Source-linked AI summary
Contextuality, Cohomology and Paradox
Samson Abramsky, Rui Soares Barbosa, Kohei Kishida, Raymond Lal, Shane Mansfield
TL;DR
Contextuality has been studied in quantum mechanics, but general mathematical treatments also reveal it in fields such as databases and constraints. This paper relates contextuality to logical paradoxes and shows that cohomological obstructions witness contextuality for a large class of models, including examples from stabiliser quantum mechanics.
Problem
Contextuality has often been treated through concrete quantum examples, while general mathematical work has shown that it also occurs in settings such as databases and constraints.
Source: §1 · p. 1Method
The paper connects contextuality with paradoxical reference cycles and uses cohomological obstructions to study All-vs-Nothing arguments and local-to-global extensions.
Source: §1 · p. 2, §5.3 · p. 14Results
A large class of models has logical or strong contextuality witnessed by cohomology, including many constructions from stabiliser quantum mechanics.
Source: Discussion · p. 17Takeaways & Limitations
Contradictory reference cycles and contextuality share a structure of local consistency with global inconsistency.
Source: §1 · p. 2Cohomological obstructions provide a sufficient but not generally necessary condition for contextuality, so false positives can occur.
Source: §5.3 · p. 15
Abstract
from arXiv · showhide
Contextuality is a key feature of quantum mechanics that provides an important non-classical resource for quantum information and computation. Abramsky and Brandenburger used sheaf theory to give a general treatment of contextuality in quantum theory [New Journal of Physics 13 (2011) 113036]. However, contextual phenomena are found in other fields as well, for example database theory. In this paper, we shall develop this unified view of contextuality. We provide two main contributions: firstly, we expose a remarkable connection between contexuality and logical paradoxes; secondly, we show that an important class of contextuality arguments has a topological origin. More specifically, we show that "All-vs-Nothing" proofs of contextuality are witnessed by cohomological obstructions.
1 Introduction
The paper frames contextuality as a general local-to-global phenomenon found beyond quantum mechanics, and develops its connection to logical paradoxes. It argues that contradictory reference cycles have the same local-consistency/global-inconsistency structure as contextuality.
1 Introduction
- General mathematical treatments have helped reveal contextuality as a pervasive phenomenon beyond quantum mechanics, including in databases and constraints [3, 5, 8, 1, 4].The paper notes that earlier quantum work was largely concrete and example-driven.Source: §1 · p. 1
- Contextuality can be understood as locally consistent data that cannot be reconciled into a global assignment, a structure found in quantum theory, databases, and constraint satisfaction.In quantum theory this corresponds to failure of a hidden-variable explanation; for databases, failure of the universal relation assumption; and for constraint satisfaction, absence of a solution.Source: §1 · pp. 1–2
- Contradictory cycles of references, such as Liar cycles [10, 30], have the same local-consistency/global-inconsistency structure as contextuality.The paper presents this as a direct connection between quantum contextuality and semantic paradoxes.Source: §1 · p. 2
Mathematical structure
The paper extends the cohomological treatment of contextuality to settings with algebraically structured outcomes. It formulates All-vs-Nothing arguments through linear equations and connects their logical inconsistency to cohomological obstructions.
Mathematical structure
- Sheaf theory [17] provides the mathematical setting for moving from local to global, and sheaf cohomology is used to characterise contextuality.The cohomological approach was initiated in [5] and is extended here to cases where observation outcomes have algebraic structure, such as Pauli spin observables with eigenvalues in Z2.Source: Mathematical structure · p. 2
- The paper gives a more general formulation of All-vs-Nothing arguments [20] as locally consistent but globally inconsistent systems of linear equations.It applies this formulation to an extensive class of arguments in the stabiliser fragment of quantum mechanics, relevant to quantum error correction [23] and measurement-based quantum computation [28].Source: Mathematical structure · p. 2
- All such All-vs-Nothing arguments are witnessed by the cohomological obstruction to extending local sections to global ones, previously studied in [5].The paper characterises the obstruction using the connecting homomorphism of the long exact sequence.Source: Mathematical structure · p. 2
- The main theorem establishes a hierarchy of probability-model properties relating algebraic, logical, and topological structures.The full version of the paper is cited as [2].Source: Mathematical structure · p. 2
2 The many faces of contextuality
The section introduces a network measurement scenario and distinguishes contextuality visible in probability distributions from a stronger form visible in their supports. In the Hardy example, a few possible and impossible outcomes suffice to rule out a classical explanation, although the behaviour can occur in quantum mechanics.
