Photon distillation creates a low-error single photon from multiple imperfect ones via interference and partial measurements. We present a cost-efficient version with constant gate success probability, regardless of indistinguishability error reduction, ideal for photonic quantum computing architectures.
Photonic quantum computers use the bosonic statistics of photons to construct, through quantum interference, the large entangled states required for measurement-based quantum computation. Therefore, any which-way information present in the photons will degrade quantum interference and introduce errors. While quantum error correction can address such errors in principle, it is highly resource-intensive and operates with a low error threshold, requiring numerous high-quality optical components. We experimentally demonstrate scalable, optimal photon distillation as a substantially more resource-efficient strategy to reduce indistinguishability errors in a way that is compatible with fault-tolerant operation. Photon distillation is an intrinsically bosonic, coherent error-mitigation technique which exploits quantum interference to project single photons into purified internal states, thereby reducing indistinguishability errors at both a higher efficiency and higher threshold than quantum error correction. We observe unconditional error reduction (i.e., below-threshold behaviour) consistent with theoretical predictions, even when accounting for noise introduced by the distillation gate, thereby achieving actual net-gain error mitigation under conditions relevant for fault-tolerant quantum computing. We anticipate photon distillation will find uses in large-scale quantum computers. We also expect this work to inspire the search for additional intrinsically bosonic error-reduction strategies, even for fault-tolerant architectures.
The foundation of fault-tolerant linear optical quantum computing is built upon the interference of identical photons to create high-fidelity entangled states. Unfortunately, fabrication limitations cause sources to emit photons that are only partially indistinguishable, leading to computational errors. To address this, the authors introduce a photon-distillation scheme based on multiphoton-interference linear optics. This protocol mitigates arbitrary indistinguishability errors in a single round. A scheme combining conventional quantum error correction and photon distillation has a higher error threshold than quantum error correction alone, and requires fewer optical components.
This thesis presents the development of resource-efficient photon-distillation schemes designed to produce highly indistinguishable photons, a key requirement for large-scale photonic quantum computing. Although photons do not naturally interact, their indistinguishability enables effective interactions through quantum interference, allowing entanglement generation using only linear optical elements and measurements. In practice, however, photons are never perfectly identical. Small differences introduce decoherence, limiting the scalability and reliability of photonic quantum technologies. Addressing indistinguishability errors is therefore essential for fault-tolerant quantum computing. The thesis first reviews the importance of photon indistinguishability and the need to mitigate related errors. It then examines experimental methods for characterizing imperfections and introduces a general model of imperfect photon states. This model captures indistinguishability errors alongside additional experimental noise from photon loss and multiphoton events. A key insight is that loss and multiphoton noise together can act as an effective source of indistinguishability error. To quantify this effect, the work defines an effective indistinguishability metric, enabling practical evaluation of photon-distillation schemes under realistic conditions. The core theoretical contribution is the development of photon-distillation protocols based on multiphoton Fourier interference. Unlike earlier approaches, these schemes can asymptotically suppress indistinguishability errors to arbitrary levels while requiring only linear scaling in photon resources relative to the desired error reduction. The analysis further suggests that photon distillation, when combined with quantum error correction, can significantly reduce the overall resource overhead for fault-tolerant photonic quantum computing. Experimentally, a three-photon Fourier-based distillation protocol was demonstrated on a high-fidelity photonic processor using a resonantly driven quantum-dot single-photon source. After subtracting excess noise contributions, a 2.2-fold reduction in indistinguishability error was achieved; when excess noise was retained, a 1.2-fold net reduction was observed. These results confirm a clear below-threshold improvement in photon quality. Additional analysis suggests that optimized distillation schemes could reduce the photon cost of a logical qubit by up to a factor of four. The thesis concludes by addressing insertion loss and outlining design strategies and future directions for integrating photon-distillation components into scalable quantum architectures.
Fault-tolerant linear optical quantum computation relies on interference between identical photons to generate entanglement. Unfortunately, photons in the lab tend to be partially distinguishable, generating less entanglement and causing unheralded errors. The authors introduce families of distillation schemes that use n-photon interference and postselection to filter out ``bad'' photons and reduce distinguishability by a factor of n, with resource costs scaling only linearly in n. Along the way, the team also resolves an open problem regarding n-mode Fourier interferometers, namely that the Zero Transmission Law characterizes all suppression if and only if n is a prime power.
We introduce state-of-the-art protocols to distill indistinguishable photons, reducing error rates by a factor of n, using only O(n) photons. We discuss the protocols, resource requirements, and applications to fault-tolerant quantum computation. The United States Government retains, and by accepting the article for publication, the publisher acknowledges that the United States Government retains, a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this work, or allow others to do so, for United States Government purposes.
Distributed quantum computing interconnects small, high-quality nodes through optical links, but this architecture carries a pronounced asymmetry: in-node gates and measurements are cheap and high-fidelity, whereas inter-node communication relies on a low-coherence communication qubit and faulty photonics. Previous approaches overcame the noisy link by placing several high-quality data qubits in each node and consuming them for Bell pair and GHZ state distillation. Here we show that distillation can be avoided altogether. The key observation is that we can engineer a communication error bias, where photonic Bell pairs suffer frequent phase errors but only rare bit-flip errors. We design the syndrome-extraction circuits so that this phase noise appears solely as a measurement error that does not propagate to the data qubits, and is therefore suppressed by simply repeating the measurement; letting the error-correcting code itself, rather than a dedicated distillation subroutine, to purify the link. This dramatically reduces the need for ancillary nuclei: Floquet codes require only a single data qubit per node, while general stabilizer codes require just one additional ancilla. We demonstrate high error-correction thresholds throughout this regime, and we identify lattice surgery as inherently robust for this setting, enabling logical operations at a threshold close to that of quantum memory. As a result, the performance of the quantum computer is limited by the high-quality data qubits, while the requirements on photon indistinguishability and coherence of the communication qubit are substantially relaxed.