Performance Enhancement of a Spin-Wave-Based Reservoir Computing System Utilizing Different Physical Conditions

We numerically study how to enhance reservoir computing performance by thoroughly extracting the spin-wave device potential for higher-dimensional information generation. The has a 1-input exciter and 120-output detectors on top of continuous magnetic garnet film transmission. For various nonlinear fading-memory dynamic phenomena distributing in space, small in-plane fields are used prepare stripe domain structures damping constants at sides bottom explored. ferromagnetic resonant frequency relaxation time spin precession clearly characterizes change dynamics with field constant. common input signal is 1-GHz cosine wave random 6-valued amplitude modulation. A basic 120-dimensional output vector obtained from time-series signals under each three conditions. Then, 240- 360-dimensional vectors also constructed concatenating two ones, respectively. In autoregressive moving average (NARMA) prediction tasks, computational enhanced as dimension becomes higher significantly low error achieved tenth-order NARMA task using optimum results clear evidence that collection diverse efficiently increases dimensionality integrated state (i.e. reservoir-state richness) thereby contributes high performance. This paper demonstrates enhancement through configuration settings practical approach on-chip devices numbers real nodes.