Driven Abrikosov Vortices for Narrowband Electromagnetic Field Sensing and Reservoir Computing

Pinned Abrikosov vortices are shown to enable detection of ultra-low electric and magnetic fields with frequency selectivity beyond that of superconducting quantum interference devices (SQUIDs). Arrays of pinned vortices are further demonstrated to function as dynamical reservoirs for neuromorphic computing and, when coupled to a convolutional recurrent neural network (C-RNN), perform diverse computational tasks including feature extraction and classification. Key mathematical findings include closed-form solutions for the nonlinear resonances of pinned vortices, validated through numerical simulations of vortex dynamics under combined impurity scattering and external field effects. Simulations of high-temperature type-II superconductors with material defects indicate field sensitivities on the order of fT/√Hz, comparable to low-temperature SQUIDs. The frequency-selective response provides at least a threefold improvement over state-of-the-art detectors, lowering the noise floor by suppressing flicker noise. Processing the same sensed data through the vortex-array reservoir yielded 97% classification accuracy, demonstrating the dual capability of vortex arrays as precision sensors and computational reservoirs. These results establish a pathway for integrating superconducting vortex physics and electromagnetic field responses with neuromorphic computing architectures.