New Opportunities in Physics‐based Computing: Magnonics, Spintronics, Photonics and Beyond

Physics-based computing explores alternatives to traditional transistor-based electronics, leveraging nonlinear phenomena in various physical domains to solve real-world problems.
This approach extends the concept of analog computing, offering potential for sustainable, energy-efficient solutions in information and communication technologies. The demand for machine learning has surged, driven by advancements like ChatGPT, but the energy cost of digital, transistor-based computing remains high, particularly in large-scale data centers. In contrast, physics-based computing, such as physical reservoir computing, harnesses nonlinear dynamics in systems like magnetic skyrmions, photons, and spin waves to enable more efficient data classification, pattern recognition, and time series predictions. By operating in the nonlinear regime, these physical systems convert input signals into measurable outputs, offering a promising route to edge computing applications in fields like medical diagnostics, au-tonomous mobility, and smart energy grids. The strength of physical computing lies in its ability to process complex data with minimal energy consumption, without relying on traditional deep learning algorithms. Magnonics, spintronics, and photonics are key fields in this emerging area, offering systems that can serve as powerful, interconnected computational reservoirs. Spintronics, harnessing the electrons’ spin, enables the design of energy-efficient neural networks, while magnonics and photonics provide high-frequency platforms for wave-based computing. The integration of these fields, alongside other platforms such as superconductors, presents new opportunities for developing complex, scalable computing systems that can meet the growing demands of machine learning while advancing sustainable technologies.

The conference language will be English. The Wilhelm and Else Heraeus-Foundation bears the cost of full-board accommodation for all participants.