Extensible universal photonic quantum computing with nonlinearity

Technical Summary: Extensible Universal Photonic Quantum Computing with Nonlinearity

Problem Statement
Universal quantum computing requires an architecture capable of executing both linear circuits and strong nonlinear operations. While photonic systems have achieved significant milestones in scalability, programmability, and linear optical operations, they have historically lacked a scalable method to integrate strong nonlinearities. Single-photon-level nonlinearities are intrinsically difficult to realize and even more challenging to implement across all modes in large-scale linear photonic circuits. Consequently, most existing optical experiments remain non-universal, restricted to linear operations, and unable to perform essential tasks such as bosonic error correction or complex many-body quantum simulations that require non-Gaussian resources.

Methodology
The authors present Clavina, an extensible, all-optical quantum photonic platform designed to overcome the integration bottleneck. The architecture draws inspiration from classical integrated circuit design, utilizing a modular, "jigsaw-like" approach where diverse functional units can be connected to a central control unit in a plug-and-play manner.

Key components of the methodology include:

• Central Control Unit: Manages phase locking, synchronization, and provides long-fiber delay lines acting as a "cache" for time-bin encoded information.
• Scalable Linear Network: A multi-core time-bin interferometer (Linear Operation Unit or LOU) that supports large-scale, fully programmable linear transformations.
• Modular Nonlinearity: The system integrates two distinct types of nonlinear resources:
  1. Input-side Nonlinearity: A heralded photon-number-state generator (producing Fock states) coupled with an inline squeezer.
  2. Gate-side Nonlinearity: An addressable Kerr gate module, realized via interferometrically combined photon-addition and -subtraction sequences.
• Time-Bin Encoding: Information is encoded in a single spatial mode but reused across sequential time bins, allowing the same physical modules to be applied throughout the circuit, thereby reducing hardware costs and enabling scalability.

Key Contributions and Results
The paper demonstrates the successful integration of these components to achieve a universal gate set and execute previously inaccessible tasks:

1. Scalability and Stability: The platform maintains high stability over nearly two hours, as evidenced by fidelity tracking. A 100-mode Gaussian boson sampling experiment confirmed the extraction of participation coefficients, and the system demonstrated sustained multimode entanglement across 8,000 temporal modes.
2. Non-Gaussian State Engineering:
   • The system generates photon-number-squeezed states near-deterministically (effective output rate $\approx$ 0.85 MHz).
   • Using an inline squeezer and a two-round breeding process within the LOU, the authors prepared small Schrödinger cat states and expanded them into larger amplitude cat states without post-selection.
   • A "compass state" (a four-component coherent state superposition) was also generated.
3. Quasi-Deterministic GKP State Generation: By interfering cat states in the LOU and utilizing real-time feed-forward based on homodyne detection outcomes, the team achieved the quasi-deterministic generation of optical Gottesman-Kitaev-Preskill (GKP) states. These states are critical resources for bosonic quantum error correction. The generated states exhibited a characteristic grid pattern in the Wigner function and stabilizer amplitudes ($\langle \hat{S}_x \rangle \approx 0.106$, $\langle \hat{S}_{|1_L\rangle} \rangle \approx 0.206$) indicating appreciable lattice-wide phase coherence.
4. Many-Body Quantum Simulation: The authors simulated the Bose-Hubbard model by connecting the programmable linear network to a tunable Kerr gate. This allowed for the encoding of boson-lattice parameters and the observation of quantum dynamics beyond the hard-core boson limit (i.e., sites with more than one boson). The system successfully tracked occupation probabilities for different initial states ($|200\rangle$ and $|110\rangle$) as the interaction-to-tunneling ratio ($U/J$) and evolution time were varied, matching ideal numerical simulations.

Significance and Claims
The paper claims that Clavina establishes a viable route toward universal photonic quantum computing by resolving the long-standing bottleneck of integrating nonlinear resources with scalable linear circuitry.

• Universality: By combining Gaussian resources (linear optics, squeezing) with non-Gaussian resources (Fock states, Kerr gates), the platform satisfies the requirements for a universal gate set in the continuous-variable setting.
• Error Correction: The quasi-deterministic generation of GKP states represents a crucial step toward fault-tolerant quantum computing, as these states are the leading resource for bosonic error correction.
• Simulation Capability: The ability to simulate the Bose-Hubbard model with tunable interactions and multi-occupancy sites demonstrates that photonic hardware is no longer limited to linear operations. This offers a distinct advantage over superconducting platforms, where on-site interactions are often non-tunable and restricted to the hard-core limit.
• Extensibility: The modular architecture allows for future upgrades, such as the integration of deterministic photon adders/subtractors via light-matter interfaces or direct Kerr-like interactions in photonic crystal microcavities, without requiring a complete redesign of the system.

The authors conclude that while challenges regarding photon loss and noise remain, the Clavina architecture provides a blueprint for a scalable, fault-tolerant, and universal photonic quantum computer.