LIGGGHTS quickstart installation guide

Posted on 28, Aug 2013

If you want LAMMPS, install LIGGGHTS instead, it is considerably easier. However, all of the required information is not in one place, it is scattered in random PDF files.



FFTW and OpenMPI are probably in the packaging repository for your system, check there first, otherwise follow the instructions on the project pages. Pleasingly the LIGGGHTS code and a mirror of the current version of LAMMPS is maintained on the CFDEMproject github page:

$> git clone LIGGGHTS
$> cd LIGGGHTS/src
$> make

This will list a vast array of preconfigured build targets and environments. Here we choose the standard LAMMPS packages and exclude the ones that require special attention or are targeted for advanced use:

$> make yes-standard no-poems no-reax no-meam no-kim no-gpu

If you are on a multi-core desktop or seek to run the code on a cluster, choose openmpi :

$> make openmpi

else you can build the serial/single process option:

$> make serial

There is not an installation target so for the time being the official guide suggests a symbolic link:

$> sudo ln -s /<absolute path to src>/LIGGGHTS/src/lmp_openmpi /usr/bin/liggghts

or for the serial version:

$> sudo ln -s /<absolute path to src>/LIGGGHTS/src/lmp_serial /usr/bin/liggghts

Running the examples

Both the LAMMPS and LIGGGHTS examples are very instructive. The LAMMPS examples/tutorials can be found here:


likewise, the LIGGGHTS specific examples/tutorials can be found here:


Running the LIGGGHTS meshGran example looks like this:

$> cd LIGGGHTS/examples/LIGGGHTS/Tutorials_public/meshGran
$> liggghts < in.meshGran

For both the serial and openmpi builds this will run on a single core/thread. For multiple cores/threads mpirun needs to be invoked and the number of processors set explicitly:

$> mpirun -n 4 liggghts < in.meshGran

Copyright, Christopher Poole 2008-2016.