Navigating Free Energy Landscapes in the Age of Machine Learning: From Static Mechanisms to Reaction Dynamics

M. Podewitz

Computational chemistry plays a central role in accelerating discovery by exploring reaction mechanisms and guiding catalyst design. Achieving transferable insights, however, requires advances in system representation, computational strategy, and data-driven analysis. The field is moving beyond static, single-structure studies toward ensemble-based explorations of chemical compound space, revealing new structure–reactivity relationships and reactivity patterns.

This talk will highlight methodological developments for investigating catalytic mechanisms across complex chemical spaces. I will discuss advances in static and dynamic modelling and demonstrate how explainable machine learning (ML) enables the analysis and interpretation of high-dimensional simulation data, uncovering mechanistic insights inaccessible to conventional approaches. Together, these developments show how integrating ML with multiscale modelling expands our ability to navigate free energy landscapes and achieve a more predictive understanding of catalytic reactivity.