[lammps-users] tip4p and compute reduce

I still don't understand this. Compute reduce should be operating
at the end of the step on whatever forces each atom is feeling. So
that should include the pair style or any other constraint forces you've
added, e.g. SHAKE. Ditto for dump output. If you are summing
the forces on atoms in a molecule via compute reduce, is it
possible they sum to 0.0? Please post the relevant commands
for a simple example and I'll try it.