Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Appendix B (B5)–(B8): exact fidelity-preserving teleportation and dense coding

Open
QAlgorithms.Papers.Independent.Support50.P11.teleportation_dense_coding_conversions

by Goku · Oct 23, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let d≥1d\ge1d≥1. Use the explicitly defined Weyl Bell measurement, one shared normalized Bell state and controlled Weyl corrections to define the EA teleportation and dense-coding codes. Let PPP be input/output dephasing followed by permutation averaging, and UUU the Haar twirl with inverse output conjugation. Define Td(M)=U∘T∘P[M]T_d(M)=U\circ T\circ P[M]Td​(M)=U∘T∘P[M] and Dd(N)=P∘D∘U[N]D_d(N)=P\circ D\circ U[N]Dd​(N)=P∘D∘U[N] by the actual CPTP constructions. Let C(λ)C^{(\lambda)}C(λ) be the d2d^2d2-ary symmetric classical channel and X(λ)X^{(\lambda)}X(λ) the isotropic ddd-dimensional channel parameterized by entanglement fidelity. For λ∈[0,1]\lambda\in[0,1]λ∈[0,1], with λ=1\lambda=1λ=1 when d=1d=1d=1,

Td(C(λ))=X(λ),Dd(X(λ))=C(λ).T_d(C^{(\lambda)})=X^{(\lambda)},\qquad D_d(X^{(\lambda)})=C^{(\lambda)}.Td​(C(λ))=X(λ),Dd​(X(λ))=C(λ).

For every CPTP MMM on d2d^2d2 dimensions and NNN on ddd dimensions, the converted maps are CPTP and admit actual EA-code realizations, and

Td(M)=X(Ps(M)),F(Td(M))=Ps(M),Dd(N)=C(F(N)),Ps(Dd(N))=F(N).T_d(M)=X^{(P_s(M))},\quad F(T_d(M))=P_s(M),\quad D_d(N)=C^{(F(N))},\quad P_s(D_d(N))=F(N).Td​(M)=X(Ps​(M)),F(Td​(M))=Ps​(M),Dd​(N)=C(F(N)),Ps​(Dd​(N))=F(N).

This retains the exact endpoint fidelities and explicitly repairs the printed twirls.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P11
Formal statement
noncomputable section

open scoped Matrix ComplexConjugate ComplexOrder ENNReal

namespace QAlgorithms.Papers.Independent.Support50.P11

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

open QAlgorithms.Papers.Independent.Information100.P21 (pt)

open QAlgorithms.Papers.Independent.Dependency50.P09 (tensorAction)

open QAlgorithms.Papers.Independent.Information100.P76 (swapMatrix)

open QAlgorithms.Papers.Independent.Information100.P69 (replacer)

open QAlgorithms.Papers.Independent.Gap20.P19 (wernerIsometry)

attribute [local instance] Classical.propDecidable

attribute [local instance] Classical.decEq

/-- PDF pp.14--15, (B5)--(B8): actual Bell-protocol CPTP conversions, exact
optimalCodeFidelity preservation, and realization in EA. Corrected B1 includes dephasing;
corrected B2 uses inverse output conjugation via accepted haarTwirl. -/
theorem teleportation_dense_coding_conversions
    (d : ℕ) (hd : 0 < d) :
    IsForwardCode d .EA (teleportCode d hd) ∧ IsForwardCode (d * d) .EA (denseCode d hd) ∧
    (∀ lambda : ℝ, 0 ≤ lambda → lambda ≤ 1 → (d = 1 → lambda = 1) →
      teleportConversion d hd (symmetricClassical (d * d) lambda) = fidelityDepolarizing d lambda ∧
      denseConversion d hd (fidelityDepolarizing d lambda) = symmetricClassical (d * d) lambda) ∧
    (∀ M : Superop (Fin (d * d)) (Fin (d * d)), IsChannel M →
      IsChannel (teleportConversion d hd M) ∧
      teleportConversion d hd M = fidelityDepolarizing d (uniformClassicalSuccess M) ∧
      channelFidelity d (teleportConversion d hd M) = uniformClassicalSuccess M ∧
      ∃ Z : Superop (Fin d × Fin (d * d)) (Fin (d * d) × Fin d),
        IsForwardCode d .EA Z ∧ link d M Z = teleportConversion d hd M) ∧
    (∀ N : Superop (Fin d) (Fin d), IsChannel N →
      IsChannel (denseConversion d hd N) ∧
      denseConversion d hd N = symmetricClassical (d * d) (channelFidelity d N) ∧
      uniformClassicalSuccess (denseConversion d hd N) = channelFidelity d N ∧
      ∃ Z : Superop (Fin (d * d) × Fin d) (Fin d × Fin (d * d)),
        IsForwardCode (d * d) .EA Z ∧ link (d * d) N Z = denseConversion d hd N) := by sorry

end QAlgorithms.Papers.Independent.Support50.P11

end
Source
On the power of PPT-preserving and non-signalling codes, arXiv:1406.7142v2, PDF p. 14, Appendix B (B5)–(B8): exact fidelity-preserving teleportation and dense coding
Read-back

