At Wed, 4 Nov 2009 15:12:14 -0000,
Roberts Ian wrote:
> I have a query about the implementation of the Sobol sequence in GSL.  I
> have been looking at using the capability and have compared it against
> the routine in numerical recipes.  In GSL the first 1024 points in the
> Sobol sequence (page 248 of the manual I have) do not resemble the
> equivalent data in Numerical recipes in Fortran (page 300) when they
> reference the same background papers.  From my own personal perspective
> (I am not an expert in the area) the sequence in Numerical Recipes
> appears to be more "random" than the GSL implementation which has
> clustering and a set pattern.

Hello,

I see what you mean... according to
http://www.nr.com/forum/showthread.php?t=361 the plots in NR are a
projection of a 3d sequence and are not actually taken directly from
the 2d Sobol sequence.  I haven't checked that but it sounds
plausible.

-- 
Brian Gough

GNU Scientific Library -
http://www.gnu.org/software/gsl/


_______________________________________________
Help-gsl mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-gsl

Reply via email to