2 The many faces of contextuality
- Alice and Bob repeatedly choose local bit registers to measure, and a common target compiles their joint outcomes into probability distributions.Alice accesses a1 and a2, Bob accesses b1 and b2; Figure 1 depicts the agents and a source.Source: §2 · pp. 2–3, Figure 1 · p. 3
2.1 Logical forms of contextuality
- The section focuses on contextuality at the level of distribution supports, a stronger form than contextuality exhibited at the probability-distribution level [3, 2].The support-based view connects contextuality directly with logic.Source: §2.1 · p. 3
- In the Hardy table, only four entries are specified, yet those support constraints rule out a classical explanation for behaviour realizable in quantum mechanics [13], yielding a stronger form of Bell’s theorem [6].The table entries distinguish positive-probability outcomes from impossible ones; Figure 2 pairs the Hardy paradox with the PR box.Source: §2.1 · p. 3, Figure 2 · p. 3
What do “observables” observe?
Non-contextuality models each measurement outcome as fixed independently of which other measurements are performed. The Hardy-table constraints leave no such assignment, so the model is contextual even using only possibilities rather than probabilities.
What do “observables” observe?
- Non-contextuality requires a function assigning each measurement an outcome independently of the other measurements performed.The passage presents this as the classical view that observables reflect objective properties of the system, while noting that actual observations of microphysical reality make it untenable.Source: What do “observables” observe? · p. 4
- The Hardy-table constraints force λ(a1)=0, λ(b1)=0, λ(b2)=1, and λ(a2)=0, which makes (a2,b1)=(0,0) possible despite the table forbidding it.Thus no assignment function λ is consistent with the specified entries.Source: What do “observables” observe? · p. 4
- Hardy models are contextual in a logical sense stronger than the probabilistic form because the inference uses only which outcomes are possible.The argument does not require full probability values.Source: What do “observables” observe? · p. 4
Strong contextuality
Logical contextuality occurs when a supported local assignment cannot extend to a compatible global assignment; strong contextuality is the stricter case where no global assignment is compatible with the support.
Strong contextuality
- Strong contextuality means no global assignment is consistent with the support at all.The Hardy paradox is not strongly contextual.Source: Strong contextuality · p. 4
- Logical contextuality requires a supported local assignment that cannot extend to a compatible global assignment.In the Hardy case, the assignment a1 7→0, b1 7→0 has this property.Source: Strong contextuality · p. 4
- Strong contextuality arises for entangled quantum states with three or more parties, while no bipartite quantum-realisable behaviour is strongly contextual.The PR box is strongly contextual and satisfies the no-signalling principle.Source: Strong contextuality · p. 4
Visualizing contextuality
Bundle diagrams make the fibred structure of contextuality visible by representing variables, their possible values, and possible joint outcomes. The diagrams distinguish the Hardy model's logical contextuality from the PR box's strong contextuality.
Visualizing contextuality
- Bundle diagrams visualize contextuality by showing the fibred structure of the examples and bridge to the sheaf-theoretic formalism.The base consists of variables, with fibres of values and edges linking jointly possible outcomes.Source: Visualizing contextuality · pp. 4–5
- In the Hardy bundle, a global path exists, but none includes the red edge, showing logical contextuality without strong contextuality.A global assignment is a closed path traversing every fibre exactly once.Source: Visualizing contextuality · p. 5
- The PR box bundle has no univocal closed path, reflecting its strong contextuality.The figure caption identifies the two diagrams as the Hardy table and PR box represented as bundles.Source: Visualizing contextuality · p. 5, Figure 3 · p. 5
Contextuality, logic and paradoxes
The paper connects contextuality's local-consistency/global-inconsistency pattern to logical paradoxes by representing Liar cycles as inconsistent Boolean equations. It identifies the length-4 Liar cycle with the PR box and notes that the cohomological analysis of paradoxes remains future work.
