Thanks for your reply. The total energies are as in the figure attached, They seem diverge gradually as you said in the reply, and the difference is probably caused by the round-off error. But I still have a question. Does the nsq method change the order of the neighbor list? when I use nsq instead , it produces the same result as using bin with the same skin value. If the answer is yes, then why we can get the same results with a different neighbor list order and the round off error exiting ? I’m not quit familiar with how the neighboring works, so still a little confused~
Running the same simulation with Nsq vs bin for
the neigh style, will produce neighbor lists that
have the same atoms, but in a different order.
So they might lead to the same kind ofround-off differences.
So your question is a good one. I'll look into the
differences when I have time. Steve Stuart's group
is checking the AIREBO implementation in LAMMPS to
see if they can find any problems. I think they have
flagged a few things with spline coeffs not matching some
of the details in their various papers. So I may wait
until they give us back any changes.
