Dear LAMMPS users and developers,
We just posted a new LAMMPS release with new features and bugfixes.
The LAMMPS developers and contributors have been particularly busy over Christmas break, so this has quite a few changes. It should be of particular interest to people using new Nvidia GPUs based on the Hopper (sm_90) and Ada (sm_89) architecture and people that have updated to the latest CUDA toolkit release (12.0), as both the GPU package and the KOKKOS package have received updates and bug fixes to enable or improve compatibility with either. Please let us know in case you run into problems.
Another notable addition is the implementation of pair, bond, angle, and dihedral styles using the Lepton library from OpenMM, which allows entering custom potentials as expression strings. In addition there are improvements to tabulation and new commands to generate tables from angle and dihedral styles. Together with the Python code in the tools/tabulate folder we hope that this should significantly simplify the situation for people that need to use custom potential functions that are not (yet) available in LAMMPS. Feedback, constructive criticism, or contributions to expand the functionality are welcome.
Some of you may have noticed that automatic download of library sources from GitHub were temporarily broken when GitHub admins changed the compression library. We have made modifications that should make such downloads more resilient. If this breaks anything, please let us know ASAP.
As part of the new release we also updated the theme files for creating the PDF and documentation website. If you notice that anything does not display as it should (and did previously), please let us know, too.