Grain Boundary Physics, Machine Learning, and the "SOAP" Formalism

Gus Hart
Brigham Young University

We demonstrate the utility of the smooth overlap of atomic positions (SOAP) formalism for prediction of grain boundary energies, mobilities, and classification. The SOAP basis provides a representation that enables machine learning to be effective despite a paucity of data due to the extreme expense of grain boundary simulations.


Back to Workshop III: Collective Variables in Quantum Mechanics