I have multiple related questions about the low discrepancy or quasi-random sequences provided in GSL.
1) When I generate the Sobol sequence for 2 dimensions, it does not start with 0,0. I don't know if it should or shouldn't, but the Joe and Kuo<http://web.maths.unsw.edu.au/~fkuo/sobol/> generator starts with 0,0. 2) When I try the Niederreiter algorithm gsl_qrng_niederreiter_2 for 100 points and dimension 2, I get the same sequence as for the Sobol, except it starts with 0,0 (and of course stops one short). Is that supposed to happen? 3) Joe and Kuo's Sobol generator goes to dimension 21201 (GSL's goes to dimension 40). Would it be better to use theirs in GSL instead? The code is available on their website (BSD license), but it is in C++. I am using GSL 1.14 on Debian squeeze. I don't have C code examples to show because I'm using GSLL (Common Lisp interface). Thank you. Liam _______________________________________________ Help-gsl mailing list [email protected] https://lists.gnu.org/mailman/listinfo/help-gsl
