Machine Learning Models of Many-body Atomic and Electronic Interactions

We kindly invite you to the Theory Seminar of the NOMAD Laboratory at the FHI of the Max-Planck-Gesellschaft and IRIS-Adlershof of the Humboldt-Universität by
Prof. Boris Kozinsky (Harvard University):

• on May 30, 2023, at 2:15 pm CEST
• at IRIS-Adlershof, Zum Großen Windkanal 2, 12489 Berlin
  
• Seminar Room at Foyer (ground floor), room number 2.049

To advance the capability of density functional theory (DFT), we introduce non-local charge density descriptors that satisfy exact scaling constraints and learn exchange functionals called CIDER, that are orders of magnitude faster than hybrid functionals but similar in accuracy. To accelerate molecular dynamics, we use machine learning to capture the potential energy surfaces obtained from quantum calculations. We developed NequIP and Allegro, the first deep equivariant neural network interatomic potential models, whose symmetry-preserving layer architecture achieves state-of-the-art data efficiency and accuracy for simulating dynamics. To enable autonomous active learning of reactive systems, we developed the FLARE algorithm that constructs accurate and uncertainty-aware Bayesian force fields on-the-fly from a molecular dynamics simulation, using Gaussian process regression.

Boris Kozinsky studied at MIT for his B.S. degrees in Physics, Mathematics, and Electrical Engineering and Computer Science, and received his PhD degree in Physics from MIT. He then worked at Bosch Research where he established the atomistic computational materials design team. In 2018, he started a group at Harvard University, which develops computational and machine learning methods and uses them for understanding electronic and atomistic mechanisms of transport, phase transformations and reactions in materials.

For questions, please contact the following email address:
office.nomad@fhi.mpg.de