I'm not sure about your particular cluster setup, but for ones based on
slurm I typically use a loop to automate running large simulations in
multiple sub-sections. The advantage to that is that as long as enough
steps are completed to generate a restart file, then even if one simulation
fails (in a continuable manner) the next one automatically picks up where
the last one left off.
E.g. If you wanted to divide your 10 ns simulation into .5 ns chunks:
(I'm using the naming conventions that get produced in charmm-gui here)
echo "running production"
step="step4.2_equilibration"
step_template=step5_production.mdin
for ii in `seq -f "%03g" 0 1 19`
do
prev_step=$step
step="step5.${ii}_production"
cp $step_template ${step}.mdin
echo "running $step"
echo "$AMBERHOME/bin/pmemd.cuda -O -i $step.mdin -c $prev_step.rst7 "
echo " -p $toppath -r $step.rst7 -ref $prev_step.rst7 "
echo " -o $step.out -x $step.nc "
$AMBERHOME/bin/pmemd.cuda -O -i $step.mdin -c $prev_step.rst7 \
-p $toppath -r $step.rst7 -ref $prev_step.rst7 \
-o $step.out -x $step.nc
#mv restraints.dat ${step}.restraints.dat
done
echo "done"
In the case where some of the simulations fail, you can simply run
additional simulations at the end to recover the lost time.
On Mon, Jul 9, 2018 at 2:14 AM, 鲍劲霄 <m15295783798.163.com> wrote:
> Dear Amber!
>
> I'm running a membrane protein MD of my own complex following the Lipid14
> tutorial with amber16 pmemd.cuda. I got the error message "ERROR:
> Calculation halted. Periodic box dimensions have changed too much from
> their initial values..." after about 800ps during equilibration. I know
> that I can restart the calculation using the latest restart file to
> continue the 10ns equilibration. However, it is not very convenient for me
> to do this automatically on the computer cluster, and it's too slow to run
> on cpu. So, I'm wondering if it's safe for me to split the whole 10ns
> equilibration into something such as 500ps*20, or should I restart the
> calculation as few as possible?
>
> Thank you in advance.
> Jingxiao Bao
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Jul 09 2018 - 10:00:04 PDT
