bug-gsl
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-gsl] Error in coulomb.c


From: Patrick Alken
Subject: Re: [Bug-gsl] Error in coulomb.c
Date: Thu, 20 Jun 2013 09:18:47 -0600
User-agent: Mozilla/5.0 (X11; Linux i686 on x86_64; rv:17.0) Gecko/20130509 Thunderbird/17.0.6

This domain of arguments is for x > 2*eta, but in test_coulomb.c there are several tests in this range of arguments which seem to pass ok.

On 06/19/2013 07:56 PM, Alexey A. Illarionov wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

Nope, I did not. I will try to look closer to this on weekend, but I
doubt that we could obtain reliable values in that domain of arguments.


Patrick Alken wrote:
Hi, thanks for the report. Do you have an example of calling the
gsl_sf_coulomb_wave_FG_e function with input arguments that lead to
the wrong output? This would be useful in making test cases for
this issue.

Patrick

On 06/18/2013 11:58 PM, Alexey A. Illarionov wrote: Accidentally
found a bug in specfun/coulomb.c . Trivial error in solution of
quadratic equation. None of the current test cases are influenced
by this bug since the value of C (actually N) determines only the
condition of applicability of "continued fraction+recursion"
algorithm.

=== modified file 'specfunc/coulomb.c' --- specfunc/coulomb.c
2007-07-02 18:34:24 +0000 +++ specfunc/coulomb.c    2012-10-26
04:27:18 +0000 @@ -1121,7 +1121,7 @@ *           we must go at
least as low as lam_G */ const double SMALL =
GSL_SQRT_DBL_EPSILON; -    const double C = sqrt(1.0 +
4.0*x*(x-2.0*eta)); +    const double C = 0.5 * sqrt(1.0 +
4.0*x*(x-2.0*eta)); const int N = ceil(lam_F - C + 0.5); const
double lam_0   = lam_F - GSL_MAX(N, 0); const double lam_min =
GSL_MIN(lam_0, lam_G);

-- С уважением, Алексей Александрович Илларионов.
- -- С уважением, Алексей Александрович Илларионов.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)

iQEcBAEBAgAGBQJRwmFkAAoJEEBWYSFzoNKecpwH/jVeqJ+KtkfNo7QZJwAEpJNj
Rs7joTxt7+n320C/qGNY/k5ivgCkJAUwgi0khweXl04uRqJBsHO/jHwbdbx9cW6t
kZZTde/EEUf8g8OA7/x9WDvxpO3D1NgTJbBJnSmqs3J4q77oztBCWCgxuOWWLaks
KNNQH1CW5yV2+hWA7CbL30rDSqwco94Fy2gMGgxUVPx8GHjBcgmaEM34mKDIM2VH
WuhK6CC2tekIEalPUbvnz2KDYOc12Fv9wSctrEt+srBHPxrG5HU9LQmAufVneHfw
b+xNEKgWTGRxjGxfFffA35JKoGYq4sak3okthtHHLxNOsdF/LpacGunWqrOdp40=
=51hh
-----END PGP SIGNATURE-----




reply via email to

[Prev in Thread] Current Thread [Next in Thread]