I wanted to find out if anyone plans to work on extending the compute centroid/stress/atom (used for calculating per atom virial stresses which in turn are used for computing heat fluxes) to FFs involving long-range dispersion/coulombic interactions. I am interested in using it for the lj/cut/coul/long pairstyle in particular. My model also requires the use of the k_space style command.
I am currently working on a project to decompose heat fluxes into molecular contributions and for now I am stuck with the compute stress/atom which when used to compute heat flux, has been reported to give inaccurate results for the heat flux contributions from angles and dihedrals when compared to using the compute centroid/stress/atom.
Your best chance to make your need known and get any feedback on it, would be to submit a “Feature request” issue in the LAMMPS GitHub page at: Issues · lammps/lammps · GitHub
Hi! Feel free to get in touch with me over PM on this forum or at GitHub to discuss this.
Measuring heat fluxes in molecular systems is tricky, and measuring it with Ewald/periodic-meshed long-range Coulombic interactions is double tricky. The latter in particular is a reef on which entire armadas of rubber ducks have crashed and burned to little noticeable progress.
Try the method of planes, and if that doesn’t work, ask yourself whether the thing you’re trying to measure in theory even really corresponds to some physical thing that can be experimentally manipulated in the lab.