Ellipsoidal particles in spherical solvent (LAMMPS)
Ellipse sample from LAMMPS packages
Ellipsoidal particles in spherical solvent, 2d system. The example used here is identical to the version released with LAMMPS, with the following exceptions:
- Time-step is halved.
- Run times increased to 1,000,000.
- Saves a movie.
Prepare Example Directory
First, make sure we are at the root of our home directory
user $cdNow, make an example directory, and copy the ellipse example directory into it.
user $mkdir examplesuser $cp -r {{HPC-Examples-Dir}}Parameter unnamed not defined. For correct usage see template's documention./lammps/ellipse examples/user $cd examples/ellipseuser $lsellipse.gayberne.lmp
We also need a script to submit the job to the cluster. Lets start with the sample script in
by copying it to our new ellipse example directory.
user $cp {{HPC-Base-Dir}}Parameter unnamed not defined. For correct usage see template's documention./torque-lmp.sh .user $lsellipse.gayberne.lmp torque-lmp.sh
The torque-lmp.sh script should look something like this:
torque-lmp.sh(original)#PBS -l walltime=10:00:00,mem=15gb,nodes=1:ppn=16 ##PBS -M EmailAddress ##PBS -m bae #PBS -q default #PBS -N Jobname #PBS -o myoutput.txt #PBS -j oe . $(which lmp-config) # The LAMMPS input file must have a '.lmp' extension (e.g., mysimulation.lmp). INPUTFILE=mysimulation doLAMMPS
Modify Queue Submission Script
The submission script we just copied has default settings that have to be modified for this simulation. Lucky, most of the defaults work for the ellipse simulation (1 node, 16 processes per node, etc.) and all we have to change is the name of the job and the lammps input file. Changing the job name to ellipse-lmp and the input file to ellipse.gayberne results in a submission script that looks like this:
torque-lmp.sh(edited)#PBS -l walltime=10:00:00,mem=15gb,nodes=1:ppn=16 ##PBS -M EmailAddress ##PBS -m bae #PBS -q default #PBS -N ellipse-lmp #PBS -o myoutput.txt #PBS -j oe . $(which lmp-config) # The LAMMPS input file must have a '.lmp' extension (e.g., mysimulation.lmp). INPUTFILE=ellipse.gayberne doLAMMPS
Submit Job
The job is now ready to be sent to the cluster.
user $qsub torque-lmp.shuser $qstatJob ID Name User Time Use S Queue ------------------------- ---------------- --------------- -------- - ----- 14222.flory ellipse-lmp user 00:00:35 R default
The job should take about 4 minutes. Check the job status with qstat to see when it is finished; when the job is no longer listed it has finished.
Check Results
When the job completes you should find the following files:
user $ls -lhtotal 22M -rw-r--r-- 1 user user 1.8K Dec 20 17:25 ellipse.gayberne.lmp -rw-rw-r-- 1 user user 1.5M Dec 20 17:29 ellipse.gayberne.out -rw-rw-r-- 1 user user 398 Dec 20 17:26 log.cite -rw-rw-r-- 1 user user 20M Dec 20 17:29 movie.mkv -rw------- 1 user user 730 Dec 20 17:29 myoutput.txt -rwxr-xr-x 1 user user 293 Dec 20 16:14 torque-lmp.sh
General information about the job can be found in myoutout.txt, if any errors occurred they should be listed in this file. A successful run will look like this:
user $cat myoutput.txt-------------- lmp-config (BEGIN) -------------- Start time : Tue Dec 20 17:25:51 EST 2016 PBS_JOBID : 14222.flory.usm.edu Working Directory: /home/user/examples/ellipse MPI command : /home/user/shared/gnu-openmpi/bin/mpirun LAMMPS command : /home/user/shared/gnu-openmpi/bin/lmp LAMMPS script : /home/user/examples/ellipse/ellipse.gayberne.lmp Processes : 16 Controlling Node : n005.cluster.com Nodes : n005.cluster.com (16) Command: mpirun lmp -in "ellipse.gayberne.lmp" -log "ellipse.gayberne.out" -screen "none" Stop time : Tue Dec 20 17:29:39 EST 2016 Elapsed time : 0 days 00h 03m 48s -------------- lmp-config ( END ) --------------
The results from the LAMMPS simulation can be found in ellipse.gayberne.out, and movie.mkv is the movie.
To watch the movie, copy it to your computer. It should be about 1 minute 23 seconds long. During the first half of movie the box is shrinking (NPT ensemble), but the movie is also zooming in, making it look like the particles are getting larger.