Distinguished Seminar in Computational Science and Engineering
March 31, 2022
How machine learning is taking over molecular modelling
Gábor Csányi
Professor of Molecular Modelling
Department of Engineering
University of Cambridge
Recorded Seminar YouTube Link:
https://youtu.be/HSZPMU1KaOo
Abstract:
Over the past decade a revolution has taken place in how we do large scale molecular dynamics. While previously first principles accuracy was solely the purview of explicit electronic structure methods such as density functional theory, the new approaches have allowed the extension of highly accurate, first principles simulations to the atomic scale, where electrons are not treated explicitly any more, and therefore hundreds of thousands of atoms can be simulated. These quantum mechanically accurate force fields and interatomic potentials are fitted to electronic structure data and use techniques inspired by those in machine learning and artificial intelligence research: neural networks, kernel regression, etc. It is a quickly moving field, and – having learned key lessons about representation, symmetry and regularisation – there appears to be some semblance of convergence between the diverse methods, which now also include polynomial expansions carried to high dimension, and message passing neural networks. Other applications such as structure classification, generative modelling, and even modelling the full quantum mechanical wave functions are also beginning to emerge.
Bio:
Gabor Csanyi is Professor of Molecular Modeling at the University of Cambridge. He is developing algorithms and data driven numerical methods for atomic scale problems in materials science and chemistry. His particular focus is to extend the notion of first principles modelling beyond the length and times scales accessible by explicit electronic structure methods, but is also interested in statistical mechanics and sampling problems.
He received his bachelor degree in applied mathematics from Cambridge in 1994, and obtained his doctorate in computational physics from the Massachusetts Institute of Technology in 2001 having worked on computational electronic structure problems. He was a postdoc in the group of Mike Payne in the Cavendish Laboratory before joining the faculty of the Engineering Laboratory at Cambridge in 2007. He received an Early Career Leverhulme Fellowship and the F. W. Bessel Award of the Humboldt Foundation.
How machine learning is taking over molecular modelling
Gábor Csányi, University of Cambridge