Re: [Bug-gsl] Bug in gsl_ran_exponential?

[Brian Gough]

At Wed, 16 Aug 2006 17:08:12 +0200,
Dr. Hans Ekkehard Plesser wrote:
> This means that gsl_rng_uniform_pos cannot return 0, even though
> 
>       p(x) = 1/mu exp(-x/mu)
> 
> has its maximum p(x=0) = 1/mu at x=0.

Thanks for the bug report.  I think it might be a good idea to make
sure that 0 is sampled---the following might work:

        double u = gsl_rng_uniform (r);
        return -mu * (u == 0.0 ? 0.0 : log (u));

Maybe there are some more of these lurking in the other routines too?

-- 
Brian Gough

Network Theory Ltd,
Publishing Free Software Manuals --- http://www.network-theory.co.uk/