I am trying to simulate a charged polymer in water, but I keep running into errors like this one: “Bond atoms 235 236 missing on proc 1 at step 851199.”
Depending on how big my system is, I will see this error several hundred thousand to several million timesteps in. I am able to restart the simulation just before it crashed using the read_restart command, and it will run for an additional several hundred thousand to several million timesteps before crashing (well-past the point at which it originally crashed).
I have read previous posts about this issue and have tried increasing the “ghost cutoff” distance by using the comm_modify command to increase the cutoff distance from the original 17.0 up to 25.0, without any improvements. The thermodynamic properties (all of the energies, pressure, and temperature) do not show any large fluctuations before the error appears. I have visualized the system, and the bond atoms that cause the error don’t appear abnormal at any point in the simulation.
I typically see the error for bonds between aromatic carbon and hydrogen atoms, but I have checked the force field parameters for these atoms and they seem to be correct. All of my force field parameters are directly reproduced from another paper (I am currently just trying to reproduce their results).
I have attached an input script and data file for a simple system which shows the error: a dimer, with no water molecules. I also have the log and trajectory files if that would be helpful.
I am using Lammps version 31 March 2017.
pss_dimer_nowater.dat (5.14 KB)
pss_dimer.in (662 Bytes)