TheoremBase

The Mehler Law Flow: Duality, Finite Second Moments, Invariance of the Gaussian Measure, Generation by the Noise Ornstein-Uhlenbeck Functional, and Decrease of Relative Entropy

For a probability measure μ\mu with finite second moment, the Mehler law flow μPt\mu P_t stays in P2(X)\mathcal{P}_2(X), is a flow leaving γc\gamma_c invariant, is dual to the Mehler semigroup on cylindrical functions of at most quadratic growth, has time derivative given by minus the noise Ornstein-Uhlenbeck functional, and does not increase relative entropy with respect to γc\gamma_c.

Statement

In the setting of Gaussian Analysis Relative to a Diagonal Gaussian Reference Measure with Noise Weights: Standing Notation, with the set FCpol(X)\mathcal{F}C_{\mathrm{pol}}(X) of continuous cylindrical functions of polynomial growth and their representations, the Mehler semigroup PtP_{t} and the Mehler law flow (μPt)t≥0(\mu P_{t})_{t\ge0} of The Mehler Law Flow of a Probability Measure on a Hilbert Space; the Ornstein-Uhlenbeck operator LaL^{a}; and, for ν∈P2(X)\nu\in\mathcal{P}_{2}(X), n∈Nn\in\mathbb{N} and g∈Cb2(Rn)g\in C^{2}_{b}(\mathbb{R}^{n}), the set of Bounded Twice Continuously Differentiable Functions with Bounded First and Second Partial Derivatives on Euclidean Space §bounded with q=nq=n, the noise Ornstein-Uhlenbeck functional Lνa(g)L^{a}_{\nu}(g). Let μ∈P2(X)\mu\in\mathcal{P}_{2}(X) and let s,t∈Rs,t\in\mathbb{R} with 0≤s0\le s and 0≤t0\le t.

1. (Flow) μPt∈P2(X)\mu P_{t}\in\mathcal{P}_{2}(X), μP0=μ\mu P_{0}=\mu, (μPs)Pt=μPs+t(\mu P_{s})P_{t}=\mu P_{s+t}, and γcPt=γc\gamma_{c}P_{t}=\gamma_{c}.

2. (Duality) Let F∈FCpol(X)F\in\mathcal{F}C_{\mathrm{pol}}(X) have a representation (n,ψ,B,q)(n,\psi,B,q) with q≤2q\le2. Then FF is integrable with respect to μPt\mu P_{t}, PtFP_{t}F is integrable with respect to μ\mu, and

∫XF d(μPt)=∫XPtF dμ.\int_{X}F\,d(\mu P_{t})=\int_{X}P_{t}F\,d\mu .

3. (The operator and the functional) For every ν∈P2(X)\nu\in\mathcal{P}_{2}(X), n∈Nn\in\mathbb{N} and g∈Cb2(Rn)g\in C^{2}_{b}(\mathbb{R}^{n}), the function La(g∘pn)L^{a}(g\circ p_{n}) is integrable with respect to ν\nu, and

Lνa(g)=−∫XLa(g∘pn) dν.L^{a}_{\nu}(g)=-\int_{X}L^{a}(g\circ p_{n})\,d\nu .

4. (Generator of the law flow) Let n∈Nn\in\mathbb{N} and g∈Cb2(Rn)g\in C^{2}_{b}(\mathbb{R}^{n}), and let Φ(r)=∫Xg∘pn d(μPr)\Phi(r)=\int_{X}g\circ p_{n}\,d(\mu P_{r}) for real r≥0r\ge0. Then

lim⁡h→0Φ(t+h)−Φ(t)h=−LμPta(g),\lim_{h\to0}\frac{\Phi(t+h)-\Phi(t)}{h}=-L^{a}_{\mu P_{t}}(g),

the limit being taken over real h≠0h\ne0 with t+h≥0t+h\ge0.

5. (Relative entropy) If μ\mu has finite relative entropy with respect to γc\gamma_{c}, then so has μPt\mu P_{t}, and H(μPt ∣ γc)≤H(μ ∣ γc)H(\mu P_{t}\,|\,\gamma_{c})\le H(\mu\,|\,\gamma_{c}).

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…