OctoInformBeta

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

Proposition 4.4: The full kernel gives the shifted divergence-curl energy of Helmholtz extensions of multiplied Neumann eigenfunction gradients

A boundary identity for Lipschitz multipliers.

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

A scalar multiplier of a Neumann eigenfunction gradient gives a vector field whose energy involves both divergence and curl. The field considered here is the componentwise Helmholtz extension of that product.✦ Its boundary values are parametrized on the unit circle.

An admissible domain Ω⊂ℝ2\Omega\subset\mathbb R^2 is a nonempty bounded simply connected open subset of the plane with smooth boundary.✦ The boundary ∂Ω\partial\Omega of Ω\Omega is smooth when every p∈∂Ωp\in\partial\Omega has an open neighborhood UU and a smooth defining function ρ:U→ℝ\rho:U\to\mathbb R with ρ(p)=0\rho(p)=0, Dρ(p)≠0D\rho(p)\ne0 and Ω∩U={ρ<0}∩U\Omega\cap U=\{\rho<0\}\cap U.✦ Write D={x∈ℝ2:|x|<1}D=\{x\in\mathbb R^2:|x|<1\} and S1={s∈ℝ2:|s|=1}S^1=\{s\in\mathbb R^2:|s|=1\}. The identification x1+ix2↔(x1,x2)x_1+ix_2\leftrightarrow(x_1,x_2) is an isometry and identifies the complex unit circle with S1S^1. The angular parametrization θ↦(cos⁡θ,sin⁡θ)\theta\mapsto(\cos\theta,\sin\theta), of period 2π2\pi, pushes angular Lebesgue measure forward to the arc length measure dσd\sigma on S1S^1.✦

The space Cc∞(Ω)C_c^\infty(\Omega) consists of smooth real functions with compact closed support contained in Ω\Omega.✦ A vector field ξ∈L2(Ω;ℝ2)\xi\in L^2(\Omega;\mathbb R^2) is the weak gradient of v∈L2(Ω)v\in L^2(\Omega) when the identity (1) holds for every φ∈Cc∞(Ω)\varphi\in C_c^\infty(\Omega) and e∈ℝ2e\in\mathbb R^2; here ∂eφ=Dφ[e]\partial_e\varphi=D\varphi[e] and φe\varphi e is the corresponding test vector field. ∫Ωv∂eφdx=−∫Ωξ⋅(φe)dx.\int_\Omega v\,\partial_e\varphi\,dx =-\int_\Omega \xi\cdot(\varphi e)\,dx.(1)✦ The Sobolev space H1(Ω)H^1(\Omega) is the weak gradient graph in L2(Ω)×L2(Ω;ℝ2)L^2(\Omega)\times L^2(\Omega;\mathbb R^2), with the Euclidean product norm.✦ The value and gradient maps extract vv and ξ\xi from the Sobolev class (v,ξ)(v,\xi). The embedding of Cc∞(Ω)C_c^\infty(\Omega) sends φ\varphi to (φ,∇φ)(\varphi,\nabla\varphi), and H01(Ω)H_0^1(\Omega) is the closure of its image in H1(Ω)H^1(\Omega); its value and gradient maps are the restrictions of these maps.

The first positive Neumann value μ\mu of Ω\Omega is the infimum of the mean-zero Rayleigh quotients in (2). μ=inf⁡{∫Ω|∇v|2dx∫Ωv2dx:v∈H1(Ω),∫Ωvdx=0,∫Ωv2dx>0}.\mu=\inf\left\{ \frac{\int_\Omega |\nabla v|^2\,dx}{\int_\Omega v^2\,dx}: v\in H^1(\Omega),\quad \int_\Omega v\,dx=0,\quad \int_\Omega v^2\,dx>0\right\}.(2) The real first Neumann eigenspace VV consists of functions uu smooth on Ω¯\overline\Omega, with u∈H1(Ω)u\in H^1(\Omega), mean zero and weak gradient equal to their gradient, satisfying (3). ∫Ω∇u⋅∇vdx=μ∫Ωuvdx(v∈H1(Ω)).\int_\Omega\nabla u\cdot\nabla v\,dx =\mu\int_\Omega uv\,dx \qquad(v\in H^1(\Omega)).(3)✦ For u∈Vu\in V and x∈Ω¯x\in\overline\Omega, the gradient ∇Ω¯u(x)\nabla_{\overline\Omega}u(x) is the vector representing the differential of uu within Ω¯\overline\Omega by the Euclidean inner product.

