Hi I simulate the electro-osmotic flow (EOF) by LAMMPS. first, create saltwater 1 molar(NaCl) and add sodium ion and all system charge neutralization with an additional charge on the channel atoms uniformly. In all cases, the system equilibrates for 2 ns, after that at 0.025 volts/Angstrom electric field run10 ns to ensure that a steady-state reach and then an additional 30 ns execute for collecting data and statistical averaging. in my simulation: all pair style is Lj/Charmm/Coul/long 8 11 time step 1fs water model is SPC/E update neighborhood list every timestep without delay We have not any problems when Na+ ions are low concentration. The problem makes cause when we have more than 60 Na+ ions in solution. In the latter case "lost atom" is observed. changing timestep, increasing simulation box length were not removed the error.
Dear Seyed,
if you are using NVT/Nph or NpT ensembles, check please the damping paramater;
For exmaple:
fix NaCl all npt temp temp ${initial_temperature} ${final_one} ${damping_for_temperature (in time units)} ${initial_pressure} ${final_value} ${damping_for_pressure (in time units)}
Note that the damping of pressure for above mentioned example is very sensitive to too small values, especially if system is too large.
PS: In your case I suppose you do not using a NVE statistical ensemble, correct?
I do use LAMMPS for Granular.
I will tell you my experience, might be helpful for you as well.
First, need to find out from where these atoms are getting lost - Crossing any boundary (fix BC cause this)
Second is to check on timestep. As you mentioned changing the timestep didn't make any changes, this may not be the reason.
Also you mentioned - increasing the concentration of particles gave this error. It is highly likely that -- pair potential is not defined well.
In case if you want to ignore lost atom (if it is not high), you can always use thermo_modify lost ignore