Photon / Photonic Qubit

Photons are the natural carriers of quantum information over long distances and require no cryogenic infrastructure, making them attractive for both quantum computing and quantum networking. Photonic qubits can encode a qubit in several ways. Polarization encoding uses horizontal and vertical polarization as the |0> and |1> basis states, manipulated with wave plates and polarizing beam splitters. Dual-rail (path) encoding represents the qubit as a single photon in one of two spatial modes (two distinct fiber paths or waveguide arms), making the qubit robust to global phase noise. Time-bin encoding uses early and late time slots within a single spatial mode, which is especially practical for fiber transmission. Each encoding scheme has different trade-offs in how easily single-qubit gates and two-qubit interactions can be implemented.

Performing deterministic two-qubit gates on photons is the central challenge of photonic quantum computing. Photons do not naturally interact with each other; in linear optical quantum computing (LOQC), two-qubit gates are implemented probabilistically using beam splitters, phase shifters, and photon-number-resolving detectors. The KLM (Knill-Laflamme-Milburn) scheme, proposed in 2001, showed that universal quantum computing with linear optics and single-photon sources is in principle possible, but requires a large overhead of ancilla photons and postselection. More recent architectures, such as those pursued by PsiQuantum, use photonic fusion-based quantum computing where entangled resource states are generated and fused with probabilistic gates, hiding the probabilistic nature behind a fault-tolerant error-correcting code. Xanadu takes a different approach with continuous-variable (Gaussian boson sampling) photonics encoded in squeezed light modes. QuiX Quantum builds programmable photonic integrated circuits for near-term boson sampling experiments.

Photon detection is a critical bottleneck. Superconducting nanowire single-photon detectors (SNSPDs) achieve detection efficiencies above 95% but require cooling to around 1-4 K, reintroducing some cryogenic complexity at the detector stage even though the photonic circuit itself runs at room temperature. Photon loss in waveguides and fiber, and the difficulty of generating indistinguishable single photons on demand, are the primary sources of error in photonic systems. Boson sampling, a computational task involving many photons interfering in a linear optical network, has been used as an early demonstration of quantum advantage with photonics; while boson sampling is not a universal quantum computation, it provides a near-term benchmark for photonic hardware progress.