gsl_sf_coupling_3j, BUG #7

[G.W.M. Vissers]

>  > I saw at documents.wolfram.com/v4/RefGuide/ThreeJSymbol.html that 
>  > mathematica's 3j symbol is built upon binomial coefficients. I implemented 
>  > their expression in the following routine:
>  > 
>  > int
>  > gsl_sf_coupling_3j_e_new (int two_ja, int two_jb, int two_jc,
>  >                           int two_ma, int two_mb, int two_mc,
>  >                           gsl_sf_result * result)
> 
> Thanks.  
> 
> It looks good to me but maybe the original author of the gsl
> 3j functions has some comments.
> 
> Brian
> 
> p.s. Did you look at the papers mentioned in the BUGS file?
> 
> http://www.ks.uiuc.edu/Publications/Papers/abstract.cgi?tbcode=SCHU76B
> http://www.ks.uiuc.edu/Publications/Papers/abstract.cgi?tbcode=SCHU75A
> http://www.ks.uiuc.edu/Publications/Papers/abstract.cgi?tbcode=SCHU75

I have to admit that although this function does better than the original 
one, it is still not as stable as one would like it to be. I have printed 
the papers mentioned, and will take a better look at them tonight. I 
expect that for large j the recurrence relations in there may be worthwile 
to implement. Maybe later this week ;-) 

Gé