BUG in gsl_sf_bessel_zero_J1_impl

BUG in gsl_sf_bessel_zero_J1_impl

[Gerard Vermeulen (CRTBT)]

Concerns version 0.6:

result.val is about 0.3 after the call
gsl_sf_bessel_zero_J1_impl(0, &result);
instead of 0, as I would expect.

Higher zeros are OK.

I also point out that gsl_sf_bessel_zero_J1_e & friends are mentioned
in the documentation, but are not implemented, yet.

Best regards -- Gerard
-- 
Gerard Vermeulen                           |  TEL: 33 (0)4 76 88 12 53
CRTBT                                      |   or  33 (0)4 76 88 12 61
BP 166 -- 38042 Grenoble Cedex 9 -- France |  Fax: 33 (0)4 76 87 50 60
http://www-crtbt.polycnrs-gre.fr/he3pol

Re: BUG in gsl_sf_bessel_zero_J1_impl

[Gerard Jungman]

"Gerard Vermeulen (CRTBT)" wrote:
> 
> Concerns version 0.6:
> 
> result.val is about 0.3 after the call
> gsl_sf_bessel_zero_J1_impl(0, &result);
> instead of 0, as I would expect.
> 
> Higher zeros are OK.

Yes, this was never accounted for, under the assumption
that the behaviour for s=0 was "undefined" or something.
Anyway, I agree that was silly, so I fixed it to
return 0.0 for s=0.


> I also point out that gsl_sf_bessel_zero_J1_e & friends are mentioned
> in the documentation, but are not implemented, yet.

gsl_sf_bessel_zero_J1_e() was inadvertently left unimplemented;
I fixed that. I'm not sure what you mean by "friends", as
both the gsl_sf_bessel_zero_J0_e() and gsl_sf_bessel_zero_Jnu_e()
implementations are present.


The fixes are now checked into CVS.

Thanks for the report!

-- 
G. Jungman