n-Locality Inequality in Quantum Networks

Updated 26 January 2026

n-Locality Inequality is a generalization of bilocality that requires independent hidden variables across n sources to rule out classical models in quantum networks.
It employs nonlinear combinations of correlators—using n-th roots—to quantify quantum violations and differentiate them from standard Bell constraints.
Experimental implementations using photonic and continuous-variable systems demonstrate its robust performance against noise and detection loopholes in device-independent protocols.

Bilocality inequality is a nonlinear network Bell-type constraint devised to detect nonclassical correlations in quantum networks with multiple independent sources. It strengthens standard locality (Bell) constraints by enforcing independence of hidden variables assigned to distinct sources, with profound implications for entanglement-swapping schemes and device-independent quantum network protocols. Violations of bilocality inequalities rule out all bilocal hidden-variable explanations and establish network nonlocality as a quantum resource. Below we systematically detail the definition, mathematical structure, quantum violations, experimental methodology, robustness, and generalizations of the bilocality inequality, referencing key results from quantum theory and experiment.

1. Bilocal Hidden-Variable Model and Mathematical Formulation

The canonical bilocality scenario involves three parties—Alice (A), Bob (B), and Charlie (C)—connected linearly by two independent quantum sources. Source S₁ emits to A and B; source S₂ emits to B and C. The bilocal hidden-variable (HV) model postulates mutually independent hidden variables λ₁ and λ₂ for S₁ and S₂, such that the joint probability of outcomes $a,b,c$ under settings $x,y,z$ factorizes: $P(a,b,c | x,y,z) = \int dλ_1\, dλ_2\, ρ(λ_1)\, ρ(λ_2)\, P(a|x,λ_1)\, P(b|y,λ_1,λ_2)\, P(c|z,λ_2).$ The independence assumption $ρ(λ_1,λ_2) = ρ(λ_1)\, ρ(λ_2)$ is critical; relaxing it reduces the constraint to standard Bell locality (Sun et al., 2018).

From this structure, the bilocality inequality is derived. One defines two tripartite correlators: $I = \frac{1}{4} \sum_{x,z} \langle A_x B^0 C_z \rangle,\quad J = \frac{1}{4} \sum_{x,z} (-1)^{x+z} \langle A_x B^1 C_z \rangle,$ where $B^0$ and $B^1$ correspond to suitable groupings of Bob's measurement outcomes depending on the protocol. The bilocal parameter is

$B := \sqrt{|I|} + \sqrt{|J|}.$

All bilocal HV models satisfy the nonlinear boundary $B \leq 1$ (Saunders et al., 2016, Sun et al., 2018, Mukherjee et al., 2014, Branciard et al., 2011).

2. Quantum Violation and Measurement Strategies

Quantum networks can validate nonbilocality by exceeding the bilocal bound. Optimal measurement involves: (i) preparation of two independent maximally entangled states at S₁ and S₂ (typically singlets $|\psi^-\rangle$ ), (ii) Alice and Charlie each perform CHSH-optimal projections (e.g., $A_0 = C_0 = (\sigma_z + \sigma_x)/\sqrt{2}, A_1 = C_1 = (\sigma_z - \sigma_x)/\sqrt{2}$ ), and (iii) Bob executes a Bell-state measurement.

Calculations for perfect Bell pairs yield $I = J = 1/2$ , such that $B_\mathrm{Q}^\mathrm{max} = \sqrt{2} > 1$ , violating the bilocal constraint. For arbitrary two-qubit sources $ρ_{AB}$ and $ρ_{BC}$ (correlation matrices $T^A$ , $T^C$ ), the maximal quantum value is

$B_\mathrm{Q}^\mathrm{max} = \sqrt{ \sqrt{t_1^A t_1^C} + \sqrt{t_2^A t_2^C} },$

with $t_k^A, t_k^C$ the leading eigenvalues of $T^A T^A{}^T$ and $T^C T^C{}^T$ (Andreoli et al., 2017, Kundu et al., 2020, Gisin et al., 2017).

For partial Bell-state measurements grouping outcomes, one finds weaker violations ( $B \approx 1.225$ for the "13"-outcome case) but improved experimental accessibility (Saunders et al., 2016).

3. Experimental Realization and Loophole Closure

State-of-the-art experiments implement bilocality tests in photonic networks employing independent entangled photon-pair sources with stringent source-independence enforcement. Independence is assured by active phase randomization—each pump diode is reset below-threshold for each trial ( $\sim$ ns timescales), ensuring no shared phase reference for spontaneous parametric down-conversion (SPDC) (Sun et al., 2018). Locality constraints are satisfied by arranging spacelike separation between pump emissions and quantum random number generator (QRNG) events across all nodes.

Experimentally, bilocality parameters— $B_{13} = 1.181 \pm 0.004$ —have been recorded, violating the bound by $45 \sigma$ , while simultaneous CHSH values ( $S = 2.652 \pm 0.059$ ) exclude standard locality by $11 \sigma$ (Sun et al., 2018). Crucially, noise scans reveal that bilocality violations persist in regimes where CHSH violations are extinguished, confirming bilocality inequality's superior noise tolerance (Saunders et al., 2016).

