Bond atoms * * missing on proc * at step *

Dear lammps-users,

I have met a bond atom missing error. From the result file I found this error was triggered at 193 step. And the lammps manual tell me that the overlaping atoms may result in it. However, I think my structures was relaxed before running. I have use the command change_box and displace_atoms in my input. I don’t know how to fix it. Could you give me some suggestions? Thanks in advance.

Best regard.

Dongbo Wang

Dear Dongbo,

Usually, bond atom missing on proc error occurs due bad force field. What potential are you using? When you say that your structures was relaxed, did you relax using NVT or NVE ensemble? If yes, was the potential energy constant with respect to time? Also, did you receive any warnings such as inconsistent image flag, as this warning does not result in error but when you run a new simulation, the atoms that are straddling the periodic boundaries will be misplaced due to wrong periodic boundaries thereby resulting in bonds located far away at least one periodic length apart.

I am not sure what system which you are simulating, but as for myself, when I relax polymers, I usually equilibrate using the NPT ensemble. If you do have overlapping atoms during relaxation, I suggest using soft potential such as DPD or soft cosine potential to gradually remove any overlaps under NVT or NVE ensemble and then switch to the actual potential that you need to use.


Masato Koizumi

the information you provide is too vague and thus no specific recommendations can be given. both commands, change_box and displace_atoms, can cause problems due to overlapping atoms if used carelessly.