Our Mission
We bring computational modeling of materials closer to experimental reality by developing materials-inspired methods. To achieve it we want simple methods that work for complex materials. We are most interested in energy materials and predicting their electronic, optical, and dynamical properties. To reach our goals, we combine quantum-mechanical methods, molecular dynamics, and computational techniques such as machine learning, see research section.
Recent Group News
20.03.2026: We developed physics-informed Hamiltonian learning with a model that we call HAMSTER: see Martin’s paper in Nature Communications and NAT school news item.
15.03.2026: Manjari Jain and Tomer Amit join our group as post-doctoral fellows, welcome to the team!
03.02.2026: We propose machine-learning for Raman spectroscopy to discover highly conductive battery materials in Manuel’s study published in AI for Science, see also NAT school news item.
Theory of Functional Energy Materials (TheoFEM) 