OctoInformBeta

OpenAI preprint, September 2026 · §4. Finiteness of approximate representation matches

Lemma 4.2: A positive Kazhdan gap controls energies uniformly over arbitrary unitary families on the other factor

A tensor gap for weakly mixing representations.

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

Tensor energy measures the failure of a vector in a Hilbert tensor product to be fixed by a finite family of tensor operators. Let Γ\Gamma be a group, let S⊆ΓS\subseteq\Gamma be finite and nonempty, and let κ>0\kappa>0. All Hilbert spaces are complex, with inner products linear in the second variable.

The pair (S,κ)(S,\kappa) has a Kazhdan energy gap if every unitary representation ρ:Γ→𝒰(H)\rho\colon\Gamma\to\mathcal U(H) on a Hilbert space HH, with no nonzero invariant vector, satisfies (1) for every η∈H\eta\in H: κ∥η∥2≤∑s∈S∥ρ(s)η−η∥2.\kappa\|\eta\|^2 \leq \sum_{s\in S}\|\rho(s)\eta-\eta\|^2.(1)✦

A unitary representation σ:Γ→𝒰(K)\sigma\colon\Gamma\to\mathcal U(K) is weakly mixing if KK has no nonzero finite-dimensional linear subspace invariant under every σ(g)\sigma(g), g∈Γg\in\Gamma.✦

For complex inner product spaces EE and KK, the Hilbert tensor product E⊗ˆKE\widehat\otimes K is the completion of their algebraic tensor product in the Hilbert tensor norm.✦ Unitaries u∈𝒰(E)u\in\mathcal U(E) and v∈𝒰(K)v\in\mathcal U(K) on complex inner product spaces EE and KK determine the tensor unitary action u⊗vu\otimes v, the continuous linear operator on E⊗ˆKE\widehat\otimes K given by (2): (u⊗v)(x⊗y)=ux⊗vy(x∈E,y∈K).(u\otimes v)(x\otimes y)=ux\otimes vy \qquad (x\in E,\ y\in K).(2)✦

The second factor uses σ\sigma itself; the paper uses the conjugate representation in its matching convention.

For a unitary SS-tuple V=(Vs)s∈SV=(V_s)_{s\in S} on a Hilbert space, its energy operator L(V)L(V) is defined by (3), where II is the identity operator: L(V)=∑s∈S(Vs−I)*(Vs−I).L(V)=\sum_{s\in S}(V_s-I)^*(V_s-I).(3)

Use the constants in (4): R=4|S|,b=κ2R=κ24|S|.R=4|S|,\qquad b=\frac{\kappa^2}{R}=\frac{\kappa^2}{4|S|}.(4)

The Claim

Theorem 1. Suppose (S,κ)(S,\kappa) has a Kazhdan energy gap, with SS finite and nonempty and κ>0\kappa>0. Let σ:Γ→𝒰(K)\sigma\colon\Gamma\to\mathcal U(K) be weakly mixing, and let U=(Us)s∈ΓU=(U_s)_{s\in\Gamma} be an arbitrary family of unitaries on a Hilbert space EE. For every ξ∈E⊗ˆK\xi\in E\widehat\otimes K, the bound is (5): b∥ξ∥2≤∑s∈S∥(Us⊗σ(s))ξ−ξ∥2.b\|\xi\|^2 \leq \sum_{s\in S}\|(U_s\otimes\sigma(s))\xi-\xi\|^2.(5)✦

In the paper, the result reads: if Γ,S,κ\Gamma,S,\kappa satisfy the Kazhdan energy gap and σ:Γ→𝒰(K)\sigma\colon\Gamma\to\mathcal U(K) has no nonzero finite-dimensional subrepresentation, then every Hilbert space EE and every unitary SS-tuple (Us)s∈S(U_s)_{s\in S} on EE, with no relations required among the UsU_s, satisfy L((Us⊗σ(s))s∈S)≥bIL((U_s\otimes\sigma(s))_{s\in S})\geq bI.

What Is Checked

The claim is stated for complex Hilbert spaces and the tensor unitary action on their completed Hilbert tensor product, under a positive Kazhdan energy gap on a nonempty finite set and the absence of nonzero finite-dimensional invariant subspaces in σ\sigma.