On Tue, Mar 5, 2013 at 2:46 PM, Emilio Angelina <
emilioluisangelina.hotmail.com> wrote:
>
> Dear All
> I need to restart a molecular dynamics calculation that was stopped due to
> a power outage.
> What restart/coordinate file should I put in the command line when calling
> pmemd (i.e., in the -c flag), the input rst/inpcrd file used in the last
> simulation or the output restart file from the dynamics I want to restart?
>
The output restart file from the dynamics you want to restart. Otherwise
you effectively 'rerun' all of the dynamics you just finished (before the
power outage).
> Similarly, what reference file should I put in the command line when
> calling pmemd (i.e., in the -ref flag), the rst/inpcrd file from the
> original simulation or the output restart file from the dynamics I want to
> restart?
>
Depends. If it's a 'true' restart, you need to use the original -ref file.
They will not be very different for the restrained atoms, though, so this
is not likely too important. (The answer, of course, depends on how you're
using the restraints -- if it's just to fix the protein while you relax
there's really no 'wrong' answer).
> If the second choice is the correct one, another question is how to
> generate the output restart file from the trayectory since the current .rst
> written to disk is truncated.
>
Uh oh. No can do, unless you set ntwv=-1. By default, trajectories don't
have velocity information, so you can't generate a restart from the
trajectory. You will have to re-run this particular part of the simulation.
In the future, if you set ntwr to a negative number (e.g., -10000), it will
write out a separate restart file every 10000 steps, so you will always be
able to retrieve the 'last' restart file that is not truncated. This is
typically a good idea on systems where you know that file buffers have a
tendency _not_ to flush reliably.
Alternatively you can run shorter simulations and restart more frequently
(I rarely run for longer than 1 or 2 ns, for instance).
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 Mar 05 2013 - 12:30:03 PST