4. Resource Theory: Negative Quasi-Probabilities and Operational Approach

A joint operational approach reveals that the violation of bilocality fundamentally requires negative entries in joint quasi-probability distributions (JQD) or quasi-stochastic maps applied at processing nodes. Allowing any negative component enables violations up to the algebraic maximum $B=2$ (“bi-PR-box” regime) (Onggadinata et al., 2023). In contrast, ordinary bilocal models—factorized JPD and positive-processing—cannot achieve $B>1$ .

Explicit constructions using CHSH-type JQDs in each source and parameterized coupler maps demonstrate exactly when nonbilocality emerges: $B = 2 \sqrt{ \eta\, \mu_1\, \mu_2 },$ for source parameters $\mu_{1,2}$ and processing parameter $\eta$ ; violation obtains when $\eta\, \mu_1\, \mu_2 > 1/4$ (Onggadinata et al., 2023).

5. Robustness Against Noise and Detection Loopholes

Bilocality inequalities exhibit enhanced robustness to both white noise and detector inefficiency compared to standard Bell inequalities. For networks of two Werner states of visibility $v$ , violation persists down to $v > 1/\sqrt{2} \simeq 0.707$ , whereas CHSH swapped-pair tests require $v > 2^{-1/4} \simeq 0.84$ (Saunders et al., 2016, Mukherjee et al., 2014, Branciard et al., 2011, Gisin et al., 2017). Detection-efficiency thresholds for loophole-free bilocality tests can drop as low as $\eta_\mathrm{biloc} = 2/3$ under symmetric measurement assumptions, outperforming the standard CHSH threshold $\eta_\mathrm{CHSH} \simeq 0.828$ (Saunders et al., 2016, Branciard et al., 2011).

Noise scaling laws for bilocal parameters are explicit: $B_{14}(v)=\sqrt{2 v},\quad B_{13}(v)=\sqrt{1.5 v},\quad B_{\mathrm{CHSH}}(v)=2 v$ (Saunders et al., 2016, Sun et al., 2018).

6. Generalizations and Network Nonlocality

Bilocality extends naturally to $n$ -party, $n$ -source quantum networks. The $n$ -locality inequality generalizes the square root to the $n$ th root: $|I|^{1/n} + |J|^{1/n} \leq 1,$ where $I, J$ are suitable concatenations of $n$ -fold correlators (Mukherjee et al., 2014, Tavakoli et al., 2014, Andreoli et al., 2017, Kundu et al., 2020). In star networks, maximal quantum violations scale as

$N_{\mathrm{star}}^\mathrm{max} = \sqrt{ \left( \prod_{i=1}^{n} t_1^{A_i} \right)^{1/n} + \left( \prod_{i=1}^{n} t_2^{A_i} \right)^{1/n} }$

subject to the sources' correlation matrices.

Recent research has treated asymmetric bilocality scenarios with unequal measurement settings per party and analyzed trilocal extensions, employing sum-of-squares (SOS) techniques for dimension-independent Tsirelson bounds and deriving explicit white-noise thresholds ( $v_c$ ), e.g., $v_c = \sqrt{2/3}$ for $n=3$ (Sasmal et al., 2023).

7. Extensions: Continuous-Variable Networks and Entropic Approaches

Continuous-variable (CV) quantum networks admit bilocality violations using pseudospin measurements on TMSV or non-Gaussian states. All nonzero squeezing leads to violation, saturating the algebraic bound in the infinite limit. Non-Gaussian resources, such as entangled coherent states and photon-subtracted states, produce maximal nonbilocality even at zero squeezing (Chakrabarty et al., 9 Jun 2025).

An alternative entropic framework posits linear bilocality witnesses in Shannon entropies: $H(A_0, C) \leq H(A_0, B) + H(C|A_1, B)$ and related relations. Entropic witnesses can be more noise-tolerant or sensitive to outcome alphabet size than standard correlator inequalities, and have enabled proofs of activation and monogamy within bilocal networks (Chaves et al., 2016).

Select Table: Bilocality Inequality—Quantum Violation Thresholds

Protocol/Source	Bilocal Violation Threshold	CHSH Violation Threshold
Werner state (DV)	$v > 1/\sqrt{2}$	$v > 2^{-1/4}$
Detection Efficiency	$\eta > 2/3$	$\eta > 0.828$
Partial BSM ("13")	$v > 2/3$	$v > 2^{-1/4}$

The bilocality inequality is a stringent diagnostic tool for network nonlocality, ruling out models with independent hidden variables. Its violation certifies the presence of quantum resources impossible to emulate in any bilocal classical theory. Beyond fundamental insights, bilocality tests underpin scalable, loophole-closed, device-independent quantum networks and inform advances in quantum communication, random number generation, and distributed quantum computation (Sun et al., 2018, Saunders et al., 2016, Sasmal et al., 2023, Chakrabarty et al., 9 Jun 2025).