OctoInformBeta

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

Lemma 4.2: Under a subcritical norm bound, positivity and scalar boundary representations accompany a bound on the total interaction

Boundary kernels from a weighted Green 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 extension on the disk combines the Poisson kernel with the resolvent of a weighted Green operator. The resulting interior evaluation and boundary interaction are described by kernels on the disk and its boundary.

The Euclidean plane is ℝ2\mathbb R^2.✦ 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) The complex coordinate isometry is z↦(ℜz,ℑz)z\mapsto(\Re z,\Im z) from ℂ\mathbb C to ℝ2\mathbb R^2✦; its restriction to the complex unit circle is the complex boundary identification.✦ The angular boundary parametrization a:ℝ/(2πℤ)→S1a:\mathbb R/(2\pi\mathbb Z)\to S^1 is a(θ)=(cos⁡θ,sin⁡θ)a(\theta)=(\cos\theta,\sin\theta).✦ Boundary arc length dsds is the pushforward of angular length by aa.✦

Let dxdx denote area measure and let ν\nu be a measure satisfying the domination condition (2), for a finite nonnegative constant CC: ν≤Cdx|D.\nu\le C\,dx|_D.(2) Write ⟨u,v⟩ν=∫uvdν\langle u,v\rangle_\nu=\int u v\,d\nu for the real L2(ν)L^2(\nu) inner product.

The Kernels and the Extension

Under the complex coordinate identification, the complex and planar disk Green kernels GG, Poisson kernels PP, and boundary interactions bb have the formulas (3), away from their singular diagonals: G(x,y)=14πlog⁡(1+(1−|x|2)(1−|y|2)|x−y|2),x,y∈D,x≠y,P(s,x)=1−|x|22π|s−x|2,s∈S1,x∈ℝ2,x≠s,b(s,t)=1π|s−t|2,s,t∈S1,s≠t.\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,\\ P(s,x)&=\frac{1-|x|^2}{2\pi|s-x|^2}, &&s\in S^1,\ x\in\mathbb R^2,\ x\ne s,\\ b(s,t)&=\frac{1}{\pi|s-t|^2}, &&s,t\in S^1,\ s\ne t. \end{aligned}(3) Their diagonal values are G(x,x)=P(s,s)=b(s,s)=0G(x,x)=P(s,s)=b(s,s)=0.

The Green integral operator 𝒢\mathcal G on L2(D,dx)L^2(D,dx) is the continuous linear L2L^2 kernel operator in (4).✦ The weighted L2L^2 inclusion i:L2(D,dx)→L2(ν)i:L^2(D,dx)\to L^2(\nu) is the continuous linear change-of-measure map.✦ The weighted Green operator TT on L2(ν)L^2(\nu) is the composition with the Hilbert adjoint i*i^* specified there: (𝒢f)(x)=∫DG(x,y)f(y)dy,T=i𝒢i*.(\mathcal G f)(x)=\int_D G(x,y)f(y)\,dy,\qquad T=i\mathcal G i^*.(4) The Neumann series resolvent 𝒮(B)\mathcal S(B) of a continuous real linear operator BB on a real Banach space is the operator series in (5); write S=𝒮(T)S=\mathcal S(T): 𝒮(B)=∑n=0∞Bn.\mathcal S(B)=\sum_{n=0}^{\infty}B^n.(5)✦

For boundary data h:S1→Eh:S^1\to E in a real normed vector space EE, the Poisson extension AhAh is the function in (6): Ah(x)=∫S1P(s,x)h(s)ds.Ah(x)=\int_{S^1}P(s,x)h(s)\,ds.(6)✦ For bounded measurable real boundary data h:S1→ℝh:S^1\to\mathbb R, the L2L^2 Poisson extension Ah∈L2(ν)Ah\in L^2(\nu) is the equivalence class of the Poisson extension of hh.✦

Using the same Green formula for planar arguments, the smoothed Poisson kernel P1:S1×ℝ2→ℝP_1:S^1\times\mathbb R^2\to\mathbb R is the integral in (7): P1(s,x)=∫DG(x,y)P(s,y)dν(y).P_1(s,x)=\int_D G(x,y)P(s,y)\,d\nu(y).(7)✦ The Green row gpg_p at p∈Dp\in D is the class of x↦G(x,p)x\mapsto G(x,p) in L2(ν)L^2(\nu).✦ The L2L^2 smoothed Poisson kernel usu_s at s∈S1s\in S^1 is the class of x↦P1(s,x)x\mapsto P_1(s,x) in L2(ν)L^2(\nu).✦

