DScribe: Library of Descriptors for Machine Learning in Materials Science
This package addresses the need for accessible and efficient descriptor implementations to accelerate machine learning applications in materials science, though it is incremental as it compiles existing methods.
DScribe is a software library that provides ready-to-use feature transformations for machine learning in materials science, enabling faster property prediction by implementing popular descriptors like Coulomb matrix and SOAP, and it is demonstrated for formation energy and ionic charge prediction tasks.
DScribe is a software package for machine learning that provides popular feature transformations ("descriptors") for atomistic materials simulations. DScribe accelerates the application of machine learning for atomistic property prediction by providing user-friendly, off-the-shelf descriptor implementations. The package currently contains implementations for Coulomb matrix, Ewald sum matrix, sine matrix, Many-body Tensor Representation (MBTR), Atom-centered Symmetry Function (ACSF) and Smooth Overlap of Atomic Positions (SOAP). Usage of the package is illustrated for two different applications: formation energy prediction for solids and ionic charge prediction for atoms in organic molecules. The package is freely available under the open-source Apache License 2.0.