Re: [Help-gsl] underflow in coulomb_bound.c

[Ivan Liu]

Hello,

I think I found what causes this underflow message. It comes from 

gsl_sf_hydrogenicR(n,l,Z,r);

where l > 0 AND r = 0. Looking at the analytical form of the radial
Hydrogenic wavefunction R_nl, in such situations, R_nl is always zero.
Which of course gives underflow error.

So I suggest that, in the underflow check of gsl_sf_hydrogenicR, one
should exclude such situations, and perhaps, which may not be needed,
assigns the result to zero directly.

Regards,
Ivan Liu


On Fri, 11 Feb 2005 16:06:20 +0000, Brian Gough
<address@hidden> wrote:
> Ivan Liu writes:
>  > My program kept being interrupted by the underflow check when I
>  > compute the coulomb wavefunction with gsl_sf_hydrogenicR. If I turn
>  > of the error handler then the program returns the seemly correct
>  > values.
> 
> Can you send a small standalone program which shows the
> problem--thanks.  (Looking at the function briefly it might be a bug
> in the way underflow is tested).
> 
> --
> Brian Gough
> 
> Network Theory Ltd,
> Commercial support for GSL --- http://www.network-theory.co.uk/gsl/
>