Software for atomistic ML

The recent revolution in ML interatomic potentials has largely been driven by the development of open-source software packages. More efforts still are needed to make sure these approaches are widely useful to the community.

In our group, we create both specialised software to underpin our research programme, and user-friendly packages for downstream applications. To date, we have released the following software packages:

GitHub | autoplex documentation | first publication

GitHub | graph-pes documentation

GitHub | first publication

 

Additional tools developed by (current or former) members of the group:

  • load-atoms: a Python package for downloading, inspecting and manipulating datasets of atomic structures.
  • augment-atoms:  a tool for augmenting datasets of atomic configurations via a model-driven, GPU-accelerated, rattle-relax-repeat procedure 
  • CHIC: a Python package for coarse-graining hybrid and inorganic crystals