Linear-optical generation of hybrid GKP entanglement from small-amplitude cat states

Hybrid bosonic codes combining bosonic codes with photon states offer a promising pathway for fault-tolerant quantum computation. However, the efficient generation of such states in optical setups remains technically challenging due to the requirement for complex non-Gaussian resources. In this paper, we propose a novel scheme to efficiently generate hybrid entangled states between a GKP qubit and a photon-number state using small-amplitude cat states as the primary resource. We apply a breeding process using small-amplitude cat states to increase the non-Gaussianity of the input states. This method requires only linear optical elements and homodyne measurements. Furthermore, we demonstrate that this protocol can be extended to generate hybrid qudit states. This scheme has the potential to provide a resource-efficient and experimentally attractive route toward implementing hybrid quantum error correction.

💡 Research Summary

In this work the authors present a fully linear‑optical scheme for generating hybrid entanglement between a Gottesman‑Kitaev‑Preskill (GKP) logical qubit and a discrete photon‑number state, using only small‑amplitude Schrödinger cat states as the non‑Gaussian resource. The central idea is to replace the demanding large‑amplitude cat or highly squeezed GKP states traditionally required for fault‑tolerant bosonic codes with readily producible odd‑cat states |C⁻_α⟩ (α ≪ 1). In the limit of very small α these states approximate a single photon with fidelity approaching unity, which dramatically reduces the experimental overhead for preparing the initial resource.

The protocol proceeds in three spatial modes. Modes 1 and 2 are initially prepared in odd cat states |C⁻α⟩ and |C⁻{√2α}⟩ respectively, while mode 3 is left in vacuum. A first 50:50 beam splitter (B₂₃) mixes modes 2 and 3; under the small‑α approximation the output becomes a superposition of a single photon in either mode 2 or 3, entangled with the cat in mode 1. A displacement D₁(α) is then applied to mode 1, after which a second beam splitter (B₁₂) interferes modes 1 and 2. Conditional homodyne detection of mode 2 with outcome p = 0 projects the remaining two modes onto

|ψ_o⟩ ∝ (|3β⟩₁ − 2|β⟩₁ + |−β⟩₁)|0⟩₃ + (|2β⟩₁ − |0⟩₁)|1⟩₃, β = α/√2.

The first bracket is an approximate GKP logical‑0/1 basis (a superposition of displaced coherent states), while the second bracket encodes a photon‑number qubit (|0⟩ or |1⟩). Thus the output is a hybrid entangled state linking a continuous‑variable logical qubit to a discrete photon‑number degree of freedom.

A detailed fidelity analysis reveals a non‑trivial trade‑off: as α→0 the small‑α approximation becomes exact but the target state collapses to the photon‑1 component, causing the overall fidelity to drop to ≈0.4. Numerical optimization shows a maximum fidelity F ≈ 0.964 at α ≈ 0.455, where the balance between non‑Gaussianity and overlap with the ideal GKP‑photon state is optimal. The authors also study the effect of a finite homodyne acceptance window |p| ≤ v_up. Enlarging v_up dramatically raises the success probability (up to ≈40 % for v_up ≈ 0.5) while only modestly degrading the average fidelity (≈0.90). Conversely, a tight window (v_up ≈ 0.1) yields an average fidelity >0.99 but with a lower success probability (~12 %).

To further enhance the non‑Gaussian character of the inputs, the paper introduces a “cat‑breeding” stage. Multiple small‑amplitude cat states are interfered on a network of beam splitters, with intermediate homodyne detections conditioning the output. After j breeding iterations the resulting state |~0(j)⟩_L exhibits increased photon‑number variance and a Wigner function with deeper negative regions, indicating stronger non‑Gaussianity. Using this bred state as the input to the main circuit yields higher output fidelities without requiring larger initial α. The authors provide explicit analytic expressions for the bred states after two iterations and illustrate the improvement with Wigner‑function plots.

Beyond qubits, the scheme is generalized to hybrid qudits. By starting from an entangled GKP‑qudit (a d‑dimensional logical subspace encoded in a continuous‑variable mode) together with a path‑entangled photon‑number state across two auxiliary modes, the same linear‑optical network plus homodyne and photon‑number detections produces a state of the form

|~0⟩_L|0⟩ + |1⟩_L|1⟩ + … + |(d‑1)⟩_L|d‑1⟩,

thereby establishing a direct bridge between continuous‑variable error‑correcting codes and discrete photon‑number encodings. The paper details the qutrit (d = 3) case, showing the required sequence of beam splitters, homodyne measurements, and a final photon‑number projection.

Experimental feasibility is discussed qualitatively. The required components—high‑quality 50:50 beam splitters, precise displacement operations (realizable with electro‑optic modulators), high‑efficiency homodyne detectors, and low‑loss optical paths—are all within current state‑of‑the‑art photonic laboratories. While the analysis neglects photon loss, detector inefficiency, and mode‑mismatch, the authors argue that typical homodyne efficiencies (>90 %) and propagation losses (<0.1 %) would still allow the protocol to achieve the reported fidelities and success rates. They suggest that future work should incorporate realistic noise models and explore fault‑tolerance thresholds when the hybrid states are used within larger error‑correction circuits.

In summary, the paper demonstrates that small‑amplitude cat states, when combined with linear optics and homodyne conditioning, suffice to generate high‑quality hybrid GKP‑photon entanglement. The cat‑breeding extension further reduces the resource overhead, and the method naturally scales to higher‑dimensional hybrid codes. This approach offers a promising, experimentally accessible pathway toward fault‑tolerant quantum computation and long‑distance quantum communication using hybrid bosonic error‑correcting codes.