Conserved En. with v-rescale thermostat drifts

User discussions
1

GROMACS version: 2019.6
GROMACS modification: No

After running NVT ensemble with v-rescale thermostat, I get a drift and my conserved energy decreases. I am wondering if someone can tell me whether or not this drift is too much? I see that total energy and potential energy are nicely flat.

I am attaching my job script, my log file, and a picture of my conserved energy.

When I run gmx energy -f ener.edr I get:
Conserved En.
6.57672e+07 1200 2492.56 -8513.98 (kJ/mol)

cons
cons792×612 4.91 KB

#!/bin/bash
#SBATCH --job-name="production run"
#SBATCH --partition=sched_mit_buehler
#SBATCH --constraint=centos7
#SBATCH --mem=50G
#SBATCH -N 20
#SBATCH --ntasks 640
#SBATCH --time=12:00:00
#SBATCH --output=output_prod.txt
#SBATCH --error=error_prod.txt
#SBATCH --mail-type=ALL
#SBATCH --mail-user=is1@wellesley.edu

echo
echo "============================ Messages from Goddess ============================"
echo " * Job starting from: "`date`
echo " * Job ID           : "$SLURM_JOBID
echo " * Job name         : "$SLURM_JOB_NAME
echo " * Working directory: "${SLURM_SUBMIT_DIR/$HOME/"~"}
echo "==============================================================================="
echo

module purge
#module load /home/istewart/gromacs/gromacs-2019.6
source ~/gromacs/2019.6/install/bin/GMXRC
source /home/istewart/plumed2-2.8.2/sourceme.sh
module load engaging/openmpi/2.0.3

gmx_mpi=/home/istewart/gromacs/2019.6/install/bin/gmx_mpi

if [ -n "$SLURM_CPUS_PER_TASK" ]; then
     ntomp="$SLURM_CPUS_PER_TASK"
else
     ntomp="1"
fi


# setting OMP_NUM_THREADS to the value used for ntomp to avoid complaints from gromacs
export OMP_NUM_THREADS=$ntomp

mpirun -np $SLURM_NTASKS $gmx_mpi mdrun -ntomp $ntomp -s 2019.tpr -v -o trajectory


echo "============================ Messages from Goddess ============================"
echo " * Job ended at     : "`date`
echo "==============================================================================="
echo

md (1).log (1.5 MB)

2

Additionally, here is the mdp for the NVT

Thanks!
eq1_10.mdp (4.2 KB)