Random and whrandom

[Michael Ackerman]

I happen to recall a reference to this problem. In 'Functional C' by
Hartel & Muller, they write (p.3): "The theory of random number
generators tells us that it is incorrect to use alternate numbers from
one random number generator to generate the values of two dice (Jain,
1991). Instead, two independent random generators must be used." (Do
they mean _any_ random number generator?) The reference is to R. Jain's
'The Art of Computer Systems Performance Analysis', publ. John Wiley, to
which I don't have access.

-- Michael

Alex Martelli wrote:
> 
> "Tim Roberts" <timr at probo.com> wrote in message
> news:ei8q6t0ohuded1oiet6qe79e1eco529ote at 4ax.com...
>     [snip]
> > whrandom is a linear congruential random number generator.  The properties
> 
> Well, not exactly -- it's a Wichmann-Hill (or rather 'the' W-H) RNG,
> that is, 3 small linear congruential RNGs whose current values are
> (normalized and) summed.  Still, that's doesnt affect your main
> points, I think.
> 
> Alex
> 
> --
> http://mail.python.org/mailman/listinfo/python-list