Presentation
4 October 2022 Frustrated arrays of nanomagnets for efficient reservoir computing (Conference Presentation)
Alexander Edwards, Dhritiman Bhattacharya, Peng Zhou, Nathan R. McDonald, Lisa Loomis, Clare D. Thiem, Jayasimha Atulasimha, Joseph S. Friedman
Author Affiliations +
Abstract
We present initial experimental results and simulate a nanomagnet reservoir computer (NMRC) solving tasks requiring high memory content, with an area-energy-delay product ten-million times lower than CMOS systems. We manufactured a small nanomagnet reservoir demonstrating a frustrated state. We evaluated the performance on two novel tasks. Our results indicate the reservoir’s short-term memory capabilities and ability to integrate information from multiple concurrent inputs. In the end, our system saw a reduction in area by a factor of 50,000, in energy by a factor of 60, and in period by a factor of four as compared with an equivalent CMOS reservoir.
Conference Presentation
© (2022) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Alexander Edwards, Dhritiman Bhattacharya, Peng Zhou, Nathan R. McDonald, Lisa Loomis, Clare D. Thiem, Jayasimha Atulasimha, and Joseph S. Friedman "Frustrated arrays of nanomagnets for efficient reservoir computing (Conference Presentation)", Proc. SPIE PC12205, Spintronics XV, PC122050N (4 October 2022); https://doi.org/10.1117/12.2633534
Advertisement
Advertisement
KEYWORDS
Manufacturing

Computer simulations

Electron beam lithography

Machine learning

Magnetism

Back to Top