Simulation got stuck after this message "Per MPI rank memory allocation (min/avg/max) = 68.21 | 68.21 | 68.21 Mbytes"

simulation got stuck after this message “Per MPI rank memory allocation (min/avg/max) = 68.21 | 68.21 | 68.21 Mbytes”, can anyone help


I think what Axel means is that there is insufficient information provided to give any meaningful help. Please follow the guidelines for the forum, see: Guidelines for posting questions to the LAMMPS forum and Please Read This First: Guidelines and Suggestions for posting LAMMPS questions.

Hi Sunita,
It is always better to share the log file (not the input file) as it includes additional information to solve your problem. Nevertheless, are you sure that the simulation was not running after this message? I mean, did you consider being “stuck” as the screen output is not being updated? because the simulation might be running (for example, at the equilibration stage) and you are just not seeing. (Please share the log file in case you want a better answer)

From what I can see here, your simulation is still running (you are on the 0 step), but it will take some time to evolve (In fact, I believe it will take a “very long” time for you to finish the simulation. Correct me if I am wrong, but I assume you are not using any parallelization and running it on a personal computer).
Also, there are some choices in your input file that I couldn’t understand. Why are you using an nve fix to equilibrate your system? I recommend you check Mark Tschapp’s tutorial and replicate what he did to have a better understanding of your simulations. You can find the link below:

This output provides two important bits of information.

  1. You have been giving us an inaccurate description of your problem. Your output does not stop at the “Per MPI ranks memory allocation” lines, but prints the title line of the thermodynamic output and the data for the initial step. This means that in contrast of your subject line and first post, LAMMPS did not get stuck during the “setup” phase of the run, but has completed it and is instead stuck in the “run” phase. This is a very important distinction and important to pinpoint where to look for issues. If you want to succeed in doing research it is important to be precise and if you want to get competent help, it is important to not mislead people. Specifically when you are in a forum where the people responding are volunteering their time, you will notice that people will start to ignore you, if they cannot trust the information you provide. While this may be a minor thing right now, it is always a good idea to follow best practices and make them a habit from the beginning.

  2. Your pressure and potential energy values are “nan”, i.e. “not a number”. That means that your system setup is bogus (i.e. geometry or force field settings or both) and thus you do not get any meaningful forces which will then taint your positions. This will lead to a deadlock at some point during time integration. If you had chosen fix npt for example, then your run would have stopped with an error of the kind “Non-numeric pressure - simulation unstable” since the adjustments for a variable cell have an additional sanity check that is missing when you are running with fix nve. You cannot recover from this, but have to correct your input. Most common for this kind of error is that you have atoms that are on top of each other, e.g. due to periodic boundary conditions, and thus when computing the force you get a division by zero (since the distance between two atoms after applying PBC is zero) and thus “nan” for energy, force, and stress.

Hi again,
1- Dr. Kohlmeyer has already answered your question.
2- I wanted to kindly ask you to continue trying to debug your input and learn from the process. What Mark Tschapp made is a great tutorial, and many people (including me) learned how to run LAMMPS from it. Running LAMMPS (or any atomistic simulation) is a complicated problem that needs a good amount of knowledge and experience (e.g., the learning curve is not comparable to FEM simulations). So, I highly recommend you spend some time learning the basics before trying to produce any scientific results. (So better to “DUPLICATE” Mark Tschapp’s simulations and not edit them before learning how lammps works)

1 Like

Hi Axel,

I appreciate your response.

While I might not have a full understanding of all the suggestions, I’m committed to learning and working through the debugging process.

Thank you once again for your assistance