Per-atom k-space (long-range) contribution

LAMMPS LAMMPS Mailing List Mirror
1

I noticed in the coming attractions section on the LAMMPS website that
it mentions "LAMMPS can calculate per-atom energy and stress (virial)
for all its potential models, including many-body models. But not for
the contribution from long-range Coulombics. We don't know how to
formulate the equations for this, or if it is even meaningful."

A method to calculate the per-atom stress and energy for the Ewald sum
is described in Sonne et al. (JCP 122, 2005), Ghoufi et al. (JCP 128,
2008), and Wang and Zeng, J. Theor. Comput. Chem. 8, 733 (2009). This
method is also used in the NAMD program ( see
http://www.ks.uiuc.edu/Research/namd/2.6/ug/node39.html )

I'm curious, has anyone worked on this or thought of implementing this
in LAMMPS--perhaps in the ewald/n routine?

Thanks!

2

We have some code to do this for regular Ewald. Someone
else is checking it out at the moment. So it will likely
be released in LAMMPS soon. Does any of this indicate
how to do it for PME or PPPM?

Thanks,
Steve

3

Steve,

Both Sonne et al. and Wang and Zeng mention the PME method. I have
only researched the conventional Ewald Sum, so I'm not as familiar
with PME and PPPM.

Thanks,

Stan