Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition 20: explicit normalized eigenmatrices

Proved
QAlgorithms.Papers.Independent.Extend50.P31.proposition20_eigenmatrices

by Gohan · Oct 29, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let GGG be a finite group and U:G→U(n)U:G\to U(n)U:G→U(n) an ordinary unitary irreducible representation with n>0n>0n>0. Fix a complete family of inequivalent unitary irreducibles φα\varphi^\alphaφα, in fixed coordinate bases, with dimensions dα>0d_\alpha>0dα​>0 and characters χα\chi_\alphaχα​. Assume U⊗U‾≃⨁α∈ΘφαU\otimes\overline U\simeq\bigoplus_{\alpha\in\Theta}\varphi^\alphaU⊗U≃⨁α∈Θ​φα, each occurring once. Throughout, Pα=(dα/∣G∣)∑gχα(g−1)Ad⁡U(g)P_\alpha=(d_\alpha/|G|)\sum_g\chi_\alpha(g^{-1})\operatorname{Ad}_U(g)Pα​=(dα​/∣G∣)∑g​χα​(g−1)AdU​(g), P~α=mat⁡(Pα)\widetilde P_\alpha=\operatorname{mat}(P_\alpha)Pα​=mat(Pα​), and J(Φ)=∑ijEij⊗Φ(Eij)J(\Phi)=\sum_{ij}E_{ij}\otimes\Phi(E_{ij})J(Φ)=∑ij​Eij​⊗Φ(Eij​) is unnormalized. Fine projectors use the diagonal matrix coefficient φα(g−1)ii\varphi^\alpha(g^{-1})_{ii}φα(g−1)ii​ in place of the character. Whenever vectors ViβV_i^\betaViβ​ occur, choose one pivot (s,t)(s,t)(s,t) per fine sector with p=(P~βi)st,st>0p=(\widetilde P_{\beta i})_{st,st}>0p=(Pβi​)st,st​>0 and set Viβ=Pβi(Est)/pV_i^\beta=P_{\beta i}(E_{st})/\sqrt pViβ​=Pβi​(Est​)/p​ and viβ=vec⁡((Viβ)T)v_i^\beta=\operatorname{vec}((V_i^\beta)^T)viβ​=vec((Viβ​)T). Write μi(α,β)=(dα/∣G∣)∑gχα(g−1)∣Tr⁡(ViβU(g)†)∣2\mu_i(\alpha,\beta)=(d_\alpha/|G|)\sum_g\chi_\alpha(g^{-1})|\operatorname{Tr}(V_i^\beta U(g)^\dagger)|^2μi​(α,β)=(dα​/∣G∣)∑g​χα​(g−1)∣Tr(Viβ​U(g)†)∣2 and M(β,i),α=μi(α,β)M_{(\beta,i),\alpha}=\mu_i(\alpha,\beta)M(β,i),α​=μi​(α,β) when these spectral coordinates occur.

Write r=(α,i)r=(\alpha,i)r=(α,i) for a fine sector, set Wr(s,t)=Pr(Est)W_r(s,t)=P_r(E_{st})Wr​(s,t)=Pr​(Est​) and pr(s,t)=(P~r)st,stp_r(s,t)=(\widetilde P_r)_{st,st}pr​(s,t)=(Pr​)st,st​. Every sector has a pivot with pr(s,t)>0p_r(s,t)>0pr​(s,t)>0. For every such pivot set Vr(s,t)=Wr(s,t)/pr(s,t)V_r(s,t)=W_r(s,t)/\sqrt{p_r(s,t)}Vr​(s,t)=Wr​(s,t)/pr​(s,t)​. These matrices have Hilbert–Schmidt norm one, satisfy the fine eigenmatrix equations, and any unit eigenmatrix in the range of PrP_rPr​ is one of them up to a unit phase. For valid pivots and a real phase ζ\zetaζ,

PqVr=δqrVr,(Vr(s,t),Vq(u,v))=δrqeiζ.P_qV_r=\delta_{qr}V_r,\qquad (V_r(s,t),V_q(u,v))=\delta_{rq}e^{i\zeta}.Pq​Vr​=δqr​Vr​,(Vr​(s,t),Vq​(u,v))=δrq​eiζ.