A closed disk conformal chart is a smooth conformal diffeomorphism Φ:D→Ω\Phi:D\to\Omega with smooth inverse, whose complex derivative dd and inverse complex derivative d′d' are nowhere zero and bounded by positive constants MM and MinvM_{\mathrm{inv}}, respectively. Its differential acts by complex multiplication by dd, and its inverse differential acts by multiplication by d′d'. The chart has globally Lipschitz extensions FF and FinvF_{\mathrm{inv}} of Φ\Phi and Φ−1\Phi^{-1}, mapping the respective closures into one another and satisfying Finv(F(z))=zF_{\mathrm{inv}}(F(z))=z on D¯\overline D and F(Finv(x))=xF(F_{\mathrm{inv}}(x))=x on Ω¯\overline\Omega. The conformal area measure mΦm_\Phi and potential measure ν\nu are the weighted area measures in (4). dmΦ(x)=|d(x)|2dx|D,dν(x)=max⁡(μ,0)dmΦ(x).dm_\Phi(x)=|d(x)|^2\,dx\big|_D, \qquad d\nu(x)=\max(\mu,0)\,dm_\Phi(x).(4)

The Energy and Its Boundary Kernels

Let LL be a real constant, and assume 0<μ<L0<\mu<L and the Dirichlet lower bound (5). L∥j∥L2(Ω)2≤∥∇j∥L2(Ω)2(j∈H01(Ω)).L\|j\|_{L^2(\Omega)}^2\le\|\nabla j\|_{L^2(\Omega)}^2 \qquad(j\in H_0^1(\Omega)).(5) The Helmholtz energy form aμa_\mu on H1(Ω)H^1(\Omega) is the bilinear form defined by (6), and the Dirichlet Helmholtz form is its restriction to H01(Ω)H_0^1(\Omega). aμ(v,w)=∫Ω∇v⋅∇wdx−μ∫Ωvwdx.a_\mu(v,w)=\int_\Omega\nabla v\cdot\nabla w\,dx -\mu\int_\Omega vw\,dx.(6)✦ The Helmholtz forcing of v∈H1(Ω)v\in H^1(\Omega) is the continuous linear functional on H01(Ω)H_0^1(\Omega) that sends jj to −aμ(v,j)-a_\mu(v,j); it depends continuously and linearly on vv. The Helmholtz correction Cμ:H1(Ω)→H01(Ω)C_\mu:H^1(\Omega)\to H_0^1(\Omega) is the map that sends vv to the inverse Dirichlet Helmholtz operator applied to the Helmholtz forcing of vv.✦ With II the identity on H1(Ω)H^1(\Omega) and ι:H01(Ω)↪H1(Ω)\iota:H_0^1(\Omega)\hookrightarrow H^1(\Omega) the inclusion, the Helmholtz extension HμH_\mu is the map defined by Hμ=I+ιCμH_\mu=I+\iota C_\mu.✦

