AI for Science seminar with Tristan Bereau, University of Heidelberg.
Overview
- Date:Starts 13 February 2025, 15:00Ends 13 February 2025, 16:30
- Seats available:40
- Location:
- Language:English
Zoom password: ai4science
The on-site event will be followed by fika in the Analysen coffee area (fika from 16:00-16:30).
Abstract:
Thermodynamic integration (TI) offers a rigorous method for estimating free-energy differences by integrating over a sequence of interpolating conformational ensembles. However, TI calculations are computationally expensive and typically limited to coupling a small number of degrees of freedom due to the need to sample numerous intermediate ensembles with sufficient conformational-space overlap.
We propose to perform TI along an alchemical pathway represented by a trainable neural network, which we term Neural TI. Critically, we parametrize a time-dependent Hamiltonian interpolating between the interacting and non-interacting systems, and optimize its gradient using a denoising-diffusion objective.
The ability of the resulting energy-based diffusion model to sample all intermediate ensembles allows us to perform TI from a single reference calculation. We apply our method to Lennard-Jones fluids, where we report accurate calculations of the excess chemical potential, demonstrating that Neural TI is capable of coupling hundreds of degrees of freedom at once. Extensions to solvation free energies for atomistic systems will be discussed.
About the speaker:
Tristan Bereau is a computational physicist working at the interface between multiscale modeling and machine learning for soft matter and biomolecules.
He earned a Ph.D. in Physics at Carnegie Mellon University in 2011. After a postdoc at the University of Basel, he led an Emmy Noether research group at the Max Planck Institute for Polymer Research. He then moved to the University of Amsterdam as an assistant professor in chemistry and computer science, followed by a role in Industry.
Tristan serves on the editorial boards of the journals Machine Learning: Science & Technnology and Computational Science and Engineering. He is currently a professor at the Institute for Theoretical Physics at the University of Heidelberg.
Structured learning
This theme focuses on how to make use of structure in data to build machine learning (ML) and artificial intelligence (AI) systems which are safer, more trustworthy and generalize better. Structure includes the relationship between data, in time and space, and how the predictions change when data is transformed in specific ways, for example rotated or scaled. These topics are abstract and general but have a direct impact on the use of AI and ML in the sciences and in applications such as drugs and materials design, or medical imaging.
