Cluster-State Witnesses of Finite-Speed Hidden Influences

Technical Summary: Cluster-State Witnesses of Finite-Speed Hidden Influences

Problem Statement
While Bell experiments have successfully ruled out local common-cause explanations for quantum correlations, they do not strictly exclude hidden influences that propagate at a finite speed $v$ where $c < v < \infty$ within a preferred frame. Previous work established that in multipartite scenarios, such finite-speed models ($v < \infty$) inevitably lead to superluminal signaling if they are to explain quantum violations, or they must be instantaneous ($v = \infty$). However, experimental verification of this constraint has been hindered by the complexity of existing witnesses. Prior constructions, such as the four-qubit witness by Bancal et al. and the three-qubit example by Barnea et al., suffered from small violation gaps (e.g., a 2.9% gap relative to the bound) and required complex states or high white-noise thresholds (e.g., 0.9720), making them impractical for near-term devices.

Methodology
The authors formulate the constraint imposed by finite-speed hidden influences as a projected-polytope separation problem. They define the Hidden-Influence Causality (HIC) set, $HIC_E = NS \cap CL(BC|E)$, where $NS$ is the set of no-signaling distributions and $CL(BC|E)$ represents distributions where the correlation between two "blind" late parties ($B$ and $C$) is Bell-local conditional on the outcomes of an "early" side ($E$).

To test this, the authors construct "no-BC" witnesses, which are linear functionals assembled solely from marginal data involving either $B$ or $C$ combined with the early parties, explicitly omitting any terms containing both $B$ and $C$. The methodology involves:

1. State Selection: Utilizing linear cluster states ($|LC_4\rangle$ and $|LC_5\rangle$), which are experimentally friendly and well-understood via stabilizer formalism.
2. Witness Construction: Combining a distributed CHSH block on the $B$-side (using rotated measurements) with stabilizer-based "lock" blocks on the $C$-side.
3. Optimization and Certification: The authors optimize the relative weights of these blocks to maximize the violation relative to the HIC bound. They use integer Farkas certificates to prove the exact support bounds of the projected polytope and perform exact rational-rank and relative-interior checks to certify that the derived inequalities are facets of the projected HIC polytope.

Key Contributions and Results
The paper presents two new, compact witnesses that significantly improve upon previous finite-speed hidden-influence tests:

1. The Four-Qubit Witness ($LC_4$):

   • Configuration: Uses a four-qubit linear cluster state with early side $E=\{A, D\}$ and blind pair $\{B, C\}$.
   • Structure: Composed of six correlators (two- and three-body terms) with no four-body terms.
   • Bound and Violation: The HIC bound is $S_4 \le 6$. The quantum value for the cluster state is $S_4 = 4 + 2\sqrt{2} \approx 6.8284$.
   • Performance: This yields an ideal white-noise visibility threshold of $\nu^*_4 \approx 0.8787$, a substantial improvement over the previous best of 0.9526.
   • Geometry: The inequality is certified as a facet of the projected no-BC HIC polytope (dimension 43 within a 44-dimensional space).

2. The Five-Qubit Witness ($LC_5$):

   • Configuration: Extends the system to five qubits with early side $E=\{A, D, E\}$ and blind pair $\{B, C\}$.
   • Structure: Composed of 14 terms, including four-body correlators. It employs a "dichotomic completion" technique to handle cases where early outcomes might not align, ensuring the witness remains valid across the full HIC set.
   • Bound and Violation: The HIC bound is $S_5 \le 10$. The quantum value is $S_5 = 6 + 4\sqrt{2} \approx 11.6569$.
   • Performance: This achieves an even lower ideal white-noise threshold of $\nu^*_5 \approx 0.8579$.
   • Geometry: Certified as a facet of the projected polytope (dimension 133 within a 134-dimensional space).

Significance and Claims
The paper claims that linear cluster states serve as "certifiable and experimentally friendly resources" for testing finite-speed hidden influences. By achieving significantly larger violation margins and lower noise thresholds than previous methods, these witnesses make feasible the experimental demonstration that finite-speed causal models are incompatible with quantum mechanics (unless they allow for superluminal signaling).

The authors emphasize that while the ideal white-noise thresholds are theoretical benchmarks, the reduction in complexity (using familiar cluster states and fewer correlators) and the improved robustness against noise make these targets accessible for current quantum platforms. The work provides a concrete path to falsifying all models of direct causation with finite speed $v < \infty$, reinforcing the conclusion that any falsifiable direct causation in spacetime must be excluded, just as local common causation was excluded by Bell's theorem. The paper modestly notes that while it identifies two specific facets, a complete characterization of the projected polytope's facet structure remains an open question for future work.