TheoremBase

Intrinsic Test Functions on the Torus Wasserstein Space

An intrinsic test function on a set of probability measures on the torus is a continuous function that is differentiable along couplings at every measure of the set with gradient in the tangent space, the gradients varying continuously along couplings of vanishing torus cost.

Statement

In the setting of Optimal Transport on the Flat Torus: Standing Notation, let K⊆P(Td)\mathcal{K}\subseteq\mathcal{P}(\mathbb{T}^{d}) and φ:P(Td)→R\varphi:\mathcal{P}(\mathbb{T}^{d})\to\mathbb{R}. Continuity of φ\varphi is understood between the metric space (P(Td),WT)(\mathcal{P}(\mathbb{T}^{d}),W_{\mathbb{T}}) of Optimal Transport on the Flat Torus: Standing Notation §measures and R\mathbb{R} with the absolute-value metric. Differentiability along couplings and the gradient along couplings ∇φ(μ)\nabla\varphi(\mu) are those of that definition, and TμT_{\mu} is the tangent space.

(Intrinsic test function) The function φ\varphi is an intrinsic test function on K\mathcal{K} if it has the following three properties.

(a) φ\varphi is continuous on P(Td)\mathcal{P}(\mathbb{T}^{d}).

(b) For every μ∈K\mu\in\mathcal{K}, φ\varphi is differentiable along couplings at μ\mu and ∇φ(μ)∈Tμ\nabla\varphi(\mu)\in T_{\mu}.

(c) Let μ∈K\mu\in\mathcal{K}, let (μn)n∈N(\mu_{n})_{n\in\mathbb{N}} be a sequence in K\mathcal{K}, and let γn∈Π(μn,μ)\gamma_{n}\in\Pi(\mu_{n},\mu) for n∈Nn\in\mathbb{N} be such that (IT(γn))n∈N(I_{\mathbb{T}}(\gamma_{n}))_{n\in\mathbb{N}} converges to 00. Then the sequence of real numbers

(∫Rd+d∥∇φ(μn)(pr1(w))−∇φ(μ)(pr2(w))∥2 γn(dw))n∈N\Bigl(\int_{\mathbb{R}^{d+d}}\bigl\lVert\nabla\varphi(\mu_{n})(\mathrm{pr}_{1}(w))-\nabla\varphi(\mu)(\mathrm{pr}_{2}(w))\bigr\rVert^{2}\,\gamma_{n}(dw)\Bigr)_{n\in\mathbb{N}}

converges to 00; each term is a real number, being at most 2∥∇φ(μn)∥μn2+2∥∇φ(μ)∥μ22\lVert\nabla\varphi(\mu_{n})\rVert_{\mu_{n}}^{2}+2\lVert\nabla\varphi(\mu)\rVert_{\mu}^{2} by the inequality of Pairs of Euclidean Points: Coordinate Projections, Pairings, the Product Measure on a Euclidean Space, Borel Norm Functions and Finite Sets §functions and the change-of-variables formula, and it does not depend on the representatives chosen, by The Lebesgue Integral and Null Sets: Almost-Everywhere Comparison, Markov's Inequality, and Dominated Convergence Almost Everywhere §comparison.

Citations

Loading…

Dependencies

Loading…

Related

0 relations

Curated associations between results. These are editable and subjective — they do not replace the dependency graph, which is derived from the references in the text.

No relations recorded yet.

Comments

Log in to comment.

Loading…