Experimental certification of multipartite Bell correlations using only few-body symmetric correlations

AI summary

Overview

Research area: Experimental quantum information — multipartite entanglement and Bell nonlocality in photonic systems.

Technical level: Intermediate (the core argument is accessible; the witness construction and polytope optimization details are advanced).

Scope: An experimental demonstration that Bell correlations in six- and eight-photon Dicke states can be certified using only permutation-invariant two- and four-body correlators, without ever measuring full-body correlations.

What This Paper Is About

Standard multipartite Bell inequalities (such as the Mermin-Ardehali-Belinskii-Klyshko family) require measuring correlations across all parties at once, which becomes experimentally impractical as systems grow. The authors ask whether Bell correlations can instead be certified from a fixed, small correlation order — only two-body and four-body symmetrized correlators — no matter how many parties there are. They answer experimentally by preparing postselected six- and eight-photon Dicke states and violating tailored permutation-invariant witnesses built from exactly four symmetric correlators.

Key Contributions

1. Few-body witnesses with dramatically improved noise tolerance. By adding four-body correlators to the permutation-invariant (PI) ansatz while keeping two measurement settings per party, the authors obtain a witness for six qubits that tolerates a white-noise fraction of p_c ≈ 0.301, compared with p_c ≈ 0.048 for the best two-body PI inequality with the same two settings per party.

2. Experimental certification of Bell correlations in six- and eight-photon Dicke states. Violations of the witness bounds by 32 standard deviations (N = 6) and 8.9 standard deviations (N = 8) are observed, from states with fidelities 0.933(4) and 0.916(11) to the half-excitation Dicke states. The authors describe the eight-photon result as, to their knowledge, the first certification of Bell correlations in an eight-photon system within this framework.

3. A scalable, correlation-order-fixed construction. The witnesses for N = 6 and N = 8 use the same four symmetric correlators (S_00, S_11, S_0011, S_1111), differing only in their coefficients. Extending the family numerically up to N = 10 shows the two- and four-body family retains p_c ≈ 0.37, whereas two-body families with two or three settings drop below 0.1 for N ≥ 6.

4. Demonstrated robustness to measurement misalignment and to reduced state fidelity. Six deliberately displaced measurement settings all retain violations between 6.8σ and 14σ, and states prepared at lower fidelities of 0.797(5) and 0.638(5) reproduce the predicted hierarchy of witness robustness.

Main Findings

• Benchmark six-photon witness: ℐ₁ = S_00 + S_11 − (1/4) S_0011 + (1/6) S_1111 ≥ −18, with quantum value β_Q ≈ −25.764 at shared angles θ ≈ (3.550, 5.730), p_c ≈ 0.301 and critical fidelity F_c ≈ 0.70.

• Comparison with the best two-body inequality: ℐ_T = 15 S_00 + 6 S_01 − S_11 ≥ −120 has β_Q = −126.041 at (3.264, 2.165) and tolerates only p_c ≈ 0.048 (F_c ≈ 0.95). At fidelity 0.933(4), ℐ₁ gives a violation factor R = 1.334(10) while ℐ_T gives R = 0.998(14), statistically compatible with its classical bound.

• Second six-photon witness: ℐ₂ = 15 S_00 − 6 S_11 − (5/4) S_0011 + (1/8) S_1111 ≥ −135, with β_Q = −157.907 at (2.864, 0.749), p_c ≈ 0.145 and F_c ≈ 0.86; measured R = 1.086(12), a 6.9σ violation at the highest fidelity.

• Eight-photon witness: ℐ₃ = (3/2) S_00 + (9/2) S_11 − (1/4) S_0011 + (1/4) S_1111 ≥ −48, with β_Q = −62.014 at (0.369, 5.844), p_c ≈ 0.226 and F_c ≈ 0.78. Measured ℐ₃ = −55.9(9), i.e. R = 1.165(19), an 8.9σ violation.

