OctoInformBeta

OpenAI preprint, September 2026 · §3. A spectral test for representation matching

Kazhdan energy, Laplacians and matching: Kazhdan gaps and bounded positive Laplacians for unitary tuples in a supplied Hilbert realization

Displacement energy defines strict representation 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

A unitary family in a C*-algebra can be compared with a finite-dimensional group representation through the displacement of vectors in their Hilbert tensor product. Let Γ\Gamma be a group, S⊆ΓS\subseteq\Gamma a finite subset, and AA a unital complex C*-algebra. Fix a family u:Γ→𝒰(A)u\colon\Gamma\to\mathcal U(A) of unitaries and a unital **-representation ρ:A→ℬ(E)\rho\colon A\to\mathcal B(E) on a complex Hilbert space EE. Inner products are linear in the second variable. Let κ,a∈ℝ\kappa,a\in\mathbb R and write R=4|S|R=4|S|.

A finite-dimensional unitary representation π\pi of Γ\Gamma is a homomorphism π:Γ→𝒰(Hπ)\pi\colon\Gamma\to\mathcal U(H_\pi) on a nonzero finite-dimensional complex Hilbert space HπH_\pi.✦ Two finite-dimensional unitary representations π\pi and τ\tau of Γ\Gamma are unitarily equivalent when there is a unitary T:Hπ→HτT\colon H_\pi\to H_\tau satisfying Tπ(g)=τ(g)TT\pi(g)=\tau(g)T for every g∈Γg\in\Gamma.✦ The set ℛ(Γ)\mathcal R(\Gamma) is the set of nonzero finite-dimensional unitary representations of Γ\Gamma modulo unitary equivalence; [π][\pi] denotes the class of a representation π\pi.✦

A finite-dimensional unitary representation π\pi of Γ\Gamma on HπH_\pi is irreducible when its only π(Γ)\pi(\Gamma)-invariant linear subspaces are {0}\{0\} and HπH_\pi.✦

For complex inner product spaces EE and FF with their induced norms, the Hilbert tensor product E⊗ˆFE\widehat\otimes F is the completion of E⊗ℂFE\otimes_{\mathbb C}F in the Hilbert tensor norm.✦ For bounded operators T∈ℬ(E)T\in\mathcal B(E) and V∈ℬ(F)V\in\mathcal B(F), the tensor product operator T⊗ˆVT\widehat\otimes V is the continuous extension specified by (1). (T⊗ˆV)(ξ⊗η)=Tξ⊗Vη(ξ∈E,η∈F).(T\widehat\otimes V)(\xi\otimes\eta)=T\xi\otimes V\eta \qquad(\xi\in E,\ \eta\in F).(1)✦

The library uses π\pi directly in the second tensor factor; passage from the paper’s initial representation to this notation takes its dual.

The Claim

Definition 1. A pair (S,κ)(S,\kappa) consisting of a finite subset SS of a group Γ\Gamma and a real number κ\kappa has a Kazhdan gap when every unitary representation σ\sigma of Γ\Gamma on a complex Hilbert space HH with no nonzero invariant vectors satisfies (2). κ∥ξ∥2≤∑s∈S∥σ(s)ξ−ξ∥2(ξ∈H).\kappa\|\xi\|^2\leq\sum_{s\in S}\|\sigma(s)\xi-\xi\|^2 \qquad(\xi\in H).(2)✦

For an arbitrary family v:Γ→Av\colon\Gamma\to A, the displacement Laplacian LS(v)L_S(v) is the element of AA given by (3). LS(v)=∑s∈S(vs−1)*(vs−1).L_S(v)=\sum_{s\in S}(v_s-1)^*(v_s-1).(3)✦ The order bounds for an arbitrary family vv and a unitary family uu are (4) and (5), respectively. 0≤LS(v).0\leq L_S(v).(4)✦ LS(u)≤R1A.L_S(u)\leq R1_A.(5)✦

