Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Tensor powers retain the private-state structure with constructed data

Proved
QAlgorithms.Papers.Independent.Gap48.P42.private_tensor_closure

by Goku · Oct 22, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let d,a,bd,a,bd,a,b be positive integers, let ρ\rhoρ be a positive semidefinite trace-one shield state on Ca⊗Cb\mathbb C^a\otimes\mathbb C^bCa⊗Cb, and let each UiU_iUi​ be a unitary on that joint shield. Write

γ(d)=1d∑i,j=0d−1∣ii⟩⟨jj∣⊗UiρUj†,\gamma^{(d)}=\frac1d\sum_{i,j=0}^{d-1}|ii\rangle\langle jj|\otimes U_i\rho U_j^\dagger,γ(d)=d1​i,j=0∑d−1​∣ii⟩⟨jj∣⊗Ui​ρUj†​,

with the bipartition AA′:BB′AA':BB'AA′:BB′ and local shield dimensions a,ba,ba,b. All logarithms and resource rates are in bits.

For every positive integer mmm, use the same fixed bijection between key tuples i∈{0,…,d−1}m\mathbf i\in\{0,\ldots,d-1\}^mi∈{0,…,d−1}m and labels in {0,…,dm−1}\{0,\ldots,d^m-1\}{0,…,dm−1} on both owners. Construct ρm=ρ⊗m\rho_m=\rho^{\otimes m}ρm​=ρ⊗m and Vi=⨂t=1mUitV_{\mathbf i}=\bigotimes_{t=1}^m U_{i_t}Vi​=⨂t=1m​Uit​​. Then ρm\rho_mρm​ is a density operator, each ViV_{\mathbf i}Vi​ is unitary, and after regrouping all key and shield registers within each owner,

regroup⁡((γ(d))⊗m)=1dm∑i,j∣ii⟩⟨jj∣⊗ViρmVj†.\operatorname{regroup}\bigl((\gamma^{(d)})^{\otimes m}\bigr) =\frac1{d^m}\sum_{\mathbf i,\mathbf j}|\mathbf i\mathbf i\rangle\langle\mathbf j\mathbf j|\otimes V_{\mathbf i}\rho_mV_{\mathbf j}^{\dagger}.regroup((γ(d))⊗m)=dm1​i,j∑​∣ii⟩⟨jj∣⊗Vi​ρm​Vj†​.

Thus the resulting state has private key dimension dmd^mdm with the computed shield and twists. This structural identity supplies the tensor-power input to the cost bound.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Gap48_P42
Formal statement
namespace QAlgorithms.Papers.Independent.Gap48.P42

open scoped BigOperators Matrix ComplexConjugate ComplexOrder ENNReal

noncomputable section

attribute [local instance] Classical.propDecidable Classical.decEq

/-- PDF p.4, observation immediately before Property 1a: the actual regrouped
m-copy matrix is private with computed shield rho^tensor m and tensor twists.
The conclusion certifies normalization and both unitary identities of those
computed twists; no unrelated private target is existentially introduced. -/
theorem private_tensor_closure {d a b : ℕ}
    (hd : 0 < d) (ha : 0 < a) (hb : 0 < b)
    (rho : Matrix (Fin a × Fin b) (Fin a × Fin b) ℂ) (hrho : IsDensityMatrix rho)
    (U : Fin d → Matrix.unitaryGroup (Fin a × Fin b) ℂ) (m : ℕ) (hm : 0 < m) :
    IsDensityMatrix (Dependency50.P42.dimensionedPower rho m) ∧
    (∀ key, (tensorTwists U m key)ᴴ * tensorTwists U m key = 1 ∧
      tensorTwists U m key * (tensorTwists U m key)ᴴ = 1) ∧
    (fun x y : Fin (d ^ m * a ^ m) × Fin (d ^ m * b ^ m) =>
      Dependency50.P42.dimensionedPower (bipartitePrivate rho U) m
        (regroupIndex d a m x.1, regroupIndex d b m x.2)
        (regroupIndex d a m y.1, regroupIndex d b m y.2)) =
      bipartitePrivateRaw (Dependency50.P42.dimensionedPower rho m) (tensorTwists U m) := by sorry

end

end QAlgorithms.Papers.Independent.Gap48.P42
Source
Quantum states representing perfectly secure bits are always distillable, arXiv:quant-ph/0602176v2, PDF p. 4, III, observation preceding Property 1a
Read-back

