We are a group that combines the fields of quantum chemistry and machine learning to explore the mechanisms behind photochemical reactions. Our aim is to understand the unique reactivities and selectivities that arise upon photoexcitation of molecules. We use TD-DFT and MCSCF calculations, as well as non-adiabatic molecular dynamics simulations, to investigate these mechanisms. Our work involves independent research and collaborations with partners worldwide.
In our research group, we are dedicated to tackling the intricate challenges that arise when dealing with mechanistic modeling of multivariate data, especially those exceeding three dimensions. The underlying complexities can often obscure the mechanisms at play. Our mission is clear: to unravel this complexity. We focus on extracting invaluable mechanistic insights from multivariate spectroscopic data, with a specific emphasis on time-resolved spectra. To achieve this, we harness the power of chemometric and machine learning techniques. Through methods like multivariate curve resolution, principal component analysis, and neural networks, we aim to decode the underlying mechanisms.
Our group specializes in the challenging realm of predicting photoinduced phenomena: While machine learning architectures achieve accurate predictions, building vast chemical databases from quantum chemical reference calculations poses limitations. Our mission is to transfer knowledge gained from simulations into processing experimental data, and vice versa, enabling us to systematically calculate relevant structures. Through automated knowledge transfer between quantum chemistry and multivariate modeling of experimental data, we develop fast workflows for understanding and optimizing excited state properties. Our goal is to construct comprehensive databases and data-driven strategies for designing molecules and materials with desired excited state properties, such as optical characteristics.