TheoremBase

A Convergent Series of Squared Wasserstein Distances to Measures of Bounded Second Moment is an Intrinsic Test Function

A convergent positive series of squared Wasserstein distances to measures with uniformly bounded second moments defines an intrinsic test function on any set with the map property; its gradient is the corresponding series of displacement fields, bounded by twice the series of distances, and its translation Hessian is twice the total weight times the identity.

Statement

In the setting of The Intrinsic Calculus on the Wasserstein Space: Standing Notation, intrinsic test functions on a subset of P2(Rd)\mathcal{P}_{2}(\mathbb{R}^{d}), their gradients along couplings and their translation Hessians are those of that definition; optimal maps are those of the setting and the map property is that of that definition, and for an optimal map SS from μ\mu to ν\nu the class id−S∈L2(μ;Rd)\mathrm{id}-S\in L^{2}(\mu;\mathbb{R}^{d}) is that of The Optimal Map as a Square-Integrable Vector Field: Integrability, Transport Cost and Uniqueness of the Class §square-integrable; convergent series of real numbers and convergent series in the real Hilbert space L2(μ;Rd)L^{2}(\mu;\mathbb{R}^{d}) are those of those items; and IdI_{d} is the identity matrix of Real Matrices, Symmetric Matrices and the Semidefinite Ordering: Standing Notation §matrices.

Let Q⊆P2(Rd)Q\subseteq\mathcal{P}_{2}(\mathbb{R}^{d}) have the map property, let B∈RB\in\mathbb{R}, let (μk)k∈N(\mu_{k})_{k\in\mathbb{N}} be a sequence in P2(Rd)\mathcal{P}_{2}(\mathbb{R}^{d}) with M2(μk)≤BM_{2}(\mu_{k})\le B for every k∈Nk\in\mathbb{N}, and let (ck)k∈N(c_{k})_{k\in\mathbb{N}} be a sequence of positive real numbers whose series converges, with sum c=∑k=1∞ckc=\sum_{k=1}^{\infty}c_{k}. Then the following hold.

1. (Convergence) For every μ∈P2(Rd)\mu\in\mathcal{P}_{2}(\mathbb{R}^{d}) the series ∑k=1∞ck W2(μ,μk)2\sum_{k=1}^{\infty}c_{k}\,W_{2}(\mu,\mu_{k})^{2} and ∑k=1∞ck W2(μ,μk)\sum_{k=1}^{\infty}c_{k}\,W_{2}(\mu,\mu_{k}) converge.

In the remaining clauses ψ:P2(Rd)→R\psi:\mathcal{P}_{2}(\mathbb{R}^{d})\to\mathbb{R} is the function, defined by clause 1, given by

ψ(μ)=∑k=1∞ck W2(μ,μk)2.\psi(\mu)=\sum_{k=1}^{\infty}c_{k}\,W_{2}(\mu,\mu_{k})^{2}.

2. (Test function) ψ\psi is an intrinsic test function on QQ.

3. (Gradient) Let μ∈Q\mu\in Q and, for every k∈Nk\in\mathbb{N}, let SkS_{k} be an optimal map from μ\mu to μk\mu_{k}. Then the series ∑k=1∞2ck (id−Sk)\sum_{k=1}^{\infty}2c_{k}\,(\mathrm{id}-S_{k}) converges in L2(μ;Rd)L^{2}(\mu;\mathbb{R}^{d}) to the gradient along couplings ∇ψ(μ)\nabla\psi(\mu), which exists by clause 2, and

∥∇ψ(μ)∥μ≤2∑k=1∞ck W2(μ,μk).\lVert\nabla\psi(\mu)\rVert_{\mu}\le2\sum_{k=1}^{\infty}c_{k}\,W_{2}(\mu,\mu_{k}).

4. (Translation Hessian) Hψ(μ)=2c IdH_{\psi}(\mu)=2c\,I_{d} for every μ∈P2(Rd)\mu\in\mathcal{P}_{2}(\mathbb{R}^{d}).

Proofs

Log in to submit a proof.

Loading...

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…