Skip to the manuscript

MATH-PROGRAMME · Documentary Treatment · Domain 01 · UC

The Element in Half the Worlds

Frankl’s conjecture and the structure of union-closed families

A family may close perfectly under union and still conceal the element that should appear in at least half its sets.

Open conjectureUC-DOC-WP01No proof claimed

A note to the reader

## Wonder first; the universal quantifier always visible A finite family of sets can obey one simple closure law and still resist a universal abundance theorem. Frankl’s conjecture says that some element belongs to at least half of the sets in every finite nonempty union-closed family. It is an **Open conjecture**. The plates are mnemonic maps, not proof diagrams. Definitions, exact quantifiers, theorem labels, finite-certificate scopes, source links, and the claim ledger govern the mathematics. **Edition status:** Open conjecture; full-tier documentary exposition; no proof claim.
Open conjecture · Frankl

Let $\mathcal F$ be a finite nonempty union-closed family. Then there is an element $x\in\operatorname{supp}(\mathcal F)$ such that $2\operatorname{freq}_{\mathcal F}(x)\ge |\mathcal F|$.

Claim boundary

This is a source-normalized illustrated documentary. It does not prove Frankl's conjecture, establish a new universal frequency bound, certify unreviewed proof claims, or make a novelty, priority, or public-release claim.

Plate IThe Garden That ClosesPairwise union is local. The conjectured abundant element is a global witness.

Chapter I

## The garden that closes under union A family $\mathcal F$ is union-closed when $A,B\in\mathcal F$ implies $A\cup B\in\mathcal F$.
Definition

The support is $\operatorname{supp}(\mathcal F)=\bigcup_{A\in\mathcal F}A$. The frequency of $x$ is $\operatorname{freq}_{\mathcal F}(x)=|\{A\in\mathcal F:x\in A\}|$.

Closure describes what happens after choosing two members. Frankl’s statement asks for a single coordinate whose incidence count reaches half the entire family. The top union belongs to every finite nonempty union-closed family, but that alone does not identify an abundant element.
Plate IIThe Half-Way BalanceThe threshold is sharp: powersets attain equality for every element.

Chapter II

## The half-way balance In the powerset $\mathcal P(U)$, each element occurs in exactly half the subsets. Thus no universal theorem can demand more than one half. Double-counting incidences gives $$\sum_x\operatorname{freq}_{\mathcal F}(x)=\sum_{A\in\mathcal F}|A|.$$
Established elementary terrain

Powersets attain the threshold sharply. A union-closed family containing a singleton $\{x\}$ satisfies Frankl for $x$ through the injection $A\mapsto A\cup\{x\}$ from sets missing $x$ to sets containing it.

The identity controls an average. It does not force the maximum frequency to reach one half unless additional structure supplies the missing bridge.

Chapter III

## Small worlds, exact ledgers For a universe of size $n$, exact enumeration can test union closure and every frequency profile. The programme’s independently replayed certificate finds no nontrivial counterexample for universes $n\le4$.
Exact bounded result

The $n\le4$ statement is exact finite verification for those universes.

Bounded-computation guardrail

Finite replay validates definitions, software, and bounded claims. It does not prove the conjecture for arbitrary finite universes.

Plate IVThe Lattice CathedralSet-family and lattice formulations require an explicit correspondence.

Chapter IV

## Mirrors, intersections, and lattices Taking complements turns unions into intersections. Inclusion turns a union-closed family into a finite join-semilattice with $A\vee B=A\cup B$. These viewpoints reveal atoms, irreducibles, upper cones, deletion operations, and minimal-counterexample constraints.
Formal bounded terrain

The programme records Lean-checked lattice theorems and hybrid packages through the governed Union-Closed theorem spine. Each result retains its precise hypotheses and certificate boundary.

Correspondence guardrail

Order duality, complement duality, ideal-family averaging, and set-family frequency are related but not interchangeable without a complete translation theorem.

Chapter V

## The average-size door Reimer’s theorem gives a lower bound on the average size of members of a union-closed family. Combined with incidence double-counting, it guarantees some element with nontrivial frequency, but the support-size denominator prevents the argument from reaching the universal half threshold.
Imported established theorem · Reimer

The exact source statement controls this edition. The theorem is a major structural constraint, not a proof of Frankl’s conjecture.

Plate IIIThe Entropy BridgeDimension-free positive constants are breakthroughs; they remain below one half.

Chapter VI

## Entropy enters the garden Choose a uniform random member $A\in\mathcal F$ and write its membership indicators as coordinates. Entropy decomposes the uncertainty of the random set, while coupled copies and union closure create inequalities among the coordinate marginals.
Imported theorem terrain

Gilmer established the first dimension-free positive constant; Sawin, Yu, Cambie, Alweiss–Huang–Sellke, Liu, and others refined the entropy and coupling landscape. Exact constants and assumptions remain source-specific.

Entropy guardrail

A positive constant below one half, a numerically optimized coupling, or a barrier for one entropy ansatz does not settle the universal conjecture.

Plate VIIslands of TheoremThe frontier advances without erasing the open boundary.

Chapter VII

## The theorem frontier Known terrain includes elementary families, bounded exact verification, average-size inequalities, rigorous dimension-free constants, structural restrictions on minimal counterexamples, and formalized lattice results.
Still open universally

No admitted result proves that every finite nonempty union-closed family has an element in at least half its sets.

Recent posted proof claims remain current-awareness entries until their theorem statements, dependencies, and complete proofs are independently audited. Repository merge, citation count, or public attention does not change mathematical status.
Plate VThe Certificate ForgeA certificate proves only the claim encoded by its contract.