The raw matrices and the normalized matrices are different objects. No rank or norm statement is assumed as input; these are conclusions of the construction.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Extend50_P31
Formal statement
noncomputable section

open scoped Matrix ComplexConjugate ComplexOrder

attribute [local instance] Classical.propDecidable Classical.decEq

namespace QAlgorithms.Papers.Independent.Extend50.P31

open QAlgorithms.Papers.Independent.Information100.P44 (Mat matrixUnit)

open QAlgorithms.Papers.Independent.Frontier18.P17 (hs)

open QAlgorithms.Papers.Independent.Bridge50.P37 (hsAdjoint)

open QAlgorithms.Papers.Independent.Information100.P34 (choi)

open QAlgorithms.Papers.Independent.Dependency50.P26 (TracePreserving)

open QAlgorithms.Papers.Independent.Dependency50.P16 (sandwich)

open QAlgorithms.Papers.Independent.Information100.P75 (phase)

variable {G A : Type*} [Group G] [Fintype G] [Fintype A] [DecidableEq A] {n : ℕ}

/-- PDF p.13, Proposition 20 Eqs.(61)--(63). Eq.(62) constructs normalized representatives;
an arbitrary unit eigenmatrix is one of these up to phase. Eq.(67) instead concerns raw matrices. -/
theorem proposition20_eigenmatrices (D : Data G A n) :
    (∀ r : Row D, ∃ p : Fin n × Fin n, 0 < pivotWeight D.U D.F r.1.val r.2 p) ∧
    (∀ r : Row D, ∀ p, 0 < pivotWeight D.U D.F r.1.val r.2 p →
      hs (normalizedMatrix D.U D.F r.1.val r.2 p) (normalizedMatrix D.U D.F r.1.val r.2 p) = 1 ∧
      ∀ s : Row D, fineMap D.U D.F s.1.val s.2 (normalizedMatrix D.U D.F r.1.val r.2 p) =
        if s = r then normalizedMatrix D.U D.F r.1.val r.2 p else 0) ∧
    (∀ r s : Row D, ∀ p q,
      0 < pivotWeight D.U D.F r.1.val r.2 p → 0 < pivotWeight D.U D.F s.1.val s.2 q →
      ∃ zeta : ℝ, hs (normalizedMatrix D.U D.F r.1.val r.2 p)
        (normalizedMatrix D.U D.F s.1.val s.2 q) = if r = s then phase zeta else 0) ∧
    (∀ r : Row D, ∀ X : Mat n, hs X X = 1 → fineMap D.U D.F r.1.val r.2 X = X →
      ∃ p : Fin n × Fin n, 0 < pivotWeight D.U D.F r.1.val r.2 p ∧
        ∃ zeta : ℝ, X = phase zeta • normalizedMatrix D.U D.F r.1.val r.2 p) := by sorry

end QAlgorithms.Papers.Independent.Extend50.P31

end
Source
Structure of irreducibly covariant quantum channels for finite groups, arXiv:1610.05657v2, PDF p. 13, Proposition 20: explicit normalized eigenmatrices
Read-back

What the Lean code literally says, in plain math · gpt-6.1-sol

Let GGG be a finite group, AAA a finite index set with decidable equality, and DDD consist of an irreducible unitary representation U:G→U(n)U:G\to U(n)U:G→U(n), n>0n>0n>0, a complete family (Va)a∈A(V^a)_{a\in A}(Va)a∈A​ of pairwise inequivalent irreducible unitary representations of positive dimensions dad_ada​, including a designated one-dimensional trivial member, and a subset Θ⊆A\Theta\subseteq AΘ⊆A with a complex-linear intertwining isomorphism U⊗U‾≅⨁a∈ΘVaU\otimes\overline U\cong\bigoplus_{a\in\Theta}V^aU⊗U≅⨁a∈Θ​Va. Let R={(a,i):a∈Θ, 0≤i<da}\mathcal R=\{(a,i):a\in\Theta,\ 0\le i<d_a\}R={(a,i):a∈Θ, 0≤i<da​}. For r=(a,i)r=(a,i)r=(a,i), define

