LAMMPS

Last Update: 27 Nov. 2024

LAMMPS is an open source general-purpose classical molecular dynamics application software. It can perform dynamics calculations for various systems such as soft matter (polymers, colloids, biomolecules, etc.), solids (metals, semiconductors, etc.), and mesoscopic systems, and can also be used as a general particle simulator.

LAMMPS

Available users

Installed Packages:
ASPHERE BODY CLASS2 COLLOID COMPRESS CORESHELL DIPOLE EXTRA-FIX EXTRA-PAIR
GRANULAR INTEL KSPACE MANYBODY MC MISC MOLECULE MPIIO OPT PERI PYTHON QEQ
REAXFF REPLICA RIGID SHOCK SPIN SRD

Module

Usage

Setup Environment

LAMMPS CPU

$ module load intel
$ module load impi
$ module load lammps

For more information, please see

$ lmp_mpi -h

LAMMPS GPU

$ module load gcc-toolset
$ module load cuda
$ module load ompi-cuda
$ module load lammps-cuda

Batch processing script description example (Hybrid parallel)

#!/bin/bash

#PJM -L rscgrp=a-batch
#PJM -L node=2
#PJM --mpi proc=60
#PJM -L elapse=2:00:00
#PJM -j

module load intel
module load impi
module load lammps
export OMP_NUM_THREADS=4
mpiexec lmp_mpi -in LAMMPS.in > output

• 2 nodes, 60 processes per node, 2 threads per process
• Loading an input file LAMMPS.in
• The calculation results shown as standard output are written to output.

Batch processing script description example (GPU)

#!/bin/bash

#PJM -L rscgrp=b-batch
#PJM -L node=1
#PJM --mpi proc=4
#PJM -L elapse=2:00:00
#PJM -j

module load gcc-toolset
module load cuda
module load ompi-cuda
module load lammps-cuda

mpirun lmp_mpi -sf gpu -pk gpu 4 -in LAMMPS.in > output

• 1 nodes, 4 processes per node, 4 GPU