betaFeedback

WorksMathematics

Packaging Systems: Intrinsic Structure, Recovery Ranks, and Globalization Theorems

Abstract

We define packaged structures through carriers, internal units, incidence, candidate spaces, and selected families. A realized packaging system additionally retains a context and formation datum related to its output by a possibly partial, multivalued realization relation. We establish comparison-relative recoverability and non-collapse criteria and an intrinsic refinement-completion theory. For atlas specializations, exact extension becomes a specialization-relative packaging completion problem. Under additional recovery and elimination axioms, we establish general globalization sufficiency criteria and sharp classifications for three classes of compatible finite atlases on a fixed cover H\mathcal H. A geometric obstruction rank ν(H)ν(\mathcal H) yields thresholds 2k+1<ν(H)2k+1<ν(\mathcal H) for empty-preserving closures of local Carathéodory number at most k≥1k\ge1; d<ν(H)d<ν(\mathcal H) for complete functional-dependency theories of local minimum Armstrong size at most d≥3d\ge3; and 3<ν(H)3<ν(\mathcal H) for transitive relations, including equivalence relations. For nonempty finite label sets and empty-preserving closures, minimum commutant realization dimension equals minimum Armstrong size. We also determine intrinsic recovery ranks and construct canonical clique completions under bounded conformality without chordality. Explicit obstructions establish sharpness; a transport hypothesis is essential for the general recovery–obstruction converse. Complexity bounds apply locally, not necessarily to global extensions.

Introduction

We develop an axiomatic theory of packaged structures and packaging systems. A packaged structure records package carriers, internal units, incidence, local candidate spaces, and selected candidate families. A realized packaging system also retains the context and formation datum associated with a particular packaged output. Its realization relation need not be total or single-valued. Within this framework, we study intrinsic completion, recoverability, and comparison-relative non-collapse, and establish exact-globalization criteria for several classes of local structures.

The terminology is motivated by our earlier work on packaged entangled states and symmetry packaging [1, 2, 3]. Related constructions arise in field theory, lattice-gauge simulation, and gauge-invariant quantum information [4, 5, 6]. Those works use specific physical or algebraic formation laws. Here we define the general framework and study which structural and local-to-global results it supports.

Packaged structures and the globalization problem

A realized packaging system is a triple

P=(Ξ,P,P),Ξ⊢P⇝P\mathbb{P} = (\Xi, P, \mathcal{P}), \qquad\Xi\vdash P \rightsquigarrow\mathcal{P}

where Ξ\Xi is a packaging context, P∈FormΞP \in\mathrm{Form}_{\Xi} is a formation datum, and the realization judgment means (P,P)∈JΞ(P,\mathcal{P}) \in J_{\Xi}. In a packaging context, a formation datum may realize no packaged structure or several distinct packaged structures; different data may also realize the same structure. A realized system requires an actual realization, whereas its packaged endpoint can be studied independently of the formation datum. Restriction and gluing laws are introduced only in the relevant specializations.

The intrinsic structure of a packaged output raises questions independent of its formation mechanism. For refinement-packaged structures, selected local refinements may agree exactly on every pairwise overlap of package interiors without admitting a global refinement with the prescribed local traces. This leads to the intrinsic completion problem.

The later globalization theorems concern additional mathematical specializations. In particular, closure and functional-dependency atlases are defined on covers of descriptor-label sets, which need not coincide with package interiors. Their restriction and compatibility laws belong to the corresponding specializations and are not primitive axioms of packaging. For the finite globalization results, let VV be a finite label set and let H=(Ij)j∈J\mathcal{H}=(I_j)_{j\in J} be a finite cover of VV.

The universal exact-globalization problem fixes a cover H\mathcal{H} and a class of local structures, and asks whether every compatible atlas in that class has a global structure whose restriction to each original chart is exactly the prescribed local structure. Only the local atlas varies; the cover and the local class are fixed.

The obstruction rank ν(H)\nu(\mathcal{H}) measures how the cover geometry can prevent exact globalization. The primal graph joins two labels whenever they belong to a common chart. We call a clique guarded if all its labels occur together in one chart. If the primal graph has an induced cycle of length at least four, then ν(H)=3\nu(\mathcal{H})=3. Otherwise, ν(H)\nu(\mathcal{H}) is the size of the smallest unguarded clique, with value ∞\infty if every clique is guarded. The globalization theorems relate this rank to the complexity of the local structures.

Principal results

We first establish what the packaging axioms allow and what they do not force. Deterministic formation is a proper special case, as is refinement packaging on any nonempty structural domain. We also give exact tests for recoverability and for failure of a proposed reduction. In the access–commutant example, the commutant determines its represented source but need not be an algebra quotient of that source.

