OctoInformBeta

OpenAI preprint, September 2026 · §2. Spectral phase and its trace

§2 (displayed): faithfulness of phase: For unitaries in the regular matrix algebra and for the left regular representation of a group

Zero phase detects the identity.

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

Let AA be a group and let DD be a finite, nonempty set of matrix coordinates. The question is whether the spectral phase of a regular operator can distinguish a nonidentity group element from the identity. The regular Hilbert space ℋA\mathcal H_A of AA and the matrix Hilbert space ℋD\mathcal H_D with coordinate set DD are the spaces of square-summable complex families given by ℋA=ℓ2(A,ℂ),ℋD=ℓ2(D×A,ℂ).\mathcal H_A=\ell^2(A,\mathbb C), \qquad \mathcal H_D=\ell^2(D\times A,\mathbb C).(1)✦ Here ℓ2(α,ℂ)\ell^2(\alpha,\mathbb C) consists of the square-summable complex families on an index set α\alpha. For a complex Hilbert space 𝒦\mathcal K, the bounded operators ℬ(𝒦)\mathcal B(\mathcal K) are its continuous complex-linear maps to itself.

For an index set α\alpha and i∈αi\in\alpha, the delta vector δi∈ℓ2(α,ℂ)\delta_i\in\ell^2(\alpha,\mathbb C) is the function δi(j)={1,j=i,0,j≠i.\delta_i(j)= \begin{cases} 1,&j=i,\\ 0,&j\ne i. \end{cases}(2)✦ For a bijection e:α→βe:\alpha\longrightarrow\beta, the linear coordinate permutation Pe:ℓ2(α,ℂ)→ℓ2(β,ℂ)P_e:\ell^2(\alpha,\mathbb C)\longrightarrow \ell^2(\beta,\mathbb C) is the complex-linear equivalence defined by (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)).(3) The left regular operator L(a)L(a) on ℋA\mathcal H_A and the diagonal right translation R(a)R(a) on ℋD\mathcal H_D, for a∈Aa\in A, are the coordinate permutations specified, for x∈Ax\in A and d∈Dd\in D, on the delta vectors of (2) by L(a)δx=δax,R(a)δ(d,x)=δ(d,xa−1).L(a)\delta_x=\delta_{ax}, \qquad R(a)\delta_{(d,x)}=\delta_{(d,xa^{-1})}.(4)

For a set 𝒮⊆ℬ(𝒦)\mathscr S\subseteq\mathcal B(\mathcal K), the operator centralizer 𝒮′\mathscr S' of 𝒮\mathscr S is the unital complex star subalgebra of bounded operators on 𝒦\mathcal K commuting with every member of 𝒮\mathscr S and with the adjoint of every member of 𝒮\mathscr S.✦ 𝒮′={T∈ℬ(𝒦):TV=VT for every V∈𝒮}.\mathscr S' =\{T\in\mathcal B(\mathcal K):TV=VT\text{ for every }V\in\mathscr S\}.(5) The regular matrix algebra M⊆ℬ(ℋD)M\subseteq\mathcal B(\mathcal H_D) is the commutant of the diagonal right translations in (4): M={R(a):a∈A}′.M=\{R(a):a\in A\}'.(6)✦

The Spectral Phase

The fractional phase h:ℂ→ℝh:\mathbb C\longrightarrow\mathbb R is the fractional part of the principal argument divided by 2π2\pi: h(z)=arg⁡z2π−⌊arg⁡z2π⌋,h(e2πix)=x−⌊x⌋(x∈ℝ),h(1)=0.\begin{aligned} h(z)&=\frac{\arg z}{2\pi} -\left\lfloor\frac{\arg z}{2\pi}\right\rfloor,\\ h(e^{2\pi i x})&=x-\lfloor x\rfloor\quad(x\in\mathbb R), \qquad h(1)=0. \end{aligned}(7)✦

