(no subject)

Hi,

I have problem with restarting the system.
I have a 2-D system with periodic boundary condition. (it’s like a parallelogram plate)
Im using fix npt in order to make system stable at the 300K temperature.
so the boundary is moving and the system starting to shrink until it gets to equilibrium.
If I stop the simulation in the early steps the restart is working well, but I cannot restart the simulation after adequate time laps which I want.
If I dont want to restart the system the system is going to be stable, but when I want to restart it usually it gives me this error:
ERROR on proc 0: Bond atoms 400 401 missing on proc 0 at step 100003 (neigh_bond.cpp:55)
(some of the particles going out of the plain)

The potentials I have are:
-Lennard-Jones
-Improper~harmonic
-bond~harmonic

I need some suggestion for solving this problem.

Thank you for your time.
Sincerely,
Nima

It's hard (for me at least) to tell what went wrong from that error
message. Perhaps you can try restarting the simulation using another
method:

1)
There is a "read_dump" command

see
http://lammps.sandia.gov/doc/read_dump.html

2)
Try creating a new data file using "read_restart", "write_data" and
then use "read_data" to load it:

http://lammps.sandia.gov/doc/write_data.html
(don't use restart2data)

3)
If you are really desperate, try the "dump2data.py" script (attached)
Syntax:

dump2data -last old_file.data < dump_file > new_file.data

This is a slow script and it is not fully tested. I have not tested
it with triclinic or 2D data, but I'm using it for my own simulations.
(I've had trouble with restart files in the past, so I was looking
for an alternative.)

There's probably a way to do this using pizza.py as well.
There are several ways to do this. I just wanted a one-line bash
command that would extract coordinates from a dump file and create a
data file.

-- documentation for dump2data.py --

With "dump2data.py" you can select which snapshot you want to read from using
-t 600000
(or use "-last" for the last frame)

You can center the coordinate data using
-center

(There is also a "-scale X" command, but it is not well tested.)

The atom style is "full" by default. You can change this using (for example),
-atomstyle "dipole"
   (I have not yet tested hybrid styles, but I hope they work.
    Surround them in quotes, such as "hybrid dipole ellipsoid")

You can output to XYZ format or raw coordinate data format using
"-xyz" or "-raw".

If you want to create multiple data files at regular time intervals use:

dump2data.py -interval 20000 old_file.data < dump_file

(I think) this will generate a series of files named
"old_file.data.1","old_file.data.2","old_file.data.3" ...
with the coordinates extracted from frames whose timestep is a
multiple of 20000.

Velocities:
If you need the velocities, then they must be present in your dump
file, and there must also be a "Velocities" section in the data file
(eg "old_file.data")

This is experimental software, so don't use this software unless you
have no other options. Try using "read_dump" or
"read_restart","write_data","read_data" instead.

Somebody will probably have replied to your post by the time I finish
typing this.
Good luck.

Andrew

dump2data.py (57.1 KB)

I should add that in my earlier reply, I assumed the problem only
occurs when you interrupt the simulation and restart it.

Is this assumption correct? What happens if the simulation is not
interrupted? Are you able to run the simulation for longer (say to
200000 timesteps) without getting any error messages?

i don't think this is a problem with restarting, but a problem with
bad dynamics. the writing of restart only forces it to happen, since
it enforces a neighborlist rebuild and check for consistency.

if you believe there is a bug in LAMMPS, please produce a minimal(!)
but complete(!) input deck that properly reproduces it.

axel.

Thanks,
yes your assumption is right, I have this issue when I want to interrupt the simulation and restart it, if I dont stop the simulation it will continues for a large timesteps without any problem.
It was good to know that restart2data is not very precise and also about the alternatives I have same problem with read_restart.

But now my problem has been resolved I reduced the initial area of the system by changing the initial system configuration.

Best,
Nima