ABSTRACT
Devices based on arrays of interconnected magnetic nano-rings with emergent magnetization dynamics have recently been proposed for use in reservoir computing applications, but for them to be computationally useful it must be possible to optimise their dynamical responses. Here, we use a phenomenological model to demonstrate that such reservoirs can be optimised for classification tasks by tuning hyperparameters that control the scaling and input-rate of data into the system using rotating magnetic fields. We use task-independent metrics to assess the rings' computational capabilities at each set of these hyperparameters and show how these metrics correlate directly to performance in spoken and written digit recognition tasks. We then show that these metrics, and performance in tasks, can be further improved by expanding the reservoir's output to include multiple, concurrent measures of the ring arrays' magnetic states.
ABSTRACT
Hydrogen detection is essential for its implementation as an energy vector. So far, palladium is considered to be the most effective hydrogen sensing material. Here we show that palladium-capped hafnium thin films show a highly reproducible change in optical transmission in response to a hydrogen exposure ranging over six orders of magnitude in pressure. The optical signal is hysteresis-free within this range, which includes a transition between two structural phases. A temperature change results in a uniform shift of the optical signal. This, to our knowledge unique, feature facilitates the sensor calibration and suggests a constant hydrogenation enthalpy. In addition, it suggests an anomalously steep increase of the entropy with the hydrogen/metal ratio that cannot be explained on the basis of a classical solid solution model. The optical behaviour as a function of its hydrogen content makes hafnium well-suited for use as a hydrogen detection material.
