Rerunning a simulation changes the energy components

GROMACS version:2022 single
GROMACS modification: No

I have simulations for a glycosilated protein dimer. There is two types of sugar and two glycosilation by monomer so the dimer has 4 sugars.
I performed a first simulation with:

gmx mdrun -s changeMDP.tpr -deffnm MDRUN

For the first frame I get this in the .log file (the .edr is the same):

Energies (kJ/mol)

       Bond            U-B    Proper Dih.  Improper Dih.      CMAP Dih.
1.07905e+04    2.87134e+04    3.29585e+04    1.53244e+03   -1.42534e+03
      LJ-14     Coulomb-14        LJ (SR)   Coulomb (SR)   Coul. recip.
8.73737e+03    1.30080e+05    1.16829e+06   -1.25230e+07    3.58665e+04
  Potential    Kinetic En.   Total Energy  Conserved En.    Temperature

-1.11074e+07 1.97155e+06 -9.13586e+06 -9.13539e+06 3.02867e+02
Pressure (bar) Constr. rmsd
3.65355e+02 5.01554e-06

When I rerun the same simulation with the same .tpr with:

gmx mdrun -s changeMDP.tpr -rerun MDRUN.xtc -deffnm RERUN

For the first frame of the .log file I obtain:

Energies (kJ/mol)

       Bond            U-B    Proper Dih.  Improper Dih.      CMAP Dih.
2.59153e+04    4.00655e+04    3.37678e+04    2.62683e+03   -1.33921e+03
      LJ-14     Coulomb-14        LJ (SR)   Coulomb (SR)   Coul. recip.
9.13910e+03    1.30032e+05    1.23807e+06   -1.25310e+07    4.21402e+04
  Potential

-1.10106e+07

Why do I obtain different results using -rerun using the same .tpr file ?

Thanks a lot !

EDIT:
Here is the content of the .mdp used for the creation of the .tpr:
integrator = md
dt = 0.002
nsteps = 1000000000 ; 1000 ns
nstlog = 10000
nstxout = 0
nstvout = 0
nstfout = 0
nstcalcenergy = 100
nstenergy = 10000
nstxout-compressed = 10000
compressed-x-precision = 100
;
cutoff-scheme = Verlet
nstlist = 20
rlist = 1.2
coulombtype = pme
rcoulomb = 1.2
vdwtype = Cut-off
vdw-modifier = Force-switch
rvdw_switch = 1.0
rvdw = 1.2
;
tcoupl = Nose-Hoover
tc_grps = Protein_with_sugar non_protein_non_sugar
tau_t = 1.0 1.0
ref_t = 300 300
;
pcoupl = Parrinello-Rahman
pcoupltype = isotropic
tau_p = 5.0
compressibility = 4.5e-5
ref_p = 1.0
;
constraints = h-bonds
constraint_algorithm = LINCS
continuation = yes
;
nstcomm = 100
comm_mode = linear
comm_grps = Protein_with_sugar non_protein_non_sugar
;
refcoord_scaling = com
energygrps = Protein_with_sugar non_protein_non_sugar

Hi Jaquieries,

I am having exactlty the same problem as you report. Evaluating the energy of a single frame using the same tpr file with rerun vs regular run result in slightly (15%) different potential energies.

Could someone let us know if we should specify a special any setting in the input script, or is this expected behaviour?

Many thanks to the community in advance

Hi adq,
I didn’t find any solution for this… I also had a strange result in the decomposition, some of the 1-4 Electrostatic and hydrophobic contribution were taking into account pairs of atoms that were extremely far from each other and were not linked in 1-4. I verified the simulation and the topolgy and theses atoms are not linked in any way.
Do you find this bug in your results ? ( do you have computed the decomposition for groups that cannot interact in 1-4 but you still obtain a value ?)

Hi,

No, I don’t think I experience the second bit. Did you try the run and the rerun both without any constrains? I thought it would be related to this. May be some experts can comment?

Thanks!

Hello,

this sounds indeed strange. Could you open a bug report at
Issues · GROMACS / GROMACS · GitLab and upload a minimum tpr and
trajectory that reproduces the issue?

We have recently fixed some issues with energy contributions from
excluded atom pairs, and if something is going wrong there we need to know.

As an additional test, could you check this with the most recent 2022.2
patch?

Thanks for your help!

Cheers

Paul