Standard library logic error when restarting colvars run

User discussions
,
1

GROMACS version: 2023.3
GROMACS modification: No

Dear GROMACS forum,

I’m running an MD simulation of ethane in argon gas, where I use the colvars module to explore energy profile of the H-C-C-H dihedral. This are the ABFsettings:

colvar {
  name phi
  width 2.
  lowerBoundary 0.
  upperBoundary +360.
  # H C C H
  dihedral {
    group1 { atomNumbers 3}
    group2 { atomNumbers 1}
    group3 { atomNumbers 2}
    group4 { atomNumbers 6}
  }
  extendedLagrangian on
  extendedFluctuation 0.1
}
abf {
 colvars phi
 fullSamples 200
}

Launching the MD run works fine, but once it finished I noticed that, during the simulation time, half of the range of the dihedral angle has not been explored.

Therefore, I tried to extend the simulation as follows:

gmx mdrun -s abf_dihedral.tpr -cpi abf_dihedral.cpt -deffnm abf_dihedral -colvars abf_dihedral.dat -colvars_restart abf_dihedral_state.dat

This is the error I get, which happens right after the program version is stated in the output:

Standard library logic error (bug):
(exception type: St12out_of_range)
basic_string::erase: __pos (which is 18446744073709551615) > this->size()
(which is 18)

Am I making any mistake in my workflow or is this a bug? I would appreciate your input in troubleshooting this.

Thank you in advance for your help and let me know if further details or files are needed.

Mattia

2

Hi Mattia, there is certainly a bug in that the code should be able to handle a potential error in the config without crashing. We’ll take a look.

Please also note that the dihedral function has by default a range of -180° to 180°.
https://colvars.github.io/gromacs-2023/colvars-refman-gromacs.html#sec:cvc_dihedral
If you want to use 0-360 try the wrapAround keyword inside dihedral, with a value of 180.0:
https://colvars.github.io/gromacs-2023/colvars-refman-gromacs.html#sec:cvc_periodic

Lastly, GROMACS 2024 beta now supports Colvars natively:
https://colvars.github.io/gromacs-2024/colvars-refman-gromacs.html#sec:colvars_mdengine_parameters
You may want to give it a try.

Giacomo

3

Hi Mattia,

Thanks for reporting the issue, this looks indeed like a bug.
Could you share the tpr file so we could try to reproduce the error ?

Thanks,
Hubert

4

Hello everyone,

thank you for your kind responses and useful inputs.

@giacomo.fiorin thank you for the suggestion of the wrapAround command. I’m trying to compare the PES from QM calculations, and unfortunately in ORCA the dihedral range can be only between 0 and 360.

@Hub I am attaching the tpr file as required, plus the colvars input and state file. Let me know if you need anything else for debugging purposes.
abf_dihedral.dat (326 Bytes)
abf_dihedral_state.dat (9.4 KB)

Mattia

5

This is the tpr file (renamed with .dat extension due to the forum extension limitation).
abf_dihedral.tpr.dat (5.9 KB)

6

Hi Mattia,

Thanks for the file. I’m able to reproduce the error.
The reason is the name of the colvars restart file : -colvars_restart abf_dihedral_state.dat.

In the colvars/gromacs code, we assume the name of this restart file is always in the form of X.colvars.state.dat so in your case abf_dihedral.colvars.state.dat.
[Edit: you need in fact to have this file copied twice : one with the name abf_dihedral.colvars.state.dat to be read by mdrun and one with the name abf_dihedral.colvars.state to be parsed by colvars]

It will not happen in GMX 2024 because we handle the restart information differently.

Anyway, I will report the issue for GROMACS 2023 and below to have a more meaningful error message and remove the need to have 2 copies of the same file

2 Likes