TheoremBase

A Toolkit for Penalised Comparison on the Noise Wasserstein Space: Constant Test Functions and Linear Combinations of Test Functions, the Identity as a Unique Noise-Optimal Map, and Discrepancies Along the Push-Forward Under (id, id) and Along Glued Couplings

Constant functions are noise intrinsic test functions with zero gradient, and real linear combinations of noise intrinsic test functions are again such functions, with the corresponding linear combination of gradients; the identity is a noise-optimal map from a measure to itself and the push-forward of the measure under (id, id) is its only noise-optimal self-coupling; the discrepancy of two noise fields along that coupling is their squared distance; and the square roots of discrepancies satisfy a triangle inequality along a glued composite coupling.

Statement

In the setting of First-Order Equations on the Noise Wasserstein Space Relative to a Noise Penalty Pair: Standing Notation, let id\mathrm{id} be the identity map of XX (First-Order Equations on the Noise Wasserstein Space Relative to a Noise Penalty Pair: Standing Notation §maps). For ν∈P(X)\nu\in\mathcal{P}(X), the zero element 0ν0_{\nu} of the real Hilbert space L2(ν;Xa)L^{2}(\nu;X^{a}) of Probability Measures on a Hilbert Space Transported in the Noise Norm: Standing Notation §fields is the class of the constant map with value 0X0_{X}, by The Space of Square-Integrable Hilbert-Valued Maps is a Real Hilbert Space: Coordinates and Synthesis §hilbert, 0X0_{X} being the zero vector of XaX^{a} by The Noise Space is a Real Hilbert Space: Orthonormal Basis, Continuous Embedding, Partial Sums, Closed Balls and Borel Measurability §hilbert; the map (id,id):X→X×X(\mathrm{id},\mathrm{id}):X\to X\times X of Borel Probability Measures on a Real Hilbert Space with an Orthonormal Basis: Standing Notation §pairs is Borel by Borel Sets of a Hilbert Space with an Orthonormal Basis: Coordinates, Determination by Finite-Dimensional Projections, and Pairs §pairing, and (id,id)#ν(\mathrm{id},\mathrm{id})_{\#}\nu is its push-forward of Borel Probability Measures on a Real Hilbert Space with an Orthonormal Basis: Standing Notation §pushforward. For ν,μ∈P(X)\nu,\mu\in\mathcal{P}(X), π∈Π(ν,μ)\pi\in\Pi(\nu,\mu), q∈L2(ν;Xa)q\in L^{2}(\nu;X^{a}) and η∈L2(μ;Xa)\eta\in L^{2}(\mu;X^{a}), the discrepancy ∫X×X∣q(x)−η(y)∣a2 π(dz)\int_{X\times X}|q(x)-\eta(y)|_{a}^{2}\,\pi(dz) is the nonnegative real number of that clause, and square roots are the nonnegative square roots of Existence and Uniqueness of the Nonnegative Square Root. Noise intrinsic test functions, noise-optimal maps with their displacements, noise-optimal couplings and uniquely noise-mapped pairs are those of First-Order Equations on the Noise Wasserstein Space Relative to a Noise Penalty Pair: Standing Notation §gradients and First-Order Equations on the Noise Wasserstein Space Relative to a Noise Penalty Pair: Standing Notation §maps. In claim 5, X(3)X_{(3)}, the maps q1,q2,q3q_{1},q_{2},q_{3} on it, gluings and the composite coupling are those of Gluing Two Couplings on a Hilbert Space over a Common Middle Marginal, and the Triangle Inequalities for the Quadratic and Noise Costs; the letters q1,q2,q3q_{1},q_{2},q_{3} always denote these maps, and the letter qq without index a noise field. Then the following hold.

1. (Constant test functions) Let Q⊆PρaQ\subseteq\mathcal{P}^{a}_{\rho}, let c∈Rc\in\mathbb{R}, and let φc:Pρa→R\varphi_{c}:\mathcal{P}^{a}_{\rho}\to\mathbb{R} be the constant function with value cc. Then φc\varphi_{c} is a noise intrinsic test function on QQ, and ∇φc(μ)=0μ\nabla\varphi_{c}(\mu)=0_{\mu} for every μ∈Q\mu\in Q.

