arXiv preprint introduces process-tensor diagnostics for information scrambling and loss in quantum reservoir computing

The study derives a classical-quantum state from the process tensor of typical QRC protocols and expresses input-subsystem mutual informations as Holevo quantities. Numerical sweeps across Hamiltonian regimes and measurement strengths then yield two scalar diagnostics: one for nonlocal scrambling inside the substrate and one for irreversible loss to the environment. These quantities are compared directly to task performance on standard time-series benchmarks, revealing that moderate scrambling without excessive loss correlates with higher accuracy.

Prior QRC literature has focused on fading memory and linear separability while treating scrambling as an implicit side effect. This work supplies explicit, computable metrics that separate scrambling from loss and ties both to the underlying open-system dynamics. The approach also clarifies why certain chaotic or strongly dissipative Hamiltonians degrade performance even when classical echo-state properties appear intact.

Because the diagnostics are derived from the process tensor, they apply to any Markovian or non-Markovian reservoir whose dynamical map can be reconstructed. Future experiments could extract the same quantities from randomized benchmarking or process tomography on superconducting or photonic platforms, providing a concrete route to hardware-aware QRC design.

The main limitation remains that all results are obtained from classical simulation of small qubit registers; experimental noise and finite sampling will alter the observed Holevo values.