National University of Singapore

LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator) is a classical molecular dynamics program from Sandia National Laboratories.  It has potentials for soft materials (biomolecules, polymers) and solid-state materials (metals, semiconductors) and coarse-grained or mesoscopic systems. It can be used to model atoms or, more generically, as a parallel particle simulator at the atomic, meso, or continuum scale.

LAMMPS verion 27Aug2011 was compiled on atlas4 cluster and version 22March2013 was compiled and run on all other clusters. You can submit and run LAMMPS jobs in serial mode using a single processor or in parallel mode using message-passing techniques on the HPC systems. Serial batch jobs should be submitted to the serial queues like linux64, atlas4_large, atlas5_large or atlas6_large using the executable lmp_serial. You can also submit parallel batch jobs to the parallel queues like atlas4_parallel, atlas5_parallel, atlas6_parallel or atlas7_parallel using the executables lmp_parallel. Some examples are given below.

Submitting serial jobs to linux64 queue need to specify the execution hosts:
> bsub -q linux64 -m atlas4cluster -o your_job.eo “lmp_serial < in.your_job > out.your_job”
> bsub -q linux64 -m atlas5cluster -o your_job.eo “lmp_serial < in.your_job > out.your_job”
> bsub -q linux64 -m cheeteahcluster3 -o your_job.eo “lmp_serial < in.your_job > out.your_job”

Submitting serial jobs to other serial queues:

> bsub -q atlas4_large -o your_job.eo “lmp_serial < in.your_job > out.your_job”
> bsub -q atlas5_large -o your_job.eo “lmp_serial < in.your_job > out.your_job”
> bsub -q atlas6_large -o your_job.eo “lmp_serial < in.your_job > out.your_job”
Submitting parallel jobs to the parallel queues:

>bsub -q atlas4_parallel -o your_job.eo -n 8 -a mvapich "mpirun.lsf lmp_parallel < in.your_job > out.your_job"
>bsub -q atlas5_parallel -o your_job.eo -n 8 -a mvapich "mpirun.lsf lmp_parallel < in.your_job > out.your_job"
>bsub -q atlas6_parallel -o your_job.eo -n 8 -a mvapich "mpirun.lsf lmp_parallel < in.your_job > out.your_job"
>bsub -q atlas7_parallel -o your_job.eo -n 8 -a mvapich "mpirun.lsf lmp_parallel < in.your_job > out.your_job"

All batch jobs should be submit from the high performance workspace: /hpctmp.

Please contact us at should you have any queries.