Coordination Funds

• Applicant: Dr. Markus Schmitt
• Subject Area: Theoretical Condensed Matter Physics ; Experimental Condensed Matter Physics
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 550495627

Modern platforms for quantum simulation enable the investigation of complex quantum states in unprecedented detail. Together with the continuous development of numerical approaches, these groundbreaking experiments have made it possible to address long-standing problems and they revealed novel phenomena of quantum many-body physics. A particularly promising aspect of cutting-edge experimental and computational methods is the possibility to investigate and often directly probe distinctive peculiarities of many-particle quantum mechanics, namely (i) the inherently high-dimensional representation of quantum information in the form of classical data, (ii) the inevitable feedback affecting the quantum state upon measurements, and (iii) the generation of entanglement in quantum dynamical processes. Despite this impressive progress, new challenges have emerged simultaneously. For example, measurements from quantum simulators resolve fluctuations of individual degrees of freedom. It is an open question how the comprised information about high-order correlations can be optimally exploited for physical insight. Moreover, the precise manipulation capabilities - especially the new paradigm of quantum-classical feedback - demand optimal control strategies, which are hard to find. Many of these challenges match the proven strengths of modern machine learning (ML) algorithms for pattern recognition, dimensional reduction, and strategy discovery. It is the central goal of this Research Unit (RU) to push the fundamental research of complex quantum systems in both theory and experiment through a collaborative interdisciplinary approach at the interface of quantum physics and ML. Concretely, the initiative aims to find new ways of revealing entanglement structures and non-local order for the characterization of complex quantum states. In view of complex quantum dynamics, the research questions concern information spreading and equilibration in isolated systems as well as controlled entanglement generation, e.g., via measurement-based state preparation. In order to advance the state of the art, it is the goal of the Research Unit to devise novel machine learning-based solutions on various fronts, including readout and data processing strategies for information retrieval from quantum simulators, new computational approaches for non-equilibrium dynamics based on neural network wave function representations, and alternative frameworks for quantum interactive dynamics based on reinforcement learning. By addressing a variety of physical problems, the collaborative effort aims to illuminate where ML can make a difference as a facilitator for scientific progress in the field. Finally, the RU is expected to have a lasting impact on the research community at the interface of ML and complex quantum systems by organizing regular scientific events, fostering a diverse research environment, and contributing to standardizing the exchange of research data.

• DFG Programme: Research Units
• Subproject of: FOR 5919: Machine learning for complex quantum states