Atom coordinates revisited and most likely off topic

After equilibrating my MD cell using NPT, I dump xs ys zs coordinates.
I convert xs ys and zs using the box hi/lo information (dump file) into coordinates in the MD cell. The purpose of the NPT is to reduce the volume of the system to the volume at 298K and 1ATM. The system volume reduces from approximately 80A per side (cube geometry) to ~40A per side (cube geometry). I get a density for PDMS of 0.96 g/cc which is reported experimentally.

When I substitute the new coordinates from the dump file back into my original system model I get bonds stretching across the MD cell. This is due to parts of molecules wrapping around (PACMAN) from the one side of the cube to the opposite side. I can see where this due to PBC and the associated MD cell images and can be thought of as other images of the molecules moving into the 000 MD cell as the cell dimensions decrease with time.

Is there a way to use LAMMPS to “squash” molecules into a desired volume so all atoms of the original molecules occupy the same but smaller cell?

Thanks in advance,


I’m not sure what you are asking. There are several options for dumping coords,

e.g. x, xs, xu, xsu, which involve wrapping. In a data file, it doesn’t matter

whether coords are inside the (periodic) simulation box or not. LAMMPS does

the wrapping. What does matter in a data file is that the image flags are

correct for atoms in the same molecule. If you let LAMMPS write the data

files, or dump the image flags and paste the correct columns into a data

file you create yourself, everything should be fine.



Thanks for the response.
I’ll incorporate the new coordinates and image flags
into the original LAMMPS data file along with
new atoms, etc.