GROMACS version: 2019
GROMACS modification: No
Dear GROMACS Users,
I’m running a long MD simulation on a GPU-compiled version of GROMACS 2019 and I’m doing a lot of restarts from checkpoints.
I’m simulating a simple protein in water system with the AMBER99 force-field and the TIP3P water model.
I noticed that the energy terms in the .log files (as well as the values extracted from the .edr files) are quite unphysical.
These are the values printed in my .log files:
–
md.part0001.log
** LJ (SR) Coulomb (SR) Coul. recip. Potential Kinetic En.**
** 1.22012e+05 -1.31707e+06 3.10615e+03 -1.03329e+06 2.01252e+05**
** Total Energy Conserved En. Temperature Pressure (bar) Constr. rmsd**
** -8.32035e+05 -1.04222e+07 2.98612e+02 3.46212e+01 2.71697e-05**
–
md.part0002.log
** LJ (SR) Coulomb (SR) Coul. recip. Potential Kinetic En.**
** 1.24139e+05 -1.32000e+06 3.18068e+03 5.84956e+13 2.01311e+05**
** Total Energy Conserved En. Temperature Pressure (bar) Constr. rmsd**
** 5.84956e+13 5.84956e+13 2.98700e+02 5.40603e+01 2.81131e-05**
–
md.part0003.log
** LJ (SR) Coulomb (SR) Coul. recip. Potential Kinetic En.**
** 1.22739e+05 -1.31664e+06 3.14486e+03 -5.84742e+22 2.01861e+05**
** Total Energy Conserved En. Temperature Pressure (bar) Constr. rmsd**
** -5.84742e+22 -5.84742e+22 2.99515e+02 2.61837e+02 2.63085e-05**
–
md.part0004.log
** LJ (SR) Coulomb (SR) Coul. recip. Potential Kinetic En.**
** 1.25545e+05 -1.32108e+06 3.07703e+03 -1.03395e+06 2.01959e+05**
** Total Energy Conserved En. Temperature Pressure (bar) Constr. rmsd**
** -8.31990e+05 -4.04757e+07 2.99662e+02 3.04066e+02 2.75968e-05**
–
md.part0005.log
** LJ (SR) Coulomb (SR) Coul. recip. Potential Kinetic En.**
** 1.23713e+05 -1.31941e+06 3.23093e+03 1.28042e+20 2.02622e+05**
** Total Energy Conserved En. Temperature Pressure (bar) Constr. rmsd**
** 1.28042e+20 1.28042e+20 3.00645e+02 1.21910e+02 2.81484e-05**
–
md.part0006.log
** LJ (SR) Coulomb (SR) Coul. recip. Potential Kinetic En.**
** 1.24454e+05 -1.32043e+06 3.10584e+03 -1.03458e+06 2.02373e+05**
** Total Energy Conserved En. Temperature Pressure (bar) Constr. rmsd**
** -8.32205e+05 -4.98755e+07 3.00275e+02 1.91567e+02 2.79327e-05**
–
md.part0007.log
** LJ (SR) Coulomb (SR) Coul. recip. Potential Kinetic En.**
** 1.22467e+05 -1.31767e+06 3.16859e+03 -1.03328e+06 2.02975e+05**
** Total Energy Conserved En. Temperature Pressure (bar) Constr. rmsd**
** -8.30304e+05 -5.79141e+07 3.01168e+02 -9.24781e+01 2.91182e-05**
–
md.part0008.log
** LJ (SR) Coulomb (SR) Coul. recip. Potential Kinetic En.**
** 1.24472e+05 -1.31978e+06 3.16369e+03 6.64432e+22 2.02319e+05**
** Total Energy Conserved En. Temperature Pressure (bar) Constr. rmsd**
** 6.64432e+22 6.64432e+22 3.00195e+02 9.59445e+01 2.81613e-05**
–
md.part0009.log
** LJ (SR) Coulomb (SR) Coul. recip. Potential Kinetic En.**
** 1.24374e+05 -1.31900e+06 3.15745e+03 -1.03241e+06 2.01351e+05**
** Total Energy Conserved En. Temperature Pressure (bar) Constr. rmsd**
** -8.31063e+05 -7.48710e+07 2.98760e+02 9.82902e+01 2.82952e-05**
Apart from the highlighted values, the energy looks reasonable for a protein in water.
Visual inspecting the structure I don’t see any clash or evident problem.
I’m using the -noappend flag for gmx mdrun.
I have already simulated around 800 ns … so I’d like to avoid starting another simulation from the beginning, but this is the only solution that comes to my mind.
Your suggestions are greatly appreciated.
Thank you in advance!
Samuele