Implementing a Magnonic Reservoir Computer Model Based on Time-Delay Multiplexing

Recently we demonstrated experimentally that microwave oscillators based on the time delay feedback provided by traveling spin waves could operate as reservoir computers. In present paper, extend this concept adding feature of multiplexing made available large propagation times/distances waves. The system utilizes nonlinear behavior propagating magnetostatic surface in a yttrium-iron garnet thin film and inherent active ring configuration to process dependent data streams. Higher dimensionality is obtained through time-multiplexing method, emulating "virtual" neurons temporally separated spin-wave pulses circulating below auto-oscillation threshold. To demonstrate efficacy concept, computer evaluated short-term memory parity check benchmark tasks, physical parameters are tuned optimize performance. By incorporating reference line mix input signal directly onto output, both amplitude phase nonlinearity can be exploited, resulting significant improvement task. We also find fading capacity easily controlling gain. Finally, show addition second configured transmit backward volume partly compensate dispersive pulse broadening enhance ring.