Downloading many-qubit entanglement from continuous-variable cluster states

Many-body entanglement is an essential resource for many quantum technologies, but its scalable generation has been challenging on qubit platforms. However, the generation of continuous-variable (CV) entanglement can be extremely efficient, but its utility is rather limited. In this work, we propose a scheme to combine the best of both qubit and CV approaches: a systematic method to download useful many-qubit entanglement from the efficiently generated CV cluster states. Our protocol is based on one-bit teleportation of the qubit correlation encoded in the displaced Gottesman-Kitaev-Preskill basis. To characterize the practical performance of our scheme, we develop an equivalent circuit to map dominant CV errors to single-qubit preparation errors. Particularly, we relate finite squeezing error to qubit erasure, and show that only 5.4 dB squeezing is sufficient to implement robust qubit memory or quantum computation (QC), and 11.9 dB for fault-tolerant QC. Our protocol can be implemented with the operations that are common in many bosonic platforms.

💡 Research Summary

The paper introduces a hybrid approach that leverages the scalability of continuous‑variable (CV) cluster states and the computational universality of discrete‑qubit cluster states. The authors propose a three‑step “downloading” protocol that extracts an N‑qubit cluster state with the same graph as an N‑mode CV cluster, using only operations that are routinely available in many bosonic platforms (conditional displacements, homodyne q‑measurements, and single‑qubit phase corrections).

Protocol Overview

1. Prepare an N‑mode CV cluster |G_CV⟩ by initializing each mode in a p‑quadrature eigenstate (ideally |0⟩ₚ) and applying CV CPHASE (ˆC_CVZ) gates according to a graph G.
2. Couple each mode to an auxiliary physical qubit (initialized in |+⟩) via a conditional‑displacement (CD) gate, ˆC_D = I⊗|0⟩⟨0| + e^{−i√π p̂}⊗|1⟩⟨1|. In the displaced Gottesman‑Kitaev‑Preskill (GKP) basis this gate implements a CNOT.
3. Measure every CV mode in the q‑quadrature. The measurement outcomes q = (q₁,…,q_N) determine a set of phase‑shifts ϕ = √π A q (A is the adjacency matrix). Applying ˆR_Z(ϕ_i)=e^{−iẐ_iϕ_i/2} to each qubit yields a state proportional to the desired qubit cluster |G⟩.

Mathematically, ˆR_Z⟨q|ˆC_D|G_CV⟩|+⟩^{⊗N} ∝ |G⟩, establishing the first main result: the qubit entanglement is “hidden” in the CV cluster and can be teleported to the ancilla qubits.

Error Mapping and Equivalent Circuit
Realistic CV clusters cannot be prepared in the ideal |0⟩ₚ state; instead they are approximated by squeezed thermal states ρ_sq = S(r) ρ_th( n̄ ) S†(r). The authors develop an equivalent circuit that maps the imperfections of the CV resource onto single‑qubit preparation errors. After the CD gate and q‑measurement, each qubit is in a superposition
|Ψ(q)⟩ ∝ ψ(q) |0⟩ + e^{−i p₀√π} ψ(q−√π) |1⟩,
where ψ(q) is the wavefunction of a finitely squeezed vacuum. Two error mechanisms appear:

1. Amplitude imbalance (γ = |ψ(q−√π)/ψ(q)| ≠ 1) caused by finite squeezing.
2. Random phase (dephasing) due to p‑displacements p₀ from thermal noise.

The amplitude imbalance is known once q and the squeezing r are known. The authors propose a weak‑measurement POVM that either restores the qubit to an equal superposition (success) or projects it onto a computational basis state (failure). Failure corresponds to a qubit erasure (loss) from the cluster. Averaging over measurement outcomes gives a deletion probability
p_del = erf( e^{−r₀} √π / 2 ).

Thus finite‑squeezing errors are converted into qubit erasures. Surface‑code based fault‑tolerant quantum computing (FTQC) tolerates up to 24.9 % loss, which translates to a squeezing requirement of 11.9 dB. Simpler tasks such as quantum memory or non‑fault‑tolerant computation tolerate up to 50 % loss, requiring only 5.4 dB squeezing—levels already achieved in recent experiments.

Thermalization, Loss, and Detection Inefficiency
Thermal noise (non‑zero n̄) introduces a dephasing channel with rate p_Φ = (1−e^{−πσ²/2})/2, where σ² is the variance of the Gaussian p‑displacement. The weak‑measurement correction remains effective because it is diagonal in the computational basis.

Channel loss and detector inefficiency are modeled as random q‑ and p‑displacements before ideal homodyne detection. p‑displacements only add a global phase after q‑measurement, while q‑displacements affect the CD gate and manifest as correlated dephasing among multiple qubits in the equivalent circuit. Correlated noise is harder to correct, so the authors suggest adding anti‑correlation through additional Gaussian operations (beam splitters, mode‑dependent squeezing, modified CPHASE strength). By carefully engineering these ancillary operations, the effective CV cluster after loss and inefficiency can be made thermal, so that the downloaded errors remain independent qubit dephasing rather than correlated.

Relaxed Conditional Displacement Requirement
The protocol originally assumes a conditional displacement of exactly √π. The authors note that a weaker displacement of √π/R can be applied repeatedly R times (using R copies of the mode or sequential interactions) to achieve the same logical effect, easing experimental constraints.

Implications
The work provides a concrete bridge between the ease of generating massive CV entanglement and the utility of qubit cluster states. By translating dominant CV imperfections into well‑studied qubit error models (erasure and dephasing), the authors enable the use of existing high‑threshold error‑correction codes. The squeezing thresholds identified (5.4 dB for non‑FT tasks, 11.9 dB for FTQC) are within reach of current optical and microwave platforms, suggesting near‑term experimental demonstrations. Moreover, the protocol’s reliance on standard bosonic operations (conditional displacements, homodyne detection) makes it broadly applicable across photonic, superconducting, and trapped‑atom systems, opening a pathway toward scalable, hybrid quantum processors that combine the best of continuous‑variable generation and discrete‑variable computation.