Em stops working (Steepest Descents converged to Fmax < 10)

User discussions
1

GROMACS version: 2022.4
GROMACS modification: Yes/No

I am simulating a molecule (without protein) in vacuum after obtaining an ITP file of a specific molecule from an automatic topology builder program.

In the mdp file, I specified nsteps = 10000, emtol = 10.0 and then did energy minimization. It fails to run all 10000 steps and gets stuck in the middle (2016 steps) as shown in the message below.

Steepest Descents converged to Fmax < 10 in 2016 steps
Potential Energy = -2.1984878e+03
Maximum force = 9.8807087e+00 on atom 156
Norm of force = 3.4880597e+00
Finished mdrun on rank 0 Fri Sep 15 11:02:38 2023

I’m getting a “bond length not finite” error in the MD step after NVT, NPT, and subsequently. I think I’m getting this error because I don’t have enough EM. I’m checking the ITP file of the molecule now. Could you please check the em.mdp file I’m using is right, or advise what I can check to resolve this issue.

Thanks.

em.mdp (1.2 KB)

2

Energy minimization is doing what it’s supposed to. The values of nsteps and emtol are the exit conditions; if one or the other is satisfied, mdrun stops. You’ve satisfied the emtol value, so there’s no point running more iterations of minimization.

3

Then, is the reason why energy minimization stops with emtol that not the system is weird but it’s done enough energy minimization to the value specified by emtol?

4

Right, mdrun has satisfied the convergence criterion you defined, so it stops.