Re: [Bug-gsl] gsl_ran_gamma wrong for certain parameters

[Brian Gough]

Tino Kluge writes:
 > gsl_ran_gamma doesn't cope with parameters a,b producing a mean of 1
 > and very small standard deviations. See the programme below. I'm
 > using fedora core precompiled packages (gsl-1.6-2 and
 > gsl-devel-1.6-2).

Thanks for the bug report.  The algorithm uses some integer
calculations which overflow for a>2^32=4x10^9, so that's why the
problem occurs there.  I'll see about fixing that.

In the meantime the alternative gsl_ran_gamma_mt() function in gsl-1.8
does not have the problem.

-- 
best regards,

Brian Gough

Network Theory Ltd,
Publishing the GSL Manual - http://www.network-theory.co.uk/gsl/manual/