Quantum Reservoir Computing Approach to Error-Mitigated Compilation

We investigate the implementation of reservoir-computing (RC) models in superconducting quantum processing architectures. Inspired by the formal model of quantum reservoir computing [1] and by recent results that indicate that disordered quantum interactions can be trained to synthesize quantum gates [2] we study an “echo state network” defined in a noisy Hilbert space, where quantum operations can act as a non-linear dynamic reservoir acting on a smaller computational space. The model is tested numerically and experimentally on gate synthesis problems of various complexities and results on achievable gate fidelities as a function of the reservoir power are discussed. This RC-gate-synthesis investigation is a precursor to data-driven error-mitigated compilation of complex algorithms on NISQ machines with an acceptable overhead in terms of number of ancillary qubits.

[1] Fujii, K., & Nakajima, K. (2017). Harnessing disordered-ensemble quantum dynamics for machine learning. Physical Review Applied, 8(2), 024030.

[2] Ghosh, S., Krisnanda, T., Paterek, T., & Liew, T. C. (2021). Realising and compressing quantum circuits with quantum reservoir computing. Communications Physics, 4(1), 1-7.