OctoInformBeta

OpenAI preprint, September 2026 · §5. From an ultrapower embedding to a finite contradiction

Lemma 5.2: Induction supplies the dual tensor factor for the initial matching assertion

Invariant tensors and coordinate matching.

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 comparison is between invariant tensors for a group representation and the quadratic displacement of coordinate unitary tuples in a norm ultraproduct.

Let Γ\Gamma and Δ\Delta be groups. A finite-dimensional unitary representation π\pi of Γ\Gamma is a homomorphism into the unitaries of a nonzero, finite-dimensional complex Hilbert space HπH_\pi.✦ All inner products are linear in the second variable. The contragredient representation π*\pi^* of Γ\Gamma is the representation on the continuous complex-linear dual Hπ*H_\pi^* given by (1), for gg in Γ\Gamma, ℓ\ell in Hπ*H_\pi^*, and vv in HπH_\pi. (π*(g)ℓ)(v)=ℓ(π(g−1)v).\bigl(\pi^*(g)\ell\bigr)(v)=\ell\bigl(\pi(g^{-1})v\bigr).(1)✦

Fix an ultrafilter ω\omega on ℕ\mathbb N and complex C*-algebras AnA_n. The bounded sections ℓ∞(An)\ell^\infty(A_n) are the uniformly bounded coordinate families, with supremum norm and coordinatewise operations.✦ The norm ultraproduct ∏ωAn\prod_\omega A_n of the C*-algebras AnA_n is the quotient of the bounded-section algebra ℓ∞(An)\ell^\infty(A_n) by equality modulo norm-null families: ∏ωAn=ℓ∞(An)/∼ω,x∼ωy⇔limn→ω∥xn−yn∥=0.\prod_\omega A_n=\ell^\infty(A_n)/{\sim_\omega}, \qquad x\sim_\omega y\ \Longleftrightarrow\ \lim_{n\to\omega}\|x_n-y_n\|=0.(2) A coordinate unitary sequence is the bounded section (un)n(u_n)_n associated to unitaries un∈Anu_n\in A_n.✦

The initial matching setting has a unital injective full-group-algebra map into (2), coordinate unitaries un(g)∈Anu_n(g)\in A_n, and a unitary representation σ\sigma of Γ\Gamma on a Hilbert space KK. Let ρn:An→B(En)\rho_n:A_n\to B(E_n) be the supplied unital Hilbert realizations, and let S⊆ΓS\subseteq\Gamma be finite. Write unu_n for the tuple g↦un(g)g\mapsto u_n(g).

Tensor Energy and Induction

For Hilbert spaces EE and FF, the Hilbert tensor product E⊗FE\otimes F is the completion of the algebraic tensor product E⊗algFE\otimes_{\mathrm{alg}}F in its natural inner product norm.✦ For bounded operators UU and VV, the tensor product operator U⊗VU\otimes V is the continuous extension of x⊗y↦Ux⊗Vyx\otimes y\mapsto Ux\otimes Vy; when UU and VV are unitary this is their tensor product unitary.

The matching unitary tuple Un,πU_{n,\pi} for unu_n and π\pi on En⊗HπE_n\otimes H_\pi consists of the operators Un,π(g)=ρn(un(g))⊗π(g)U_{n,\pi}(g)=\rho_n(u_n(g))\otimes\pi(g).✦ For a real threshold aa, the quadratic displacement energy ℰS,n,π\mathcal E_{S,n,\pi} of the tuple Un,πU_{n,\pi} on En⊗HπE_n\otimes H_\pi and the strict matching condition Match⁡ρn,S,a(un,π)\operatorname{Match}_{\rho_n,S,a}(u_n,\pi) are defined by ℰS,n,π(x)=∑s∈S∥Un,π(s)x−x∥2,Match⁡ρn,S,a(un,π)⇔there is x∈En⊗Hπ with ∥x∥=1 and ℰS,n,π(x)<a.\begin{aligned} \mathcal E_{S,n,\pi}(x) &=\sum_{s\in S}\|U_{n,\pi}(s)x-x\|^2,\\ \operatorname{Match}_{\rho_n,S,a}(u_n,\pi) &\ \Longleftrightarrow\ \text{there is }x\in E_n\otimes H_\pi \text{ with }\|x\|=1\text{ and }\mathcal E_{S,n,\pi}(x)<a. \end{aligned}(3) The convention in (3) uses π\pi directly in the second tensor factor; the paper uses the conjugate representation, so its initial representation is replaced by its dual in this convention.

