Tokyo University of Science Develops AI Design for 2D Magnonic Crystals, Widening 4th-5th Band Gaps
Updated
Updated · Mirage News · Jul 28
Tokyo University of Science Develops AI Design for 2D Magnonic Crystals, Widening 4th-5th Band Gaps
3 articles · Updated · Mirage News · Jul 28
Summary
A Tokyo University of Science team reported an inverse-design framework that uses AI-guided topology optimization to find 2D magnonic crystal structures with larger complete magnonic band gaps.
The method pairs a genetic algorithm with frequency-domain Landau-Lifshitz-Gilbert simulations, letting researchers screen many candidate unit-cell layouts faster and more reliably than conventional time-domain simulations.
In a bicomponent europium oxide-iron crystal, the optimization converged on a non-intuitive topology that outperformed conventional geometries, with the widest gap appearing between the 4th and 5th bands.
Explainable machine-learning analysis showed the design landscape becomes more non-convex at higher-order bands, indicating multiple viable solutions and helping clarify why those bands are harder to control.
The July 28 Small Structures study points to broader use in spintronic logic, multi-frequency signal processing and energy-efficient computing hardware, including future data-center applications.