Let e1,e2e_1,e_2 be the standard orthonormal basis. For a vector field X=(X1,X2)∈H1(Ω)2X=(X_1,X_2)\in H^1(\Omega)^2, write ∂iXj\partial_i X_j for its weak derivatives; its L2L^2 vector value is X1e1+X2e2X_1e_1+X_2e_2. Its divergence is div⁡X=∂1X1+∂2X2\operatorname{div}X=\partial_1X_1+\partial_2X_2, and its curl is curl⁡X=∂1X2−∂2X1\operatorname{curl}X=\partial_1X_2-\partial_2X_1. For vector fields X,Y∈H1(Ω)2X,Y\in H^1(\Omega)^2, the divergence-curl energy QQ and its shifted form ℰμ\mathcal E_\mu are the bilinear forms given by (7). Q(X,Y)=∫Ω(div⁡Xdiv⁡Y+curl⁡Xcurl⁡Y)dx,ℰμ(X,Y)=Q(X,Y)−μ∫ΩX⋅Ydx.\begin{aligned} Q(X,Y)&=\int_\Omega\bigl( \operatorname{div}X\operatorname{div}Y+ \operatorname{curl}X\operatorname{curl}Y\bigr)\,dx,\\ \mathcal E_\mu(X,Y)&=Q(X,Y)-\mu\int_\Omega X\cdot Y\,dx. \end{aligned}(7) For u∈Vu\in V, let Ui∈H1(Ω)U_i\in H^1(\Omega) be the Sobolev class of the iith component of ∇Ω¯u\nabla_{\overline\Omega}u.✦ Multiplication by a globally Lipschitz real function b:ℝ2→ℝb:\mathbb R^2\to\mathbb R sends v∈H1(Ω)v\in H^1(\Omega) to bvbv, with the weak product rule.✦ For a globally Lipschitz real function b:ℝ2→ℝb:\mathbb R^2\to\mathbb R and s∈S1s\in S^1, define the extended field, boundary multiplier and boundary gradient by (8). (Xb)i=Hμ(bUi),β(s)=b(F(s)),g(s)=∇Ω¯u(F(s)).(X_b)_i=H_\mu(bU_i),\qquad \beta(s)=b(F(s)),\qquad g(s)=\nabla_{\overline\Omega}u(F(s)).(8) For i=1,2i=1,2, the coordinate embedding Ai:L2(Ω)→L2(Ω;ℝ2)A_i:L^2(\Omega)\to L^2(\Omega;\mathbb R^2) is the continuous linear map sending a scalar function ff to the vector field feife_i. For i=1,2i=1,2, the coordinate projection πi:L2(Ω;ℝ2)→L2(Ω)\pi_i:L^2(\Omega;\mathbb R^2)\to L^2(\Omega) is the continuous linear map sending ZZ to Z⋅eiZ\cdot e_i; thus ∂iv=πi(∇v)\partial_i v=\pi_i(\nabla v) for v∈H1(Ω)v\in H^1(\Omega).

For x,y∈Dx,y\in D and s,t∈S1s,t\in S^1, the disk Dirichlet Green kernel GG, Poisson kernel PP and bare boundary interaction JJ are the real kernels in (9), using the complex coordinate identification above, with G(x,x)=0G(x,x)=0 and J(s,s)=0J(s,s)=0 on the diagonals. G(x,y)=14πlog⁡(1+(1−|x|2)(1−|y|2)|x−y|2),P(s,x)=1−|x|22π|s−x|2,J(s,t)=1π|s−t|2.\begin{aligned} G(x,y)&=\frac1{4\pi}\log\left( 1+\frac{(1-|x|^2)(1-|y|^2)}{|x-y|^2}\right),\\ P(s,x)&=\frac{1-|x|^2}{2\pi|s-x|^2},\qquad J(s,t)=\frac1{\pi|s-t|^2}. \end{aligned}(9) Let 𝒢\mathcal G be the continuous Green integral operator on L2(D,dx)L^2(D,dx), acting by integration against GG.✦ The weighted inclusion j:L2(D,dx)→L2(ν)j:L^2(D,dx)\to L^2(\nu) is the change-of-measure map, and the weighted Green operator is T=j𝒢j*T=j\mathcal G j^*. Its Neumann series resolvent is ℛT=∑n=0∞Tn\mathscr R_T=\sum_{n=0}^\infty T^n.✦ The smoothed Poisson kernel P1P_1 and its L2(ν)L^2(\nu) class psp_s are defined by (10). P1(s,x)=∫DG(x,y)P(s,y)dν(y),ps=[P1(s,⋅)]∈L2(ν).P_1(s,x)=\int_D G(x,y)P(s,y)\,d\nu(y), \qquad p_s=[P_1(s,\cdot)]\in L^2(\nu).(10) The boundary remainder RR on S1×S1S^1\times S^1 is the real kernel given by the sum of the Poisson overlaps and the resolvent interaction in (11); the full boundary kernel NN on S1×S1S^1\times S^1 is the real kernel given there. R1(s,t)=∫DP(s,x)P(t,x)dν(x),R2(s,t)=∫DP(s,x)P1(t,x)dν(x),Rtail(s,t)=⟨ps,ℛTpt⟩L2(ν),R(s,t)=R1(s,t)+R2(s,t)+Rtail(s,t),N(s,t)=J(s,t)+R(s,t).\begin{aligned} R_1(s,t)&=\int_D P(s,x)P(t,x)\,d\nu(x),\\ R_2(s,t)&=\int_D P(s,x)P_1(t,x)\,d\nu(x),\\ R_{\mathrm{tail}}(s,t)&=\langle p_s,\mathscr R_Tp_t\rangle_{L^2(\nu)},\\ R(s,t)&=R_1(s,t)+R_2(s,t)+R_{\mathrm{tail}}(s,t),\qquad N(s,t)=J(s,t)+R(s,t). \end{aligned}(11)

