Download file (7.51 MB)
Download as zip (7.24 MB)

Collection

Citation

Description:

The edge of chaos (EoC) refers to a dynamical regime near the boundary between ordered and chaotic states, optimizing information processing in nonlinear dynamical systems. In physical reservoir computing (PRC), the EoC is crucial for high computational performance. Spin-wave interference–based PRC (SWI-PRC) demonstrated high performance, but the EoC's role hasn't been experimentally clarified. This study investigates a SWI-PRC system on a YIG single crystal, tuned by external magnetic field and input pulse interval adjustments. Using phase-space reconstruction and Lyapunov analysis, we evaluate the maximum Lyapunov exponent (λ_max) and identify the EoC operating point. Additionally, we find an additional optimal regime on the chaotic side of the transition (the edge of chaos on the chaotic side). Performance evaluation through nonlinear waveform transformation tasks reveals up to 97% accuracy near the EoC, linking the λ_max to computational performance. These findings suggest that high performance in SWI-PRC stems from dynamically optimized points, offering design insights to enhance information-processing capabilities in edge computing applications.

Rights:

• Creative Commons BY Attribution 4.0 International
  

Keyword: reservoir computing, spin wave, edge of chaos

Date published: 2026-04-30

Publisher: IOP Publishing

Journal:

• Japanese Journal of Applied Physics (ISSN: 13474065) vol. 65 issue. 8 080903

Funding:

• Ministry of Education, Culture, Sports, Science and Technology JPMXP1225NM5247
• Japan Society for the Promotion of Science 20241712

Manuscript type: Author's version (Accepted manuscript)

MDR DOI: https://doi.org/10.48505/nims.6297

First published URL: https://doi.org/10.35848/1347-4065/ae5da2

Related item:

Other identifier(s):

Contact agent:

Updated at: 2026-05-18 08:42:49 +0900

Published on MDR: 2026-05-18 10:23:33 +0900