OctoInformBeta

OpenAI preprint, September 2026 · Hilbert multipliers and the absence of critical points

Section 6: proof of Theorem 1.1: For nonzero first Neumann eigenfunctions on bounded smooth simply connected planar domains

Gradient traces exclude interior critical points.

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

The question concerns interior zeros of the gradient of a first Neumann eigenfunction and the dimension of its boundary gradient traces. An admissible domain Ω\Omega is a nonempty bounded open simply connected subset of the Euclidean plane ℝ2\mathbb R^2 with smooth boundary.✦ Smooth boundary means that each p∈∂Ωp\in\partial\Omega has an open neighborhood WW and a smooth real defining function ρ\rho on WW satisfying (1). ρ(p)=0,Dρ(p)≠0,Ω∩W={x∈W:ρ(x)<0}.\rho(p)=0,\qquad D\rho(p)\ne0,\qquad \Omega\cap W=\{x\in W:\rho(x)<0\}.(1)✦

The complex coordinate isometry ι:ℂ→ℝ2\iota:\mathbb C\to\mathbb R^2 identifies the conformal coordinate plane with the Euclidean plane. For a∈ℂa\in\mathbb C, complex scalar multiplication MaM_a is the real linear map specified in (2); the unit disk DD and its boundary S1S^1 use this identification. ι(ξ+iη)=(ξ,η),Maι(z)=ι(az),D={x∈ℝ2:|x|<1},S1={x∈ℝ2:|x|=1}.\begin{gathered} \iota(\xi+i\eta)=(\xi,\eta),\qquad M_a\iota(z)=\iota(az),\\ D=\{x\in\mathbb R^2:|x|<1\},\qquad S^1=\{x\in\mathbb R^2:|x|=1\}. \end{gathered}(2)

A closed disk conformal chart for Ω\Omega is a smooth conformal diffeomorphism Φ:D→Ω\Phi:D\to\Omega with smooth inverse and nonzero complex derivatives d,d′d,d' obeying (3) for positive constants B,BinvB,B_{\mathrm{inv}}. DΦ(x)=Md(x),0<|d(x)|≤B(x∈D),DΦ−1(y)=Md′(y),0<|d′(y)|≤Binv(y∈Ω).\begin{gathered} D\Phi(x)=M_{d(x)},\quad 0<|d(x)|\le B\quad(x\in D),\\ D\Phi^{-1}(y)=M_{d'(y)},\quad 0<|d'(y)|\le B_{\mathrm{inv}}\quad(y\in\Omega). \end{gathered}(3) Its forward and inverse extensions F,G:ℝ2→ℝ2F,G:\mathbb R^2\to\mathbb R^2 are globally Lipschitz with respective nonnegative constants L,LinvL,L_{\mathrm{inv}}, agree with Φ,Φ−1\Phi,\Phi^{-1} on their interiors, and map D¯,Ω¯\overline D,\overline\Omega into one another as mutual inverses. The boundary restriction h=F|S1h=F|_{S^1} is the circle boundary homeomorphism from S1S^1 onto ∂Ω\partial\Omega.✦

Eigenfunctions and Boundary Traces

An H1H^1 weak gradient of a real function vv on Ω\Omega is a vector field gg such that v,gv,g are square integrable and satisfy (4) for every smooth compactly supported test function φ\varphi whose closed support lies in Ω\Omega, and every e∈ℝ2e\in\mathbb R^2. ∫ΩvDφ[e]=−∫Ω⟨g,e⟩φ.\int_\Omega v\,D\varphi[e] =-\int_\Omega \langle g,e\rangle\varphi.(4)✦

The set ℛΩ\mathcal R_\Omega of mean-zero Rayleigh values consists of the quotients in (5), where gg is an H1H^1 weak gradient of vv. The first positive Neumann value μ\mu of Ω\Omega is the infimum of ℛΩ\mathcal R_\Omega. ℛΩ={∫Ω|g|2∫Ωv2:g is a weak gradient of v,∫Ωv=0,∫Ωv2>0},μ=inf⁡ℛΩ.\mathcal R_\Omega= \left\{\frac{\int_\Omega|g|^2}{\int_\Omega v^2}: g\text{ is a weak gradient of }v,\ \int_\Omega v=0,\ \int_\Omega v^2>0\right\}, \qquad \mu=\inf\mathcal R_\Omega.(5)

