Hardy-Type Nonlocality

• Hardy-type nonlocality is a framework that uses specific measurement constraints to demonstrate a nonzero quantum event alongside zero-probability outcomes, defying local hidden-variable models.
• It generalizes Hardy’s original paradox to multipartite, high-dimensional, and temporal settings, providing a logical alternative to statistical Bell inequality violations.
• These arguments facilitate device-independent self-testing and randomness certification, offering robust tools for experimental quantum information processing and foundational studies.

Hardy-type nonlocality refers to a family of logical, inequality-free arguments demonstrating that certain quantum correlations cannot be simulated by any local (and sometimes even partially nonlocal) hidden-variable models. These arguments generalize Hardy's original 1992–3 paradox for two qubits to multipartite, high-dimensional, and even temporal quantum scenarios. Unlike Bell-inequality violations, Hardy-type contradictions are established directly via constraints on possible and impossible measurement outcomes, providing a strictly logical manifestation of nonlocality that is significant both conceptually and operationally.

1. Definition and Structure of Hardy-type Arguments

A Hardy-type nonlocality proof specifies a set of joint outcome constraints in a given Bell scenario such that:

• There is at least one outcome ("Hardy-success" event) with strictly positive probability ($p_H>0$).
• A set of other specified outcomes must have exactly zero probability under quantum predictions.

For the canonical bipartite 2-2-2 case (two parties, two measurements each, two outcomes per measurement), Hardy’s conditions can be expressed as follows:

• $P(a=+1,b=+1|x=1,y=1)=p_H>0$,
• $P(a=-1,b=+1|x=2,y=1)=0$,
• $P(a=+1,b=-1|x=1,y=2)=0$,
• $P(a=+1,b=+1|x=2,y=2)=0$.

No local-hidden-variable (LHV) model can satisfy these constraints with $p_H>0$, but quantum correlations can, up to a maximal value $$p_H^{\mathrm{QM}}=\frac{5\sqrt{5}-11}{2} \approx 0.09$$ (Choudhary et al., 2016).

The tripartite (three-party) scenario generalizes this logic, as in:

• $p_H = P(a=+1, b=+1, c=+1|x=0, y=0, z=0) > 0$
• Four zero-probability constraints on appropriate marginal events (Adhikary et al., 28 Apr 2025).

2. Logical Nonlocality and Possibilistic Characterization

In the “possibilistic” or logical nonlocality framework, outcomes are treated as simply possible or impossible (binary support). Hardy’s paradox is logically nonlocal if no global deterministic assignment of measurement outcomes (across all possible measurements) is compatible with the observed pattern of possibility and impossibility.

In $(2,2,l)$ or $(2,k,2)$ scenarios, the occurrence of Hardy’s pattern subsumes all nonlocality proofs without inequalities—every nonlocal model in these settings must (coarse-grained) exhibit a Hardy paradox (Mansfield, 2016, Mansfield et al., 2011). This provides a necessary and sufficient criterion for logical nonlocality and supports efficient polynomial-time algorithms for deciding nonlocality in these cases.

3. Multipartite and Generalized Hardy-type Nonlocality

Hardy-type nonlocality arguments extend gracefully to more parties and higher dimensions:

• In the $N$-party scenario, the “generalized Hardy” patterns demand one strictly positive joint probability and $2N-1$ or more zeros, arranged such that classical bi-locality models cannot reproduce $q>0$ unless the correlation is genuinely $N$-partite nonlocal (Bhattacharya et al., 2015, Yu et al., 2013).
• These arguments can be constructed for arbitrary local dimension, and for generalized (no-signaling) theories, the maximal Hardy success probability is $1/3$ for $N=3,4$ with evidence suggesting $q^{\mathrm{GNST}}=1/3$ for all $N$ (Bhattacharya et al., 2015).
• In tripartite quantum theory, the maximal Hardy success probability is numerically lower (e.g., for three qubits, $p_H^{\mathrm{QM}}\approx0.0158$), reflecting the restrictive nature of quantum correlations compared to no-signaling polytope bounds (Bhattacharya et al., 2015).

A major insight is that, in multipartite experiments, Hardy-type nonlocality certifies genuine multiway nonlocality: any observed $q>0$ is inconsistent with all hybrid bi-local models that allow nonlocality in two parties at a time (Bhattacharya et al., 2015, Yu et al., 2013, Adhikary et al., 28 Apr 2025).

4. Device-independent Self-Testing and Randomness Certification

Hardy-type arguments enable device-independent self-testing—a process by which measurement statistics alone suffice to certify both the quantum state and measurement observables (up to local isometries). In the tripartite scenario:

• The four zero constraints from the Hardy test confine the shared pure state to a specific subspace, and every nonzero $p_H$ self-tests a unique three-parameter family of genuine entangled states (Adhikary et al., 28 Apr 2025).
• Unlike Bell inequalities, which typically self-test at unique extremal violation points, Hardy relations allow self-testing for a continuum of nonmaximal points, expanding experimental flexibility and accommodating nonideal noise (Adhikary et al., 28 Apr 2025, Sasmal et al., 2020).