Liar cycles
- A Liar cycle of length N chains statements asserting that the next statement is true, with the final statement asserting the first is false.For N = 1, this is the classic Liar sentence.Source: Contextuality, logic and paradoxes · p. 5
- Replacing the cycle's truth assertions with Boolean equations yields an inconsistent system, although every subset of up to n −1 equations is consistent.The equations are x1 = x2, …, xn−1 = xn, and xn = ¬x1.Source: Contextuality, logic and paradoxes · p. 6
Contextuality and the PR box
- The length-4 Liar cycle corresponds exactly to the PR box, with matching correlations and anticorrelation among the variables.The stated correspondence is x1 ∼a2, x2 ∼b1, x3 ∼a1, and x4 ∼b2.Source: Contextuality, logic and paradoxes · p. 6
Contradiction and contextuality
- The Liar-cycle contradiction follows two branches: deriving contradiction from assuming S1 true and from assuming S1 false.The broader contextuality analysis treats local consistency that cannot be assembled into a globally consistent assignment as a pattern that can also apply to logical paradoxes.Source: Contextuality, logic and paradoxes · pp. 5–6
Scope
- The paper discusses one specific case and says the analysis can generalize to a large class of examples, with an extended treatment planned for future work.The planned treatment will apply sheaf-cohomology tools developed later in the paper.Source: Contextuality, logic and paradoxes · p. 6
3 Sheaf formulation of contextuality
The sheaf formulation represents measurement contexts and possible outcomes as a presheaf over variables, then defines contextuality through compatible families and global sections. For no-signalling probability tables, their supports induce empirical models to which this characterization applies.
Measurement scenarios
- The formalism models measurements with a finite variable set X, a cover of maximal jointly measurable contexts M, and a common outcome set O.Together these form a measurement scenario ⟨X, M, O⟩.Source: §3 · pp. 6–7
The sheaf of events
- The event sheaf assigns outcome valuations to variable subsets, with restriction maps taking valuations to smaller subsets.A section over all of X is a global section.Source: §3 · p. 7
Empirical models
- Possibility tables can be represented by a subpresheaf whose sections are the possible outcomes; its context supports must be nonempty and restrictions beneath the cover surjective.Surjectivity beneath the cover corresponds to the underlying empirical model satisfying no-signalling.Source: §3 · p. 7
- An empirical model is defined as a subpresheaf satisfying conditions E1, E2, and E3.The present paper uses “empirical model” for the support presheaf rather than the probability table itself.Source: §3 · p. 7
Characterizing contextuality
- Logical contextuality means some local section belongs to no compatible family, while strong contextuality means this holds for every local section, equivalently that no global section exists.For no-signalling probability tables, the distributions' supports induce empirical models, making these characterizations applicable.Source: §3 · p. 7
4 All-vs-Nothing arguments
All-vs-Nothing arguments identify local equations that are jointly inconsistent with any global assignment. The section extends this approach from quantum parity arguments to general linear theories and relates it to affine closures.
4.1 All-vs-Nothing for quantum theory
- Mermin’s and other known All-vs-Nothing arguments arise from AvN triples, whose conditions generate inconsistent parity constraints.For the GHZ example, each variable occurs twice, making the summed left sides even while the right sides sum to an odd value.Source: §4.1 · p. 8
- Every state stabilized by a subgroup generated by an AvN triple admits an All-vs-Nothing argument; independent check vectors yield a stabilized subspace of dimension 2n−3.The triple forces odd parity for one measurement, while any global assignment implies even parity.Source: §4.1 · pp. 8–9
4.2 Generalized All-vs-Nothing arguments
- Generalized All-vs-Nothing arguments replace parity equations with R-linear theories; an inconsistent such theory implies strong contextuality.The mod-3 example motivates moving beyond Z2, and the ring structure yields a more general formulation.Source: §4.2 · pp. 9–10
4.3 Affine closures
- When R is a field, affine subspaces are exactly the solution sets of linear theories; this equality need not hold for general rings.The paper defines the affine closure context by context to obtain an empirical model.Source: §4.3 · pp. 10–11
- For any empirical model, AvN_R implies strong contextuality of its affine closure, and the converse holds when R is a field.The affine closure can have a global assignment even when the original model is strongly contextual.Source: §4.3 · p. 11
5 Cohomology witnesses contextuality
This section develops cohomological obstructions as witnesses of contextuality, using Čech and relative cohomology to test whether local sections extend to compatible global families. The obstruction is sufficient, but not generally necessary, for contextuality.
