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.
The localized Wasserstein gradient flow exists, satisfies EVI, and is unique
Statement
Assume AC and the feature-measure energy hypotheses. Let and let have finite energy and be concentrated on . There is a unique curve in that starts at , solves the continuity equation with velocity equal almost everywhere to the negative minimum-norm element of , and satisfies Writing , it also satisfies the evolution variational inequality for each fixed and almost every (with the exceptional null set allowed to depend on ). Any two such localized flows obey .
Statement references
Feature-measure energy and its regularity assumptions
Wasserstein gradient flow for a feature-measure energy
The localized feature energy is semiconvex along generalized geodesics
The selected pointwise field is a strong Wasserstein subgradient
The Hilbert lift induces the localized Wasserstein continuity equation
Facts & Assumptions
Given: AC, , , and the model regularity hypotheses.
The localized lift is a proper lower-bounded semiconvex energy on a separable Hilbert space; its gradient flow exists uniquely for all time and satisfies the Hilbert energy identity. A lower-bounded semiconvex Hilbert energy generates a unique gradient flow The localized lifted feature energy is semiconvex
The Hilbert flow pushes forward to a finite-action continuity-equation solution and an Wasserstein curve with the displayed minimum pointwise velocity. The Hilbert lift induces the localized Wasserstein continuity equation
The model energy has the all-coupling strong subgradient inequality with coefficient . The selected pointwise field is a strong Wasserstein subgradient
Along a finite-action continuity equation, squared distance to a fixed measure has derivative bounded above by the velocity-pairing with any optimal coupling. Derivative of squared W2 distance along a finite-action continuity equation
Any two localized continuity-equation solutions with these velocities satisfy the stated W2 stability estimate. One-sided semiconvex characteristics give W2 uniqueness and stability
AC is the inherited axiom in [F1]–[F5]. The Axiom of Choice
Proof
Given: AC, , and the hypotheses in the Statement.
Form the label Hilbert space and the initial vector . This Euclidean Borel label space is countably generated, so its -vector space is separable under the generalized label-space interface. It has finite because . [F1] yields the unique Hilbert gradient flow for all . Define . By [F2] this curve solves the localized continuity equation with the prescribed minimum-norm velocity and is in .
The Hilbert energy identity in [F1] is . By [F2], almost everywhere and . The change-of-variables identity for the pushforward therefore gives the displayed Wasserstein energy identity.
Fix and a time where [F4] applies. Choose an optimal coupling of . The derivative estimate [F4] gives By [F3], . Combining the inequalities proves the EVI.
If and are two such solutions, [F5] gives their stability estimate; equal initial data imply for every , hence equality of the measures and uniqueness. The construction in step 1.1 proves existence.
Depends on
- Feature-measure energy and its regularity assumptions
- Wasserstein gradient flow for a feature-measure energy
- The localized feature energy is semiconvex along generalized geodesics
- The selected pointwise field is a strong Wasserstein subgradient
- The Hilbert lift induces the localized Wasserstein continuity equation
- One-sided semiconvex characteristics give W2 uniqueness and stability
- Derivative of squared W2 distance along a finite-action continuity equation
- Superposition for finite-action Euclidean continuity equations
- Finite-action continuity equations generate AC2 Wasserstein curves
- AC2 curves admit a unique minimum-action velocity field
- The Axiom of Choice
- A lower-bounded semiconvex Hilbert energy generates a unique gradient flow
- The localized lifted feature energy is semiconvex
Used by
Dependency tree · two levels
88 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, Gigli, and Giuseppe Savaré, Gradient Flows in Metric Spaces and in the Space of Probability Measures (standard reference, not scraped)