Gaussian Dressing of the Wick-Square Cost: the Riccati Coefficients, the Dressed Variances, the Quadratic Profile, the Shifted Pair and the Shifted Operator
Solving a scalar Riccati equation in each coordinate gives a quadratic profile and narrower dressed variances; the Gaussian free-energy pair of the dressed variances is the original pair shifted by the profile, and the lifted Wick-square operator shifted by the profile is the lifted Ornstein-Uhlenbeck operator relative to the dressed Gaussian with the running cost raised by an explicit constant.
1. (The Riccati coefficients) For every i∈[d] there is exactly one bi∈R with
2θbi2+(2λ0+cia)bi=βiandcia+θbi>0,
namely bi=θ−1((λ0/2+a/ci)2+2θβi−λ0/2−a/ci), the radicand being positive. We write b=(b1,…,bd).
2. (The dressed variances) The vector c′=(c1′,…,cd′) with ci′=a(a/ci+θbi)−1 is a variance vector, called the dressed variances, and for every x∈Rd
where Sc′, ∣x∣c′2 and, below, Zc′ are the scaling map, weighted square and normalizing constant of c′.
3. (The quadratic profile) The function u0:Rd→R, u0(x)=21∑i=1dbixi2, is of class C2 on Rd with Du0(x)=(b1x1,…,bdxd) and ∂j∂iu0(x) equal to bi if i=j and to 0 otherwise. The function
Φ0:P2(Rd)→R,Φ0(μ)=∫Rdu0dμ,
called the quadratic profile, is an intrinsic test function on P2(Rd); for every μ∈P2(Rd), ∇Φ0(μ) is the class Du0 in L2(μ;Rd) of the gradient map of u0, and HΦ0(μ) is the diagonal matrix with diagonal entries b1,…,bd.
4. (The dressed pair) Let (D,DΣ,E,Σ) and (D′,DΣ′,E′,Σ′) be the Gaussian free-energy pairs with temperature a and variances c and c′ respectively. Then D′=D, DΣ′=DΣ,
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.