Lost atoms in a Kinetic Monte Carlo simulation

Hello LAMMPS users,

I am using the LAMMPS version 12Dec2018. I am currently attempting to do a KMC simulation to calculate the diffusion coefficient of an Inconel-Ni alloy using LAMMPS. I am getting the ‘Lost atoms’ error after a few iterations. I have run the same structure with a normal MD simulation at 1000 K, and the simulation ran without any problems. Since I don’t have any fixes in my code and I have periodic boundaries, I am not sure why atoms are being lost. The neighbor list cutoff distance also seems to be fine. I am attaching the input file along with this. All I am doing is switching two atoms when the Metropolis criterion is met, and running one iteration to get the energy of the structure after the atom switching. How can atoms be lost in such a case? Thank you.

KMC_LATEST.in (18.8 KB)

1 Like

You are using an almost 4 years old version of LAMMPS. Nobody will look into fixing any bugs or other issues unless you can confirm that the same issue happens with a current version.

Your input file is far too complex and convoluted for making sense of it easily. Unless you figure out a way to reproduce the situation with a minimal input, nobody will try to reproduce your issue, which is a prerequisite for serious debugging.

I can only speculate, but I see the command run 1 pre no post no. Using “pre no” may not be allowed if you change atom positions, since you would need to construct a new neighbor list and re-do other setup tasks.

To add to Axel’s comment, please explain in words what you mean by doing a KMC simulation with LAMMPS to measure diffusion. I see a lengthy double loop in your script - what are you trying to do with it. As Axel, said your input script is 500 lines and far too complicated for anyone else to easily understand.


I also do KMC to calculate the diffusion coefficient. Have you solved your problem now? Would you mind talking to me?