My research focuses on developing computational methods that help design molecules that can actually be made, understood, and used in practice. Modern artificial intelligence can suggest enormous numbers of candidate compounds, but many are unrealistic from a chemical perspective. A central part of my work therefore combines machine learning with quantum-chemical modeling to evaluate whether proposed molecules are synthesizable before experimental effort is invested. By integrating physical chemistry directly into data-driven models, we make computational predictions more reliable and chemically meaningful.
Another major research direction concerns cyclodextrins and related host–guest systems. Cyclodextrins are ring-shaped sugar molecules that can encapsulate other compounds inside their cavities and are widely used in formulation science, drug delivery, and environmental chemistry. Using molecular simulations and theoretical models, we study how these complexes form and how their binding properties can be tuned. This work helps connect atomic-level understanding with practical applications such as selective pollutant capture and improved pharmaceutical formulations.
More broadly, I develop machine-learning approaches for molecular design that combine cheminformatics, statistical modeling, and physical simulations. The goal is to build predictive pipelines that can propose new molecules with desired properties while remaining consistent with chemical feasibility. Rather than replacing chemical insight, these models are designed to work together with theory and experiment to accelerate discovery.
Underlying these activities is the idea of a “virtual laboratory,” where chemical systems are explored through simulations performed on high-performance computers. Many of the processes we study require quantum-chemical descriptions at the atomic level, and part of my research therefore involves developing new computational methods capable of treating increasingly complex molecular systems. These tools are applied in close collaboration with experimental researchers in Denmark and internationally, ensuring that computational predictions remain connected to measurable chemistry and practical applications.
.
