[lammps-users] prd with pair_style eam/cd


I have used both the "prd" command and "pair_style eam/cd" separately with great success. However, whenever I use these two together in the same simulation, the cluster I am working on gives me the MPI error:

[c24-02-33.internal.acs.unt.prv:20602] *** An error occurred in MPI_Bcast
[c24-02-33.internal.acs.unt.prv:20602] *** on communicator MPI COMMUNICATOR 4 SPLIT FROM 0
[c24-02-33.internal.acs.unt.prv:20602] *** MPI_ERR_TRUNCATE: message truncated
[c24-02-33.internal.acs.unt.prv:20602] *** MPI_ERRORS_ARE_FATAL (goodbye)

This error only occurs whenever I try to run the simulation for a large number of timesteps (< 10000). My script looks something like:

Could be due to any number of things. Since you are running a loop
of calculations, I would try to narrow it down to which one is failing,
run it by itself, and look at the thermo output for all the replicas to
see if anything is amiss. It should be one particular event that
had a problem, which could be that an atom was lost or something
else to foul up communication between the replicas.