5.1 Čech cohomology
- The Čech cohomology group is the quotient of cocycles by coboundaries, built from cochains on intersecting contexts and restriction maps.A 0-cocycle is a family of local sections agreeing on overlaps.Source: §5.1 · p. 12
5.2 Relative cohomology and obstructions
- The analysis focuses on connected covers, since scenarios can be reduced to connected components.For connected covers, compatibility across intersecting contexts makes the empty-intersection exception irrelevant.Source: §5.2 · p. 12
- Relative cohomology defines the obstruction for a local section as the connecting homomorphism’s image in the first cohomology of the relative presheaf.The construction assumes the presheaf is flasque beneath the cover.Source: §5.2 · pp. 12–13
- For a connected cover, a local section’s obstruction vanishes exactly when it extends to a compatible family agreeing with that section.The obstruction lies in the first cohomology group, which captures these extension failures in sheaf cohomology.Source: §5.2 · pp. 13–14
5.3 Witnessing contextuality
- The construction turns a set-valued empirical model into an abelian presheaf by taking formal linear combinations over a ring.The free module embeds each outcome as a formal linear combination, allowing cohomology to be applied to the model.Source: §5.3 · p. 14
- A nonzero cohomological obstruction implies logical contextuality, and if it is nonzero for every local section, strong contextuality.The converse does not hold in general: formal linear combinations can yield compatible families that are not genuine global sections.Source: §5.3 · pp. 14–15
- Under a ring homomorphism R′→R, contextuality detected over R is also detected over R′.The proposition states that a vanishing obstruction over R′ implies vanishing over R.Source: §5.3 · p. 15
- When an obstruction vanishes, compatibility forces the resulting formal linear combinations to be affine, with coefficients summing to one.The shared coefficient sum follows from agreement on restrictions to the empty context.Source: §5.3 · p. 15
6 Cohomology and AvN arguments
The section shows that all-vs-Nothing models have cohomological obstructions witnessing strong contextuality, with an analogous result for logical contextuality. It also states that the known quantum examples of strong contextuality are captured because they are of AvN type.
6 Cohomology and AvN arguments
- For AvNR models, cohomological obstructions witness strong contextuality, and it suffices to use cohomology with coefficients in the ring R itself.The proof argues contrapositively, relating a vanishing obstruction to a compatible family and then to failure of strong contextuality for the affine model.Source: §6 · pp. 15–16
- An analogous result covers logical contextuality by asking whether a global assignment can extend a specified local assignment.This adapts the notion of an inconsistent theory to a partial assignment.Source: §6 · p. 16
- All known quantum examples of strong contextuality are of AvN type and are therefore captured by cohomology.The results also establish that a cohomological obstruction witnesses genuine contextuality.Source: §6 · p. 17
Discussion
The discussion presents cohomology as a witness of contextuality for a large class of models, including constructions from stabiliser quantum mechanics. It identifies a precise characterization of what cohomology detects as future work.
Discussion
- Cohomology witnesses logical or strong contextuality for a large class of models, including constructions from stabiliser quantum mechanics.The results generalise earlier work, while a precise characterization of what cohomology detects remains open.Source: Discussion · p. 17
- The first implication in Theorem 21 can be reversed when R is a field; with quotient-module equational consistency, reversal extends to general rings.The stated role of a field is to allow coordinatization.Source: Discussion · p. 17
- A detailed cohomological treatment of logical paradoxes is proposed as future work.The paper says its methods can be applied to an elaborated version of the paradox treatment.Source: Discussion · p. 17