Welcome to the Theory Division

In the theory division at the Max Planck Institute for the Science of Light, we focus on the intersection between machine learning and physics, especially in the areas of quantum physics and optics. Pushing the boundaries of artificial scientific discovery, we tackle challenges like having the computer discover novel experimental designs, finding the most informative measurement sequences, and deriving new hypotheses from observations. Work of this type in the division has contributed to proposing new quantum optics experiments and quantum error correction strategies. We aim for interpretability and generalizability, employing and embedding existing physics knowlegde in machine learning models, covering areas like topological transport, quantum computing, or nonlinear optics. In the second major direction of the division, we explore new neuromorphic computing setups. Research in that field is driven by the current unsustainable scaling of resource requirements for training and deploying ever-larger neural networks. The goal is to design alternatives to digital neural networks, based on analog  hardware that is much closer to the microscopic physics. This promises to show a high degree of parallelism, speed, and energy efficiency. We pay particular attention to developing efficient means of physics-based training and applying them to new platforms, e.g. based on integrated photonics.