[lammps-users] Bad initial geometry (ERROR on proc 8: Bond atoms 3571 3573 missing on proc 8)

EmersonPL · April 6, 2021, 12:04pm

Dear colleagues

I’m using Lammps 7 Aug 2019 version.

I am trying to carry out polymer chain simulations, but all simulations have crashed.

An example of an error I’m getting is as follows:
ERROR on proc 8: Bond atoms 3571 3573 missing on proc 8 at step 3731172 (…/ntopo_bond_all.cpp:63)

Looking at previous answers on the Lammps forum, I noticed that this kind of error is due to bad dynamics. I checked the points below (solutions from previous answers) and I still haven’t been very successful:

I used the command “neigh_modify delay 0 every 1 check yes”;
I checked the force field parameters (maybe the published ones are wrong);
I reduced the timestep.

Now I am looking at the issue of initial geometry, which I think should be the (real) problem.

First, I build a polymeric chain in a given software:

Later, I built my system with “n” polymer chains (i.e., my initial geometry used in Lammps):

From the second figure (e.g., my initial geometry), I perform a minimization step, 2 ns of equilibration nvt, 3 ns of equilibration npt and 5 ns of production npt. Example below:

My question is, the most correct thing would be to minimize the energy of a single polymeric chain, apply a equilibration step (Langevin dynamics at 5000 K as in the example of the Lammps tutorials https://icme.hpc.msstate.edu/mediawiki/index. php / LAMMPS_Polymer.html) and only then, build my system with “n” polymeric chains?

Sorry if my question was too out of scope for the mailing list.

Any suggestion is appreciated.

Thanks in advance for your time.

Best regards,

Emerson Parazzi Lyra

PhD candidate

School of Chemical Engineering

University of Campinas
500 Albert Einstein Ave, Campinas, SP, 13083-852, Brazil

Email: [email protected]

Orcid ID: https://orcid.org/0000-0002-7969-3764

akohlmey · April 6, 2021, 2:07pm

if the error happens at step 3731172, then the issue is unlikely the initial geometry, unless there are issues with the topology information (assignment of bonds, angles, dihedrals and atom types) itself. you should look up the atom IDs to see which bond is stretched too long and also visualize your trajectory. that usually gives some significant hints on what is wrong, e.g. if the topology information is wrong.

outside of incorrect topology data, it still could be an issue of the first field settings or how it is applied. also, with hydrogen atoms in the system, a time step of 0.5 fs is fairly aggressive. something like 0.25fs would be more conservative or fix shake on the hydrogen atom bonds only and 1fs would be more conservative, as well.

axel.

EmersonPL · April 6, 2021, 2:25pm

Thanks for the advice and your time, Dr. Axel Kohlmeyer!

I will investigate the problem further based on your observations.