I was wondering if, as Axel suggested, you had any update or comment concerning the erroneous computation of reax/c/species vector of molecule’s id when used with multiprocessors. Specifically, this command gives erroneous ids, i.e. same molecule id for atoms belonging to different molecules. For serial computations, using atom_modify sort 0 solves the issue, though not for parallel simulations. Tested with lammps-7Aug19; compiled with the make mpi and make intel_cpu_intelmpi flags.
Thank you very much for your time.