For a unitary family U:Γ→𝒰(H)U\colon\Gamma\to\mathcal U(H) on a complex Hilbert space HH, the quadratic displacement energy ℰS(U,x)\mathcal E_S(U,x) of a vector x∈Hx\in H is the real number given by (6). ℰS(U,x)=∑s∈S∥Usx−x∥2.\mathcal E_S(U,x)=\sum_{s\in S}\|U_sx-x\|^2.(6)✦ The bottom Rayleigh energy λS(U)\lambda_S(U) of a unitary family U:Γ→𝒰(H)U\colon\Gamma\to\mathcal U(H) on a complex Hilbert space HH is the real number given by the infimum in (7), with the empty infimum taken to be zero. λS(U)=inf⁡{ℰS(U,x):x∈H,∥x∥=1}.\lambda_S(U)=\inf\{\mathcal E_S(U,x):x\in H,\ \|x\|=1\}.(7)✦

For a unitary family u:Γ→𝒰(A)u\colon\Gamma\to\mathcal U(A), a unital representation ρ:A→ℬ(E)\rho\colon A\to\mathcal B(E) and a finite-dimensional unitary representation π\pi of Γ\Gamma on HπH_\pi, the matching tuple UπU^\pi is the family of unitaries on E⊗ˆHπE\widehat\otimes H_\pi given by (8). Ugπ=ρ(ug)⊗ˆπ(g)(g∈Γ).U_g^\pi=\rho(u_g)\widehat\otimes\pi(g) \qquad(g\in\Gamma).(8)✦ Strict matching Match⁡a(ρ,u;π)\operatorname{Match}_a(\rho,u;\pi) of a finite-dimensional unitary representation π\pi of Γ\Gamma with a unitary family u:Γ→𝒰(A)u\colon\Gamma\to\mathcal U(A) at a real threshold aa in a unital Hilbert realization ρ:A→ℬ(E)\rho\colon A\to\mathcal B(E) is the condition in (9). Match⁡a(ρ,u;π)quad⇔∃x∈E⊗ˆHπ,∥x∥=1 and ℰS(Uπ,x)<a.\operatorname{Match}_a(\rho,u;\pi)quad\Longleftrightarrow\quad \exists x\in E\widehat\otimes H_\pi,\quad \|x\|=1\ \text{ and }\ \mathcal E_S(U^\pi,x)<a.(9)✦ The matching irreducible classes constitute the subset ℳa(ρ,u)⊆ℛ(Γ)\mathcal M_a(\rho,u)\subseteq\mathcal R(\Gamma) defined by (10). ℳa(ρ,u)={[π]:π is irreducible and Match⁡a(ρ,u;π)}.\mathcal M_a(\rho,u)= \{[\pi]:\pi\text{ is irreducible and }\operatorname{Match}_a(\rho,u;\pi)\}.(10)✦

In the paper, the result reads: for a finitely generated discrete property (T) group Γ\Gamma, choose a nonempty finite generating set SS containing a Kazhdan set and 0<κ≤R=4|S|0<\kappa\leq R=4|S| such that every unitary representation σ\sigma without nonzero invariant vectors satisfies ∑s∈S∥σ(s)ξ−ξ∥2≥κ∥ξ∥2\sum_{s\in S}\|\sigma(s)\xi-\xi\|^2\geq\kappa\|\xi\|^2 for every vector ξ\xi; for a unitary SS-tuple zz in a nonzero unital C*-algebra, put L(z)=∑s∈S(zs−1)*(zs−1)L(z)=\sum_{s\in S}(z_s-1)^*(z_s-1), so 0≤L(z)≤R10\leq L(z)\leq R1; for a nonzero finite-dimensional unitary representation π\pi, define d(z,π)=min⁡spec⁡L((zs⊗π(s)¯)s∈S)d(z,\pi)=\min\operatorname{spec}L((z_s\otimes\overline{\pi(s)})_{s\in S}); and fix b=κ2/(4|S|)b=\kappa^2/(4|S|) and 0<a<min⁡(κ,b)0<a<\min(\kappa,b).

What Is Checked

The definitions are stated for arbitrary real κ,a\kappa,a, a finite subset SS of a group, a unitary family uu in a unital complex C*-algebra and a supplied unital Hilbert realization ρ\rho, with π\pi nonzero and finite-dimensional.

For unitary operators TT and VV, the completed tensor product operator in (1) is a unitary on the Hilbert tensor product.✦