TheoremBase

Gaussian Integration by Parts for Functions of Finitely Many Coordinates Relative to a Diagonal Gaussian Measure on a Hilbert Space

For a function of the first n coordinates with a gradient of at most linear growth, the integral of a coordinate derivative against a diagonal Gaussian measure equals the integral of the function times the coordinate divided by its variance, and the derivative along a Cameron-Martin vector integrates to the integral of the function times the Paley-Wiener functional.

Statement

In the settings of Borel Probability Measures on a Real Hilbert Space with an Orthonormal Basis: Standing Notation and Differential Calculus and Convexity on Euclidean Open Sets: Standing Notation, with the coordinates xkx_{k}, the coordinate maps pnp_{n} and the Euclidean norm ∥⋅∥\lVert\cdot\rVert of Borel Probability Measures on a Real Hilbert Space with an Orthonormal Basis: Standing Notation §coordinates, and integrals and integrability as in Measure Spaces and the Lebesgue Integral: Standing Notation, let cc be a variance sequence, γc\gamma_{c} the diagonal Gaussian measure on XX with variances cc, HcH_{c} the Cameron-Martin space of cc, and, for h∈Hch\in H_{c}, ℓh\ell_{h} the Paley-Wiener functional of hh relative to cc.

Let n∈Nn\in\mathbb{N}, let M∈RM\in\mathbb{R} be nonnegative, and let ψ:Rn→R\psi:\mathbb{R}^{n}\to\mathbb{R} be of class C1C^{1} on Rn\mathbb{R}^{n}, where the class C1C^{1} and the partial derivatives ∂iψ\partial_{i}\psi are understood as in Differential Calculus and Convexity on Euclidean Open Sets: Standing Notation §derivatives with q=nq=n, the set Rn\mathbb{R}^{n} being open by claim 1 of Euclidean Space is Open in Itself, and CkC^k Maps are Continuous, and suppose that

∣∂iψ(y)∣≤M(1+∥y∥)for all y∈Rn and i∈[n].|\partial_{i}\psi(y)|\le M\bigl(1+\lVert y\rVert\bigr)\qquad\text{for all }y\in\mathbb{R}^{n}\text{ and }i\in[n].

Let φ=ψ∘pn:X→R\varphi=\psi\circ p_{n}:X\to\mathbb{R} and, for i∈[n]i\in[n], φi=(∂iψ)∘pn:X→R\varphi_{i}=(\partial_{i}\psi)\circ p_{n}:X\to\mathbb{R}.

1. (Integrability) The following functions are Borel and integrable with respect to γc\gamma_{c}: the functions φ\varphi and φi\varphi_{i} (i∈[n]i\in[n]); the functions x↦xkφ(x)x\mapsto x_{k}\varphi(x) (k∈Nk\in\mathbb{N}); and the functions ℓhφ\ell_{h}\varphi (h∈Hch\in H_{c}).

2. (Coordinate directions) For every i∈[n]i\in[n],

∫Xφi dγc=1ci∫Xxi φ(x) γc(dx).\int_{X}\varphi_{i}\,d\gamma_{c}=\frac{1}{c_{i}}\int_{X}x_{i}\,\varphi(x)\,\gamma_{c}(dx).

For every k∈Nk\in\mathbb{N} with k>nk>n,

∫Xxk φ(x) γc(dx)=0.\int_{X}x_{k}\,\varphi(x)\,\gamma_{c}(dx)=0.

3. (Cameron-Martin directions) For every h∈Hch\in H_{c}, with coordinates hih_{i},

∑i=1nhi∫Xφi dγc=∫Xℓh φ dγc.\sum_{i=1}^{n}h_{i}\int_{X}\varphi_{i}\,d\gamma_{c}=\int_{X}\ell_{h}\,\varphi\,d\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…