From: Boost C++ Libraries (noreply_at_[hidden])
Date: 2013-06-08 11:04:53
#8621: erf function gives wrong results with pgcpp - PGI 10.4
-------------------------------------------------+-------------------------
Reporter: Jacques Desfosses | Owner: johnmaddock
<jacques.desfosses@â¦> | Status: new
Type: Bugs | Component: math
Milestone: To Be Determined | Severity: Problem
Version: Boost 1.53.0 | Keywords: erf pgi
Resolution: |
-------------------------------------------------+-------------------------
Comment (by johnmaddock):
Thanks, the problem value is:
{{{ Q[0] = -8.7793466483067508e+269 }}}
With Q being initialized like this:
{{{
static const T Q[] = {
1L,
BOOST_MATH_BIG_CONSTANT(T, 53, 1.84759070983002217845),
BOOST_MATH_BIG_CONSTANT(T, 53, 1.42628004845511324508),
BOOST_MATH_BIG_CONSTANT(T, 53, 0.578052804889902404909),
BOOST_MATH_BIG_CONSTANT(T, 53, 0.12385097467900864233),
BOOST_MATH_BIG_CONSTANT(T, 53, 0.0113385233577001411017),
BOOST_MATH_BIG_CONSTANT(T, 53, 0.337511472483094676155e-5),
};
}}}
Can you take a look and see if the compiler has a problem with
initialization from a long int? Initialization from long and long long is
used all over the place in Boost.Math, so while I could easily patch that
case, it would be hard to get all of them :-(
Besides it really should work!
-- Ticket URL: <https://svn.boost.org/trac/boost/ticket/8621#comment:8> Boost C++ Libraries <http://www.boost.org/> Boost provides free peer-reviewed portable C++ source libraries.
This archive was generated by hypermail 2.1.7 : 2017-02-16 18:50:13 UTC