For refinement-packaged structures, we construct an intrinsic completion operator and a completion space G(P)\mathcal{G}(\mathcal{P}). A static isomorphism induces a bijection between completion spaces. We give two finite examples with the same package–unit incidence and matching types of local selections. In both examples, the selected refinements agree on every pairwise overlap. Yet one example has exactly one global completion, while the other has none.

We classify universal exact globalization for three classes of local mathematical structures. For a class K\mathcal{K}, write UGK(H)\mathrm{UG}_{\mathcal{K}}(\mathcal{H}) when every compatible K\mathcal{K}-atlas on H\mathcal{H} admits an exact global extension. The three classes are specified as follows:

  • UGcar≤k\mathrm{UG}_{\mathrm{car}\leq k} concerns finite empty-preserving closures of Carathéodory number at most kk;

  • UGarm≤d\mathrm{UG}_{\mathrm{arm}\leq d} concerns complete functional-dependency theories of minimum Armstrong size at most dd, allowing empty premises;

  • UGtrans\mathrm{UG}_{\mathrm{trans}} concerns transitive binary relations.

The same transitive criterion holds for preorders, partial orders, strict partial orders, and equivalence relations. The complexity bounds apply to local structures; a global extension need not satisfy them.

Theorem 1.1 (Globalization obstruction spectrum). For every finite cover H\mathcal{H}, every integer k≥1k \ge1, and every integer d≥3d \ge3,

UG⁡car≤k(H)⟺2k+1<ν(H),\operatorname{UG}_{\mathrm{car}\leq k}(\mathcal{H}) \Longleftrightarrow2k+1 < \nu(\mathcal{H}),
UG⁡arm≤d(H)⟺d<ν(H),\operatorname{UG}_{\mathrm{arm}\leq d}(\mathcal{H}) \Longleftrightarrow d < \nu(\mathcal{H}),
UG⁡trans(H)⟺3<ν(H).\operatorname{UG}_{\mathrm{trans}}(\mathcal{H}) \Longleftrightarrow3 < \nu(\mathcal{H}).

Every failed inequality admits an explicit finite obstruction supported on primitive cover geometry.

The globalization classifications arise within restriction-equipped atlas specializations of the packaging framework. The atlas realization bridge (Theorem 4.2) represents compatible atlases as packaging systems and identifies their exact extensions with a specialization-relative completion space. Its nonemptiness requires further mathematical structure: § 10 formulates recovery and elimination axioms that give general globalization sufficiency, and isolates the obstruction-transport hypotheses needed for sharp converses. The three displayed thresholds are established by verifying suitable recovery mechanisms and constructing explicit obstructions in the respective classes; they are not consequences of the primitive realization axioms alone.

Proposition 8.55 distinguishes extension of an individual atlas from universal exact globalization and shows that the universal answer depends on the class of local structures. Native dimension two is exceptional: universal exact globalization holds on every finite cover (Proposition 6.4). In higher dimensions, a global commutant realization can require more dimensions than its local realizations; Proposition 6.7 exhibits an increase from three to four dimensions.

The first two thresholds differ because the two closure classes recover composed implications in different ways. For a nonempty finite attribute set and an empty-preserving closure cc, we prove

dcomm(c)=arm⁡(c),d_{\mathrm{comm}}(c) = \operatorname{arm}(c),

where the left-hand side is the minimum dimension of a commuting self-adjoint realization and the right-hand side is the minimum Armstrong size.

For the native class CommCldd_d, Theorem 6.32 establishes original-chart recovery for composable local implications whose combined support is a primal clique, provided every primal clique of size at most dd is guarded. The recovered implication has a nonempty premise drawn from the original premises, of size at most d−1d-1, and is valid in an original chart. Thus, unlike ordinary Horn composition [7], the theorem controls both premise size and the location of the recovered consequence.

For d≥3d \ge3, the intrinsic recovery ranks satisfy

ipr⁡(CommCldd)=d,ipr⁡(Cd−1)=2d−1,\operatorname{ipr}(\mathrm{CommCld}_d) = d,\qquad\operatorname{ipr}(\mathcal{C}_{d-1}) = 2d-1,

where Cd−1\mathcal{C}_{d-1} is the class of finite empty-preserving closures of Carathéodory number at most d−1d-1. Thus bounding the minimum Armstrong size yields a lower recovery rank than bounding the Carathéodory number alone.