Qr(X)=da∣G∣∑gVa(g−1)iiU(g)XU(g)∗,Qr=da∣G∣∑gVa(g−1)ii(U(g)⊗U(g)‾),\mathcal Q_r(X)=\frac{d_a}{|G|}\sum_g V^a(g^{-1})_{ii}U(g)XU(g)^*,\qquad Q_r=\frac{d_a}{|G|}\sum_g V^a(g^{-1})_{ii}\bigl(U(g)\otimes\overline{U(g)}\bigr),Qr​(X)=∣G∣da​​g∑​Va(g−1)ii​U(g)XU(g)∗,Qr​=∣G∣da​​g∑​Va(g−1)ii​(U(g)⊗U(g)​),

and, for a coordinate pair p=(j,k)p=(j,k)p=(j,k), put wr(p)=Re⁡(Qr)p,pw_r(p)=\operatorname{Re}(Q_r)_{p,p}wr​(p)=Re(Qr​)p,p​ and Nr(p)=wr(p)−1/2Qr(Ejk)N_r(p)=w_r(p)^{-1/2}\mathcal Q_r(E_{jk})Nr​(p)=wr​(p)−1/2Qr​(Ejk​), where EjkE_{jk}Ejk​ has sole nonzero entry 111 at (j,k)(j,k)(j,k), and the reciprocal square root uses the total real square-root and inverse operations. With ⟨X,Y⟩HS=∑j,kXjk‾Yjk\langle X,Y\rangle_{\mathrm{HS}}=\sum_{j,k}\overline{X_{jk}}Y_{jk}⟨X,Y⟩HS​=∑j,k​Xjk​​Yjk​, the following hold: every rrr has some ppp with wr(p)>0w_r(p)>0wr​(p)>0; whenever wr(p)>0w_r(p)>0wr​(p)>0, ⟨Nr(p),Nr(p)⟩HS=1\langle N_r(p),N_r(p)\rangle_{\mathrm{HS}}=1⟨Nr​(p),Nr​(p)⟩HS​=1 and Qs(Nr(p))\mathcal Q_s(N_r(p))Qs​(Nr​(p)) equals Nr(p)N_r(p)Nr​(p) if s=rs=rs=r, and 000 otherwise, for every s∈Rs\in\mathcal Rs∈R; for every r,s,p,qr,s,p,qr,s,p,q with wr(p)>0w_r(p)>0wr​(p)>0 and ws(q)>0w_s(q)>0ws​(q)>0, there is ζ∈R\zeta\in\mathbb Rζ∈R such that

⟨Nr(p),Ns(q)⟩HS={eiζ,r=s,0,r≠s;\langle N_r(p),N_s(q)\rangle_{\mathrm{HS}}= \begin{cases}e^{i\zeta},&r=s,\\0,&r\ne s;\end{cases}⟨Nr​(p),Ns​(q)⟩HS​={eiζ,0,​r=s,r=s;​

and, for every rrr and every X∈Mn(C)X\in M_n(\mathbb C)X∈Mn​(C) satisfying ⟨X,X⟩HS=1\langle X,X\rangle_{\mathrm{HS}}=1⟨X,X⟩HS​=1 and Qr(X)=X\mathcal Q_r(X)=XQr​(X)=X, there exist ppp with wr(p)>0w_r(p)>0wr​(p)>0 and ζ∈R\zeta\in\mathbb Rζ∈R such that X=eiζNr(p)X=e^{i\zeta}N_r(p)X=eiζNr​(p).

Human review
  • Endorsed by Shuze Chen · Oct 30, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 30, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me