Machine Learning

The results of crystal structure prediction studies provide a rich source of information on structure-stability relationships and, when coupled with property simulations, the relationships between solid state structure and function. This is an attractive application of machine learning methods, as a means of accelerating the methods and improving reliability of the results. Unsupervised learning approaches are also being developed to identify structure in crystal structure prediction datasets, which can help in interpreting results and maximising what we learn from each study.

Olga Egorova, Roohollah Hafizi, David C Woods and Graeme M. Day

J. Phys. Chem. A (2020) accepted for publication

Jack Yang, Sandip De, Josh E. Campbell, Sean Li, Michele Ceriotti, Graeme M. Day

Chemistry of Materials, 30, p.4361-4371 (2018)

David McDonagh, Chris-Kriton Skylaris and Graeme M. Day

Journal of Chemical Theory and Computation, 15, 2743-2758 (2019)

Felix Musil, Sandip De, Jack Yang, Joshua E. Campbell, Graeme M. Day and Michele Ceriotti

Chemical Science, 9, 1289-1300 (2018)