For locally propagation-recoverable closure classes, controlled propagation gives clique-locality under guarding and chordality: the global closure restricts to each primal clique as the closure generated by the original rules supported there (Theorem 4.17). Without chordality, bounded conformality still yields compatible least clique fillers that may be adjoined simultaneously without changing the globally generated closure (Theorem 4.18). The transitive case instead uses composition along paths, with cycle-based obstructions (Theorem 7.12).

More generally, the intrinsic propagation–recovery rank, defined through rooted circuits, gives a sufficient criterion for universal exact globalization. For a restriction-stable class C\mathcal{C} of finite empty-preserving closures with finite rank p:=ipr⁡(C)≥3p := \operatorname{ipr}(\mathcal{C}) \ge3, the inequality p<ν(H)p < \nu(\mathcal{H}) implies UGC⁡(H)\operatorname{UGC}(\mathcal{H}) (Theorem 8.43). A converse requires additional obstruction-transport hypotheses, verified by explicit constructions for the bounded-Carathéodory and bounded-Armstrong classes. Theorem 8.41 provides a counterexample showing that restriction and relabeling stability alone do not suffice for the converse.

Relation to previous work

Our results draw on closure systems, database dependencies, hypergraphs, and amalgamation. The comparisons below distinguish the mathematical objects being extended and the quantifiers governing extension, rather than terminology alone. The primitive packaged coordinates admit a faithful relational presentation; our non-collapse claims therefore concern specified comparisons, not absolute non-representability.

Sheaf-type gluing and local equivalence relations provide established approaches to compatible local data [8, 9, 10, 11]. By contrast, the primitive packaging framework specifies package-side structure and a realization relation without assuming restriction maps; these enter only in appropriate specializations.

Finite closure systems, implicational bases, rooted circuits, and abstract convexity have a substantial theory [12, 13, 14, 15]. Gluing closure and convexity structures has also been studied [16, 17]. For a specified finite closure atlas, the least generated global closure and an exact extension criterion are considered in [18]. Our bounded-Carathéodory theorem instead fixes the cover and classifies when every compatible atlas satisfying the local complexity bound has an exact global extension.

Functional dependencies, Armstrong relations, and partition semantics are classical [19, 20, 21, 22]. Local and global consistency under functional dependencies has been studied for database schemes [23, 24, 25, 26]. Functional dependence in locally consistent KK-relations is studied in [27]. Our bounded-Armstrong theorem classifies universal exact extension for compatible complete dependency theories subject to a local minimum-Armstrong-size bound.

Related local-to-global methods occur in constraint consistency, relational elimination, and consistency witnesses for annotated relations [28, 29, 30, 31, 32]. Our recovery invariant measures how much of a clique must lie in one original chart before a composed consequence can be recovered there.

For transitive structures, related results arise in model-theoretic amalgamation [33, 34]. Our theorem classifies finite covers on which every compatible transitive atlas has an exact global extension. The cover criteria use chordality and bounded conformality, notions familiar from hypergraph acyclicity [35, 36, 37, 38]. Classical α\alpha-acyclic database hypergraphs are characterized by chordality and full conformality; in the three principal exact-globalization classifications here, chordality remains necessary but only class-dependent bounded conformality is required. The semantic objects and extension questions differ from relational-instance consistency.

Organization

The paper proceeds in three stages:

  1. Axiomatic foundations. Section 2 develops the primitive packaging axioms, formation and realization relations, comparison-relative recovery and non-collapse criteria, and examples of distinct native formation laws.

  1. Intrinsic structure and atlas specialization. Section 3 establishes intrinsic refinement completion, coherence, and invariance results. § 4 realizes compatible atlases as packaging systems and proves controlled propagation, canonical clique completion, and sharp globalization for bounded-Carathéodory closures.

  1. Further specializations and synthesis. § 5 relates minimum commutant realization dimension to minimum Armstrong size. § 6 proves native clique resolution and the bounded-Armstrong globalization classification. § 9 develops transitive and equivalence-relation amalgamation. § 10 formulates recovery-equipped packaging classes, develops intrinsic recovery ranks and obstruction transport, and synthesizes the sharp class-specific results.

The primitive packaging axioms, intrinsic completion properties, and additional restriction and recovery laws of the specializations play distinct logical roles throughout. In particular, the atlas bridge records realized packaging systems, while exact globalization requires the further hypotheses verified in the respective classes.

Foundations of packaging systems

This section defines the objects used throughout the paper: packaged structures, packaging contexts, formation data, and realized packaging systems. We keep two questions separate: whether an output is allowed by a context and whether a formation datum actually realizes it.

The main foundational result, Theorem 2.21, shows that the possible selections on a fixed scaffold form a Boolean algebra, while the formation relation can still be arbitrary. Further results identify proper specializations and make precise which comparisons can show non-collapse. Minimal-cover selection and access–commutant formation give concrete examples of different formation laws. Together, these results supply the language needed for the later completion and globalization theorems.

