Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Relative error of N to L mixed-state cloning

Open
QAlgorithms.Papers.Independent.Intakenext100.P97.n_to_l_cloning_lowerBound

by Gohan · Oct 31, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

All states are positive semidefinite, trace-one matrices, including pure and rank-deficient states. Write F(ρ,σ)F(\rho,\sigma)F(ρ,σ) for squared Uhlmann fidelity, f=F(ρ1,ρ2)f=\sqrt{F(\rho_1,\rho_2)}f=F(ρ1​,ρ2​)​, and D(ρ,σ)=1−F(ρ,σ)=sin⁡Δ(ρ,σ)D(\rho,\sigma)=\sqrt{1-F(\rho,\sigma)}=\sin\Delta(\rho,\sigma)D(ρ,σ)=1−F(ρ,σ)​=sinΔ(ρ,σ).

Let d>1d>1d>1, let ρ1≠ρ2\rho_1\ne\rho_2ρ1​=ρ2​ be density operators on H=CdH=\mathbb C^dH=Cd, and let N,M≥1N,M\ge1N,M≥1 be integers, with L=N+ML=N+ML=N+M. An arbitrary finite-dimensional environment EEE and normalized ancillas Υj\Upsilon_jΥj​ on H⊗M⊗EH^{\otimes M}\otimes EH⊗M⊗E are allowed. Fix a common unitary VVV on H⊗L⊗EH^{\otimes L}\otimes EH⊗L⊗E, and set ρ~j=Tr⁡E[V(ρj⊗N⊗Υj)V†]\widetilde\rho_j=\operatorname{Tr}_E[V(\rho_j^{\otimes N}\otimes\Upsilon_j)V^\dagger]ρ​j​=TrE​[V(ρj⊗N​⊗Υj​)V†] and ϕ=F(Υ1,Υ2)\phi=\sqrt{F(\Upsilon_1,\Upsilon_2)}ϕ=F(Υ1​,Υ2​)​. With

RN→L=D(ρ~1,ρ1⊗L)+D(ρ~2,ρ2⊗L)D(ρ1⊗L,ρ2⊗L),R_{N\to L}=\frac{D(\widetilde\rho_1,\rho_1^{\otimes L})+D(\widetilde\rho_2,\rho_2^{\otimes L})}{D(\rho_1^{\otimes L},\rho_2^{\otimes L})},RN→L​=D(ρ1⊗L​,ρ2⊗L​)D(ρ​1​,ρ1⊗L​)+D(ρ​2​,ρ2⊗L​)​,

both regimes satisfy

fM≤ϕ≤1⟹RN→L≥fNϕ−fL1−f2Nϕ21−f2L,0≤ϕ≤fM⟹RN→L≥0.\begin{aligned}f^M\le\phi\le1&\quad\Longrightarrow\quad R_{N\to L}\ge f^N\phi-\frac{f^L\sqrt{1-f^{2N}\phi^2}}{\sqrt{1-f^{2L}}},\\0\le\phi\le f^M&\quad\Longrightarrow\quad R_{N\to L}\ge0.\end{aligned}fM≤ϕ≤10≤ϕ≤fM​⟹RN→L​≥fNϕ−1−f2L​fL1−f2Nϕ2​​,⟹RN→L​≥0.​

Furthermore, for every density operator σ\sigmaσ on EEE, the choice Υj=ρj⊗M⊗σ\Upsilon_j=\rho_j^{\otimes M}\otimes\sigmaΥj​=ρj⊗M​⊗σ and V=IV=IV=I gives RN→L=0R_{N\to L}=0RN→L​=0. This extends the bound and the full-information example to multiple original and additional copies.

Formalization Note The ideal LLL-copy matrix is grouped as its first NNN and last MMM factors, using exact entrywise tensor powers.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P97
Formal statement
noncomputable section

open scoped Matrix ComplexConjugate

namespace QAlgorithms.Papers.Independent.Intakenext100.P97

open QAlgorithms.Papers.Independent.Information100.P67
  (squaredFidelity sineDistance receiverChannel)

open QAlgorithms.Papers.Independent.Dependency50.P48 (densityPower)

/-- PDF p.4, §IV, Eq. (23) and the full-information example after it.
There are `N ≥ 1` original copies and `M ≥ 1` additional copies, `L=N+M`.
The example uses identity evolution; arbitrary evolution could spoil the clones. -/
theorem n_to_l_cloning_lowerBound {d N M : ℕ} (hd : 1 < d)
    (hN : 1 ≤ N) (hM : 1 ≤ M)
    {E : Type*} [Fintype E] [DecidableEq E] [Nonempty E]
    (rho1 rho2 : Matrix (Fin d) (Fin d) ℂ)
    (tau1 tau2 : Matrix ((Fin M → Fin d) × E) ((Fin M → Fin d) × E) ℂ)
    (V : Matrix.unitaryGroup (((Fin N → Fin d) × (Fin M → Fin d)) × E) ℂ)
    (hrho1 : IsDensityMatrix rho1) (hrho2 : IsDensityMatrix rho2)
    (htau1 : IsDensityMatrix tau1) (htau2 : IsDensityMatrix tau2)
    (hdistinct : rho1 ≠ rho2) :
    let f := rootFidelity rho1 rho2
    let phi := rootFidelity tau1 tau2
    (f ^ M ≤ phi ∧ phi ≤ 1 →
      cloningBound N M f phi ≤ copiesError N M rho1 rho2 tau1 tau2 V) ∧
    (0 ≤ phi ∧ phi ≤ f ^ M → 0 ≤ copiesError N M rho1 rho2 tau1 tau2 V) ∧
    (∀ sigma : Matrix E E ℂ, IsDensityMatrix sigma →
      copiesError N M rho1 rho2
        (Matrix.kronecker (densityPower rho1 M) sigma)
        (Matrix.kronecker (densityPower rho2 M) sigma) 1 = 0) := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P97

end
Source
Lower bound on the relative error of mixed-state cloning and related operations, arXiv:quant-ph/0208159v2, PDF p. 4, §IV, Eq. (23) and following paragraph
Read-back

