Hi all!
I want to study the permeation of gases through 2d-membranes via molecular dynamics simulations. In my case I simulate helium atoms through nanoporous graphene membranes and I need to compute the flux of He atoms crossing from one side of the graphene surface to the other side. I use a standard Lennard-Jones potential for the C-He and the He-He interactions, and no interaction between carbons.
My problem is that when I run in parallel I obtain a very different result for the flux of He atoms, in comparison with the serial running (differences of 25%). I have tried with different timesteps but the problem stays.
If anyone could help me or can show me some documentation to learn using LAMMPS in parallel I would be very grateful.
Alfonso.
Hi all!
I want to study the permeation of gases through 2d-membranes via
molecular dynamics simulations. In my case I simulate helium atoms
through nanoporous graphene membranes and I need to compute the flux of
He atoms crossing from one side of the graphene surface to the other
side. I use a standard Lennard-Jones potential for the C-He and the
He-He interactions, and no interaction between carbons.
My problem is that when I run in parallel I obtain a very different
result for the flux of He atoms, in comparison with the serial running
(differences of 25%). I have tried with different timesteps but the
problem stays.
why should the size of the time step have an impact on the result of
these kind of studies?
If anyone could help me or can show me some documentation to learn using
LAMMPS in parallel I would be very grateful.
it is *extremely* unlikely, that your difference is due to using
LAMMPS in parallel vs. serial.
there has to be some other issue in your approach that causes such a
large systematic difference.
axel.