• Fidelity-dependence confirms the predicted ordering: at fidelity 0.797(5), ℐ₁ still violates with R = 1.105(9) while ℐ₂ gives R = 0.927(13) and no longer violates; at 0.638(5) neither detects a violation, consistent with p_c(ℐ₁) > p_c(ℐ₂) > p_c(ℐ_T).

• Family optima versus sparse witnesses: a see-saw optimization over all eight coefficients of the (6,2,2) two- and four-body family reaches p_c ≈ 0.46 (and 0.38 for N = 8), but the optimum weights all eight correlators, which would amplify statistical uncertainty; the sparse witnesses were selected for experimental practicality instead.

• System-size trend: the two- and four-body family with two settings reaches p_c ≈ 0.53, 0.46, 0.38, and 0.37 for N = 4, 6, 8, and 10 respectively, whereas two-body families with two or three settings stay below 0.1 for N ≥ 6. A third measurement setting raises the six-qubit two-body threshold only to p_c ≈ 0.093.

• Measurement-angle tolerance: for ℐ₁ at fidelity 0.933(4), the optimal point P₀ gives R = 1.334 ± 0.010 (32σ), and the displaced settings P₁–P₆ (angles between 3.197 and 3.910 for θ₀, and 5.547 to 5.976 for θ₁) yield R from 1.074 ± 0.011 to 1.168 ± 0.015, i.e. 6.8σ to 14σ.

• Measurement cost reduction: because of no-signaling and symmetrization, one N-fold coincidence event supplies up to C(N,2) samples of a two-body PI correlator and C(N,4) samples of a four-body one; only 15 permutation-related configurations are needed for ℐ₁ and ℐ₂, and 28 for ℐ₃, instead of 2^N combinations in the complete two-setting space.

• Scope of the claim: the data are postselected on N-fold coincidences, so the violation characterizes the postselected ensemble and is explicitly not a loophole-free or device-independent Bell test. The certificate rests on two stated assumptions — that the same state and same local measurement are used across all configurations, and that the postselected coincidences are a fair sample.

Methodology in Plain English

Photon pairs are produced by type-II spontaneous parametric down-conversion in a periodically poled KTiOPO₄ crystal pumped by a mode-locked Ti:sapphire laser (775 nm center wavelength, 80 MHz repetition rate). A potassium dihydrogen phosphate crystal plus a pair of Soleil-Babinet compensators corrects temporal walk-off, and five 50:50 beam splitters distribute photons into six spatial modes (with two additional beam splitters for the eight-mode case). Each mode passes through a quarter-wave plate, half-wave plate and polarizing beam splitter to implement the chosen local measurement, then into single-mode fiber and a superconducting nanowire single-photon detector of about 90% efficiency. Detecting exactly one photon in each mode postselects the Dicke state.

The theory side works as follows. Because the target Dicke state is permutation symmetric, the authors use Bell expressions built only from symmetrized correlators, summed over all tuples of distinct parties. The classical bound of such an expression is computed over the vertices of the permutation-invariant local polytope, whose size grows only polynomially with the number of parties rather than as 2^(mN). The state is a common eigenstate of the all-party σ_z and σ_x products, which forces every odd-body correlator in the x–z plane to vanish, so only even-body terms can contribute. Measurement angles and coefficients are chosen by optimizing the critical white-noise fraction p_c = 1 − β_C/β_Q against a depolarized model of the target state. Once fixed, the classical bound remains valid for arbitrary states and arbitrary local measurements; the design model only determines how large the violation is.

Experimentally, the six-photon state fidelity is characterized by an exact projector decomposition over 21 polarization settings, and the eight-photon state over 29 settings. Sixfold coincidences are recorded at 6.8 Hz and eightfold coincidences at 0.031 Hz. Because several expectation values come from the same coincidence data, statistical correlations are retained using a Poisson parametric bootstrap, with error bars from one standard deviation of bootstrap realizations.

