Alphabeta Math
LemmaStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedprecheck passjudge pass (gpt-6-sol)audited 2026-09-30
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

Composition preserves perfect satisfiability

Statement

Let G be a finite binary constraint graph over a finite alphabet Σ with ∣Σ∣≥2, and let H=G∘P be the Boolean arity-six composition defined in Composition of an edge system with an assignment tester. For this finite Boolean constraint system, write val⁡(H) for the maximum satisfied fraction, with value one when the constraint list is empty; equivalently, val⁡(H)=1−UNSAT⁡(H) under the convention of Assignment tester and rejection ratio. If val⁡(G)=1, then val⁡(H)=1, including the edgeless case.

Facts & Assumptions

Given: Fix G and its robust ordered edge circuits, with the local two-piece assignment tester P and composition H fixed as in the definition.

[F1]

The graph has finite vertex and edge sets and a finite nonempty alphabet, so its labeling set is finite; its value is the maximum over labelings, and an edgeless graph has value one. (Constraint graph and labeling value)

[F2]

The code C is injective, so every valid block has a unique decoded label. (Shared codeword blocks and edge acceptance circuits)

[F3]

The robust edge circuit accepts exactly when both blocks are valid and their decoded ordered labels satisfy the edge relation; on a loop it tests the diagonal pair. (Shared codeword blocks and edge acceptance circuits)

[F4]

The two-piece construction supplies an explicit Boolean assignment tester, of arity at most six, for the combined named list of the two pieces. (A two-piece constant-query PCP of proximity)

[F5]

By perfect completeness of an assignment tester, every accepted named input extends to an auxiliary labeling satisfying every local constraint. (Assignment tester and rejection ratio)

[F6]

Composition identifies the first and second named pieces coordinatewise with their endpoint blocks, including both pieces of a loop, and gives every other gadget variable a private edge name. (Composition of an edge system with an assignment tester)

[F7]

If the original graph is edgeless, the composition has the empty constraint list. (Composition of an edge system with an assignment tester)

[F8]

Local gadget labelings that agree on all identifications combine into one output labeling. (Composition of an edge system with an assignment tester)

[F9]

Each local constraint is copied uniformly, so an accepted local constraint remains accepted in every copy. (Composition of an edge system with an assignment tester)

[F10]

For each labeling, constraint-system value is its satisfied fraction and UNSAT⁡τ=1−val⁡τ; the overall unsatisfiability is the minimum over labelings, and the empty list has value one. (Assignment tester and rejection ratio)

Proof

Given: Assume val⁡(G)=1.

1.1F1F7F10givencases

If E(G)=∅, then G has value one by [F1], while the composition has an empty constraint list by [F7] and therefore val⁡(H)=1 by [F10] and the stated convention. It remains to consider E(G)≠∅.

1.2F1givenalgebrachoose

Because G has finitely many vertices and Σ is finite and nonempty, its set of labelings is finite and nonempty, so the maximum in [F1] is attained. Choose σ:V(G)→Σ with val⁡σ(G)=val⁡(G)=1. Since the edge set is nonempty, every edge relation is satisfied by its ordered pair of endpoint labels.

2.1F2F3step 1.2constructalgebra

For each active vertex v, assign its shared block the codeword Bv=C(σ(v)). By [F2] this is valid and decodes uniquely to σ(v). If e=(v,w), then σ satisfies its ordered relation, so [F3] says the robust circuit Ce accepts (Bv,Bw). If e is a loop, both pieces are the same block and the satisfied diagonal pair is accepted as well.

3.1F4F5step 2.1chooseconstruct

For each edge e, [F4] supplies the assignment tester on the combined named input list of its two raw pieces. The fixed input (Bv,Bw) is accepted by Ce by step 2.1. Its accepted-input clause in [F5] therefore gives at least one Boolean assignment be to the gadget's auxiliary variables satisfying every local constraint. Order that finite variable list as in the explicit tester output and take the lexicographically first such be. The edge set and each Boolean search space are finite, so this specifies the witnesses without an axiom of choice.

4.1F6F8F9step 2.1step 3.1constructalgebra

Assign each shared vertex block its fixed codeword and assign each edge-private variable its value from be. By [F6], the named pieces take the already fixed endpoint blocks, including the same block in both positions of a loop; all other variables are private to their edge. These local labelings agree on every identification, so [F8] combines them into an output labeling τ. Every local constraint accepts under its be, and [F9] preserves acceptance in every uniform copy. Thus every constraint of H accepts under τ.

5.1F10step 1.1step 4.1algebradischarge-construct∎

Every constraint of H is satisfied by τ, so val⁡τ(H)=1 and UNSAT⁡τ(H)=0 by [F10]. Hence UNSAT⁡(H)=0, and the stated convention gives val⁡(H)=1. Together with the edgeless case, this proves the claim.

Remarks

Dinur's proof of Lemma 1.8 extends each satisfied original edge to a satisfying assignment of its local gadget and uses equal gadget sizes to average the local fractions. This lemma is its perfect-completeness direction specialized to a fully satisfiable input graph. The present composition uses the two-piece Boolean tester and the explicit least-common-multiple padding in Composition of an edge system with an assignment tester. The Arora–Barak proof of Lemma 18.30 makes the same witness extension for each satisfiable qCSP cluster. Both source arguments support the construction pattern; the local tester completeness used here is supplied and proved by A two-piece constant-query PCP of proximity.

Depends on

Used by

Dependency tree · two levels

20 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources