Re: [Bug-gsl] problem in calling function gsl_sf_hyperg_2F1

[Patrick Alken]

Hello,

  The issue seems to be one of the hypergeometric series not converging 
properly, however the resulting solution doesn't seem to be affected too 
much. You can use

gsl_set_error_handler_off()

to ignore the error, and you should find the resulting value agrees 
pretty well with Mathematica. Unfortunately when you do this the error 
code is not properly passed to the user program so I've corrected this 
on the git repository.

Patrick

On 03/30/2015 11:54 PM, liweiming wrote:
> I have a problem calling function gsl_sf_hyperg_2F1 in gsl.  My orginal code is 
> :
>
> #include <gsl/gsl_sf_hyperg.h>
> #include <cmath>
> #include <iostream>
>
> int main(int argc, char * argv[])
> {
>      double result;
>      double a, b, c, x;
>      a = 3.23191;
>      b = -4.0229;
>      c = 8.02291;
>      x = 0.5;
>      result = gsl_sf_hyperg_2F1(a, b, c, x);
>      std::cout << "result = " << result << std::endl;
>      return 0;
> }
>
> My gcc version is 4.8.2. I could compile and link with no problem.
> I complied using :  g++ -o test test.cpp -lgsl -lgslcblas
>
> But on running, it returned the following message:
> $ ./test
> gsl: hyperg_2F1.c:65: ERROR: error
> Default GSL error handler invoked.
> Aborted (core dumped)
>
>
> But by definition the series should be convergent at x=0.5.
> I could get the result using maple, which returned the following :
>
> hypergeom([3.32191, -4.0229], [8.0229], .5);
>                            0.4186139005
>
> Is it because I made a mistake when calling this function ?
>
> Thank you very much !