Quantum Optical Neuromorphic Platform for Universal Quantum Information Processing using Cavity-QED Systems

Physically motivated computing hardware has recently captured interest because of the potential to leverage natural physical dynamics for computation, as well as the promises of orders of magnitude improvement in speed, scale and efficiency. We present an integrated photonic architecture that uses cavity-QED based optical nonlinearities, and show that this offers a path to deterministic high-fidelity quantum operations. The activation function relies on the mechanism of photon subtraction, followed by coherent photon addition to introduce a restricted photon-number-selective phase gate. We show analytically that the dynamics of the nonlinear activation function is confined to the single-mode subspace, which is a common limitation in many other optical nonlinearities. Through numerical simulations, we show that the proposed architecture can deterministically prepare a wide array of quantum resource states such as N00N states and GHZ states, which are essential for quantum metrology and measurement-based quantum algorithms. Moreover, we also show that the proposed architecture can also prepare the two-mode encoded version of the χ⁽²⁾ binomial bosonic code, which can protect against dephasing errors, photon loss and amplitude damping errors with a constant code rate. Both the encoding and logical operations on the code are achieved using this neural network architecture. The proposed device paves the way for near-term quantum photonic processors that enable quantum computing and simulation which can be achieved using present-day quantum hardware.