Re: [AMBER] REMD production run restart

From: Daniel Roe <>
Date: Fri, 4 Feb 2011 10:51:57 -0500


The '-A' flag tells sander to append to all output files (mdout, mdcrd
etc) which can be a bit messy if all of your file names don't match up
(plus there is potential to overwrite your input coordinates if e.g.
you accidentally give them the same name as your restart files) so I
personally wouldn't recommend it.

In my opinion your best option is to restart the run in a separate
directory using the restart files generated from the first run (i.e.
the files specified by the '-r' flag in your groupfile). So for
example, say your groupfile from the first run looks something like

-O -rem 1 -remlog rem.log -i in.000 -p topology.parm7 -c
initial.000.rst7 -o rem.out.000 -r run0.000.rst7 -x
-O -rem 1 -remlog rem.log -i in.001 -p topology.parm7 -c
initial.001.rst7 -o rem.out.001 -r run0.001.rst7 -x

Copy the restart files (in this case run0.*.rst7) to a new directory.
The groupfile for the restart run will look something like:

-O -rem 1 -remlog rem.log -i in.000 -p topology.parm7 -c run0.000.rst7
-o rem.out.000 -r run1.000.rst7 -x
-O -rem 1 -remlog rem.log -i in.001 -p topology.parm7 -c run0.001.rst7
-o rem.out.001 -r run1.001.rst7 -x

Make sure that you have ntx=5 and irest=1 set in the input file of the
restart run.


On Fri, Feb 4, 2011 at 10:21 AM, Sangita Kachhap <> wrote:
> Hello all
> I am running REMD (production run) it has stooped running due to cluster error.
> I want to restart it.
> As Amber manual there is option if use -A in command line it wiil help in
> restarting
> REMD simulation. I go through the amber mailing list there also I didnt get any
> idea.
> can anyone sugesst to how restart REMD?
> With regard
> Sangita Kachhap
> _______________________________________________
> AMBER mailing list

AMBER mailing list
Received on Fri Feb 04 2011 - 08:30:10 PST
Custom Search