For a continuous real-valued function ff on a topological space XX, its complexification fℂf_{\mathbb C} is the continuous complex-valued function fℂ(x)=f(x)f_{\mathbb C}(x)=f(x).✦

For a normal bounded operator UU on a complex Hilbert space 𝒦\mathcal K and a vector ξ∈𝒦\xi\in\mathcal K, the spectral vector functional FU,ξF_{U,\xi} is the positive real-linear functional on compactly supported continuous real-valued functions on the spectrum σ(U)\sigma(U) given by FU,ξ(f)=Re⁡⟨ξ,fℂ(U)ξ⟩.F_{U,\xi}(f) =\operatorname{Re}\langle\xi,f_{\mathbb C}(U)\xi\rangle.(8)✦ In (8), fℂ(U)f_{\mathbb C}(U) denotes continuous functional calculus. The vector spectral measure μU,ξ\mu_{U,\xi} on σ(U)\sigma(U) is the representing Riesz measure of FU,ξF_{U,\xi}: ∫σ(U)f(z)dμU,ξ(z)=FU,ξ(f).\int_{\sigma(U)}f(z)\,d\mu_{U,\xi}(z)=F_{U,\xi}(f).(9)✦ For a normal operator U∈ℬ(𝒦)U\in\mathcal B(\mathcal K) and a vector ξ∈𝒦\xi\in\mathcal K, the vector phase ℓξ(U)\ell_\xi(U) is the integral of the fractional phase (7) against the vector spectral measure μU,ξ\mu_{U,\xi}✦; for a finite family (ξj)j∈J(\xi_j)_{j\in J} in 𝒦\mathcal K, the summed vector phase ℓ(ξj)(U)\ell_{(\xi_j)}(U) is the sum of the individual vector phases: ℓξ(U)=∫σ(U)h(z)dμU,ξ(z),ℓ(ξj)(U)=∑j∈Jℓξj(U).\ell_\xi(U)=\int_{\sigma(U)}h(z)\,d\mu_{U,\xi}(z), \qquad \ell_{(\xi_j)}(U)=\sum_{j\in J}\ell_{\xi_j}(U).(10)

For a normal matrix operator U∈ℬ(ℋD)U\in\mathcal B(\mathcal H_D), its unnormalized phase trace ℓ(U)\ell(U) is the summed vector phase at the identity coefficient in each matrix coordinate: ℓ(U)=∑d∈Dℓδ(d,1)(U).\ell(U)=\sum_{d\in D}\ell_{\delta_{(d,1)}}(U).(11)✦ For a∈Aa\in A and z∈ℂz\in\mathbb C with |z|=1|z|=1, the scalar-rotated group phase ℓz(a)\ell_z(a) is the vector phase of zL(a)zL(a) at δ1\delta_1: ℓz(a)=ℓδ1(zL(a)).\ell_z(a)=\ell_{\delta_1}(zL(a)).(12)✦

The Claim

Theorem 1. Let AA be a group and let DD be a finite, nonempty coordinate set. Let II be the identity operator on ℋD\mathcal H_D. For a unitary UU in the regular matrix algebra MM, ℓ(U)=0⇔U=I,\ell(U)=0\quad\Longleftrightarrow\quad U=I,(13)✦ For an element a∈Aa\in A, ℓ1(a)=0⇔a=1.\ell_1(a)=0\quad\Longleftrightarrow\quad a=1.(14)✦

What Is Checked

The assertions (13) and (14) are stated over, respectively, unitaries of the commutant (6) with a finite, nonempty coordinate set and group elements represented by (4), with phases defined by (11) and (12).

Coordinate permutation (3) preserves square-summability and inner products; these maps are unitary coordinate permutations of the corresponding Hilbert spaces.

For every set 𝒮\mathscr S of bounded operators, an element of its commutant (5) is unitary in that algebra exactly when it is unitary as an operator on 𝒦\mathcal K.✦