LAMMPS WWW Site - LAMMPS Documentation - LAMMPS Mailing List Archives
Re: [lammps-users] Dumping Modified Copies of the Current System State Without Restart Files
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lammps-users] Dumping Modified Copies of the Current System State Without Restart Files


From: Axel Kohlmeyer <akohlmey@...24...>
Date: Tue, 11 Jul 2017 15:44:50 -0400

On Tue, Jul 11, 2017 at 2:45 PM, Ian Chesser <ichesser@...1789...> wrote:
> Hello,
>
> I am driving the motion of a grain boundary forward with a fix command corresponding to a synthetic driving force as well as an npt fix that sets the temperature of the system. Every 10000 steps, I dump the current state of the system in motion. For each snapshot of the moving boundary, I would like to quench the structure to 0 K, producing a new set of quenched boundary motion frames. It is problematic to dump both the set of dynamic and quenched structures in a loop over run steps, since lammps tracks the current state of the system, and the driven boundary motion at a finite temperature is impeded by quenching. The motion of the boundary should be independent of the quenching.
>
> One solution is to output a series of restart files, and read them into a script that quenches a given input structure.
> However, I am wondering if this process can be done in one script. Is there a way to create a modified copy of a current structure such that the current structure remains in the state it was in?

no. but you can check out fix store/state to see, if it allows you to
store all properties that you wish to restore after the temporary
modifications you made.

however, i would consider writing out restarts during a run and then
loading and quenching them at a later point in time the cleaner
approach. please recall, that you do not have to do the quenching
_immediately during_ the production calculation (after all these are
restarts), but you can just accumulate the restarts and then do a loop
over all of them later, _after_ you completed your continuous
production run. if performance and disk space is a concern, there is
another option: you would start two concurrent calculations
side-by-side, one that runs the regular LAMMPS trajectory and a
second, that uses the python script interface (or the library
interface from a small custom  c or fortran wrapper) where you have a
piece of code, that initializes a LAMMPS instance, then waits until a
suitable restart file appears, runs "clear" loads the restart and sets
up the quench run and outputs the resulting data as needed and deletes
the processed restart (if desired) and then waits for the next restart
to appear. rinse and repeat.

axel.

>
> Thank you,
> --
> Ian
>
>
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> lammps-users mailing list
> lammps-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/lammps-users



-- 
Dr. Axel Kohlmeyer  akohlmey@...24...  http://goo.gl/1wk0
College of Science & Technology, Temple University, Philadelphia PA, USA
International Centre for Theoretical Physics, Trieste. Italy.