Re: Robustness Check

[Theodore C. Belding]

Sorry, Sven. I think my last email might seem a bit short, since I dashed
it off very quickly. It's been a long day. :)

What I should have said is: 

You've done a great job with the Swarm RNG library and I know that you've
implemented long-period RNGs and many of the RNGs with splitting
facilities, plus a bunch of other cool stuff. I didn't mean to imply that
the Swarm RNG library didn't have these features, and I apologize if I
gave that impression. I added the L'Ecuyer reference as an afterthought,
just because it was handy and because it seemed that some people weren't
familiar with it.

While Swarm is important in its own right, I'm personally more concerned
with issues about RNGs in general, and that's what I tried to address in
my first email. (Partly this is to justify how we generate seeds in
Drone.)

My argument about seeding RNGs should apply to any RNG that has a
one-to-one mapping between initial seed(s) and initial state, and some
known period length within which the RNG state doesn't repeat (it doesn't
matter if individual random iterates repeat). I'm not familiar with the
details of Matsumoto's RNG but I don't see why it wouldn't apply to it as
well. I may be wrong but I'd like to see a good counter-argument. Anyway,
I thought it was a general point that needed to be made. I think the fact
that you can't guarantee non-overlapping sequences in Matsumoto's RNG
reinforces my point that it isn't essential to guarantee non-overlapping
sequences of random numbers between runs.

I may be wrong but I get the impression from your email that we don't
disagree about this. 
-Ted

--
Ted Belding                              address@hidden 
University of Michigan Center for the Study of Complex Systems
Homepage: http://www-personal.umich.edu/~streak/
PGP key:  http://www-personal.umich.edu/~streak/pgp-key.html






                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.