Skip to content
Logo: WECORE
Register / Login
Software Marketplace

LAMMPS

stable Open source
Molecular Dynamics Available

Run LAMMPS on Slurm-backed HPC at WECORE: classical molecular dynamics from atomistic to coarse-grained and mesoscale models, with spatial decomposition across MPI ranks and GPU acceleration.

Slurm GPU MPI
Overview

Understanding LAMMPS.

LAMMPS is a classical molecular dynamics code built around spatial decomposition and a very large library of interatomic potentials, from simple Lennard-Jones through EAM and ReaxFF to machine-learned potentials.

It is deliberately modular. Most groups use it as a simulation engine they script, rather than as a fixed application.

Use cases

What people run it for.

Materials under load

Follow fracture, dislocation dynamics, and mechanical response at atomistic resolution, where a continuum model stops answering the question.

Thin films and interfaces

Simulate surfaces, interfaces, and layered systems, where the structure at the boundary sets the behaviour of the whole.

Coarse-grained polymers and soft matter

Reach time and length scales an all-atom model cannot, including granular and mesoscale flows.

Reactive chemistry

Work with ReaxFF and related potentials where bonds break and form and a fixed bond topology will not do.

Run modes

How it runs on WECORE.

Slurm batch

Slurm batch jobs and arrays

Submit input scripts to the queue and let spatial decomposition spread the simulation box across MPI ranks, so a model that will not fit on one node runs across the cluster. Parameter sweeps go out as job arrays.

GPU

GPU offload for pair and neighbour work

The GPU packages move pair and neighbour computation onto accelerator nodes wherever the potential supports it. That is usually the difference between a run measured in weeks and one measured in days.

Coming soon

Bring your simulation stack to a cloud you control.

OpenFOAM, LAMMPS, and GROMACS are ready now, with the rest of the catalog rolling out. Request early access and tell us which app you need next.