The Poisson overlap R0R_0, smoothed Poisson overlap R1R_1, and resolvent boundary interaction R2R_2 are the real-valued functions on S1×S1S^1\times S^1 defined in (8): R0(s,t)=∫DP(s,x)P(t,x)dν(x),R1(s,t)=∫DP(s,x)P1(t,x)dν(x),R2(s,t)=⟨us,Sut⟩ν.\begin{aligned} R_0(s,t)&=\int_D P(s,x)P(t,x)\,d\nu(x),\\ R_1(s,t)&=\int_D P(s,x)P_1(t,x)\,d\nu(x),\\ R_2(s,t)&=\langle u_s,S u_t\rangle_\nu. \end{aligned}(8) The full boundary evaluation kernel K:D×S1→ℝK:D\times S^1\to\mathbb R, boundary remainder R:S1×S1→ℝR:S^1\times S^1\to\mathbb R, and full boundary kernel N:S1×S1→ℝN:S^1\times S^1\to\mathbb R are defined by (9): K(p,s)=P(s,p)+P1(s,p)+⟨gp,Sus⟩ν,R(s,t)=R0(s,t)+R1(s,t)+R2(s,t),N(s,t)=b(s,t)+R0(s,t)+R1(s,t)+R2(s,t).\begin{aligned} K(p,s)&=P(s,p)+P_1(s,p)+\langle g_p,S u_s\rangle_\nu,\\ R(s,t)&=R_0(s,t)+R_1(s,t)+R_2(s,t),\\ N(s,t)&=b(s,t)+R_0(s,t)+R_1(s,t)+R_2(s,t). \end{aligned}(9) For bounded measurable real boundary data hh, the full boundary extension Wh:D→ℝW_h:D\to\mathbb R is the function in (10): Wh(p)=Ah(p)+⟨gp,SAh⟩ν.W_h(p)=Ah(p)+\langle g_p,S Ah\rangle_\nu.(10)✦

The Claim

Theorem 1. Let ν\nu satisfy (2), and use the kernels (9). For every p∈Dp\in D, the boundedness assertion (11) holds: there exists Bp≥0 such that K(p,s)≤Bp(s∈S1).\text{there exists }B_p\ge0\text{ such that } K(p,s)\le B_p\quad(s\in S^1).(11)✦ The boundary remainder satisfies (12): R∈L1(S1×S1,dsdt).R\in L^1(S^1\times S^1,ds\,dt).(12)✦

Suppose in addition that ∥T∥<1\|T\|<1. The pointwise assertions are (13), (14), (15), and (16): K(p,s)>0(p∈D,s∈S1),K(p,s)>0\qquad(p\in D,\ s\in S^1),(13)✦ N(s,t)>0(s,t∈S1,s≠t),N(s,t)>0\qquad(s,t\in S^1,\ s\ne t),(14)✦ N(s,t)=N(t,s)(s,t∈S1),N(s,t)=N(t,s)\qquad(s,t\in S^1),(15)✦ R(s,t)≥0(s,t∈S1).R(s,t)\ge0\qquad(s,t\in S^1).(16)✦ For bounded measurable real functions h,kh,k on S1S^1, the scalar representations are (17) and (18): ∫S1K(p,s)h(s)ds=Wh(p)(p∈D),\int_{S^1}K(p,s)h(s)\,ds=W_h(p)\qquad(p\in D),(17)✦ ∫S1×S1R(s,t)h(s)k(t)dsdt=⟨Ah,SAk⟩ν.\int_{S^1\times S^1}R(s,t)h(s)k(t)\,ds\,dt =\langle Ah,S Ak\rangle_\nu.(18)✦ If dd is a real number with ∥T∥≤d<1\|T\|\le d<1, the total mass bound is (19): ∫S1×S1R(s,t)dsdt≤ν(D)1−d.\int_{S^1\times S^1}R(s,t)\,ds\,dt\le\frac{\nu(D)}{1-d}.(19)✦

In the paper, the result reads: for a bounded nonnegative subcritical disk density qq, with subcritical bound dd and q(D)=∫Dqdxq(D)=\int_Dq\,dx, writing II for the identity operator and using the subscript qq for the inner product weighted by qdxq\,dx, the kernels KK and NN are finite at their stated points, K(p,⋅)K(p,\cdot) is positive and bounded for fixed pp, RR is symmetric, nonnegative and integrable on S1×S1S^1\times S^1 with ∬R≤q(D)/(1−d)\iint R\le q(D)/(1-d), and Lipschitz scalar or vector boundary data satisfy Wh(p)=∫S1K(p,s)h(s)dsW_h(p)=\int_{S^1}K(p,s)h(s)\,ds and ⟨Ah,(I−T)−1Ak⟩q=∬S1×S1R(s,t)h(s)⋅k(t)dsdt\langle Ah,(I-T)^{-1}Ak\rangle_q =\iint_{S^1\times S^1}R(s,t)h(s)\cdot k(t)\,ds\,dt.

What Is Checked

The statements concern the explicit disk kernels and arc length measure, with the measure domination (2) and the operator norm hypotheses stated in the theorem.

For area measure on DD, Green rows belong to L2L^2 almost everywhere✦, and their L2L^2 norms have a uniform nonnegative bound on DD.✦

For measures η\eta and ξ\xi and an L2L^2 kernel operator with kernel JJ from L2(η)L^2(\eta) to L2(ξ)L^2(\xi), the kernel integral x↦∫J(x,y)f(y)dη(y)x\mapsto\int J(x,y)f(y)\,d\eta(y) belongs to L2(ξ)L^2(\xi) for every f∈L2(η)f\in L^2(\eta).✦