Chapter VIII

## The certificate forge The programme separates human proof, imported theorem, Lean theorem, exact finite replay, numerical evidence, and hybrid certificate. A hybrid package may combine a formal transfer theorem with an independently replayed finite fact, but its public claim must expose both components.
Certification discipline

Lean checks encoded terms under explicit imports. Exact replay checks a declared finite search space. Neither silently verifies source correspondence or an unbounded theorem.

No promotion by proximity

Many correct restricted results surrounding an open conjecture do not add up to a proof unless a complete logical bridge is supplied.

Technical appendix A

## Precise definitions and normalizations For finite $\mathcal F\subseteq\mathcal P(U)$, define $\operatorname{supp}(\mathcal F)=\bigcup\mathcal F$ and $\operatorname{freq}_{\mathcal F}(x)=|\{A\in\mathcal F:x\in A\}|$. An element is abundant when $2\operatorname{freq}_{\mathcal F}(x)\ge|\mathcal F|$. The canonical target excludes the empty family and reduces the ground set to the support.

Technical appendix B

## Elementary proof spine The top union belongs to the family by finite repeated closure. Powerset sharpness follows by toggling a chosen coordinate. The singleton theorem uses the injective map $A\mapsto A\cup\{x\}$. These are complete arguments within their scopes and suitable for direct formalization.

Technical appendix C

## Entropy method skeleton For membership indicators $(X_x)_x$, $H(A)=H((X_x)_x)\le\sum_x H(X_x)$. Coupled copies are combined through union closure. Assuming all marginals lie below a selected threshold, one seeks an entropy contradiction. The coupling, concavity estimates, and optimization domain are theorem-bearing details.

Technical appendix D

## Lattice correspondence notes Inclusion supplies a finite join-semilattice. Upper cones and irreducible elements can encode frequency-like data. Translation obligations include separation, the ground-set representation, internal versus ambient lattices, and the exact relation between lattice elements and set coordinates.

Technical appendix E

## Formal and computational inventory | Artifact class | Supports | Does not support | |---|---|---| | Lean theorem | Encoded theorem under imports | Unencoded source correspondence | | Exact finite certificate | Declared finite search | Arbitrary universes | | Hybrid package | Explicit components and bridge | A stronger unstated theorem | | Numerical optimization | Candidates and diagnostics | Rigorous universal bounds |

Technical appendix F

## Source audit ledger Reimer’s theorem and the post-2022 entropy results are literature-derived imports. Programme lattice claims are governed by Domain 01 and their individual formal or hybrid records. Recent proof claims remain `NEEDS_AUDIT` until complete independent review.
Imported-source discipline

A bibliography identifies provenance. It does not replace theorem-body verification, normalization checks, or proof.

Technical appendix G

## Claim-level trust matrix | Claim | Trust class | Qualification | |---|---|---| | Frankl’s half-frequency statement | open | no admitted proof | | Top union, powerset, singleton injection | established | elementary proofs | | No nontrivial counterexample for $n\le4$ | exact finite verification | bounded universe | | Reimer average-size theorem | imported established | source statement governs | | Dimension-free constant bounds | imported established | constants and hypotheses are source-specific | | Programme lattice results | formal or hybrid bounded | individual records govern | | 2026 posted proof claims | needs audit | not promoted here | | SVG plates | pedagogical | never proof authority |
Final claim boundary

This web edition changes presentation and collection membership, not theorem strength. Frankl’s conjecture remains open, and no interaction, plate, bounded replay, or formal special case supplies the missing universal proof.

Sources and programme crosswalk

## Governing literature and campaign record Programme links: [Domain 01](../../domains/union_closed/) · [canonical master plan](https://github.com/grandchallenge/MATH-PROGRAMME/blob/main/DOMAIN_01_UNION_CLOSED_MASTER_PLAN.md) · [UC-DOC-WP00 source lock](https://github.com/grandchallenge/MATH-PROGRAMME/tree/main/campaigns/union_closed/UC_DOC_WP00_DOCUMENTARY_SOURCE_LOCK) · [UC-DOC-WP01 admission](https://github.com/grandchallenge/MATH-PROGRAMME/tree/main/campaigns/union_closed/UC_DOC_WP01_WEB_ADMISSION) · [review records](https://github.com/grandchallenge/MATH-PROGRAMME/tree/main/reviews/union_closed)

Edition record

This is the first Wave Two admission and the first non-Millennium full-tier edition. It exercises the shared reader against finite combinatorics, exact enumeration, formal proof artifacts, hybrid certificates, and active source-audit obligations.

The web edition is derivative. The committed pointer is a source record; the checksum-locked complete illustrated source bundle is the authoritative source artifact. MathJax 3.2.2 is a version-pinned network enhancement, and the source TeX remains present when it is unavailable.

Web claim boundary: This is a source-normalized illustrated documentary. It does not prove Frankl's conjecture, establish a new universal frequency bound, certify unreviewed proof claims, or make a novelty, priority, or public-release claim.

Rendered PDF
3,343,773 bytes · 6ea03bef444f19ae8013e80c76a5112fda9c6b740d61387c2bfeea5921ac71dc · metadata_only
Complete LaTeX source
50,548 bytes · e889079fc77163e57b0c239e8f25ae29a3ded640b32120f65d1f3708c05dfdde · metadata_only
Authoritative complete illustrated source bundle
3,100,936 bytes · 3a1fcf16dee92c6bbf5fd8285702e31c828aa6d1666e5605e8981346f4bd2daf · metadata_only