Re: [AMBER] random seed not in output file -> run not reproducible

From: Jason Swails <jason.swails.gmail.com>
Date: Tue, 14 May 2013 08:02:20 -0400

On Tue, May 14, 2013 at 7:35 AM, Ross Walker <ross.rosswalker.co.uk> wrote:
>
>
> So it ONLY prints ig if you are using a thermostat based on the random
> number stream. It does not appear to be smart enough to print it if you
> have ntt=0 or 1 but hot start your velocities with irest=0 and tempi>0.0.
>
> I'll fix it in the tree and we can issue a bug fix but for your sims there
> not much that can be done I'm afraid. Sorry.
>

If it makes you feel any better, it would not have mattered, anyway.
 Even if it had printed out the 'ig' value, the printed 'ig' value would
only be correct for the master node. When the random seed is pulled from
the wallclock, every thread is given a different ig value. Since each
thread only generates velocities specific to the atoms they 'own', there
would be no way of reproducing an ig=-1 run in parallel.

HTH,
Jason

-- 
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue May 14 2013 - 05:30:02 PDT
Custom Search