How to avoid LINCS/core dumped error?

User discussions
1

GROMACS version: 2020.2
GROMACS modification: No

Hello friends,
I am trying to run my equilibration simulation on my system. However, I receive this error from gromacs.

image
image780×584 20.5 KB

Any ideas on how I can avoid this? I was told this sometimes happens due to high potential energy of system. However, the PR of my system in is the negatives. Any help is appreciated.

Best regards,
Rosa

2

This will either be due to a problem with the topology for the molecule that is exploding or close contacts of the atoms/molecules involved. Load up the starting coordinate file, look at the atoms involved, see if you can see something out of the ordinary. Also worth checking the topology settings around those atoms too.

3

Apparently you should carefully check topology file if all parameters are correct. If topology is correct, yet you are facing the problem, you might want to try one of the followings (only at equilibration step, not for production run).

  • Put very tiny LJ on hydrogen atoms, once you get stable system, remove those.

  • Try using OpenMP for gmx during equilibration, once system is stable use MPI/OpenMP.

  • Try switching to single precision version of gromacs only for equilibration, that helps sometimes. I confess, I have no idea why this trick works.

  • Run unconstraint (bond/angle) energy minimization and a short NVT equilibration.

Mentioned tricks work sporadically, but not guaranteed.

Finally, with a stable force field and correct topology, and without bad contacts between atoms, you should never experience LINCS error.

I hope you find these helpful.

Sincerely,
Masrul Huda