Minimal structural vocabulary

Before defining packaging, we specify the kinds and sizes of objects it can contain. A structural domain is a set of possible carriers; each carrier can have local candidate data. These definitions organize the data but impose no maps, quotients, or comparisons of information. This lets us distinguish properties of the general axioms from assumptions added in later examples.

Convention 2.1 (Set-theoretic size). Fix Grothendieck universes U∈VU \in V. A UU-small set belongs to UU. A UU-class is a VV-small collection of UU-small objects, not necessarily itself in UU. Unless stated otherwise, structural domains, carrier and package families, internal-unit sets, descriptor index sets and codomains, and sets carrying refinements are

U\mathcal{U}-small. Contexts, reconstruction-regime collections, and class-valued descriptor admissibility assignments are V\mathcal{V}-small. The local candidate spaces remain U\mathcal{U}-small even when the regime collections are larger. For I∈UI \in\mathcal{U}, Ref⁡(I)\operatorname{Ref}(I) is U\mathcal{U}-small. No construction takes a tagged union over a proper U\mathcal{U}-class.

Definition 2.2 (Structural domain). A structural domain C\mathcal{C} is a specified U\mathcal{U}-small set of mathematical objects. No common morphisms, topology, algebraic operations, incidence, or dynamics are assumed unless supplied by a specialization.

Definition 2.3 (Structural carrier). A structural carrier is an object K∈CK \in\mathcal{C} that may support package-side structure. A context specifies which carriers are admissible, through Car⁡Ξ⊆CΞ\operatorname{Car}_{\Xi} \subseteq\mathcal{C}_{\Xi}. A carrier need not itself be a subset of the internal-unit set or a quotient class.

Definition 2.4 (Local candidate datum). Let KK be a structural carrier and let I∈UI \in\mathcal{U} be a specified local unit set. A local candidate datum on the pair (K,I)(K,I) consists of a U\mathcal{U}-small set L(K;I)L(K;I) together with a distinguished subset

R(K;I)⊆L(K;I).R(K;I) \subseteq L(K;I).

The set L(K;I)L(K;I) is the local candidate space, and R(K;I)R(K;I) is the selected candidate family. The set L(K;I)L(K;I) need not have an order, algebra, topology, probability law, or categorical structure. We also assume no representation by maps, partitions, or equivalence relations unless a specialization supplies one. The candidate space and the selected family are both permitted to be empty.

The general candidate coordinate of a packaged structure will use only Definition 2.4. Later we specialize the local candidate space to a refinement poset on the package interior. Thus normalized refinements, quotient representatives, and their kernels are not part of the definition of an arbitrary local candidate datum.

Descriptor families are introduced for later reconstruction equipment. Their relation to refinement candidates belongs to the refinement specialization and is not required by the general candidate coordinate.

Packaged structures and packaging contexts

The basic theory separates the data retained by a packaged output from the conditions governing its realization. A packaged structure specifies carriers, units, incidence, candidates, and selections. A packaging context specifies admissible outputs and the formation data that realize them.

Packaged structure P\mathfrak{P}

Axioms (S1)–(S5) specify carriers, internal units, incidence, local candidate spaces, and selected families without imposing coherence between packages. Lemma 2.8 establishes the required set-theoretic size of the model class.

Definition 2.5 (Packaged-structure axioms and models). Fix a structural domain C\mathcal{C}. A candidate package-side datum over C\mathcal{C} is a five-tuple

P=(Pkg⁡P,EP,φP,LP,RP).\mathfrak{P} = \left(\operatorname{Pkg}_{\mathfrak{P}}, E_{\mathfrak{P}}, \varphi_{\mathfrak{P}}, L_{\mathfrak{P}}, R_{\mathfrak{P}}\right).

The packaged-structure axiom system Ax⁡PkgStr(C)\operatorname{Ax}_{\mathrm{PkgStr}}(\mathcal{C}) consists of the following five conditions:

(S1) Carrier axiom.

Pkg⁡P⊆C.\operatorname{Pkg}_{\mathcal P} \subseteq\mathcal C.

The members of Pkg⁡P\operatorname{Pkg}_{\mathcal P} are the package carriers.

(S2) Internal-unit axiom.

EP∈U.E_{\mathcal P} \in\mathcal U.

Thus the package-side internal-unit set is U\mathcal U-small.

(S3) Incidence axiom.

◃P⊆EP×Pkg⁡P.\mathrel{\triangleleft}_{\mathcal P} \subseteq E_{\mathcal P} \times\operatorname{Pkg}_{\mathcal P}.

