Regensburg 2022 – wissenschaftliches Programm

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

MA: Fachverband Magnetismus

MA 19: Poster 1

MA 19.31: Poster

Dienstag, 6. September 2022, 17:30–20:00, P2

FD micromagnetic solver for inverse-design magnonics — •Andrey Voronov¹, Qi Wang¹, Dieter Suess², Andrii Chumak¹, and Claas Abert² — ¹Nanomagnetism and Magnonics, Faculty of Physics, University of Vienna, Austria — ²Physics of Functional Materials, Faculty of Physics, Iniversity of Vienna, Austria

The idea of utilizing a collective excitation of the electron spin system in magnetic solids, so-called spin-waves, for data processing has been developing in recent years. However, the design of complex data-processing units requires elaborate and complicated investigations.

Recently, the concept of inverse-design magnonics, in which any functionality can be specified first and a feedback-based computational algorithm is used to obtain the device design, has been demonstrated numerically [1]. The same algorithm was used to design a magnonic (de-)multiplexer, a nonlinear switch, and a circulator [1].

One of the next challenges for inverse design is the computation of universal Boolean logic gates NAND and NOR. However, such gates require increasing the complexity of the structure used in [1] and the combination of the MuMux3 simulations with the direct binary search algorithm (DBS) is no longer applicable. Here I report on the use of finite difference (FD) micromagnetic solver based on the Pytorch open source machine learning framework for inverse design. The proposed algorithm greatly facilitates the design of the applied devices and is a useful tool especially for spin-wave computing elements.

[1] Wang, Q., et al (2021). Nature Communications, 12(1), 1-9.