Towards Fault-Tolerant Quantum Photonic Neural Networks

Quantum photonic neural networks (QPNNs) have emerged as a promising platform for optical quantum computation. Our architecture uses cavity-assisted light-matter interactions, and is ideally suited to operate on information encoded in the bosonic mode degree of freedom [1]. This ability to perform universal operations in encoded bases enables coherent control of bosonic error-correcting codes, which protects information from photon loss and phase errors. To detect these errors, the strong light-matter interaction provided by our optical nonlinearity can be cascaded to perform non-demolition measurements of the total photon number. Conditioned on this measurement, a unitary correction operation can be applied to correct the error. With perfect components, and a conditional routing gate, this enables the construction of error correcting circuits, robust to photon loss. By training the joint system - the computational modes of the QPNN and the correction circuitry, the circuit can be made hardware-efficient and resilient to faulty components while maintaining the fidelity of the logical operation. This work paves the way for fault-tolerance and hardware-efficient quantum computation using near-term optical hardware.

[1] Basani, J. R., Niu, M. Y., & Waks, E. (2024). Universal Logical Quantum Photonic Neural Network Processor via Cavity-Assisted Interactions. arXiv preprint arXiv:2410.02088.