For K∈Pkg⁡PK \in\operatorname{Pkg}_{\mathcal P}, its interior is

IK=Int⁡P(K):={e∈EP:e◃PK}.I_K = \operatorname{Int}_{\mathcal P}(K) := \{e \in E_{\mathcal P} : e \mathrel{\triangleleft}_{\mathcal P} K\}.

(S4) Local-candidate axiom.

LP=(LP(K))K∈Pkg⁡PL_{\mathcal P} = \left(L_{\mathcal P}(K)\right)_{K \in\operatorname{Pkg}_{\mathcal P}}

is a family of U\mathcal U-small sets indexed by Pkg⁡P\operatorname{Pkg}_{\mathcal P}, where LP(K)L_{\mathcal P}(K) is the declared local candidate space attached to (K,IK)(K,I_K).

(S5) Selection axiom.

RP=(RP(K))K∈Pkg⁡PR_{\mathcal P} = \left(R_{\mathcal P}(K)\right)_{K \in\operatorname{Pkg}_{\mathcal P}}

is a family satisfying

RP(K)⊆LP(K)(K∈Pkg⁡P).R_{\mathcal P}(K) \subseteq L_{\mathcal P}(K) \qquad(K \in\operatorname{Pkg}_{\mathcal P}).

A packaged structure is a model of (S1)–(S5); write

PkgDat⁡C=Mod⁡(Ax⁡PkgStr(C)),CP:=C.\operatorname{PkgDat}_{\mathcal C} = \operatorname{Mod}\left(\operatorname{Ax}_{\mathrm{PkgStr}}(\mathcal C)\right), \qquad\mathcal C_{\mathcal P} := \mathcal C.

The ambient domain C\mathcal C is part of the type, not a sixth coordinate. The same five-tuple with different ambient domains is regarded as differently typed (equivalently, domain-tagged).

Definition 2.6 (Package and internal unit). Let P\mathcal P be a packaged structure. A package is a carrier K∈Pkg⁡PK \in\operatorname{Pkg}_{\mathcal P}, together with its interior IKI_K, local candidate space LP(K)L_{\mathcal P}(K), and selected candidate family RP(K)R_{\mathcal P}(K). An element e∈EPe \in E_{\mathcal P} is a package-side internal unit.

Remark 2.7 (Scope of the primitive axioms). The axioms permit empty package families, empty or overlapping interiors, uncovered internal units, and empty candidate or selected families. They impose no compatibility condition between packages and do not require a global completion. Such conditions may be introduced by specialization.

Lemma 2.8 (Size of the packaged-data space). Let C\mathcal C be a U\mathcal U-small structural domain. Then PkgDat⁡C\operatorname{PkgDat}_{\mathcal C} is a V\mathcal V-small set.

Proof. Because C∈U\mathcal C \in\mathcal U, the collection P(C)\mathcal P(\mathcal C) of package-carrier sets is U\mathcal U-small. For each Pkg⁡⊆C\operatorname{Pkg} \subseteq\mathcal C and each E∈UE \in\mathcal U, the possible incidence relations form the U\mathcal U-small set

P(E×Pkg⁡).\mathcal P(E \times\operatorname{Pkg}).

Since U∈V\mathcal U \in\mathcal V, the collection of all U\mathcal U-small sets is V\mathcal V-small. Thus the possible indexed families

(L(K))K∈Pkg⁡,(R(K))K∈Pkg⁡,R(K)⊆L(K),\left(L(K)\right)_{K \in\operatorname{Pkg}}, \qquad \left(R(K)\right)_{K \in\operatorname{Pkg}}, \qquad R(K) \subseteq L(K),

form a V\mathcal V-small collection. Combining the five coordinates gives a V\mathcal V-small set of packaged structures over C\mathcal C. ∎

Packaging context Xi

Having fixed what counts as a packaged output, we specify the environment in which that output can be formed. A packaging context declares a structural domain, admitted carriers, formation data, and a primitive realization relation. Keeping this relation separate from structural compatibility is what allows partial and multivalued formation in the ensuing system axioms.

Definition 2.9 (Packaging context Ξ\Xi). A packaging context is a quadruple

Ξ=(CΞ,Car⁡Ξ,Form⁡Ξ,JΞ)\Xi= (\mathcal{C}_{\Xi}, \operatorname{Car}_{\Xi}, \operatorname{Form}_{\Xi}, J_{\Xi})

that satisfies the following conditions:

(P1)(P1) CΞ\mathcal{C}_{\Xi} is a structural domain, and

Car⁡Ξ⊆CΞ\operatorname{Car}_{\Xi} \subseteq\mathcal{C}_{\Xi}

is the set of admissible carriers.

