Subject: Evaluating gamma_incomplete() for large values
From: Raymond Toy
Date: Fri, 13 Jan 2012 11:55:19 -0800
On Fri, Jan 13, 2012 at 11:07 AM, Norman Beam <ephemeris at grayswandir.com>wrote:
> What I'm doing is somewhat silly but I ran into a limitation I can't seem
> to work around although I had assumed I could.
>
> I can't seem to get numerical results from gamma_incomplete (and thus
> cdf_poisson()) for values above about 450000.
>
> bfloat(cdf_poisson(460000,**460000));
> gamma_incomplete: series expansion failed for
> gamma_incomplete(4.60001b5, 4.6b5).
>
> Are there setting that would allow me to do this or am I bumping into some
> fundamental numerical limit?
>
>
This is a known bug in gamma_incomplete. I am sort of working on it.
Ray