Please retain the lammps user list on all replies. The information you have isn’t sufficient to determine what’s going on. The comment you have there is also confusing. The line you have in your input deck simply tells lammps to use the same coefficients for the interactions amongst all the atoms. It’s up to you to decide what that should be. As for debugging, there are a few things I always try. First, replace atc with nvt and see if the md only system is stable. Then, you should try running the case in serial before trying parallel, and without time filtering before adding it in. That said, my suspicion is you never define the velocities of the atoms. The rescaling thermostat you’re using can’t rescale zero to a finite value, hence the blow up.
Jeremy