The default mapping of LPs to MPI ranks is block mapping, as in the following example for a 6 LP model running on 4 processes:

rank0 : LPs 0,1
rank1 : LPs 2,3
rank2 : LPs 4
rank3 : LPs 5