OctoInformBeta

OpenAI preprint, September 2026 · Boundary kernels and the multiplier identity

Section 4.2: evaluation and interaction kernels: Evaluation and interaction on the unit disk with a measure dominated by area

Boundary kernels from the smoothed resolvent.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

Boundary evaluation on the disk depends on Poisson data at a boundary point and a weighted Green operator. Boundary interaction pairs data from two points of the circle, with the singular interaction of the unperturbed disk as its leading term.

Identify the Euclidean plane ℝ2\mathbb R^2 with ℂ\mathbb C through the real-linear isometry determined by the orthonormal basis (1,i)(1,i).✦ The unit disk DD and unit circle S1S^1 are the sets in (1): D={x∈ℝ2:|x|<1},S1={s∈ℝ2:|s|=1}.D=\{x\in\mathbb R^2:|x|<1\},\qquad S^1=\{s\in\mathbb R^2:|s|=1\}.(1) Let dAdA denote area measure restricted to DD, and let ν\nu be a finite measure satisfying ν≤CdA\nu\le C\,dA for a finite constant C≥0C\ge0.

In the library’s definitions, real integrals have value zero when their integrands are not integrable.

In these complex coordinates, the disk Dirichlet Green kernel GG and the disk Poisson kernel PsP_s, for s∈S1s\in S^1, are the real functions in (2); the diagonal value of GG is set to zero: G(x,y)=14πlog⁡(1+(1−|x|2)(1−|y|2)|x−y|2)(x,y∈D,x≠y),G(x,x)=0(x∈D),Ps(x)=1−|x|22π|s−x|2(x∈D).\begin{aligned} G(x,y)&=\frac{1}{4\pi} \log\!\left(1+\frac{(1-|x|^2)(1-|y|^2)}{|x-y|^2}\right) &&(x,y\in D,\ x\ne y),\\ G(x,x)&=0 &&(x\in D),\\ P_s(x)&=\frac{1-|x|^2}{2\pi|s-x|^2} &&(x\in D). \end{aligned}(2) The complex boundary interaction b:ℂ×ℂ→ℝb:\mathbb C\times\mathbb C\to\mathbb R is b(s,t)=1/(π|s−t|2)b(s,t)=1/(\pi|s-t|^2) for distinct points and zero for equal points. The planar boundary interaction on S1×S1S^1\times S^1 is its restriction through the complex coordinate isometry, as in (3): b(s,t)={1π|s−t|2,s≠t,0,s=t.b(s,t)= \begin{cases} \displaystyle\frac{1}{\pi|s-t|^2},&s\ne t,\\[4pt] 0,&s=t. \end{cases}(3)

The Operators and Smoothed Data

Let H0=L2(D,dA)H_0=L^2(D,dA) and H=L2(ν)H=L^2(\nu) be real Hilbert spaces. The weighted inclusion I:H0→HI:H_0\to H is the continuous linear change-of-measure map, and I*I^* denotes its Hilbert adjoint.✦ The Green integral operator 𝒢:H0→H0\mathcal G:H_0\to H_0 and weighted Green operator T:H→HT:H\to H are the continuous linear operators in (4): (𝒢f)(x)=∫DG(x,y)f(y)dA(y),T=I𝒢I*.(\mathcal Gf)(x)=\int_D G(x,y)f(y)\,dA(y),\qquad T=I\mathcal G I^*.(4) For a continuous linear operator AA on a real Banach space, its Neumann series resolvent 𝒮(A)\mathcal S(A) is the sum in (5), with value zero when the operator series is not summable: 𝒮(A)=∑n=0∞An.\mathcal S(A)=\sum_{n=0}^{\infty}A^n.(5)✦

For s∈S1s\in S^1, the smoothed Poisson kernel Us:ℝ2→ℝU_s:\mathbb R^2\to\mathbb R is the function obtained by integrating the Green kernel against the Poisson kernel with respect to ν\nu, as in (6): Us(x)=∫DG(x,y)Ps(y)dν(y).U_s(x)=\int_D G(x,y)P_s(y)\,d\nu(y).(6)✦ For p∈Dp\in D and s∈S1s\in S^1, the Green row gpg_p and smoothed Poisson row usu_s are the elements of HH represented by the functions in (7), where brackets denote equivalence classes modulo equality ν\nu-almost everywhere: gp=[G(⋅,p)],us=[Us].g_p=[G(\cdot,p)],\qquad u_s=[U_s].(7) ⟨⋅,⋅⟩H\langle\cdot,\cdot\rangle_H denotes the real inner product on HH.

