Max Veit
Lecturer in Digital Chemistry

Profile

Biography

Max obtained his bachelor's degree in physics and in computer science from the University of Minnesota, Twin Cities in 2014, and continued to a master's and finally a PhD at the University of Cambridge in 2018. His PhD work with Prof. Gábor Csányi focused on developing machine learning (ML) models for accurate atomistic simulation of molecular liquids. He went on to a postdoctoral position, first at EPFL, Switzerland, in the COSMO group of Prof. Michele Ceriotti, where he worked on improving the efficiency and expanding the capability of atomistic ML simulations; he then moved to Aalto University, Finland, in the DAS group led by Dr. Miguel Caro, where he worked to apply such physics-inspired ML models to practical problems that push the frontiers of chemical complexity, system size, and predictive capability currently achievable with ML atomistic simulations. Before joining the University of York in April 2025, he held a six-month UNITE! visiting professorship at the TU Darmstadt, Germany, hosted in the Materials Modelling group of Prof. Karsten Albe, where he taught a course on machine learning for materials modelling. Now, as a new lecturer in the Department of Chemistry, he looks forward to building new collaborations that will showcase the potential of Digital Chemistry (in particular, machine-learning-assisted simulation) to accelerate research in energy materials and sustainable chemistry.

Research

Overview

My research interests lie at the intersection of computational chemistry, physical chemistry, and materials science (particularly soft and amorphous materials). The question at the heart of our field is: How can we predict how a substance behaves, based only on the quantum-mechanical equations governing matter at the smallest scales? For some types of substances -- particularly, simple crystalline materials with perfectly periodic lattices, or rarified gases -- we can get quite far using analytical (theoretical) arguments alone. However, in order to make accurate predictions, we almost always need to use a computer algorithm to find an approximate solution of the Schroedinger equation. Such algorithms can deliver highly accurate results and posess extraordinary predictive power; however, they are complex and require lots of computing power, making them impractical for studying large systems or processes happening over long timescales. These issues are particularly acute when studying materials with complex or amorphous structures, which would require impractically large unit cells to capture their structures using first-principles methods.

However, machine learning techniques now offer us a shortcut to bypass the traditionally prohibitive cost of first-principles methods. We can capture both the accuracy and predictive power of first-principles methods in an efficient potential energy model, called a machine learning interatomic potential (ML-IP), that is fast enough to simulate such large, amorphous systems over the long timescales required. The field of ML-IPs has advanced and matured rapidly over the past two decades or so, but many important gaps remain and many interesting chemical systems remain out of reach of the speed and accuracy improvements promised by ML-IPs. My work focuses on expanding the capabilities of these models so that they can predict more properties relevant to experimental researchers and technological applications. I am also working on a key methodological challenge for ML-IPs: how can we make these data-driven models give accurate, physically consistent predictions of long-range interactions like electrostatics, induction, and van der Waals (dispersion)?

We work on diverse applications touching on various other research themes within the Department, from ferroelectric materials to interactions of biomolecules, but our primary focus currently is on the modelling and design of next-generation electrodes (anodes) for lithium-ion and sodium-ion batteries.

Available PhD research projects

We currently have a few available topics for potential MChem or PhD projects; if you are interested in any of the problems described above, please feel free to get in touch!

PhD positions are normally contingent on securing funding; please see the department's funding pagefor information on current funding opportunities. We will also typically advertise about one position per year through Departmental studentships; those projects will appear on the PGR projects page toward the end of the year. Finally, you may find relevant opportunities through the CDT in Process Industries: Net Zero; note this is currently only available to UK students.