Yes, restarting might be inefficient, but with lack of cluster resources
REMD cannot be performed at all in a reasonable time with proper density of
replicas over temperature parameter. And intervals may be stretched.
2011/8/23 Carlos Simmerling <carlos.simmerling.gmail.com>
> what do you mean by step by step? running a few hundred steps of MD
> then stopping and running a different replica? based on how pmemd
> works, this would be very inefficient.
>
> There are indeed plans for the future versions of amber to have native
> REMD inside pmemd.
>
> On Tue, Aug 23, 2011 at 3:24 PM, Dmitry Mukha <dvmukha.gmail.com> wrote:
> > Is there among Amber developers the idea to implement serial
> (step-by-step)
> > REMD using pmemd.cuda?
> >
> > ---------- Forwarded message ----------
> > From: Carlos Simmerling <carlos.simmerling.gmail.com>
> > Date: 2011/8/23
> > Subject: Re: [AMBER] Is it possible to do REMD for a 12000 atom molecule?
> > To: AMBER Mailing List <amber.ambermd.org>
> > your question isn't clear. REMD should not be much slower than
> > standard MD for the same system in sander. Did you compare these?
> >
> >>
> >
> > --
> > Sincerely,
> > Dmitry Mukha
> > Institute of Bioorganic Chemistry, NAS, Minsk, Belarus
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Sincerely,
Dmitry Mukha
Institute of Bioorganic Chemistry, NAS, Minsk, Belarus
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Aug 23 2011 - 13:00:03 PDT
