Machine Learning Approaches to Chemical Reactivity
Our research combines quantum chemistry and machine learning to understand, predict, and simulate chemical reactivity. We focus on transition metal chemistry and homogeneous catalysis, leveraging first-principles electronic structure methods and machine-learned interatomic potentials to bridge the gap between computational predictions and experimental observations.
Modern electronic structure methods have transformed our ability to investigate catalytic mechanisms, but many catalytic processes cannot be adequately described by static calculations alone. Molecular motion, explicit solvent interactions, and entropic effects often play a decisive role in determining reactivity and selectivity. At the same time, recent advances in machine learning have made it possible to extend atomistic simulations to significantly longer length and time scales while retaining the accuracy of quantum chemical calculations.
By combining these complementary approaches, we seek not only to explain known catalytic systems, but also to predict new reactivity and guide the rational design of next-generation homogeneous catalysts.
Research topics:
