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.
A W2 limit in the two-homogeneous case is globally minimizing
Statement
Assume AC, the full two-homogeneous hypotheses and separated compact initialization in the preceding global-minimizer theorem. If the Wasserstein gradient flow converges in to , then is a global minimizer of over finite nonnegative measures.
Statement references
Facts & Assumptions
Given: AC, the assumptions in the Statement, and .
Assume the Axiom of Choice, as stated in the hypotheses and required by the flow and coupling suppliers. The Axiom of Choice
W2 convergence implies weak convergence of the two-homogeneous projections. W2 convergence implies weak convergence of homogeneous projections
The projected-limit theorem makes its weak limit a global minimizer and states that converges to the energy of that limit. A separating two-homogeneous flow has globally minimizing projection limits
For a positive two-homogeneous feature and regularizer, whenever has finite second moment. Feature and energy identities for homogeneous projections
Bounded continuous real tests determine Borel probability measures on any metric space. Weak limits are unique
Proof
Given: AC and the data in the Statement.
Apply [F1] to obtain weak convergence .
The projected-limit theorem [F2] applies to this weak limit. If it denotes the limit by , then for every continuous test on the compact sphere both and give the limit of the same integrals . Testing shows that both have the same total mass . If , both finite nonnegative measures are zero. If , then and are Borel probability measures with equal integrals against every bounded continuous test, so [F4] gives . Viewed as a measure on , this limit globally minimizes over all finite nonnegative Borel measures, exactly as asserted by the theorem.
By [F3], the feature and regularizer integrals of equal those of its two-homogeneous projection, so . By step 2.1, this equals the global minimum of over finite nonnegative Borel measures on , and therefore is itself a global minimizer.
Depends on
Used by
Nothing in the library uses this result yet.
Dependency tree · two levels
22 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
- Lénaïc Chizat and Francis Bach, On the Global Convergence of Gradient Descent for Over-parameterized Models using Optimal Transport (standard reference, not scraped)
- Luigi Ambrosio, Nicola Gigli, and Giuseppe Savaré, Gradient Flows in Metric Spaces and in the Space of Probability Measures (standard reference, not scraped)