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

Let d,a,b,md,a,b,md,a,b,m be positive natural numbers, let ρ\rhoρ be a positive semidefinite trace-one matrix on Ca⊗Cb\mathbb C^a\otimes\mathbb C^bCa⊗Cb, and let UiU_iUi​ be unitaries on that space for i∈Fin⁡(d)i\in\operatorname{Fin}(d)i∈Fin(d). For each natural c,rc,rc,r, write χc,r:(Fin⁡(r)→Fin⁡(c))≃Fin⁡(cr)\chi_{c,r}:(\operatorname{Fin}(r)\to\operatorname{Fin}(c))\simeq\operatorname{Fin}(c^r)χc,r​:(Fin(r)→Fin(c))≃Fin(cr) for the fixed cardinality bijection, and write ℓc,e:Fin⁡(c)×Fin⁡(e)≃Fin⁡(ce)\ell_{c,e}:\operatorname{Fin}(c)\times\operatorname{Fin}(e)\simeq\operatorname{Fin}(ce)ℓc,e​:Fin(c)×Fin(e)≃Fin(ce) for the fixed local-index bijection. Define the grouped tensor power Pr(ρ)P_r(\rho)Pr​(ρ) by Pr(ρ)(x,y),(x′,y′)=∏t<rρ(χa,r−1(x)t,χb,r−1(y)t),(χa,r−1(x′)t,χb,r−1(y′)t)P_r(\rho)_{(x,y),(x',y')}=\prod_{t<r}\rho_{(\chi_{a,r}^{-1}(x)_t,\chi_{b,r}^{-1}(y)_t),(\chi_{a,r}^{-1}(x')_t,\chi_{b,r}^{-1}(y')_t)}Pr​(ρ)(x,y),(x′,y′)​=∏t<r​ρ(χa,r−1​(x)t​,χb,r−1​(y)t​),(χa,r−1​(x′)t​,χb,r−1​(y′)t​)​. Define Γ\GammaΓ by its key blocks Γ(ℓd,a(i,x),ℓd,b(j,y)),(ℓd,a(i′,x′),ℓd,b(j′,y′))=d−1(UiρUi′†)(x,y),(x′,y′)\Gamma_{(\ell_{d,a}(i,x),\ell_{d,b}(j,y)),(\ell_{d,a}(i',x'),\ell_{d,b}(j',y'))}=d^{-1}(U_i\rho U_{i'}^\dagger)_{(x,y),(x',y')}Γ(ℓd,a​(i,x),ℓd,b​(j,y)),(ℓd,a​(i′,x′),ℓd,b​(j′,y′))​=d−1(Ui​ρUi′†​)(x,y),(x′,y′)​ when i=ji=ji=j and i′=j′i'=j'i′=j′, and zero otherwise. For each κ∈Fin⁡(dm)\kappa\in\operatorname{Fin}(d^m)κ∈Fin(dm), define WκW_\kappaWκ​ on Cam⊗Cbm\mathbb C^{a^m}\otimes\mathbb C^{b^m}Cam⊗Cbm by (Wκ)(x,y),(x′,y′)=∏t<m(Uχd,m−1(κ)t)(χa,m−1(x)t,χb,m−1(y)t),(χa,m−1(x′)t,χb,m−1(y′)t)(W_\kappa)_{(x,y),(x',y')}=\prod_{t<m}(U_{\chi_{d,m}^{-1}(\kappa)_t})_{(\chi_{a,m}^{-1}(x)_t,\chi_{b,m}^{-1}(y)_t),(\chi_{a,m}^{-1}(x')_t,\chi_{b,m}^{-1}(y')_t)}(Wκ​)(x,y),(x′,y′)​=∏t<m​(Uχd,m−1​(κ)t​​)(χa,m−1​(x)t​,χb,m−1​(y)t​),(χa,m−1​(x′)t​,χb,m−1​(y′)t​)​. Define Rd,a,m:Fin⁡(dmam)→Fin⁡((da)m)R_{d,a,m}:\operatorname{Fin}(d^ma^m)\to\operatorname{Fin}((da)^m)Rd,a,m​:Fin(dmam)→Fin((da)m) by splitting xxx as (κ,α)=ℓdm,am−1(x)(\kappa,\alpha)=\ell_{d^m,a^m}^{-1}(x)(κ,α)=ℓdm,am−1​(x) and setting Rd,a,m(x)=χda,m(t↦ℓd,a(χd,m−1(κ)t,χa,m−1(α)t))R_{d,a,m}(x)=\chi_{da,m}(t\mapsto\ell_{d,a}(\chi_{d,m}^{-1}(\kappa)_t,\chi_{a,m}^{-1}(\alpha)_t))Rd,a,m​(x)=χda,m​(t↦ℓd,a​(χd,m−1​(κ)t​,χa,m−1​(α)t​)), and define Rd,b,mR_{d,b,m}Rd,b,m​ in the same way. Then Pm(ρ)P_m(\rho)Pm​(ρ) is positive semidefinite of trace 111, every WκW_\kappaWκ​ satisfies both Wκ†Wκ=IW_\kappa^\dagger W_\kappa=IWκ†​Wκ​=I and WκWκ†=IW_\kappa W_\kappa^\dagger=IWκ​Wκ†​=I, and the matrix obtained from Pm(Γ)P_m(\Gamma)Pm​(Γ) by applying Rd,a,mR_{d,a,m}Rd,a,m​ and Rd,b,mR_{d,b,m}Rd,b,m​ to its two local row indices and likewise to its column indices equals the following matrix: after splitting each local index using ℓdm,am\ell_{d^m,a^m}ℓdm,am​ or ℓdm,bm\ell_{d^m,b^m}ℓdm,bm​, its block between matching row keys κ\kappaκ and matching column keys λ\lambdaλ is (dm)−1WκPm(ρ)Wλ†(d^m)^{-1}W_\kappa P_m(\rho)W_\lambda^\dagger(dm)−1Wκ​Pm​(ρ)Wλ†​, and every other key block is zero. The conclusion is equality of the entire matrices. The cases d=1d=1d=1, a=1a=1a=1, or b=1b=1b=1 are included, while m=0m=0m=0 is excluded.

Human review
  • Endorsed by Shuze Chen · Oct 22, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 22, 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