Steve, LAMMPS users,
I understand fix/thermal conductivity can be used in a simulation with atoms having the same mass.
I was wondering about the changes/additions involved to the code in order to use the same Muller-Plathe algorithm in a simulation where there maybe different atomic masses (as well as different interatomic potentials) in the simulation domain.
If the 2 masses are different, the issue is thermostatting as I recall.
For a multi-mass system, you could always define multiple fixes,
each of them swapping on pairs with the same mass. You
should be able to use it with a system with pair hybrid (multiple
I wanted to clarify your comment below regarding thermostatting. First, I would do a NVT simulation to equilibrate the domain at a given temperature, followed by fix thermal/conducitvity in conjunction with NVE.
Is the NVT step where you mentioned there might be issues with multi-mass systems? If so, I would like to understand what the issue would be.
No, I was referring to thermostatting during the fix thermal/cond.
You can do whatever thermostatting you want outside of that.
But during a run with fix thermal/cond, if swaps are between
atoms with equal mass, then overall energy is conserved and
you should be able to run NVE, which is important b/c you don't
want to be adding/subtracting additional eng to the system. If
the masses aren't the same, my recollection is that either it's
trickier or maybe not possible to conserve energy in the swaps,
though I can't remember why not at the moment (I'm away from
my office). Thus there could be an issue with the system
staying at equilibrium.