[gmx-users] PBC "no domain decomposition" error
nedomacho at gmail.com
Wed May 13 23:07:37 CEST 2015
It is my understanding that PBC will be (and really must be) taken
into account whenever the bonds across the box are present in your
topology and periodic-molecules is enabled in your mdp. Whether the
particle population is now optimally decomposed to utilize the number of
cores you've got is another question.
jace> I ran into an issue when simulating graphene sheets with periodic boundary
jace> conditions. It said:
jace> There is no domain decomposition for 8 nodes that is compatible with the
jace> given box and a minimum cell size of 4.685 nm
jace> I figured that was due to the apparent bond length of the carbons
jace> connected across the box, so I set -rdd to be 0.2 nm and now the
jace> simulation seems to be running fine.
jace> Is the -rdd command ignoring the bonds completely for the simulation, or
jace> still using bonds across the box edges once PBC is implemented?
jace> Thank you!
More information about the gromacs.org_gmx-users