Re: [Bug-gsl] gsl_sf_coupling_3j bug report

[Brian Gough]

At Thu, 29 Sep 2011 17:09:15 +0400,
Grigory I. Rubtsov wrote:
> 
> Dear GSL developers,
> 
> Thank you for the great effort in developing easy to use and fast
> scientific library. I am using GSL in most of scientific calculations.
> 
> Recently I encounter a bug in calculation of 3j symbol for relatively large l.
> In particular:
> gsl_sf_coupling_3j_e(2*l1,2*l2,2*l3,2*m1,2*m2,2*m3,&r)
>   for l1=249, l2=248, l3=2, m1=5, m2=-6, m3=1
>   gives 0 and error 0 (correct answer should be -0.022878)
> while
>   for l1=248, l2=247, l3=2, m1=5, m2=-6, m3=1
>   gives -0.0229267 and error 2.98369e-17 (the result is correct)
> 

Thanks for the bug report. I am adding a test case and looking at your
patch.