[lammps-users] Losing Atoms

I’m trying to simulate the interaction of lj atoms with a solid wall.

I put a wall at the lower z-dimension using the wall/lj93 fix, and I also have a reflecting wall at the upper z-dimension using the wall/reflect fix.
The x, and y dimensions are periodic.

However I’m still losing atoms.

I thought periodic boundary conditions meant that atoms that leave through one end would wrap around to the other side of the cells in the periodic dimensions.

below is the input and output along with the error.


Lennard-Jones atoms interacting with 9-3 wall interaction

dimension 3
boundary p p f

atom_style atomic
neighbor 0.3 bin
neigh_modify delay 5

create geometry

lattice fcc 0.01
region box block 0 20 0 22 -1.0 10.0
create_box 1 box

mass 1 1.0

create atoms

create_atoms 1 region box

LJ potentials

pair_style lj/cut 2.5
pair_coeff * * 1.0 1.0 2.5

initial velocities

velocity all create 0.1 482748


fix 1 all nve
fix 2 all wall/lj93 zlo -2 1.0 1.0 2.5
fix 3 all wall/reflect zhi


timestep 0.0025
thermo 1000

dump 1 all atom 500 dump.lennard_jones_substrate
run 20000

Lost atoms means the processors lost track of them. This happens when the atoms move a distance larger than 1 cpu domain. You are loosing atoms because they are moving very fast - from bad contacts, probably.

Quoting Daniel Casimir <[email protected]>: