Skip to main content
Institute Lecture on "Charting the Future of Materials Modeling: From Machine Learning to Quantum Monte Carlo"
Seminar/Talk
to
Venue

P.C. Saxena Auditorium, IIT Bombay

The Indian Institute of Technology Bombay is organizing an Institute Lecture on Thursday, November 12, 2026.

The details of the lecture are provided below:

Title: 'Charting the Future of Materials Modeling: From Machine Learning to Quantum Monte Carlo'

Speaker: Prof. Georg Kresse, Professor for Computational Quantum Mechanics, University of Vienna, Austria.

---------------------------------------------------------------------------------

About the Speaker: Prof. Georg Kresse (born 1967, Vienna) is a leading physicist in computational materials science. He earned his PhD from Vienna University of Technology in 1993 and later became a professor there in Computational Quantum Mechanics. A member of the Austrian Academy of Sciences and the International Academy of Quantum Molecular Sciences, he has received numerous awards, including the START Grant, Hellmann Prize, and Kardinal-Innitzer-Preis. Kresse is the main developer of VASP, the world's most widely used quantum mechanical simulation software, with related publications among the most cited in science. His current research focuses on machine learning for advanced material predictions. With 350+ publications and an h-index over 130, he is a highly influential figure in his field. Speaker's webpage: https://cmp.univie.ac.at/people/professors/georg-kresse/

Abstract: Over the past decade, machine learning has proven revolutionary for computational materials science. By bridging the gap between first-principles accuracy and empirical force-field speeds, machine-learned force fields (MLFFs) have unlocked previously impossible simulations. Today, we can routinely perform nanosecond-scale molecular dynamics of liquid water, capture complex structural phase transitions in perovskites, and model the dynamic behavior of catalytic surfaces under finite temperatures. However, any machine learning model is ultimately bottlenecked by the quality of its underlying training data. While Density Functional Theory (DFT) has long been the workhorse of our community, standard semi-local DFT remains a mean-field approximation. It frequently falls short when describing delicate electronic correlations, non-covalent interactions, and strongly correlated materials. To build truly predictive, next-generation materials models, we must move beyond the mean-field limit. In this talk, I will outline our roadmap within the Vienna ab-initio Simulation Package (VASP) to address this critical accuracy bottleneck by making high-level, many-body electronic structure methods routine. First, I will discuss the Random Phase Approximation (RPA), an established method in VASP that seamlessly incorporates exact exchange and non-local correlation to significantly improve upon standard DFT. Finally, I will introduce our most recent and exciting development: Auxiliary-Field Quantum Monte Carlo (AFQMC). Driven by major methodological advances over the past five years, we have achieved a fully optimized, small-memory-footprint implementation of AFQMC within VASP. This breakthrough delivers true quantum-chemistry accuracy with highly favorable computational scaling, pushing the boundaries to systems of up to 100 atoms. Ultimately, by combining these highly accurate beyond-mean-field methods with efficient machine learning strategies, we hope to open the door to a new era of robust, predictive materials modeling.