For real kernels R0:S1×S1→ℝR_0:S^1\times S^1\to\mathbb R and real functions c:S1→ℝc:S^1\to\mathbb R, the boundary energy form ℬR0,c\mathcal B_{R_0,c} is the form on pairs of functions h,kh,k from S1S^1 into a real inner product space given by (12). ℬR0,c(h,k)=12∬S1×S1J(s,t)⟨h(s)−h(t),k(s)−k(t)⟩dσ(s)dσ(t)−∬S1×S1R0(s,t)⟨h(s),k(t)⟩dσ(s)dσ(t)+∫S1c(s)⟨h(s),k(s)⟩dσ(s).\begin{aligned} \mathcal B_{R_0,c}(h,k) &=\frac12\iint_{S^1\times S^1}J(s,t) \langle h(s)-h(t),k(s)-k(t)\rangle\,d\sigma(s)d\sigma(t)\\ &\quad-\iint_{S^1\times S^1}R_0(s,t) \langle h(s),k(t)\rangle\,d\sigma(s)d\sigma(t) +\int_{S^1}c(s)\langle h(s),k(s)\rangle\,d\sigma(s). \end{aligned}(12)✦

The Claim

Theorem 1. Let Ω\Omega be admissible, let Φ\Phi be a closed disk conformal chart, and assume 0<μ<L0<\mu<L and (5). For u∈Vu\in V and every globally Lipschitz real multiplier bb on ℝ2\mathbb R^2, the quantities in (8) and (11) satisfy the multiplier kernel identity (13). ℰμ(Xb,Xb)=12∬S1×S1N(s,t)(β(s)−β(t))2g(s)⋅g(t)dσ(s)dσ(t).\mathcal E_\mu(X_b,X_b) =\frac12\iint_{S^1\times S^1}N(s,t) (\beta(s)-\beta(t))^2\,g(s)\cdot g(t)\,d\sigma(s)d\sigma(t).(13)✦

Let HH be a real inner product space, let R0R_0 be an integrable symmetric real kernel on S1×S1S^1\times S^1, let c:S1→ℝc:S^1\to\mathbb R be arbitrary, and let h:S1→Hh:S^1\to H and η:S1→ℝ\eta:S^1\to\mathbb R be Lipschitz. The boundary multiplier identity is (14). ℬR0,c(ηh,ηh)−ℬR0,c(h,η2h)=12∬S1×S1(J(s,t)+R0(s,t))(η(s)−η(t))2⟨h(s),h(t)⟩dσ(s)dσ(t).\begin{aligned} &\mathcal B_{R_0,c}(\eta h,\eta h) -\mathcal B_{R_0,c}(h,\eta^2h)\\ &\qquad=\frac12\iint_{S^1\times S^1}(J(s,t)+R_0(s,t)) (\eta(s)-\eta(t))^2 \langle h(s),h(t)\rangle\,d\sigma(s)d\sigma(t). \end{aligned}(14)✦

In the paper, the result reads: for Ω\Omega as above, 0≠u∈V0\ne u\in V, g(s)=∇u(F(s))g(s)=\nabla u(F(s)) and every real Lipschitz η\eta on S1S^1, the energy is E(ηg)=12∬S1×S1N(s,t)(η(s)−η(t))2g(s)⋅g(t)dσ(s)dσ(t)E(\eta g)=\frac12\iint_{S^1\times S^1}N(s,t)(\eta(s)-\eta(t))^2 g(s)\cdot g(t)\,d\sigma(s)d\sigma(t).

What Is Checked

The multiplier kernel identity is stated over the Sobolev Helmholtz extension of a Lipschitz multiple of an eigenfunction gradient, with the conformal potential measure; the boundary identity is stated over an integrable symmetric remainder kernel and Lipschitz boundary data.

The potential density bound is the finite number μM2\mu M^2, and the potential measure is finite✦ and satisfies ν≤μM2dx|D\nu\le\mu M^2\,dx|_D.✦ The Green rows belong to L2(D,dx)L^2(D,dx) for almost every xx✦ and have a uniform norm bound on DD.✦ The associated kernel integrals send L2L^2 functions to L2L^2 functions✦. The Dirichlet Sobolev space is complete✦, and the Dirichlet Helmholtz form is coercive under the nonnegative spectral parameter condition, the strict gap μ<L\mu<L and (5).✦