Magnonic inverse-design processor

Artificial Intelligence (AI) technology has revolutionized our everyday lives and research. The concept of inverse design, which involves defining a functionality by a human and then using an algorithm to search for the device's design, opened new perspectives for information processing. A specialized AI-driven processor capable of solving an inverse problem in real-time offers a compelling alternative to the time and energy-intensive CMOS computations. Here, we report on a magnon-based processor that uses a complex reconfigurable medium to process data in the gigahertz range, catering to the demands of 5G and 6G telecommunication. Demonstrating its versatility, the processor solves inverse problems using two algorithms to realize RF notch filters and demultiplexers. The processor also exhibits potential for binary, reservoir, and neuromorphic computing paradigms.