Re: WHY!!! num_runs should be divisible by runs_per_frame * frames_per_restart -->REMD

From: Nicholus Bhattacharjee (nicholusbhattacharjee_at_gmail.com)
Date: Wed May 04 2016 - 10:07:53 CDT

Hello Peter,

Thanks for the reply. Initially I thought that num_runs, runs_per_frame and
frames_per_restart of REMD corresponds to run, dcdfreq and restartfreq
respectively of regular MD. However,from a closer look to the manual I
could understand that frames_per_restart is dependant on runs_per_frame and
hence the rule. Anyway thanks again.

On Tue, May 3, 2016 at 6:08 PM, Peter Freddolino <petefred_at_umich.edu> wrote:

> If this constraint is not satisfied, you could end up with a loss of
> synchronicity between your restart, dcd, and log files. Imagine, for
> example, that you set num_runs to 119, runs_per_frame to 10, and
> frames_per_restart to 10. Then your simulation would end after 119 runs,
> and your log file would have information through all 119 runs, but your dcd
> file would have only information through run 110, and your restart files
> would be from the end of run 100. This becomes problematic for analysis and
> even more so if you start another run from this endpoint…
>
> Best,
> Peter
>
> > On Apr 22, 2016, at 9:21 AM, Nicholus Bhattacharjee <
> nicholusbhattacharjee_at_gmail.com> wrote:
> >
> > Hello all,
> >
> > I don't understand the idea why in REMD
> >
> > num_runs should be divisible by runs_per_frame * frames_per_restart
> >
> > It looks like as if for regular MD
> >
> > run should be divisible by dcdfreq * restartfreq (which is not
> mendatory)
> >
> >
> > Am I missing something??
> >
> > Thanks for the reply.
>
>

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:22:05 CST