#fbqc

Long known that a scalable, fault-tolerant quantum computer requires millions of physical qubits and advanced error correction. Photonic quantum computing, which encodes quantum information using photons (light particles), is one of the most promising pathways and is making remarkable progress, notably with Fusion-Based Quantum Computation. Studies and technological advances bring error-corrected quantum processing with light closer than ever.

Knowing Fusion-Based Quantum Computation

FBQC, a quantum computation method, uses photons and “fusion” processes to build complicated entangled states for quantum algorithms. FBQC accumulates entangled resources using ‘fusion’ measurements, unlike other quantum computing models that use gate-by-gate applications. This method is ideal for photonic systems since photons' low noise has been crucial to the early demonstrations of superposition, entanglement, and logic gates. Large-scale photonic quantum computing has been problematic because it requires several components that outperform the most modern conventional integrated photonics, such as highly efficient single-photon detectors and complicated integrated systems.

A Photonic Platform for FBQC

The establishment of a scalable photon-based quantum computing platform using silicon photonics manufacturing is a huge accomplishment. With a fully integrated 300-mm silicon photonics process flow, this platform designed with GlobalFoundries ensures scalability and performance similar to high-volume commercial setups.

Important components and their performance are crucial for FBQC:

High-Fidelity Heralded Single-Photon Sources (HSPS): These sources are essential to FBQC and use spontaneous four-wave mixing (SFWM) to generate a photon's pair probabilistically. The platform achieved coincidence-to-accidentals ratios of up to 3,000 for single-photon production on-chip. The initial sources have a spectral purity of 99.5% ± 0.1% without filtration.

Photonic quantum computing uses high-efficiency photon detection to indicate quantum state formation. Niobium nitride (NbN) layers enable high-performance superconducting nanowire single-photon detectors (SNSPDs) with 93.4% median on-chip detection efficiency.

With an average SPAM fidelity of 99.98% ± 0.01%, the platform proficiently prepares and measures single, path-encoded qubits. Great fidelity is needed for quantum processes to be accurate.

Quantum modules must be networked to scale beyond a single chip. With a point-to-point qubit network, the platform achieved a Pauli transfer matrix fidelity of 99.72% ± 0.04% for qubits delivered over 42 meters of optical fibre. Telecommunications-wavelength photonic qubits can transmit without quantum translation.

High-Visibility HOM measures photon interference between two on-chip sources with 99.50% ± 0.25% visibility, setting a new standard for photonic quantum computation.

High-Fidelity Two-Qubit Fusion: FBQC's eponymous operation, Bell fusion, projects onto two-qubit Bell states. The platform demonstrated a fidelity of 99.22% ± 0.12% with the ideal Bell state.

Fault-Tolerant, Low-Overhead FBQC

The baseline technology performs well, but significant improvements are needed to enable “useful” fault-tolerant quantum computing, especially in component loss and deterministic operations. This is where blocklet concatenation and other advanced error correction methods help.

PsiQuantum's Daniel Litinski and colleagues have released research on FBQC-specific “blocklet concatenation” methods. These protocols aim for fault-tolerant operation with reduced overheads than surface codes.

Key elements of this improved error correction include:

Blocklets: This developing technology encodes a single logical qubit, the fundamental, protected unit of quantum information, into multiple physical qubits to reduce error risk and enable modular and scalable architectures.

Code Distance: This crucial metric reveals how few physical qubit errors are needed to cause a wrong logical state. The research conjecture is that the code distance scales favourably with the product of the inner and product code distances raised to L-2. The blocklet code design has L layers.

Numerical simulations indicate subthreshold scaling, supporting the hypothesised link between minimum-weight errors and code distance. This is important because it shows that the code can rectify errors even when the physical error rate is below a threshold, which is required for actual quantum computing. Using 8, 10, and 12-qubit resource states, researchers developed protocol families with 13.8%, 19.1%, and 11.5% erasure thresholds. Erasure threshold, the greatest physical error rate at which these codes can consistently retrieve quantum information, illustrates their practicality.

Footprint Scaling: The resource cost per logical qubit, or "footprint," scales favourably. Unlike surface codes, which require many physical qubits to encode a single logical qubit, this may save resources.

The work proposes applying similar protocols to other quantum computing platforms in addition to photonic hardware, including logical operations, decoding, and implementation.

Technological Advances for FBQC Systems

The report lists several next-generation elements and technologies needed for fault-tolerant quantum computing:

SiN Waveguides with Low Loss: SiN waveguides have a lower refractive index contrast and greater confinement and manufacturing sensitivity than silicon-on-insulator waveguides. Their multimode waveguide losses are as low as 0.5 ± 0.3 dB m⁻¹.

Cascaded Resonator Sources, also known as fabrication-tolerant photon sources, address cryogenic thermal dissipation and pump power issues. A 24-resonator device achieved >99% two-source indistinguishability over a ±400-pm resonance shift, with 99.35% purity and remarkable manufacturing stability.

Super-efficient photon-number-resolving detectors Unlike single-photon detectors, PNRDs can distinguish low photon numbers, allowing FBQC to exclude higher-order photon number states and detect unwanted events. PNRDs with waveguides can resolve four photons and have 98.9% median on-chip detection efficiency with five unit cells.

Low-Loss Fibre-to-Chip Coupling: Practical fibre networking requires reducing loss when coupling optical fibres to photonic chips. New edge coupler designs for high-numerical aperture fibre exhibit coupling losses of 52 ± 12 mdB.

Fast electro-optic switches (BTO) are needed to overcome fusion gate and spontaneous source non-determinism. Barium Titanate (BTO) as an electro-optic phase shifter with high Pockels values allows the creation of vast, low-loss switching networks. The half-wave loss-voltage product is 0.33 ± 0.02 dB.V.

The Way Forward

FBQC approaches may tolerate 1% per-qubit fusion network faults and 10% optical loss between photon generation and detection. The shown feature-complete collection of optical components with optical losses of several percent or less and fully integrated circuits with sub-percent error levels is a big step.