Bell Nonlocality in Quantum Systems

• Bell nonlocality is defined as quantum correlations in multipartite systems that cannot be described by any local hidden variable theory.
• It is demonstrated through the violation of Bell inequalities, where quantum states achieve metrics (e.g., CHSH value of 2√2) that classical models cannot reach.
• This phenomenon underpins device-independent quantum applications such as cryptography and randomness expansion, highlighting challenges in scalable and networked quantum systems.

Bell nonlocality is the empirically established phenomenon that quantum correlations between measurements on spatially separated physical systems cannot be accounted for by any theory which combines locality (no superluminal influences), realism (measurement outcomes determined by pre-existing variables), and the assumption of freedom of measurement choice. The violation of Bell inequalities—linear or nonlinear constraints satisfied by all such local realistic theories—provides a rigorous operational signature distinguishing quantum mechanics from classical models. Bell nonlocality is fundamental to the theoretical structure of quantum mechanics, quantum information science, and experimental implementations of device-independent quantum technologies.

1. Foundational Definition and Mathematical Structure

Bell nonlocality is operationally defined through observed statistical correlations in multipartite systems. For two parties, Alice and Bob, performing choices of measurements $x, y$ and registering outcomes $a, b$, a Bell-local (local hidden variable, LHV) model dictates

$$p(a, b | x, y) = \int d\lambda\, \rho(\lambda)\, p(a|x, \lambda)\, p(b|y, \lambda)$$

where $\lambda$ is a shared hidden variable, $\rho(\lambda)$ its (x,y)-independent distribution (free choice), and $p(a|x, \lambda)$ and $p(b|y, \lambda)$ are local response functions respecting locality (Brunner et al., 2013). Quantum mechanics, by contrast, predicts joint distributions

$$p_Q(a, b | x, y) = \operatorname{Tr}[\rho_{AB} (M_{a|x} \otimes M_{b|y})]$$

that can violate the aforementioned decomposition for suitable choices of $\rho_{AB}$ and local measurements $M_{a|x}$, $a, b$0.

The crucial logical implication—established by Bell’s original theorem—is that no theory respecting the above LHV structure can match quantum violations of certain linear inequalities (Bell inequalities), e.g. CHSH: $a, b$1 where $a, b$2 are correlators. Quantum mechanics attains $a, b$3 [(Zukowski, 2015); (Brunner et al., 2013)]. The precise formalism for general scenarios extends this logic to higher dimensions, more parties, and more measurement outcomes [(Brunner et al., 2013); (Zanfardino et al., 2023)].

2. Interpretational Frameworks and Causal Structure

The canonical “nonlocality” established by Bell’s theorem refers to the violation of factorization (local causality) imposed by LHV models, not to the existence of superluminal causal influences. Bell nonlocality is thus the failure of local realism—specifically, the joint assumption of locality and realism—rather than a proof of “spooky action at a distance” (Zukowski, 2015, Boughn, 2016).

Żukowski shows that a violation of Bell inequality requires, in addition to locality and freedom, counterfactual definiteness—the assignment of preexisting values to unmeasured observables. If complementarity (Bohr) is respected, so that only actually performed measurements have outcomes, quantum mechanics remains “local” in the operational and relativistic sense (Zukowski, 2015). Similarly, Boughn argues that these violations are a direct consequence of quantum superposition and entanglement; the quantum formalism itself does not entail dynamical nonlocality or superluminal causation (Boughn, 2016).

In the Everett (many-worlds) interpretation as developed by Brown and Timpson, Bell inequality violations are produced by local unitary evolution and wavefunction branching, enforcing Lorentz covariance and eliminating the need for instantaneous nonlocal influence (Brown et al., 2014). The Heisenberg-picture “local foliation” perspective reaches the same conclusion: records of measurement outcomes are created by local interactions, and the nonlocal correlations arise only when these local branches are subsequently compared (Bédard, 2024).

Alternative formulations, such as Bohmian mechanics and Many Interacting Worlds (MIW), consciously embrace ontologically nonlocal hidden-variable dynamics to restore determinism and realism, while explicitly violating the factorization assumed in Bell’s theorem (Ghadimi et al., 2018).

3. Hierarchies of Quantum Correlations and Relation to Entanglement

Bell nonlocality is strictly stronger than quantum entanglement and Einstein-Podolsky-Rosen (EPR) steering. Entanglement is necessary but not sufficient for Bell nonlocality; some entangled states admit LHV models for all possible measurements (Chen et al., 2015, Sengupta et al., 2020). EPR steering lies between these notions: if a mapped (depolarized) state is steerable, the original state is Bell nonlocal, establishing a rigorous hierarchy: Bell nonlocality $a, b$4 EPR steering $a, b$5 entanglement (Chen et al., 2015).

Resource-theoretic treatments unify entanglement and Bell nonlocality as instances of “non-LOCC (local operations and classical communication) processes.” Instantaneous classical processes not simulatable by LOSR (local operations and shared randomness) are precisely Bell nonlocal boxes; all entangled states can activate nonlocality under suitable LOCC pre-processing and measurement (Sengupta et al., 2020). Generalizations to channels and measurements (POVMs) yield systematic monotones and witnesses for quantifying nonlocality (Sengupta et al., 2020).

In qudit systems, Bell nonlocality is tightly linked to the presence of negativity in discrete Wigner functions; stabilizer states (with nonnegative Wigner representations) do not violate generalized CHSH inequalities for any dimension (Meyer et al., 2024).