This structure further enables device-independent certification of quantum randomness:

• For each nonzero $p_H$, one can compute the adversary's optimal guessing probability via semidefinite programming (NPA hierarchy), directly bounding the extractable min-entropy.
• In the optimal tripartite case, seven outcomes become equiprobable, certifying a maximal global randomness of $\log_2 7\approx 2.8073$ bits, approaching the theoretical bound for three parties (Adhikary et al., 28 Apr 2025, Sasmal et al., 2020).
• Notably, for a given nonzero Hardy parameter, there is a range (not a single value) of certifiable randomness due to the existence of many quantum extremal distributions realizing the same $p_H$ (Sasmal et al., 2020).

5. Relation to Bell Inequalities and Physical Principles

Hardy-type nonlocality arguments are qualitative (logical) proofs without inequalities but are intimately related to—and sometimes strictly stronger than—Bell inequalities:

• For two qubits, any nonzero Hardy probability implies CHSH inequality violation, with a direct quantitative relation $S=2+4p_H$ (CHSH $S$ parameter) (Xiang, 2010).
• The upper bounds on $p_H$ in quantum theory (Tsirelson-type limits), no-signaling theory, and under physical principles such as information causality or local orthogonality (LO) hierarchy, have been studied in detail:
  • Quantum: $p_H^{\mathrm{QM}}\approx0.09$
  • Local Orthogonality: $p_H^{\mathrm{LO}}\approx0.177$
  • Information causality/Macroscopic locality: $p_H\approx0.206$
  • No-signaling: $p_H^{\mathrm{NS}}=1/2$ (Das et al., 2013)
• LO is especially significant for tightly bounding the quantum region and revealing the structure of nonlocality beyond standard bipartite information principles (Das et al., 2013).

For multipartite scenarios, Hardy-type nonlocality can witness forms of nonlocality that are not accessible by standard Bell/Mermin inequalities. For instance, Hardy-type arguments detect genuine tripartite nonlocality that evades detection by the Mermin inequality, which only rules out fully local (but not bilocal hybrid) models (Adhikary et al., 28 Apr 2025).

6. High-Dimensional, Temporal, and Atypical Hardy-type Constructions

Hardy-type arguments generalize to high-dimensional local Hilbert spaces and even to temporal (Leggett–Garg–type) settings:

• For two qudits, generalized Hardy paradoxes with ordering constraints (e.g., “$A_2<B_1$ never, but $A_2<B_2$ sometimes”) yield maximal success probabilities $p_H^{(d)}$ that grow rapidly with $d$, approaching $0.42$ as $d\to\infty$ (Chen et al., 2013).
• In temporal analogues, the maximal quantum Hardy probability reaches $1/4$ for any Hilbert space dimension—a result of projective measurement structure (Choudhary et al., 2013).
• The logical (possibilistic) structure of the original Hardy paradox remains maximal for binary-outcome, two-setting scenarios; for more general settings, “stronger” logical proofs, which do not reduce to Hardy-type arguments, appear only beyond the (2,2,l) and (2,k,2) horizon (Mansfield et al., 2011, Mansfield, 2016).
• For nonmaximally entangled pure multipartite states (except for tensor products of single-qubit and bipartite maximally entangled states), logical contextuality is generic and can be algorithmically certified via Hardy-type arguments, often with just $n+2$ local observables for $n$ qubits (Abramsky et al., 2015).

7. Foundational and Experimental Significance

Hardy-type nonlocality fundamentally enriches both the conceptual and operational understanding of quantum theory:

• It exposes the logical structure of nonlocality, distinct from statistical (average) violations of inequalities and accessible even in single experimental runs (Choudhary et al., 2016).
• The method provides a versatile framework for quantum information processing—self-testing, dimension witnessing, certifiable device-independent randomness, and cryptographic primitives—often with greater robustness to nonmaximal violations and experimental imperfections than traditional Bell-test protocols (Adhikary et al., 28 Apr 2025, Sasmal et al., 2020).
• Hardy-type arguments illuminate the boundary between quantum, no-signaling, and “super-quantum” post-quantum theories. Not all parity arguments can be realized quantumly; “super-quantum” Hardy-type paradoxes are realized by extremal non-signaling boxes (e.g., PR-boxes) but are unattainable in quantum mechanics, thus providing sharp physical demarcations (Fritz, 2010).

A distinguishing feature is the detection of genuine multipartite nonlocality under strictly weaker assumptions (settings and outcome independence) than previous approaches, and the accessibility of non-exposed extremal quantum points far from those defined by linear Bell inequalities (Adhikary et al., 28 Apr 2025, Yu et al., 2013). These features make Hardy-type nonlocality indispensable both as a diagnostic tool of foundational nonclassicality and as a resource for advanced device-independent quantum technologies.