Why This Matters

Impact on research. Multipartite Bell tests have been bottlenecked by the need to measure ever-higher-order correlations. This work shows that a fixed correlation order — two- and four-body — suffices to certify Bell correlations in six- and eight-photon states, and provides numerical evidence up to N = 10 that the threshold degrades slowly while two-body approaches collapse below 0.1. The six-photon fidelity of 0.933(4) also substantially exceeds the 0.654(24) reported in the earlier reference on this state. The paper is explicit that the advantage is a reduction of required correlation order, not of photon-number detection requirements.

Potential applications (prospective, not claims of this paper):

• Scalable certification of entanglement resources in photonic quantum computing and communication hardware.
• Practical benchmarking of intermediate-scale quantum devices where full tomography or full-body correlation measurement is infeasible.
• Device-independent-flavored protocol design where only a subset of correlators is assumed trustworthy.
• Certification of symmetric states in atomic-ensemble and superconducting platforms, where related witnesses have already revealed Bell correlations.

Industry relevance. The methods target the practical economics of quantum hardware validation: fewer measurement configurations (C(N,2) rather than 2^N) means less acquisition time per certificate. This matters for groups building photonic sources of entangled states and for metrology-grade precision spectroscopy, which is reflected in the author affiliations spanning precision spectroscopy institutes and quantum-computing initiatives.

Future Directions

• Closing the loopholes. The current certification is postselected and neither loophole-free nor device-independent; moving toward fair-sampling-free or detection-loophole-free versions is the natural next step.
• Pushing to larger N experimentally. The numerics extend to N = 10 with p_c ≈ 0.37, but the experiment stops at N = 8 with a coincidence rate of 0.031 Hz, so higher photon numbers require brighter sources or better loss management.
• Using the high-threshold but dense witnesses. The family optimum reaches p_c ≈ 0.46 for N = 6 and 0.38 for N = 8 but weights all eight correlators; finding sparse witnesses that approach these optima would improve robustness without inflating statistical uncertainty.
• Extending beyond two settings or beyond four-body order. The paper notes that a third setting raises the six-qubit two-body threshold only to p_c ≈ 0.093, leaving open whether other correlator orders or asymmetric constructions offer better trade-offs.

Target Audience

Experimental photonics groups working on multiphoton entanglement and Dicke-state generation; quantum-information theorists interested in Bell inequalities, permutation-invariant polytopes and entanglement witnesses; and quantum-hardware teams seeking low-cost certification protocols for symmetric multipartite states. Readers without a background in Bell scenarios and correlation functions will need the supplemental material for the polytope construction, but the main argument is followable from the experimental results alone.

Authors’ abstract

Certifying Bell correlations in increasingly larger multipartite quantum systems is challenging because conventional Bell inequalities rely on many-body correlations, which become experimentally demanding to estimate as system size grows. Here we demonstrate that multipartite Bell correlations can be certified using few-body permutation-invariant (PI) correlations with a fixed maximal correlation order. We experimentally test few-body PI Bell correlation witnesses tailored for six- and eight-photon postselected Dicke states. The addition of a few four-body correlators is found to significantly enhance the noise tolerance compared with previously studied two-body witnesses: for instance, for six qubits, the selected witness tolerates a white-noise fraction of $p_c\approx0.301$, compared with $p_c\approx0.048$ for the best two-body PI inequality with the same two settings per party. For six and eight photons respectively, we observe violations of the witness bounds by $32$ and $8.9$ standard deviations, arising from postselected polarisation states with fidelities of $0.933(4)$ and $0.916(11)$ to the corresponding half-excitation Dicke states. The witnesses for the six- and eight-photon cases use the same symmetric two- and four-body correlators, just with different coefficients. These results demonstrate an experimentally accessible approach for certifying multipartite Bell correlations in larger systems without requiring full-body correlation measurements.