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

Let d,N,Md,N,Md,N,M be natural numbers with d>1d>1d>1, N≥1N\ge1N≥1, and M≥1M\ge1M≥1; let EEE be a finite nonempty type with decidable equality; let ρ1≠ρ2\rho_1\ne\rho_2ρ1​=ρ2​ be positive semidefinite complex d×dd\times dd×d matrices of trace 111; let τ1,τ2\tau_1,\tau_2τ1​,τ2​ be positive semidefinite matrices of trace 111 on {0,…,d−1}M×E\{0,\ldots,d-1\}^{M}\times E{0,…,d−1}M×E; and let VVV be any unitary on ({0,…,d−1}N×{0,…,d−1}M)×E(\{0,\ldots,d-1\}^{N}\times\{0,\ldots,d-1\}^{M})\times E({0,…,d−1}N×{0,…,d−1}M)×E. For K∈NK\in\mathbb NK∈N, define (ρi⊗K)a,b=∏j=0K−1(ρi)aj,bj(\rho_i^{\otimes K})_{a,b}=\prod_{j=0}^{K-1}(\rho_i)_{a_j,b_j}(ρi⊗K​)a,b​=∏j=0K−1​(ρi​)aj​,bj​​, and put

F(X,Y)=Re⁡tr⁡X YX,s(X,Y)=1−F(X,Y)2,f=F(ρ1,ρ2),ϕ=F(τ1,τ2).F(X,Y)=\operatorname{Re}\operatorname{tr}\sqrt{\sqrt X\,Y\sqrt X}, \quad s(X,Y)=\sqrt{1-F(X,Y)^2}, \quad f=F(\rho_1,\rho_2),\quad \phi=F(\tau_1,\tau_2).F(X,Y)=RetrX​YX​​,s(X,Y)=1−F(X,Y)2​,f=F(ρ1​,ρ2​),ϕ=F(τ1​,τ2​).

For auxiliary matrices η1,η2\eta_1,\eta_2η1​,η2​ on the same registers as the τi\tau_iτi​, define

Oi=Tr⁡E[V(ρi⊗N⊗ηi)V†],Ii=ρi⊗N⊗ρi⊗M,EV(η1,η2)=s(O1,I1)+s(O2,I2)s(I1,I2),O_i=\operatorname{Tr}_E[V(\rho_i^{\otimes N}\otimes\eta_i)V^\dagger], \quad I_i=\rho_i^{\otimes N}\otimes\rho_i^{\otimes M}, \qquad \mathcal E_V(\eta_1,\eta_2)= \frac{s(O_1,I_1)+s(O_2,I_2)}{s(I_1,I_2)},Oi​=TrE​[V(ρi⊗N​⊗ηi​)V†],Ii​=ρi⊗N​⊗ρi⊗M​,EV​(η1​,η2​)=s(I1​,I2​)s(O1​,I1​)+s(O2​,I2​)​,

where partial trace sums entries with equal EEE coordinates. The theorem asserts all three statements

fM≤ϕ≤1 ⟹ fNϕ−fN+M1−f2Nϕ21−f2(N+M)≤EV(τ1,τ2),0≤ϕ≤fM ⟹ 0≤EV(τ1,τ2),EI(ρ1⊗M⊗σ, ρ2⊗M⊗σ)=0for every positive semidefinite matrix σ on E of trace 1.\begin{aligned} f^M\le\phi\le1 &\ \Longrightarrow\ f^N\phi- \frac{f^{N+M}\sqrt{1-f^{2N}\phi^2}}{\sqrt{1-f^{2(N+M)}}} \le\mathcal E_V(\tau_1,\tau_2),\\ 0\le\phi\le f^M &\ \Longrightarrow\ 0\le\mathcal E_V(\tau_1,\tau_2),\\ \mathcal E_I(\rho_1^{\otimes M}\otimes\sigma,\, \rho_2^{\otimes M}\otimes\sigma)&=0 \quad\text{for every positive semidefinite matrix $\sigma$ on $E$ of trace $1$}. \end{aligned}fM≤ϕ≤10≤ϕ≤fMEI​(ρ1⊗M​⊗σ,ρ2⊗M​⊗σ)​ ⟹ fNϕ−1−f2(N+M)​fN+M1−f2Nϕ2​​≤EV​(τ1​,τ2​), ⟹ 0≤EV​(τ1​,τ2​),=0for every positive semidefinite matrix σ on E of trace 1.​

In the last equality evolution is the identity, independently of the originally supplied VVV. The boundary ϕ=fM\phi=f^Mϕ=fM belongs to both implications. Matrix square roots use real continuous functional calculus; real square roots of nonpositive numbers and division by zero return zero.

Human review
  • Endorsed by Shuze Chen · Oct 31, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 31, 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