Genarris

Genarris generates molecular crystal structures by random sampling with physical constraints on intermolecular distances. Structures are generated in all space groups compatible with the molecular point group symmetry and the requested number of molecules per unit cell, including with molecules occupying special Wyckoff positions. The input to Genarris is a 3D structure of the molecule. For a flexible molecule, Genarris can be started from a conformational ensemble. A machine learned model estimates the molecular crystal volume from the single molecule structure. The Rigid Press algorithm uses a regularized hard-sphere potential to compress the unit cell and achieve close packing, while preserving the space group symmetries. Genrarris facilitates seamless execution of user-defined workflows for processing the initial pool of generated structures and provides an interface with various DFT codes and machine-learned interatomic potentials (MLIPs) for relaxations and energy evaluations. Genarris can be used to perform crystal structure prediction, to produce curated sets of structures to serve as initial populations for global optimization algorithms, or to generate training sets for machine learning. Genarris is written in Python and distributed under a permissive BSD-3 license.