(P2)(P2)

Form⁡Ξ∈V\operatorname{Form}_{\Xi} \in\mathcal{V}

is a set, called the formation-data space.

(P3)(P3) With

PkgDat⁡Ξ={Q∈PkgDat⁡CΞ:Pkg⁡Q⊆Car⁡Ξ},\operatorname{PkgDat}_{\Xi} = \left\{ \mathcal{Q} \in\operatorname{PkgDat}_{\mathcal{C}_{\Xi}} : \operatorname{Pkg}_{\mathcal{Q}} \subseteq\operatorname{Car}_{\Xi} \right\},

the realization relation is

JΞ⊆Form⁡Ξ×PkgDat⁡Ξ.J_{\Xi} \subseteq\operatorname{Form}_{\Xi} \times\operatorname{PkgDat}_{\Xi}.

Foundational compatibility

An output can be allowed by a context without being realized by any formation datum. We call the first property compatibility and define it here. Compatibility is necessary for realization but does not imply it. This use of the word is different from agreement of local data on chart overlaps, introduced later.

References

  1. [1]Rongchao Ma. Theory of packaged entangled states. Reports in Advances of Physical Sciences, 1(3):1750005, 2017. doi:10.1142/S2424942417500050.DOI
  2. [2]Rongchao Ma. Symmetry packaging I: Irreducible representation blocks, superselection, and packaged entanglement in quantum field theory. Nuclear Physics B, 1018:117076, 2025. doi:10.1016/j.nuclphysb.2025.117076.DOI
  3. [3]Rongchao Ma. Symmetry packaging II: Packaged subspaces, packaged entanglement, and classification. Nuclear Physics B, 1018:117093, 2025. doi:10.1016/j.nuclphysb.2025.117093.DOI
  4. [4]Rongchao Ma. Packaged quantum states in field theory: No partial factorization, multi-particle packaging, and hybrid gauge-invariant entanglement, 2025. arXiv:2502.00766.arxiv.org/abs/2502.00766
  5. [5]Rongchao Ma. Packaged quantum states for quantum simulation of lattice gauge theories, 2025. arXiv:2502.14654.arxiv.org/abs/2502.14654
  6. [6]Rongchao Ma. Packaged quantum states for gauge-invariant quantum computation and communication, 2025. arXiv:2505.02205.arxiv.org/abs/2505.02205
  7. [7]Paolo Libratore. One head is better than two: A polynomial restriction for propositional definite horn forgetting. Journal of Logic, Language and Information, 34:49–88, 2025. doi:10.1007/s10849-025-09428-w.
  8. [8]Saunders Mac Lane and Ieke Moerdijk. Sheaves in Geometry and Logic: A First Introduction to Topos Theory. Universitext. Springer, New York, 1992. doi:10.1007/978-1-4612-0927-0.
  9. [9]Kimmo I. Rosenthal. Local equivalence relations. Topology and its Applications, 13(2):167–176, 1982. doi:10.1016/0166-8641(82)90018-9.DOI
  10. [10]Kimmo I. Rosenthal. Sheaves and local equivalence relations. Cahiers de Topologie et Géométrie Différentielle Catégoriques, 25(2):179–206, 1984. URL: https://www.numdam.org/item/CTGDC_1984__25_2_179_0/.
  11. [11]Anders Kock and Ieke Moerdijk. Spaces with local equivalence relations, and their monodromy. Topology and its Applications, 72(1):47–78, 1996. doi:10.1016/0166-8641(95)00101-8.DOI
  12. [12]Garrett Birkhoff. Lattice Theory, volume 25 of American Mathematical Society Colloquium Publications. American Mathematical Society, Providence, RI, 3 edition, 1967.
  13. [13]Bernhard Ganter and Rudolf Wille. Formal Concept Analysis: Mathematical Foundations. Springer, Berlin, 1999. doi:10.1007/978-3-642-59830-2.DOI
  14. [14]Nathalie Caspard and Bernard Monjardet. The lattices of closure systems, closure operators, and implicational systems on a finite set: A survey. Discrete Applied Mathematics, 127(2):241–269, 2003. doi:10.1016/S0166-218X(02)00209-3.DOI
  15. [15]Kenji Kashiwabara and Masataka Nakamura. The prime stems of rooted circuits of closure spaces and minimum implicational bases. The Electronic Journal of Combinatorics, 20(1):P22, 2013. doi:10.37236/3068.DOI
  16. [16]Gerard Sierksma. Convexity on unions of sets. Compositio Mathematica, 42(3):391–400, 1981. URL: https://www.numdam.org/item/CM_1980__42_3_391_0/.