For s,t∈S1s,t\in S^1, the Poisson overlap a0(s,t)a_0(s,t) is the real number obtained by integrating the product of the two Poisson kernels with respect to ν\nu, as in (8).✦ The smoothed Poisson overlap a1(s,t)a_1(s,t) and resolvent boundary interaction a∞(s,t)a_\infty(s,t), for s,t∈S1s,t\in S^1, are the real numbers given by the integral and inner product in (8): a0(s,t)=∫DPs(x)Pt(x)dν(x),a1(s,t)=∫DPs(x)Ut(x)dν(x),a∞(s,t)=⟨us,𝒮(T)ut⟩H.\begin{aligned} a_0(s,t)&=\int_D P_s(x)P_t(x)\,d\nu(x),\\ a_1(s,t)&=\int_D P_s(x)U_t(x)\,d\nu(x),\\ a_\infty(s,t)&=\langle u_s,\mathcal S(T)u_t\rangle_H. \end{aligned}(8)

The Definition

Definition 1. Let ν\nu be a finite measure dominated by a finite multiple of disk area measure, and use the operators and rows defined above. The evaluation kernel K:D×S1→ℝK:D\times S^1\to\mathbb R is defined by (9): K(p,s)=Ps(p)+Us(p)+⟨gp,𝒮(T)us⟩H.K(p,s)=P_s(p)+U_s(p)+\langle g_p,\mathcal S(T)u_s\rangle_H.(9)✦ The boundary remainder R:S1×S1→ℝR:S^1\times S^1\to\mathbb R is defined by (10): R(s,t)=a0(s,t)+a1(s,t)+a∞(s,t).R(s,t)=a_0(s,t)+a_1(s,t)+a_\infty(s,t).(10)✦ The full boundary kernel N:S1×S1→ℝN:S^1\times S^1\to\mathbb R is defined by (11): N(s,t)=b(s,t)+∫DPs(x)Pt(x)dν(x)+∫DPs(x)Ut(x)dν(x)+⟨us,𝒮(T)ut⟩H.N(s,t)=b(s,t)+\int_D P_s(x)P_t(x)\,d\nu(x) +\int_D P_s(x)U_t(x)\,d\nu(x) +\langle u_s,\mathcal S(T)u_t\rangle_H.(11)✦

In the paper, the result reads: for p∈Dp\in D and distinct s,t∈S1s,t\in S^1, K(p,s)=Ps(p)+∑n≥1(TnPs)(p)K(p,s)=P_s(p)+\sum_{n\ge1}(T^nP_s)(p), R(s,t)=∑n≥0∫DPsTnPtdνR(s,t)=\sum_{n\ge0}\int_D P_sT^nP_t\,d\nu, and N(s,t)=1/(π|s−t|2)+R(s,t)N(s,t)=1/(\pi|s-t|^2)+R(s,t), where iterated nonnegative integrals define these series without assuming Ps∈L2(ν)P_s\in L^2(\nu).

What Is Checked

The definitions (9), (10) and (11) are stated over the unit disk and circle with a measure dominated by a finite multiple of area, using the weighted inclusion and its Hilbert adjoint.

The Green rows are square-integrable with respect to dAdA for almost every point of DD✦, and the real values of their extended L2(dA)L^2(dA) norms have a uniform nonnegative bound on DD.✦

For a measurable L2L^2 kernel MM between measures α\alpha and β\beta, the associated kernel operator 𝒦M:L2(β)→L2(α)\mathcal K_M:L^2(\beta)\to L^2(\alpha) is continuous and linear; its integral in (12) is square-integrable with respect to α\alpha for every f∈L2(β)f\in L^2(\beta): (𝒦Mf)(x)=∫M(x,y)f(y)dβ(y).(\mathcal K_M f)(x)=\int M(x,y)f(y)\,d\beta(y).(12)✦