Projects

Reservoir Computing is a cutting-edge approach for predicting chaotic time series, a challenge with applications across science and technology. Instead of relying solely on abstract algorithms, it leverages dynamical systems that transform input data into rich, high-dimensional states. These systems can even be physical.
In our research, we explore active matter - driven, non-equilibrium model systems often resembling "swarms" - as reservoirs. Such systems naturally display emergent behaviors like flocking and produce intricate spatio-temporal patterns. By simulating these dynamics, we investigate how the underlying physics influences predictive performance within the machine learning framework.
Our goal is to harness the complexity of active matter to inspire unconventional, energy-efficient, and neuromorphic computing methods, paving the way for novel in-materio and edge computing devices.

Researcher: Mario Gaimann

We investigate computation through the lens of dynamical systems, unifying physical processes and machine learning. By treating both hardware and algorithms as evolving dynamical systems, we leverage natural physical dynamics - such as relaxation to stable states and phase transitions - as a foundation for robust, efficient computation.
Our work shows that concepts from physics - like order parameters and criticality in the Ising model - directly shape learned representations in self-supervised auto-encoders, revealing deep connections between physical phase transitions and optimization dynamics in neural networks.
By framing computation as constrained dynamical evolution, we enable abstraction across scales: algorithms emerge predictably from physical behavior, decoupled from hardware specifics. This approach, grounded in dynamical systems theory, paves the way for general-purpose physical computers that are energy-efficient, scalable, and inherently adaptive.

Researcher: Max Weinmann

Physical systems far from equilibrium-ranging from turbulent fluids and active matter to driven materials-often display complex collective behavior that cannot be captured by tracking every microscopic detail. Our research focuses on building effective coarse-grained models that reveal the essential physics of such non-equilibrium many-body systems and we develop new methods to describe their dynamics using machine learning. By combining theoretical approaches from statistical mechanics with modern data-driven methods, we aim to bridge fundamental physics with practical modeling tools.
We seek to advance the physics of coarse-graining non-equilibrium dynamics, providing insights into how information is stored, processed, and transmitted in complex physical systems. At the same time, the methods developed here contribute to the broader interdisciplinary effort of integrating physics-based understanding with machine learning, opening new opportunities for both fundamental science and applied research.

Researcher: Patrick Egenlauf