The real first Neumann eigenspace VV consists of functions u:ℝ2→ℝu:\mathbb R^2\to\mathbb R smooth on Ω¯\overline\Omega, with H1H^1 weak gradient ∇u\nabla u and zero mean, satisfying (6) for every real function vv with H1H^1 weak gradient gg. ∫Ω⟨∇u,g⟩=μ∫Ωuv.\int_\Omega\langle\nabla u,g\rangle =\mu\int_\Omega uv.(6)✦ The first interior Neumann eigenspace VintV_{\mathrm{int}} is the space of real functions u:ℝ2→ℝu:\mathbb R^2\to\mathbb R smooth on Ω\Omega, with H1H^1 weak gradient ∇u\nabla u and zero mean, satisfying (6) for every real function vv with H1H^1 weak gradient gg.

A scalar-vector pair (v,Y)(v,Y) on Ω¯\overline\Omega is a pair of functions v:Ω¯→ℝv:\overline\Omega\to\mathbb R and Y:Ω¯→ℝ2Y:\overline\Omega\to\mathbb R^2.✦ The restriction pair J(v,Y)J(v,Y) of a scalar field vv and a vector field YY is (v|Ω¯,Y|Ω¯)(v|_{\overline\Omega},Y|_{\overline\Omega}).✦ The closed Neumann eigenjet space 𝒥\mathcal J is the real linear space defined by (7). 𝒥={J(v,Y):v∈Vint,v,Y continuous on Ω¯,Y|Ω=∇v|Ω}.\mathcal J= \left\{J(v,Y):v\in V_{\mathrm{int}},\quad v,Y\text{ continuous on }\overline\Omega,\quad Y|_\Omega=\nabla v|_\Omega\right\}.(7)✦

The boundary gradient trace γJ\gamma J of an eigenjet JJ is the vector-valued function on S1S^1 defined in (8); the boundary gradient trace space 𝒯\mathcal T is its range. (γJ(v,Y))(s)=Y(h(s)),𝒯={γJ:J∈𝒥}.(\gamma J(v,Y))(s)=Y(h(s)),\qquad \mathcal T=\{\gamma J:J\in\mathcal J\}.(8) For U⊂S1U\subset S^1, the vanishing subspace 𝒯U\mathcal T_U of 𝒯\mathcal T consists of the traces that vanish throughout UU, as in (9). 𝒯U={g∈𝒯:g(s)=0 for every s∈U}.\mathcal T_U=\{g\in\mathcal T:g(s)=0\text{ for every }s\in U\}.(9)✦

The Claim

Theorem 1. Let Ω\Omega be admissible and let F,GF,G be a closed disk conformal chart. The trace space 𝒯\mathcal T is finite dimensional, with the bound (10). dim⁡ℝ𝒯≤2.\dim_{\mathbb R}\mathcal T\le2.(10)✦ For every nonempty open subset UU of S1S^1, the local nullity satisfies (11). dim⁡ℝ𝒯U≤1.\dim_{\mathbb R}\mathcal T_U\le1.(11)✦ If u∈Vu\in V is nonzero at some point of Ω\Omega, then (12) holds for each p∈Dp\in D. ∇u(F(p))≠0.\nabla u(F(p))\ne0.(12)✦

In the paper, the result reads: for boundary energy EE, boundary gradient trace gg and Hilbert coordinate multipliers bjb_j, if Φ(p)\Phi(p) were critical, the multiplier identity and Hilbert squared-distance identity would give ∑jE(bjg)=12|∇u(Φ(p))|2=0\sum_j E(b_jg)=\tfrac12|\nabla u(\Phi(p))|^2=0, every coordinate multiplier bjgb_jg would therefore remain in the boundary-gradient trace space, dense nonzero-gradient support would contradict dimension at most two, and nondense support would contradict boundary uniqueness.

What Is Checked

The noncriticality assertion is stated over an admissible domain with a closed disk conformal chart and a closure-smooth eigenfunction nonzero somewhere in Ω\Omega; the rank assertions concern boundary gradients of continuous eigenjets from VintV_{\mathrm{int}}. Boundary-gradient evaluation on 𝒥\mathcal J preserves addition and real scalar multiplication.