  17. [17]Eric Degreef and René Forneau. Gluings and closures. Compositio Mathematica, 51(3):325–332, 1984. URL: https://www.numdam.org/item/CM_1984__51_3_325_0/.
  18. [18]Jaehwan Kim. Closure atlases and local-to-global obstructions in finite closure systems, 2026. Version 1, 19 June 2026. URL: https://arxiv.org/html/2606.24909v1, arXiv:2606.24909v1.
  19. [19]Catriel Beeri, Martin Dowd, Ronald Fagin, and Richard Statman. On the structure of Armstrong relations for functional dependencies. Journal of the ACM, 31(1):30–46, 1984. URL: https://research.ibm.com/publications/on-the-structure-of-armstrong-relations-for-functional-dependencies, doi:10.1145/2422.322414.DOI
  20. [20]Heikki Mannila and Kari-Jouko Räihä. Design by example: An application of armstrong relations. Journal of Computer and System Sciences, 33(2):126–141, 1986. doi:10.1016/0022-0000(86)90015-2.DOI
  21. [21]Tony T. Lee. An algebraic theory of relational databases. Bell System Technical Journal, 62(10):3159–3204, 1983. doi:10.1002/j.1538-7305.1983.tb03470.x.DOI
  22. [22]Stavros S. Cosmadakis, Paris C. Kanellakis, and Nicolas Spyratos. Partition semantics for relations. Journal of Computer and System Sciences, 33(2):203–233, 1986. doi:10.1016/0022-0000(86)90019-X.DOI
  23. [23]Yehoshua Sagiv. A characterization of globally consistent databases and their correct access paths. ACM Transactions on Database Systems, 8(2):266–286, 1983. doi:10.1145/319983.319988.DOI
  24. [24]Kent Laver, Alberto O. Mendelzon, and Marc H. Graham. Functional dependencies on cyclic database schemes. In Proceedings of the 1983 ACM SIGMOD International Conference on Management of Data, pages 79–91. ACM, 1983. doi:10.1145/582192.582208.DOI
  25. [25]Yehoshua Sagiv and Oded Shmueli. A characterization of finite FD-acyclicity. Journal of Computer and System Sciences, 38(2):380–404, 1989. doi:10.1016/0022-0000(89)90008-1.DOI
  26. [26]Ke Wang. On characterizing boundedness of database schemes with bounded dependencies. Theoretical Computer Science, 100(2):347–364, 1992. doi:10.1016/0304-3975(92)90308-3.DOI
  27. [27]Timon Barlag, Miika Hannula, Juha Kontinen, Nina Pardal, and Jonni Virtema. Locally consistent K-relations: Entailment and axioms of functional dependence, 2025. Version 2, revised 23 February 2026. arXiv:2505.11057v2, doi:10.48550/arXiv.2505.11057.DOI
  28. [28]Peter van Beek and Rina Dechter. Constraint tightness and looseness versus local and global consistency. Journal of the ACM, 44(4):549–566, 1997. doi:10.1145/263867.263499.DOI
  29. [29]Rina Dechter and Peter van Beek. Local and global relational consistency. Theoretical Computer Science, 173(1):283–308, 1997. doi:10.1016/S0304-3975(97)86737-0.DOI
  30. [30]Albert Atserias and Phokion G. Kolaitis. Consistency of relations over monoids. Journal of the ACM, 72(3):18:1–18:47, 2025. doi:10.1145/3721855.DOI
  31. [31]Albert Atserias and Phokion G. Kolaitis. Consistency witnesses for annotated relations. SIGMOD Record, 54(2):7–17, 2025. doi:10.1145/3749116.3749118.DOI
  32. [32]Albert Atserias and Phokion G. Kolaitis. Gamma acyclicity, annotated relations, and consistency witness functions. In Balder ten Cate and Maurice Funk, editors, 29th International Conference on Database Theory (ICDT 2026), volume 365 of Leibniz International Proceedings in Informatics (LIPIcs), pages 16:1–16:18, Dagstuhl, Germany, 2026. Schloss Dagstuhl – Leibniz-Zentrum für Informatik. URL: https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.ICDT.2026.16, doi:10.4230/LIPIcs.ICDT.2026.16.DOI
  33. [33]Paolo Lipparini. The strong amalgamation property into union, 2023. Version 3, 29 January 2023; first posted in 2021. URL: https://arxiv.org/abs/2103.00563v3, arXiv:2103.00563v3.
  34. [34]Martin Otto. Amalgamation and symmetry: From local to global consistency in the finite, 2024. Version 3 (2024). URL: https://arxiv.org/abs/1709.00031v3, arXiv:1709.00031v3.
