Molecular Simulation in the Age of AI

ISC and TH Seminar
Date: Dec 13, 2023
Time: 02:00 PM - 03:30 PM (Local Time Germany)
Speaker: Dr. Edward Pyzer-Knapp
Head of Research Innovation for UK and Ireland at IBM, Visiting Professor of Industrially Applied AI at the University of Liverpool, and the Editor in Chief for Applied AI Letters.
Location: FHI library (building A)
Host: ISC and TH Department

The history of chemical discovery has been punctuated by computational and theoretical developments. Evolving from empirical observation, increasingly systematised experimentation allowed for the development of theoretical underpinnings, which in turn afforded the paradigm shifting application of computational techniques, which has since co-evolved with the development of the technologies upon which they are run. Recent years have seen the emergence of data-driven techniques and technologies for building powerful models, appearing to enable us to side-step the requirement for expensive physical simulations – replacing them with highly performant, but black-box alternatives.

The need for simulation, however, has always transcended the numerical output, score, or fitness function. Simulation is a core enabler in our ability to understand the world around us at a level not visible to us any other way. The question, therefore, that we should be asking is not whether data driven techniques will replace these approaches, but instead how can we combine the best of both to achieve deep scientific insight and understanding with the speed and complexity afforded by data-driven models. In this talk, I will discuss the major research areas in my team, namely machine learned potentials, generative AI in simulation, and Bayesian optimization, and how we combine them to try to achieve a step-change in molecular simulation.