Dear lammps users,
I have simulated TIP4P/2005. In the reference manual, it has been said that Coulomb cutoff = 8.5, however I have put 10 instead of 8.5 in my simulation. when I plot MSD versus time for 10 ns production, Instead of linear relationship, I see superdiffusive behavior. I don’t know why this has happened. I appreciate your help. can it be related to a longer cut-off?
pair_style lj/cut/tip4p/long/gpu 1 2 1 1 0.1546 10
kspace_style pppm/tip4p 1e-05

Since you are using a long-range Coulomb treatment, the Coulomb forces are essentially the same, only the division of work between real space and reciprocal space changes. The LJ contribution decreases with O(r^6) and thus rather rapidly and the difference in interactions between the different cutoffs should be small. You should see this from the total energy. It should change very little when you change the cutoff.

The first step in cases where you don’t get the expected results would be to visualize your trajectory and see if there is anything unexpected.

The second option would be to break down your simulation into 10 segments of 1ns and compute the MSD separately for each segment via using “uncompute” and “compute” to recreate the command. Then you can compare the results from the different sections and see if there is a trend and also you can compute the average MSD graph over the 10 runs and thus get an assessment of the errors and fluctuation of that property.