Let f:Γ→Δf:\Gamma\to\Delta be an injective homomorphism and π0\pi_0 a finite-dimensional unitary representation of Γ\Gamma. The image subgroup HH of Δ\Delta is H=f(Γ)H=f(\Gamma), and the unitary representation α\alpha of HH on Hπ0H_{\pi_0} is defined by α(f(g))=π0(g)\alpha(f(g))=\pi_0(g). A normalized coset transversal is a choice ri∈Δr_i\in\Delta representing each left coset i∈Δ/Hi\in\Delta/H, with rH=1r_H=1.✦ The transversal cocycle cc of the normalized coset transversal (ri)i∈Δ/H(r_i)_{i\in\Delta/H} is the function from Δ×(Δ/H)\Delta\times(\Delta/H) to HH given by c(d,i)=rdi−1dri∈H(d∈Δ,i∈Δ/H).c(d,i)=r_{di}^{-1}d r_i\in H \qquad(d\in\Delta,\ i\in\Delta/H).(4)✦ The induced representation Σf\Sigma_f of Δ\Delta along ff is the unitary representation on the square-summable direct sum Kf=ℓ2(Δ/H,Hπ0)K_f=\ell^2(\Delta/H,H_{\pi_0}) given by (Σf(d)ξ)(i)=α(c(d,d−1i))ξ(d−1i).(\Sigma_f(d)\xi)(i) =\alpha\bigl(c(d,d^{-1}i)\bigr)\xi(d^{-1}i).(5)✦

The Claim

Theorem 1. Let f:Γ→Δf:\Gamma\to\Delta be injective and let π0\pi_0 be a finite-dimensional unitary representation of Γ\Gamma. There is a nonzero tensor z0∈Kf⊗algHπ0*z_0\in K_f\otimes_{\mathrm{alg}}H_{\pi_0^*} such that (Σf(f(g))⊗π0*(g))z0=z0(g∈Γ).\bigl(\Sigma_f(f(g))\otimes\pi_0^*(g)\bigr)z_0=z_0 \qquad(g\in\Gamma).(6)✦

In the initial matching setting, let π\pi be a finite-dimensional unitary representation of Γ\Gamma and let 0<a≤R0<a\le R, where R=4|S|R=4|S|. Suppose that a nonzero z∈K⊗algHπz\in K\otimes_{\mathrm{alg}}H_\pi satisfies (σ(s)⊗π(s))z=z(\sigma(s)\otimes\pi(s))z=z for every s∈Ss\in S. Then the strict matching condition of (3) holds on an ω\omega-large set: {n∈ℕ:Match⁡ρn,S,a(un,π)}∈ω.\bigl\{n\in\mathbb N: \operatorname{Match}_{\rho_n,S,a}(u_n,\pi)\bigr\}\in\omega.(7)✦

In the paper, the result reads: under the assumed unital embedding, lim⁡n→ωd(xn,π0*)=0\lim_{n\to\omega}d(x_n,\pi_0^*)=0 and consequently {n:[π0*]∈Dn}∈ω\{n:[\pi_0^*]\in D_n\}\in\omega, where xnx_n and DnD_n are its coordinate tuples and matching sets, dd is its spectral distance, and [π0*][\pi_0^*] denotes the unitary equivalence class of π0*\pi_0^*.

What Is Checked

The assertions (6) and (7) are stated over induction along an injective group homomorphism and the unital ultraproduct matching setting, respectively, with a nonzero finite-dimensional representation in each case. The norm-null congruence respects addition and multiplication, and coordinate unitary sequences satisfy both adjoint inverse identities in the bounded-section algebra.