Alphabeta Math
TheoremStatement: 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.

Metric derivatives of AC2 Wasserstein curves

Statement

Assume the Axiom of Choice. If T>0 and μ∈AC2([0,T];P2(Rd)), then the metric derivative ∣μ˙t∣=lim⁡h→0W2(μt+h,μt)∣h∣ exists for almost every t∈(0,T), is measurable and belongs to L2(0,T). For all 0≤s≤t≤T, W2(μs,μt)≤∫st∣μ˙r∣dr. It is the least nonnegative square-integrable upper control: if g∈L2(0,T) and W2(μs,μt)≤∫stg(r)dr for all s≤t, then ∣μ˙t∣≤g(t) almost everywhere.

Facts & Assumptions

Given: AC and a curve μ∈AC2([0,T];P2(Rd)) with an L2 upper control g.

[F1]

A real absolutely continuous function on a compact interval is differentiable almost everywhere and equals its initial value plus the integral of its derivative; the indefinite integral of an L1 function has derivative equal to that integrand almost everywhere. These scalar facts require Countable and Dependent Choice. Fundamental theorem of calculus for absolutely continuous functions The indefinite integral of an L1 function is differentiable almost everywhere The Axiom of Countable Choice (ACω) The axiom of dependent choice: a relation in which every element is related to something admits an N-indexed chain

[F2]

AC implies the countable- and dependent-choice principles in [F1]. The Axiom of Choice AC implies DC implies countable choice

[F4]

Holder on the finite interval gives g∈L1(0,T) from g∈L2(0,T). Holder's inequality for integrals, including the endpoint cases

[F3]

The Wasserstein distance satisfies the triangle inequality, hence the reverse triangle inequality for distance to a fixed point. The quadratic Wasserstein distance is a metric

Proof

1.1givenF1F2F3F4

Let D=Q∩[0,T] together with 0,T. For each q∈D, set fq(t)=W2(μt,μq). The reverse triangle inequality and the assumed control give ∣fq(t)−fq(s)∣≤∫stg(r)dr, so fq is absolutely continuous. By [F1], it has a derivative almost everywhere, and because g∈L1 by Holder, at common differentiability points of fq and G(t)=∫0tg(r)dr, difference quotients give ∣fq′(t)∣≤g(t). Since D is countable, outside one null set all these derivatives exist and are bounded by g. Define m(t)=sup⁡q∈D∣fq′(t)∣ there and set m=0 on the null set. Difference quotients along h=1/n give measurable versions of the scalar derivatives, so m is measurable; also 0≤m≤g almost everywhere and hence m∈L2(0,T).

2.1F1F3step 1.1

For any s,t, continuity of the curve and density of D imply W2(μs,μt)=sup⁡q∈D∣fq(t)−fq(s)∣. The inequality from right to left is the reverse triangle inequality. For the other direction take qj∈D with qj→s; then fqj(s)→0 and fqj(t)→W2(μt,μs). For each q, [F1] and step 1.1 give ∣fq(t)−fq(s)∣≤∫st∣fq′(r)∣dr≤∫stm(r)dr. Taking the supremum proves W2(μs,μt)≤∫stm(r)dr.

3.1F1step 1.1step 2.1

The function H(t)=∫0tm(r)dr is absolutely continuous, and [F1] gives H′(t)=m(t) almost everywhere. At a time t where H is differentiable and every fq is differentiable, step 2.1 gives lim sup⁡h→0W2(μt+h,μt)∣h∣≤m(t), while W2(μt+h,μt)≥∣fq(t+h)−fq(t)∣ gives a lower limit at least ∣fq′(t)∣ for every q. Taking the supremum yields a lower limit at least m(t). Thus the metric derivative exists and equals m(t) almost everywhere. The interval bound and square integrability follow from step 2.1 and m≤g.

4.1F1step 1.1step 3.1∎

Let h be any other nonnegative L2 upper control. Repeating the scalar difference-quotient bound of step 1.1 with h gives ∣fq′(t)∣≤h(t) almost everywhere for each q. Countability of D lets these inequalities hold off a single null set, so m(t)≤h(t) there. By step 3.1, ∣μ˙t∣=m(t) almost everywhere, proving leastness.

Depends on

Used by

Dependency tree · two levels

54 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