Re: [Help-gsl] gsl_sf_bessel_il_scaled

[Brian Gough]

At Sat, 20 Nov 2010 23:22:57 +0100, Torquil Macdonald Sørensen wrote:
> Thanks Brian! Btw, would you say that any occurance of underflow
> corresponds to the output of that bessel function itself being
> smaller than 1e-308 ?
>
> If it is that simple, I will simply replace by 0 manually when I get
> those underflow errors.

That is probably ok, but I could not guarantee it in all cases.  The
code is not too complicated if you want to look at it.

To be safe I'd suggest computing with the first term of the asymptotic
expansion yourself to get an order of magnitude estimate of the
correct value.