Alphabeta Math
LemmaStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedjudge pass (gpt-6-luna)
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.

Finite maximal coupling and heat-bath stationary law

Statement

For probability vectors p,q on a finite set A, there is a coupling (X,Y) with Pr⁡(X≠Y)=12∑a∈A∣p(a)−q(a)∣. For a finite region Λ with strictly positive single-site Gibbs conditionals, the random-scan heat-bath chain has the finite-volume Gibbs law as its unique stationary distribution. This chain is an auxiliary finite-state proof device, not a physical time evolution.

Facts & Assumptions

Given: A finite state alphabet and the normalized finite-volume conditional laws of Conditional Gibbs kernel consistency.

[F1]

The finite-volume conditional Gibbs laws are positive probability kernels consistent under nested regions (Conditional Gibbs kernel consistency).

Proof

1.1givenalgebra

Put m(a)=min⁡{p(a),q(a)} and r=1−∑am(a)=12∑a∣p(a)−q(a)∣. Couple the common mass m(a) on the diagonal. If r>0, couple the residual masses p−m and q−m by their product divided by r; their supports are disjoint, so this part always disagrees. If r=0, the diagonal coupling already has the required marginals.

1.2F1algebra

On the finite configuration space AΛ, choose a site uniformly and resample its spin from the conditional Gibbs law. The finite-volume Gibbs law is invariant under each such update because it resamples from its own conditional distribution; hence it is stationary for the random-scan mixture.

2.1F1step 1.2algebra∎

Let N=∣Λ∣>0 and fix a target configuration. Since there are finitely many strictly positive conditional probabilities, their minimum p∗>0. In N updates, the probability of selecting a fixed ordering of all sites is N−N, and conditional on that ordering the probability of setting every spin to its target is at least p∗N. Thus every starting state reaches the target with probability at least ε=N−Np∗N>0. The N-step kernel has a common component εδtarget, so it contracts total variation by at most 1−ε; two stationary laws must coincide. If Λ=∅, the configuration space is a singleton and uniqueness is immediate.

Depends on

Dependency tree · 2 levels

Sources