Sparse Gaussian Process Potentials and Simulations of Solid Electrolytes

Add to your list(s) Download to your calendar using vCal

• Dr Amir Hajibabaei, University of Cambridge
• Monday 05 December 2022, 14:30-15:00
• Small Lecture Theatre, Bragg Building, Cavendish .

If you have a question about this talk, please contact Dr Christoph Schran.

We explore sparse Gaussian process regression (SGPR) method for creation of scalable kernel-based machine learning potentials. In these algorithms the potential energy is represented using a subset of training geometries called the inducing points or a set of pseudo inputs. Parsimonious sampling of the training and inducing points are the main challenges which are studied in the context of on-the-fly learning. This methodology is demonstrated with simulations of superionic diffusion in candidate solid electrolytes with emphasis on sulfides such as Li3PS4, Li7P3S11, etc.

This talk is part of the Lennard-Jones Centre series.

This talk is included in these lists:

• Hanchen DaDaDash
• Lennard-Jones Centre
• Small Lecture Theatre, Bragg Building, Cavendish

Note that ex-directory lists are not shown.