[lammps-users] mean square displacement


To calculate diffusion coefficients for the group such as water or methanol molecule, lammps calculates mean square displacements of the group, by taking arithmetic average of all atoms of the group it seems; whereas more accurate way it seems is to calculate mean sq disp of the centre of mass of such group.

How large is supposed to be an error?

Thanks in advance


The com keyword on the fix msd command will subtract
out the COM motion. I don't think you want the MSD
of the COM. That would be 0 for big groups and miss
the diffusivity of individual atoms.