lammps-users Digest, Vol 73, Issue 37

Hi everyone,

I wanted to use WCA potential in LAMMPS. I looked up ways on the internet to do so and wrote the input script as follows:

pair_style lj/cut 1.122
pair_modify shift yes

pair_coeff 1 1 1.0 1.000 1.122
pair_coeff 1 2 1.0 1.100 1.122
pair_coeff 2 2 1.0 1.200 1.122

After some runs the simulation ends and an error is displayed which says “Atoms lost”.

Please tell me if there is something wrong in the script and how can I use WCA potential correctly. Any help is appreciated. Thanks in advance.

Vyom Vats,
LAMMPS User.

Unless you are using non-periodic boundary conditions
(http://lammps.sandia.gov/doc/boundary.html), it probably means you
have a numerical error and the forces are too large.

This is one of the most common error messages in lammps and it
generally means you were not careful enough setting up your system.
It could be because the initial geometry is unfavorable, your
integration timestep is too large, (or you may have added some fix
which has the effect of adding a large amount of energy to the
system).

http://lammps.sandia.gov/threads/msg26833.html
http://lammps.sandia.gov/threads/msg18455.html
http://lammps.sandia.gov/threads/msg16594.html
http://lammps.sandia.gov/threads/msg14924.html
http://lammps.sandia.gov/threads/msg16603.html

It might fixed by increasing the neighboring frequency
(http://lammps.sandia.gov/doc/neigh_modify.html), and/or using a
smaller time step (http://lammps.sandia.gov/doc/timestep.html). If it
happens early in your simulation, it might mean you did not minimize
or equilibrate the system enough beforehand. It also never hurts to
try visualising the system up until the time it crashes if possible to
see what it is doing wrong.

Cheers
Andrew