[Bug-gsl] hyperg_2F1.c

[Lee Warren]

There is an issue with computing hypergeometric functions
of the form

gsl_sf_hyperg_2F1( a, b, c, x )

where a and b are negative integers, c is negative,
and x>=1.0

For example,

gsl_sf_hyperg_2F1(-1,-1,-1/2,1.5) throws an domain
error when in fact the series converges to -2.0.

If I comment out lines 635-637 (to remove the
error check for x), gsl_sf_hyperg_2F1 appears to
compute the result correctly.

It seems that the requirement x<1.0 is too strict
for some argument sets.

This may affect the other 2F1 hypergeometric
functions as well.



Lee Warren