Hi,
I know that lammps is using spatial decomposition techniques for
parallelization. But I was wondering if there's a flag somewhere in the
code that enables other kinds of decomposition? (i.e. atom or force
decomposition)
For my simulations,the molecules are not homogeneously distributed in
the box and the box size is too large. I end up with very inefficient
parallelization and I was wondering if this could be due to the space
decomposition techniques that lammps is using (as the web site says,
lammps is most efficient in systems where particles fill in a 3-D box
with roughly uniform density,
thanks a lot,
-sezen