  35. [35]Catriel Beeri, Ronald Fagin, David Maier, and Mihalis Yannakakis. On the desirability of acyclic database schemes. Journal of the ACM, 30(3):479–513, 1983. doi:10.1145/2402.322389.DOI
  36. [36]Ronald Fagin. Degrees of acyclicity for hypergraphs and relational database schemes. Journal of the ACM, 30(3):514–550, 1983. doi:10.1145/2402.322390.DOI
  37. [37]Johann Brault-Baron. Hypergraph acyclicity revisited. ACM Computing Surveys, 49(3):54:1–54:26, 2016. doi:10.1145/2983573.
  38. [38]Martin Otto. Highly acyclic groups, hypergraph covers, and the guarded fragment. Journal of the ACM, 59(1):5:1–5:40, 2012. doi:10.1145/2108242.2108247.DOI
  39. [39]Øystein Ore. Theory of equivalence relations. Duke Mathematical Journal, 9(3):573–627, 1942. doi:10.1215/S0012-7094-42-00942-6.DOI
  40. [40]B. A. Davey and H. A. Priestley. Introduction to Lattices and Order. Cambridge University Press, Cambridge, 2 edition, 2002. doi:10.1017/CBO9780511809088.DOI
  41. [41]Vaughan F. R. Jones. Von Neumann algebras. Lecture notes. University of California, Berkeley, 2010. Dated 29 November 2010. URL: https://math.berkeley.edu/~vfr/MATH20909/VonNeumann2009.pdf.
  42. [42]Rudolf Wille. Restructuring lattice theory: An approach based on hierarchies of concepts. In Ivan Rival, editor, Ordered Sets, volume 83 of NATO Advanced Study Institutes Series C: Mathematical and Physical Sciences, pages 445–470. D. Reidel, Dordrecht, 1982. doi:10.1007/978-94-009-7798-3_15.DOI
  43. [43]Fănică Gavril. The intersection graphs of subtrees in trees are exactly the chordal graphs. Journal of Combinatorial Theory, Series B, 16(1):47–56, 1974. doi:10.1016/0095-8956(74)90094-X.DOI
  44. [44]Chin-Wen Ho and R. C. T. Lee. Counting clique trees and computing perfect elimination schemes in parallel. Information Processing Letters, 31(2):61–68, 1989. doi:10.1016/0020-0190(89)90070-7.DOI
  45. [45]Hisayuki Hara and Akimichi Takemura. Boundary cliques, clique trees and perfect sequences of maximal cliques of a chordal graph, 2006. Version 1, 11 July 2006. URL: https://arxiv.org/abs/cs/0607055v1, arXiv:cs/0607055v1.
  46. [46]Lhouari Nourine and Simon Vilmin. Enumerating maximal consistent closed sets in closure systems. In Agnès Braud, Aleksey Buzmakov, Tom Hanika, and Florence Le Ber, editors, Formal Concept Analysis, volume 12733 of Lecture Notes in Computer Science, pages 57–73. Springer, 2021. Preprint version 1 (2021). URL: https://arxiv.org/abs/2102.04245v1, arXiv:2102.04245v1, doi: 10.1007/978-3-030-77867-5_4.
  47. [47]Christian Bliek and Djamila Sam-Haroud. Path consistency on triangulated constraint graphs. In Proceedings of the Sixteenth International Joint Conference on Artificial Intelligence, volume 1, pages 456–461, 1999. URL: https://www.ijcai.org/Proceedings/99-1/Papers/066.pdf.
  48. [48]Christophe Lecoutre, Stéphane Cardon, and Julien Vion. Second-order consistencies. Journal of Artificial Intelligence Research, 40:175–219, 2011. Journal text deposited on arXiv in 2014. URL: https://arxiv.org/abs/1401.3872v1, arXiv:1401.3872v1, doi:10.1613/jair.3180.
  49. [49]Sunil Chopra and M. R. Rao. The partition problem. Mathematical Programming, 59(1–3):87–115, 1993. doi:10.1007/BF01581239.DOI
  50. [50]Andrea Hornáková, Jan-Hendrik Lange, and Bjoern Andres. Analysis and optimization of graph decompositions by lifted multicuts. In Doina Precup and Yee Whye Teh, editors, Proceedings of the 34th International Conference on Machine Learning, volume 70 of Proceedings of Machine Learning Research, pages 1539–1548. PMLR, 2017. URL: https://proceedings.mlr.press/v70/hornakova17a.html.

Paper details

vlnc:W_t9uf3qbj
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 9

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents