Regensburg 2025 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

MA: Fachverband Magnetismus

MA 7: INNOMAG e.V. Prizes 2025 (Diplom-/Master and Ph.D. Thesis)

MA 7.1: Hauptvortrag

Montag, 17. März 2025, 15:00–15:20, H18

Realizing Reservoir Computing with skyrmions in geometrical confinements tuned by ion irradiation — •Grischa Beneke — Institut für Physik, Johannes Gutenberg-Universität Mainz, 55128 Mainz, Germany

Physical reservoir computing (RC) is a beyond von-Neumann computing paradigm that harnesses complex physical systems' dynamics for efficient information processing. Magnetic skyrmions, topological spin textures, show promise for RC systems due to their non-linear interactions and low-power manipulation capabilities. Previous spin-based RC implementations either focused on static detection or required rescaling of real-world data to match intrinsic magnetization dynamics timescales. In this thesis, we demonstrate time-multiplexed skyrmion RC by adjusting the reservoir's intrinsic timescales to match real-world temporal patterns [1]. Using hand gestures recorded via range-Doppler radar in collaboration with industry, we show that our hardware solution outperforms conventional software-based neural networks while consuming less energy. The system's ability to directly integrate sensor data without temporal conversion enables real-time applications. To structure the geometrical confinement, we investigate how high-energy ion irradiation influences magnetic properties in skyrmion-hosting multilayers. By locally modifying the perpendicular magnetic anisotropy, we create attractive and repulsive regions that enable controlled skyrmion nucleation, manipulation and confinement, enabling novel devices and, enhancing device functionality. [1] G. Beneke et al., Nat. Commun. 15, 8103 (2024).

Keywords: Skyrmions; Reservoir computing; Unconventional computing; Focused Ion Beam; Ion Irradiation