Re: [Help-gsl] ln gamma |x| too large error

[Brian Gough]

At Fri, 16 Mar 2007 10:38:48 +0000 (GMT),
Clare Quarman wrote:
> When using the special function lngamma_sgn_e, sometimes I receive an
> error, the only info is that it's from line 1249 in gamma.c.  Looking at
> this bit of gamma.c, a nearby comment states that |x| is too large for the 
> fractional part to be extracted.
> 
> I'm wondering if this restriction the same both for +ve and -ve values of
> x, and what constitutes too large a magnitude for x?
> 
> Does anyone understand what's going on here?

The gamma function has a singularity at every negative integer, and
for x < -1e15 there isn't enough precision to distinguish between
integer and non-integer values in double-precision -- that is when
this error should occur.

-- 
Brian Gough

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