[R-sig-hpc] parallel random number generators

[Dirk Eddelbuettel]

On 18 May 2009 at 18:02, A.J. Rossini wrote:
| On Sat, May 16, 2009 at 1:14 AM, Ross Boylan <ross at biostat.ucsf.edu> wrote:
| > I notice snow defaults to using rlecuyer rather than rsprng.  Should I
| > take this as a sign that the former is "better" in some sense?
| 
| Yes.  SPRNG's theoretical properties seem to be poorer from a
| theoretic sense than the STREAMS algorthim, at least from my readings,
| but there is also the problem that SPRNG is abandonmentware.

There is a sprng4 release but I never got it built. It also changed the API,
so not sure how hard it would be to update rsprng.

Dirk

-- 
Three out of two people have difficulties with fractions.