I am trying to run relaxations of graphene in Lammps with LATTE parallelized over kpoints. I currently have Lammps and LATTE built with MPI, but I am unsure of how to partition resources.
So far, I have tried running lammps in multi-partition mode, where I specify one partition has 1 processor and the second partition has X number of processors (i.e. mpirun -np 42 lmp_mpi -partition 1 41 -in input_file). In the input file I then specify:
partition yes 1 fix 1 all nve
partition yes 2 fix 2 all latte NULL
Running this command gives the following error:
"Open MPI has detected that this process has attempted to initialize
MPI (via MPI_INIT or MPI_INIT_THREAD) more than once. This is
[a34n18:76070] *** An error occurred in MPI_Init
[a34n18:76070] *** reported by process [1622214326,1]
[a34n18:76070] *** on a NULL communicator
[a34n18:76070] *** Unknown error
[a34n18:76070] *** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,
[a34n18:76070] *** and potentially your MPI job)"
I think the origin of this problem is running fix latte with lmp_mpi, because even when I do not partition, but just have one MPI task, I still get this error.
Is this the correct way to go about solving this problem, or am I on the wrong track here?