Extending Simulation on Super Computer

User discussions
1

GROMACS version:2021 and 2023
GROMACS modification: Yes/No
I am trying to extend the GROMACS simulation after completion. I need to continue with a more frequent output rate (500 steps instead of 5000) while maintaining everything is the same. So, after I finish the first 20 ns of my simulation and make sure that it reaches equilibrium, I edit the .mdp file to give me trajectory information every 500 steps, not 5000. The reason behind this is that I need to reduce the size of my .trr file.
The way I am doing it is as follows;
gmx grompp -s new.mdp -c last_state.gro -r last_state.gro -t last_state.cpt -p same_topology.top -o more_frequent.tpr
gmx mdrun -v -deffnm more_frequent
When I do so I get the below error:
Step 20000101 Pressure scaling more than 1%. This may mean your system is not yet equilibrated. Use of Parrinello-Rahman pressure coupling during equilibration can lead to simulation instability, and is discouraged.

Step 20000151 Pressure scaling more than 1%. This may mean your system is not yet equilibrated. Use of Parrinello-Rahman pressure coupling during equilibration can lead to simulation instability, and is discouraged.

Step 20000152, time 20000.2 (ps) LINCS WARNING
relative constraint deviation after LINCS:
rms 71.542221, max 1498.503662 (between atoms 18580 and 18570)
bonds that rotated more than 30 degrees:
atom 1 atom 2 angle previous, current, constraint length
48428 48427 90.0 0.0459 0.2038 0.1000
48427 48426 57.6 0.0459 0.1002 0.1000
48589 48588 90.0 0.0459 0.5917 0.1000
48590 48589 90.0 0.0459 0.6072 0.1000
…
Fatal error:
Too many LINCS warnings (29731)
If you know what you are doing you can adjust the lincs warning threshold in
your mdp file
or set the environment variable GMX_MAXCONSTRWARN to -1,
but normally it is better to fix the problem
Fatal error:
Too many LINCS warnings (29731)
If you know what you are doing you can adjust the lincs warning threshold in
your mdp file
or set the environment variable GMX_MAXCONSTRWARN to -1,
but normally it is better to fix the problem

2

Without more details, I don’t see anything that would go wrong. First, I would compare the .mdp files to make sure there are no other changes. I would also check that the last_state.gro file corresponds to the last_state.cpt file, otherwise your restraints might get quite extreme.

But, anyhow, to preserve the ensemble when you extend like this, you might want to do:
gmx mdrun -v -deffnm more_frequent -cpi last_state.cpt
as well. See the instructions about the -t option in https://manual.gromacs.org/current/onlinehelp/gmx-grompp.html.

Completely unrelated to the extension and crash, is there a specific reason why you are writing an uncompressed .trr file instead of a compressed .xtc file?

3

Thanks a lot for your kind reply. Below is the section of my .mdp file that I am changing;

nsteps = 20010000
dt = 0.001 ; 1 fs
; Output control
nstxout = 500
nstvout = 500
nstenergy = 500
nstlog = 500

However, dt stays unchanged.

The .gro and .cpt files are the files produced by the simulation after it ends. can they be uncorrelated?

I added the option -cpi last_state.cpt but I am still getting the same error.

There is no critical reason for me to keep .trr files instead of .xtc. I will move to .xtc file to save the disk space. Thanks a lot for noting that.

4

Unless the gro file has been processed somehow afterwards, it should correlate to the .cpt file, indeed.

If you are just extending the simulation you might want to use the same file as restraint configuration (the -r option) as you have in the first simulation. Otherwise your new simulation will not be an extension of the old one, but rather different. I don’t think it should cause any crashes, but it’s worth trying.

If you don’t need it to be an actual extension, but just a continuation in general, you might as well remove the -t option to gmx grompp and the -cpi option to gmx mdrun. That should also help avoiding the crashes, I think.

5

Thanks for the kind follow-up. I will try your advice and I hope it will solve the issue.
Thanks again!