2. (Linear combinations) Let Q⊆PρaQ\subseteq\mathcal{P}^{a}_{\rho}, let φ,ψ:Pρa→R\varphi,\psi:\mathcal{P}^{a}_{\rho}\to\mathbb{R} be noise intrinsic test functions on QQ, let s,t∈Rs,t\in\mathbb{R}, and let sφ+tψ:Pρa→Rs\varphi+t\psi:\mathcal{P}^{a}_{\rho}\to\mathbb{R} be the function with value sφ(ν)+tψ(ν)s\varphi(\nu)+t\psi(\nu) at ν\nu. Then sφ+tψs\varphi+t\psi is a noise intrinsic test function on QQ, and

∇(sφ+tψ)(μ)=s ∇φ(μ)+t ∇ψ(μ)for every μ∈Q,\nabla(s\varphi+t\psi)(\mu)=s\,\nabla\varphi(\mu)+t\,\nabla\psi(\mu)\qquad\text{for every }\mu\in Q,

the right-hand side being formed in the real vector space L2(μ;Xa)L^{2}(\mu;X^{a}).

3. (The identity) Let μ∈Pρa\mu\in\mathcal{P}^{a}_{\rho}. Then id\mathrm{id} is a noise-optimal map from μ\mu to μ\mu, and its displacement id−id\mathrm{id}-\mathrm{id} is 0μ0_{\mu}; the coupling (id,id)#μ(\mathrm{id},\mathrm{id})_{\#}\mu is the only noise-optimal coupling of μ\mu and μ\mu; consequently the ordered pair (μ,μ)(\mu,\mu) is uniquely noise-mapped; and Wa(μ,μ)=0W_{a}(\mu,\mu)=0.

4. (Discrepancies along (id,id)#ν(\mathrm{id},\mathrm{id})_{\#}\nu) Let ν∈P(X)\nu\in\mathcal{P}(X), let q,η∈L2(ν;Xa)q,\eta\in L^{2}(\nu;X^{a}), and let Δ=(id,id)#ν\Delta=(\mathrm{id},\mathrm{id})_{\#}\nu. Then Δ∈Πa(ν,ν)⊆Π(ν,ν)\Delta\in\Pi^{a}(\nu,\nu)\subseteq\Pi(\nu,\nu), Ia(Δ)=0I^{a}(\Delta)=0, and

∫X×X∣q(x)−η(y)∣a2 Δ(dz)=∥q−η∥ν2.\int_{X\times X}|q(x)-\eta(y)|_{a}^{2}\,\Delta(dz)=\lVert q-\eta\rVert_{\nu}^{2}.

5. (Discrepancies along a glued composite coupling) Let ν,λ,μ∈P2(X)\nu,\lambda,\mu\in\mathcal{P}_{2}(X), the set of The Second Moment of a Borel Probability Measure on a Hilbert Space and the Probability Measures with Finite Second Moment §space; let π12∈Π(ν,λ)\pi_{12}\in\Pi(\nu,\lambda) and π23∈Π(λ,μ)\pi_{23}\in\Pi(\lambda,\mu), let σ\sigma be a gluing of π12\pi_{12} and π23\pi_{23}, and let π13=(q1,q3)#σ\pi_{13}=(q_{1},q_{3})_{\#}\sigma, which belongs to Π(ν,μ)\Pi(\nu,\mu) by Gluing Two Couplings on a Hilbert Space over a Common Middle Marginal, and the Triangle Inequalities for the Quadratic and Noise Costs §composite, both applied with ν,λ,μ\nu,\lambda,\mu in place of its μ,λ,ν\mu,\lambda,\nu. Let q∈L2(ν;Xa)q\in L^{2}(\nu;X^{a}), η∈L2(λ;Xa)\eta\in L^{2}(\lambda;X^{a}) and θ∈L2(μ;Xa)\theta\in L^{2}(\mu;X^{a}). Then

∫X×X∣q(x)−θ(y)∣a2 π13(dz) ≤ ∫X×X∣q(x)−η(y)∣a2 π12(dz)+∫X×X∣η(x)−θ(y)∣a2 π23(dz).\sqrt{\int_{X\times X}|q(x)-\theta(y)|_{a}^{2}\,\pi_{13}(dz)}\ \le\ \sqrt{\int_{X\times X}|q(x)-\eta(y)|_{a}^{2}\,\pi_{12}(dz)}+\sqrt{\int_{X\times X}|\eta(x)-\theta(y)|_{a}^{2}\,\pi_{23}(dz)} .

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…