The article “A unified interpretation of stress in molecular systems” J Elast 2010 Admal & Tadmor includes an observation that the computation of the stress state via the virial underestimates the actual value b/c contributions across the cell boundaries are not included (viz. p. 121).
This matter appears to be addressed (at least to some extent) in “General formulation of pressure and stress tensor for arbitrary many-body interaction potentials under periodic boundary conditions” J Chem Phys 2009 Thompson et al. My understanding from the discussion thread “Per-atom virial formulation for many-body potentials” is that the the per-atom virial stress form from presented in the Thompson et al paper is currently implemented in LAMMPS.
It is quite clear from Thompson et al that the formulation corrects for pairwise potential effects. (“The atom-cell form of the virial is similar to the standard expression for a finite nonperiodic system, but with an explicit correction for interactions with periodic images. This correction term was formally identified by Louwerse and Baerends, but they indicated that there was no straightforward way to evaluate it for arbitrary potentials. We have been able to write the correction term directly in terms of partial forces acting on periodic image atoms. The correction term is identical to the one derived by Bekker et al. for pair potentials.”)
Is the virial calculation also corrected for PBC effects of angle and dihedral potentials?
A cursory review of compute_pressure.cpp appears to indicate that forces due to angles, dihedrals, and impropers can be included when the respective flags are thrown. I’ve not plumbed deep enough to see whether this includes ghosts.
Thanks,
Jim Moller
Mechanical and Manufacturing Engineering
Miami University
Oxford, OH USA