GC-07 Magnonic Resonators as Building Blocks of Scalable Magnonic Logic Circuits and Reservoir Computers

We report on nonlinear magnonic resonators as viable building blocks of unconventional computing architectures. Via micromagnetic simulations, we demonstrate that the spin-wave modes confined in magnonic resonators [1-3] exhibit a strongly nonlinear response owing to energy concentration when resonantly excited by incident linear spin waves [4]. We use this nonlinearity to design and optimize (i) magnonic logic gates, and (ii) nonlinear magnonic reservoirs for data processing in the time domain. The developed gates include NAND (with fidelity of better than 95%, which can be improved further albeit at the expense of unwanted phase variation), as well as AND (93%) and NOR (92%). The designed reservoirs consist of one or more resonators coupled either actively (via an additional microwave feedback ring) or passively (via multiple spin wave reflections between the resonators). Their performance is analyzed for a range of benchmark tests on analog and noisy binary data, using standard time-multiplexing protocols for realistic device parameters informed by the micromagnetic simulations. Our tests reveal that the passive reservoirs can show superior performance for simple short memory tasks, such as retrieving previous / predicting next sample, parity checks and majority gates, simple arithmetic operations. For binary inputs, fidelity of better than 99% may be achieved even for noise levels of up to 10%. At the same time, the actively coupled reservoirs perform better on more complex tasks, such as NARMA10 or NARMA20 time series prediction, while the feedback coefficient proves a useful handle for fine-tuning the reservoir. Our magnonic devices, both gates and reservoirs, may be concatenated to form more complex and powerful logic circuits and reservoir computing architectures. The research leading to these results has received funding from the UK Research and Innovation (UKRI) under the UK government’s Horizon Europe funding guarantee (Grant No. 10039217) as part of the Horizon Europe (HORIZON-CL4-2021-DIGITAL-EMERGING-01) under Grant Agreement No. 101070347. Yet, views and opinions expressed are those of the authors only and do not necessarily reflect those of the EU, and the EU cannot be held responsible for them.References: [1] V. V. Kruglyak, Appl. Phys. Lett. 119, 200502 (2021). [2] K. G. Fripp, A. V. Shytov, and V. V. Kruglyak, Phys. Rev. B 104, 054437 (2021). [3] H. Qin, et al, Nature Commun. 12, 2293 (2021). [4] K. G. Fripp, Y. Au, A. V. Shytov, and V. V. Kruglyak, Appl. Phys. Lett. 122, 172403 (2023).