TheoremBase

The Wick-Square Corrector and the Score-Paired Wick-Square Cost Relative to a Diagonal Gaussian Measure on a Hilbert Space

For diagonal Wick couplings, the Wick-square corrector is a diagonal quadratic profile, and the Wick-square cost of a measure in the score domain is defined as the pairing of the Gaussian score with the corrector's gradient.

Statement

In the setting of A Diagonal Gaussian Reference Measure on the Noise Wasserstein Space, Rescaled Heads and Gaussian Tails: Standing Notation, so that the reference measure is ρ=γc\rho=\gamma_{c}, let β,κ∈R\beta,\kappa\in\mathbb{R} be positive with ck≤κ akc_{k}\le\kappa\,a_{k} for every k∈Nk\in\mathbb{N}, and let (D,DΣ,E,Σ)(\mathcal{D},\mathcal{D}_{\Sigma},\mathcal{E},\Sigma) be the Gaussian entropy pair with temperature β\beta, whose hypothesis holds with this κ\kappa; thus Σ(ν)=β Zνa\Sigma(\nu)=\beta\,Z^{a}_{\nu} for ν∈DΣ\nu\in\mathcal{D}_{\Sigma}, where ZνaZ^{a}_{\nu} is the noise score field. ⟨⋅,⋅⟩ν\langle\cdot,\cdot\rangle_{\nu} is the inner product of L2(ν;Xa)L^{2}(\nu;X^{a}), and convergent series are those of that definition.

1. (Hypothesis on the couplings) The corrector and the cost below are defined for a sequence of Wick couplings: a real sequence w=(wk)k∈Nw=(w_{k})_{k\in\mathbb{N}} such that the series ∑k=1∞∣wk∣ ck2/ak\sum_{k=1}^{\infty}|w_{k}|\,c_{k}^{2}/a_{k} converges and there is W∈RW\in\mathbb{R} with ∣wk∣ ck≤W|w_{k}|\,c_{k}\le W for every k∈Nk\in\mathbb{N}; such a WW is called a bound for ww.

2. (The corrector) Let ww be a sequence of Wick couplings with bound WW, and let ϖ=(ϖk)k∈N\varpi=(\varpi_{k})_{k\in\mathbb{N}} with ϖk=wkck/(βak)\varpi_{k}=w_{k}c_{k}/(\beta a_{k}). Then ∑k=1∞∣ϖk∣ ck\sum_{k=1}^{\infty}|\varpi_{k}|\,c_{k} converges, being β−1\beta^{-1} times the series of clause 1, and ∣ϖk∣ ak=β−1∣wk∣ ck≤β−1W|\varpi_{k}|\,a_{k}=\beta^{-1}|w_{k}|\,c_{k}\le\beta^{-1}W, so ϖ\varpi is admissible with bound β−1W\beta^{-1}W. The Wick-square corrector with couplings ww is the diagonal quadratic profile Πw=Φϖ:Pρa→R\Pi_{w}=\Phi_{\varpi}:\mathcal{P}^{a}_{\rho}\to\mathbb{R}, which is a noise intrinsic test function on D\mathcal{D}, with gradient along noise couplings ∇Πw(ν)=Vϖ(ν)\nabla\Pi_{w}(\nu)=V_{\varpi}(\nu) at ν∈D\nu\in\mathcal{D}, by Diagonal Quadratic Profiles on the Noise Wasserstein Space: Moment Bounds, a Lipschitz Estimate, the Gradient Field, and Noise Intrinsic Test Functions §test, since D⊆Pρa\mathcal{D}\subseteq\mathcal{P}^{a}_{\rho} by The Gaussian Entropy Pair on the Noise Wasserstein Space: Relative Entropy and the Noise Score Field §penalty-domain.

3. (The score-paired Wick-square cost) Let ww be a sequence of Wick couplings. For ν∈DΣ\nu\in\mathcal{D}_{\Sigma} both Σ(ν)\Sigma(\nu) and ∇Πw(ν)\nabla\Pi_{w}(\nu) lie in L2(ν;Xa)L^{2}(\nu;X^{a}), by The Gaussian Entropy Pair on the Noise Wasserstein Space: Relative Entropy and the Noise Score Field §pair and by Diagonal Quadratic Profiles on the Noise Wasserstein Space: Moment Bounds, a Lipschitz Estimate, the Gradient Field, and Noise Intrinsic Test Functions §field as used in clause 2. The score-paired Wick-square cost with couplings ww is the function

Gw:DΣ→R,Gw(ν)=⟨Σ(ν),∇Πw(ν)⟩ν=β ⟨Zνa,∇Πw(ν)⟩ν.G_{w}:\mathcal{D}_{\Sigma}\to\mathbb{R},\qquad G_{w}(\nu)=\langle\Sigma(\nu),\nabla\Pi_{w}(\nu)\rangle_{\nu}=\beta\,\langle Z^{a}_{\nu},\nabla\Pi_{w}(\nu)\rangle_{\nu}.

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…