Genarris: Random generation of molecular crystal structures and fast screening with a Harris approximation

Abstract

We present Genarris, a Python package that performs configuration space screening for molecular crystals of rigid molecules by random sampling with physical constraints. For fast energy evaluations, Genarris employs a Harris approximation, whereby the total density of a molecular crystal is constructed via superposition of single molecule densities. Dispersion-inclusive density functional theory is then used for the Harris density without performing a self-consistency cycle. Genarris uses machine learning for clustering, based on a relative coordinate descriptor developed specifically for molecular crystals, which is shown to be robust in identifying packing motif similarity. In addition to random structure generation, Genarris offers three workflows based on different sequences of successive clustering and selection steps: the "Rigorous" workflow is an exhaustive exploration of the potential energy landscape, the "Energy" workflow produces a set of low energy structures, and the "Diverse" workflow produces a maximally diverse set of structures. The latter is recommended for generating initial populations for genetic algorithms. Here, the implementation of Genarris is reported and its application is demonstrated for three test cases.

Data-efficient machine learning for molecular crystal structure prediction

“…Nonetheless, the intermolecular contributions are arguably much more important for CSP and nal polymorph ranking, as evidenced by the wide application of CSP protocols with completely rigid molecules. 11,39,40 By tting separate models, the intermolecular contributions are not overshadowed by the intramolecular ones. Secondly, data generation for an intramolecular correction is very cheap, as it only requires calculations on the gas-phase molecule.…”

“…49 Candidate crystal structures were obtained with the Genarris package. 11 Additional tasks such as FPS and hyperparameter optimization were performed with the MLtools package available at https://github.com/ simonwengert/mltools.git.…”

“…9,10 In recent years, a range of methods have been developed for the approximate stability assessment in the initial screening step. Li et al 11 for instance evaluate stabilities of trial congurations by applying the Harris approximation to DFT, with crystal electron densities constructed from the superposition of frozen single molecule densities. Tailor-made empirical potentials have also been successfully used for the screening step, as demonstrated for instance by Neumann et al 12 in the blind tests of organic crystal structure prediction organized by the Cambridge Crystallographic Data Center 9 (CCDC).…”

GAtor: A First-Principles Genetic Algorithm for Molecular Crystal Structure Prediction

“…The initial pool for each target was generated with Genarris 107 to create a starting population of diverse, highquality structures. 107 The distribution of space groups for each initial pool is provided in the supporting information. The generated initial pool structures were locally optimized with the same DFT settings used in the GA and checked for duplicates.…”

Guest Editorial: Special Topic on Data-Enabled Theoretical Chemistry

“…An important problem is that of finding stable polymorphs of molecular crystals. Li et al 54 introduce Genarris, a Python package that does inexpensive approximate DFT calculations and analyzes results with a relative coordinate descriptor developed specifically for this task. It uses ML for clustering and can be targeted for various outcomes, ranging from random structure generation to finding a maximally diverse set of structures to seed a genetic algorithm.…”