Philip,
I think its easier if we use "j < nlocal" in your force calculation
and remove 1/2 from the heat flux formula?
Abilash.
Abilash,
Perhaps I've misunderstood your suggestions, but I don't think that's going to work, because the neighbor lists are constructed such that each pair is only visted once, with i a real atom and j>i either real or ghost. Thus atom 2 is in the neighbor list of 1, but 1 is not in the neighbor list of 2 (at least for Lennard-Jones interactions). If these two atoms interact via a periodic boundary, then in a 30-atom system we might have i=1 and j=47, where atom 47 is a copy of 2. If I restrict j<nlocal then I don't include the contribution of this pair to the sum at all, which would give incorrect results.
If you think this analysis is wrong, please let me know!
Thanks,
Philip
Philip,
I think its easier if we use "j < nlocal" in your force calculation
and remove 1/2 from the heat flux formula?
Abilash.