Alphabeta Math
TheoremStatement: AI-adaptedProof: AI-adaptedPipeline-generatedprecheck passjudge 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.

Unique finite gravitational Gibbs variational optimum

Statement

Under the hypotheses of Finite gravitational Gibbs normalization and differentiable moments, let ρ:ΓN→[0,∞) be a measurable probability density relative to νN, so ∫ΓNρ dνN=1, satisfying ∫ΓN∣ρlog⁡ρ∣ dνN<∞,∫ΓN∣HN,a∣ρ dνN<∞, with 0log⁡0=0. For any probability density σ with integrable energy, define its mean energy by ⟨HN,a⟩σ:=∫ΓNHN,aσ dνN. Define S(ρ)=−kB∫ΓNρlog⁡ρ dνN and D(ρ∥pβ):=∫ΓNρlog⁡(ρ/pβ) dνN, with 0log⁡(0/pβ):=0. Then S(ρ)−kBβ⟨HN,a⟩ρ≤kBlog⁡ZN, with equality exactly when ρ=pβ almost everywhere. Among densities with the same mean energy as pβ, this makes pβ the unique entropy maximizer.

For every bounded measurable real u with ∫pβu dνN=0, ∫HN,apβu dνN=0, and u≠0 in L2(pβdνN), the perturbation ρt=pβ(1+tu) is a probability density of the same mean energy for ∣t∣∥u∥∞≤1/2. For sufficiently small nonzero t, S(ρt)<S(pβ),D(ρt∥pβ)=t22∫pβu2 dνN+Rt,∣Rt∣≤23∣t∣3∥u∥∞∫pβu2 dνN. This is a variational claim in the stated density class, not a dynamical equilibration theorem.

Facts & Assumptions

Given: The positive normalized density pβ, its finite energy moments, and the reference-relative entropy convention.

[F1]

For the softened finite model, pβ=ZN−1e−βHN,a is positive and normalized, and its energy moments are finite. (Finite gravitational Gibbs normalization and differentiable moments)

[F2]

The phase-reference item defines S(ρ)=−kB∫ρlog⁡ρ dνN for probability densities relative to the dimensionless measure νN. (Finite gravitational phase and entropy reference)

[F3]

A nonnegative measurable function with integral zero vanishes almost everywhere. (A nonnegative measurable function has integral 0 exactly when it vanishes almost everywhere)

Proof

1.1algebra

For s>0, ϕ(s)=slog⁡s−s+1 has derivative log⁡s, is decreasing for 0<s<1, and increasing for s>1. Its continuous value at s=0 is 1, so ϕ(s)≥0 with equality exactly at s=1.

2.1F1step 1.1algebra

Since pβ>0, set r=ρ/pβ almost everywhere. The assumptions and log⁡pβ=−βHN,a−log⁡ZN make ρlog⁡(ρ/pβ) integrable. Expanding ϕ and using ∫ρ=∫pβ=1 gives D(ρ∥pβ)=∫pβϕ(r) dνN=∫ρlog⁡ρ dνN+β⟨HN,a⟩ρ+log⁡ZN≥0.

3.1F1F2F3step 2.1algebra

Rearranging step 2.1 and using the entropy convention [F2] gives the stated inequality. Equality forces ϕ(r)=0 almost everywhere for the probability measure pβdνN, so r=1 almost everywhere by [F3], and ρ=pβ. Conversely, if ρ=pβ almost everywhere then D(ρ∥pβ)=0 in step 2.1 and the inequality is an equality. At equal mean energy this is the unique entropy-maximizing density.

4.1F1F2step 2.1algebra∎

Put ψ(x)=(1+x)log⁡(1+x)−x. For ∣x∣≤1/2, ψ(0)=ψ′(0)=0 and ψ′′(x)=1/(1+x), hence ∣ψ′′(x)−1∣≤2∣x∣. Integrating this second-derivative bound twice from zero gives ∣ψ(x)−x2/2∣≤∣x∣3/3≤2∣x∣3/3. The two centering conditions make ρt normalized and energy preserving, while ∣tu∣≤1/2 makes it nonnegative. Moreover [F1] gives ∫∣HN,a∣pβ<∞ and ∫pβ∣log⁡pβ∣<∞ by log⁡pβ=−βHN,a−log⁡ZN; since 1+tu and log⁡(1+tu) are bounded, these imply ∫∣HN,a∣ρt<∞ and ∫∣ρtlog⁡ρt∣<∞. Thus ρt belongs to the stated density class. Therefore D(ρt∥pβ)=∫pβψ(tu) dνN=t22∫pβu2 dνN+Rt with the stated bound. Since u is nonzero in L2(pβdνN), the quadratic term dominates the cubic bound for sufficiently small nonzero t. The equal-energy identity in step 2.1 then gives S(ρt)<S(pβ).

Depends on

Dependency tree · 2 levels

Sources