Memory-augmented hybrid quantum reservoir computing

Reservoir computing is an effective method for predicting chaotic systems by the use of a highdimensional dynamic reservoir with fixed internal weights, while the learning phase is kept linear, which simplifies training and reduces computational complexity compared with fully trained recurrent neural networks. Quantum reservoir computing uses the exponential growth of Hilbert spaces in quantum systems, allowing greater information processing, memory capacity, and computational power. We present a hybrid neuromorphic quantum-classical approach that implements memory through classical postprocessing of quantum measurements, thus avoiding the need for multiple coherent input injections (as in the original proposal). We tested our model on two physical platforms-a fully connected Ising model and a Rydberg-atom array-and evaluated it on various benchmark tasks, including the chaotic Mackey-Glass time series prediction, where it demonstrates significantly enhanced predictive capabilities and achieves a substantially longer prediction time, outperforming previously reported approaches.