miniff -- A minimal implementation of classical and neural-network force fields in python

miniff is a minimal implementation of classical and neural-network force fields written in python. It is based on Behler-Parinello approach to fitting classical atomic energies and uses pytorch and cython under the hood. The package addresses various aspects of neural-network classical potentials, including dataset construction, computing descriptors, assembling workflows, and more. The implementation is parallelized and tested.

Note: this version includes a limited documentation.