OctoInformBeta

OpenAI preprint, September 2026 · §4. A planar surface cannot have one nontrivial boundary

§4 (proof of Theorem 4.1): the limit at the spectral cut: Right-continuity along nonnegative rotations supplies the limit at the phase cut

Boundary inequalities force the exceptional coefficient to be trivial.

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 planar surface whose other boundary words are trivial has one exceptional boundary word to consider. The question is whether the boundary phase inequalities allow that word to represent a nontrivial group element.

Let AA be an arbitrary group. The relevant phases are attached to elements of AA through their left regular operators on ℓ2(A,ℂ)\ell^2(A,\mathbb C), with the phase cut at 11.

Regular Operators

For a complex inner product space ℋ\mathcal H, a bounded operator TT on ℋ\mathcal H is a continuous complex-linear map from ℋ\mathcal H to itself.✦ The regular Hilbert space operators of AA are the bounded operators on ℓ2(A,ℂ)\ell^2(A,\mathbb C).✦

The delta vector δi\delta_i, for ii in an index set α\alpha, is the square-summable complex family defined by (1): δi(j)={1,j=i,0,j≠i.\delta_i(j)= \begin{cases} 1,&j=i,\\ 0,&j\ne i. \end{cases}(1)✦

For a bijection e:α→βe\colon\alpha\to\beta, the linear coordinate permutation PeP_e is the complex-linear equivalence between the corresponding ℓ2\ell^2 spaces given, together with its inverse, by (2).✦ The unitary coordinate permutation PeP_e associated with a bijection e:α→βe\colon\alpha\to\beta is the isometric coordinate equivalence between ℓ2(α,ℂ)\ell^2(\alpha,\mathbb C) and ℓ2(β,ℂ)\ell^2(\beta,\mathbb C) given by (2).✦ (Pef)(j)=f(e−1(j)),(Pe−1g)(i)=g(e(i)).(P_e f)(j)=f(e^{-1}(j)),\qquad (P_e^{-1}g)(i)=g(e(i)).(2)

For a∈Aa\in A, the left regular isometry L(a)L(a) on ℓ2(A,ℂ)\ell^2(A,\mathbb C) is the coordinate permutation induced by x↦axx\mapsto ax.✦ For a∈Aa\in A, the left regular operator L(a)L(a) is the bounded operator on ℓ2(A,ℂ)\ell^2(A,\mathbb C) specified on delta vectors by (3): L(a)δx=δax.L(a)\delta_x=\delta_{ax}.(3)✦

Spectral Phase

The scalar rotation r(x)r(x) through a real parameter xx is the complex number given by (4), where i\mathrm i is the imaginary unit: r(x)=exp⁡(2πix).r(x)=\exp(2\pi\mathrm i x).(4)✦

The fractional argument hh is the real-valued function on ℂ\mathbb C that assigns to each complex number the fractional part of its principal argument divided by 2π2\pi.✦ Its values, including the value at the cut, are fixed by (5): h(z)=Arg⁡(z)2π−⌊Arg⁡(z)2π⌋,h(r(x))=x−⌊x⌋,h(1)=0.\begin{aligned} h(z)&=\frac{\operatorname{Arg}(z)}{2\pi} -\left\lfloor\frac{\operatorname{Arg}(z)}{2\pi}\right\rfloor,\\ h(r(x))&=x-\lfloor x\rfloor,\qquad h(1)=0. \end{aligned}(5)✦

For a continuous real-valued function ff on a topological space XX, its complexification fℂf_{\mathbb C} is the continuous complex-valued function obtained by composing ff with the inclusion ℝ↪ℂ\mathbb R\hookrightarrow\mathbb C.✦

For a normal bounded operator UU on ℋ\mathcal H and a vector ξ∈ℋ\xi\in\mathcal H, the spectral vector functional FU,ξF_{U,\xi} is the positive real-linear functional on compactly supported continuous real functions on the spectrum σ(U)\sigma(U) given by (6).✦ Here fℂ(U)f_{\mathbb C}(U) denotes continuous functional calculus. FU,ξ(f)=Re⁡⟨ξ,fℂ(U)ξ⟩.F_{U,\xi}(f)=\operatorname{Re}\langle\xi,f_{\mathbb C}(U)\xi\rangle.(6)

The vector spectral measure μU,ξ\mu_{U,\xi} is the finite measure on σ(U)\sigma(U) obtained by Riesz representation of FU,ξF_{U,\xi}.✦ For a normal bounded operator UU on ℋ\mathcal H and a vector ξ∈ℋ\xi\in\mathcal H, the vector spectral phase Φξ(U)\Phi_\xi(U) is the real number obtained by integrating the fractional argument against the vector spectral measure μU,ξ\mu_{U,\xi}, as in (7): Φξ(U)=∫σ(U)h(z)dμU,ξ(z).\Phi_\xi(U)=\int_{\sigma(U)}h(z)\,d\mu_{U,\xi}(z).(7)✦

For a∈Aa\in A and a scalar zz of modulus one, the scalar-rotated group phase γa(z)\gamma_a(z) is the vector spectral phase at δ1\delta_1 of the unitary operator zL(a)zL(a), as in (8): γa(z)=Φδ1(zL(a)).\gamma_a(z)=\Phi_{\delta_1}(zL(a)).(8)✦

The Claim

Theorem 1. Let a∈Aa\in A, and let (xj)j≥0(x_j)_{j\geq0} be a sequence of nonnegative real numbers converging to zero. Then (9) holds: limj→∞γa(r(xj))=γa(1).\lim_{j\to\infty}\gamma_a(r(x_j))=\gamma_a(1).(9)✦

Theorem 2. Let ZZ be a finite nonempty set, let b=|Z|b=|Z|, and choose z0∈Zz_0\in Z. Let c:Z→Ac\colon Z\to A assign a boundary coefficient to each label. Suppose that, for every real ε\varepsilon with 0<ε<10<\varepsilon<1, the boundary inequality (10) holds: γc(z0)(r((b−1)ε))+(b−1)(1−ε)≤b−1.\gamma_{c(z_0)}\bigl(r((b-1)\varepsilon)\bigr) +(b-1)(1-\varepsilon)\leq b-1.(10) Then the exceptional coefficient satisfies (11): c(z0)=1.c(z_0)=1.(11)✦

What Is Checked

The continuity statement is over regular operators of an arbitrary group and nonnegative sequences tending to zero; the boundary criterion is over a finite nonempty set of labels and a coefficient assignment satisfying the inequalities for every 0<ε<10<\varepsilon<1.

A bijection preserves square-summability under the coordinate change in (2).✦ Every scalar rotation in (4) has modulus one.✦