Now that the system is equilibrated, we can begin production MD simulations, during which time we will collect ∂H/∂λ data.

#################
# PRODUCTION MD #
#################
echo "Starting production MD simulation..."

cd ../
mkdir Production_MD
cd Production_MD

gmx grompp -f $MDP/md_$LAMBDA.mdp -c ../NPT/npt$LAMBDA.gro 
-p $FREE_ENERGY/topol.top -t ../NPT/npt$LAMBDA.cpt -o md$LAMBDA.tpr

gmx mdrun -deffnm md$LAMBDA

echo "Production MD complete."

The whole workflow should take about 2.5 hours to complete on a GPU workstation. Since there are many simulations to run (21 sets of λ vectors), it is best to run these jobs on clusters where more processors are available such that jobs can be run simultaneously. Once all of the production simulations are complete, we can analyze the resulting data.