4. Geometric, Statistical, and Computational Tools

The set of Bell-local behaviors forms a convex polytope whose extremal points correspond to deterministic local assignments. Violations are detected via facet inequalities (such as CHSH) or more general forms (e.g., CGLMP for $a, b$6-outcome measurements) [(Brunner et al., 2013); (Zanfardino et al., 2023)]. Geometric approaches define nonlocality in terms of the minimal trace-norm or relative-entropy distance of a state to the polytope of local states; in the special case of Werner and Bell-diagonal states, the closest local states remain within the same family (Zanfardino et al., 2023).

Testing nonlocality rapidly becomes intractable as the number of parties and settings grows. Tensor network representations and sparse-recovery algorithms from compressed sensing provide scalable methods for nonlocality detection and quantification; these show that any nonsignaling quantum correlation admits a quasi-probability (not necessarily positive) representation over deterministic assignments (Eliëns et al., 2020).

Device-independent quantum information protocols rely on the robustness of Bell nonlocality against imperfections. Recent work develops mapping and optimization techniques (“bouncing” between Bell inequalities and Hamiltonians) to maximize certification of nonlocality in large-scale quantum simulators, explicitly for many-body spin models under realistic noise (Li et al., 2024).

5. Extension to Many-Body, Network, and Macroscopic Scenarios

The landscape of Bell nonlocality extends far beyond the bipartite or few-body scenario. In many-body systems, the design and detection capacity of Bell inequalities depend crucially on the topology of the system’s interaction graph. Detection capacity can be optimized via tropical algebra and tensor networks, with toroidal or nontrivial topologies favoring stronger Bell violations by maximizing frustration and nonclassical ground-state correlations (Emonts et al., 2024).

Multipartite states exhibit rich patterns: although strict monogamy holds for the sharing of bipartite Bell violations, multipartite settings enable polygamy and even “hyper-polygamy”—a single $a, b$7-qubit state can simultaneously violate all Bell inequalities on $a, b$8-party subsystems, for a macroscopic number of $a, b$9 (Munné et al., 9 Dec 2025).

In network scenarios with independent sources (networks of Bell tests), Bell nonlocality produces fundamentally new phenomena—bilocality, star and ring nonlocality, activation of locally hidden nonlocality, non-convexity of local sets, and complex nonlinear (rather than facet) constraints. Entropic, covariance, inflation, and semidefinite programming methods are now central to formalize and certify nonlocality in such architectures (Tavakoli et al., 2021).

Bell nonlocality is also a live subject in the macroscopic regime, both as a foundational tool for distinguishing quantum from classical behavior and as an experimental challenge: no loophole-free violation of generalized Bell inequalities has yet been established for large-spin or multi-mode bosonic systems (Dalton, 2018). Proposals employing collective spin observables and large-$$p(a, b | x, y) = \int d\lambda\, \rho(\lambda)\, p(a|x, \lambda)\, p(b|y, \lambda)$$0 generalizations of CHSH give explicit bounds, but finding suitable states to violate these inequalities at macroscopic scales remains open.

6. Experimental Realizations and Activation Protocols

Bell nonlocality underlies device-independent quantum cryptography, randomness expansion, and self-testing. Loophole-free Bell tests have been realized in small qubit/ion and photonic systems, but scalability and robustness—particularly in many-body and networked scenarios—remain major challenges (Brunner et al., 2013).

Recent experiments demonstrate activation of Bell nonlocality: states that are Bell-local in ordinary scenarios can have their nonlocality “unlocked” by embedding them in network configurations, e.g., via broadcast channels or multipartite resources, showing that nonlocality is not a static attribute but a resource that can be revealed by network context (Villegas-Aguilar et al., 2023). Intensity-only and aggregate-detection protocols allow Bell nonlocality to be revealed even without particle resolution, so long as visibility thresholds are met (Patrick et al., 2020).

7. Conceptual and Statistical Interconnections

The bounds of Bell inequalities are deeply linked to local uncertainty relations. Application of the Aharonov-Vaidman decomposition shows that the amount by which a Bell inequality can be violated is quantitatively bounded by the uncertainty (sum of variances) in the local observables. Conversely, observing a Bell violation instantly certifies a nontrivial level of local uncertainty (Hsu, 2023). This statistical linkage provides a unified perspective connecting nonlocality and operational uncertainty.

Table: Hierarchies and Measures in Nonlocality

Class/Resource	Defining Property	Detection
Entanglement	Non-separability	Entanglement witnesses
EPR Steering	Failure of LHS model	Steering inequalities
Bell Nonlocality	No LHV model for joint stats	Bell inequality violation
Nonlocal Box (PR-box)	Nonsignaling, supra-quantum	Achieves algebraic bounds

All genuine Bell-nonlocal states are necessarily steerable and entangled, but not vice versa. Moreover, nonlocality is generally measurement- and scenario-dependent—certain inequalities, measurement choices, and system sizes are required to reveal it.

---

Bell nonlocality is thus both a precise technical feature of quantum theory—absence of a local realism decomposition for observed correlations—and a resource enabling device-independent verification of quantum behaviors. Its interpretation, operational implications, and utility as a resource are central across quantum foundations and quantum technologies, with ongoing theoretical, computational, and experimental developments continually reshaping the field [(Brunner et al., 2013); (Zukowski, 2015); (Chen et al., 2015); (Sengupta et al., 2020); (Munné et al., 9 Dec 2025); (Tavakoli et al., 2021); (Bédard, 2024)].