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

For every positive natural number ddd, use indices 0,…,d−10,\ldots,d-10,…,d−1, let Δ\DeltaΔ erase all off-diagonal matrix entries, and put φd=∣Ωd⟩⟨Ωd∣\varphi_d=|\Omega_d\rangle\langle\Omega_d|φd​=∣Ωd​⟩⟨Ωd​∣, where ∣Ωd⟩=d−1/2∑j∣j,j⟩|\Omega_d\rangle=d^{-1/2}\sum_j|j,j\rangle∣Ωd​⟩=d−1/2∑j​∣j,j⟩. Label d2d^2d2 matrices by pairs c=(a,b)c=(a,b)c=(a,b), and define

Wa,b∣j⟩=e2πibj/d∣j+a mod d⟩,∣βc⟩=1d∑i,j(Wc)ij∣i,j⟩.W_{a,b}|j\rangle=e^{2\pi i bj/d}|j+a\bmod d\rangle,\qquad |\beta_c\rangle=\frac1{\sqrt d}\sum_{i,j}(W_c)_{ij}|i,j\rangle.Wa,b​∣j⟩=e2πibj/d∣j+amodd⟩,∣βc​⟩=d​1​i,j∑​(Wc​)ij​∣i,j⟩.

Define the maps

Bd(X)ce=1c=etr⁡(∣βc⟩⟨βc∣X),Cd(Y)=∑cWcYccWc†,\mathcal B_d(X)_{ce}=\mathbf1_{c=e}\operatorname{tr}(|\beta_c\rangle\langle\beta_c|X), \qquad \mathcal C_d(Y)=\sum_cW_cY_{cc}W_c^\dagger,Bd​(X)ce​=1c=e​tr(∣βc​⟩⟨βc​∣X),Cd​(Y)=c∑​Wc​Ycc​Wc†​,

where YccY_{cc}Ycc​ is the d×dd\times dd×d block indexed by the label ccc. Let TdT_dTd​ be the map that adjoins the shared state φd\varphi_dφd​, applies Bd\mathcal B_dBd​ to the first input and the first share, and applies Cd\mathcal C_dCd​ to the second input and the second share; let SdS_dSd​ be defined with Cd\mathcal C_dCd​ on the first side and Bd\mathcal B_dBd​ on the second. Thus TdT_dTd​ maps matrices on d×d2d\times d^2d×d2 to matrices on d2×dd^2\times dd2×d, and SdS_dSd​ maps matrices on d2×dd^2\times dd2×d to matrices on d×d2d\times d^2d×d2. More generally, adjoining a state σ\sigmaσ and applying local maps E,DE,DE,D means applying E⊗DE\otimes DE⊗D to the matrix with entries X(a,b),(a′,b′)σ(ℓ,r),(ℓ′,r′)X_{(a,b),(a',b')}\sigma_{(\ell,r),(\ell',r')}X(a,b),(a′,b′)​σ(ℓ,r),(ℓ′,r′)​, grouped as ((a,ℓ),(b,r))((a,\ell),(b,r))((a,ℓ),(b,r)). Call a map ZZZ from K×OK\times OK×O to I×KI\times KI×K an allowed code if it is a channel, there is a channel E0E_0E0​ satisfying tr⁡2Z(X)=E0(tr⁡2X)\operatorname{tr}_2Z(X)=E_0(\operatorname{tr}_2X)tr2​Z(X)=E0​(tr2​X) for all XXX, and ZZZ has this shared-state local form for some positive finite share dimensions, a positive semidefinite trace-one shared state, and local channels. Channels mean finite Kraus maps with ∑jAj†Aj=I\sum_jA_j^\dagger A_j=I∑j​Aj†​Aj​=I. For a map Q:I→OQ:I\to OQ:I→O, define its link with ZZZ by

LK(Q,Z)(X)bc=∑a,a′,i,j,β,γXaa′Q(Eij)βγZ(E(a,β),(a′,γ))(i,b),(j,c).\mathcal L_K(Q,Z)(X)_{bc} =\sum_{a,a',i,j,\beta,\gamma} X_{aa'}Q(E_{ij})_{\beta\gamma} Z(E_{(a,\beta),(a',\gamma)})_{(i,b),(j,c)}.LK​(Q,Z)(X)bc​=a,a′,i,j,β,γ∑​Xaa′​Q(Eij​)βγ​Z(E(a,β),(a′,γ)​)(i,b),(j,c)​.

Define

Fm(Q)=Re⁡tr⁡ ⁣[φm(id⁡m⊗Q)(φm)],um(Q)=1m∑iRe⁡Q(Eii)ii,F_m(Q)=\operatorname{Re}\operatorname{tr}\!\left[ \varphi_m(\operatorname{id}_m\otimes Q)(\varphi_m)\right], \qquad u_m(Q)=\frac1m\sum_i\operatorname{Re}Q(E_{ii})_{ii},Fm​(Q)=Retr[φm​(idm​⊗Q)(φm​)],um​(Q)=m1​i∑​ReQ(Eii​)ii​,

and the two averages

Hd(Q)(X)=∫U(d)U†Q(UXU†)U dU,Pm(Q)(X)=Δ ⁣(1m!∑πPπ†Q(PπΔ(X)Pπ†)Pπ),\mathcal H_d(Q)(X)=\int_{U(d)}U^\dagger Q(UXU^\dagger)U\,dU, \qquad \mathcal P_m(Q)(X)= \Delta\!\left(\frac1{m!}\sum_{\pi} P_\pi^\dagger Q(P_\pi\Delta(X)P_\pi^\dagger)P_\pi\right),Hd​(Q)(X)=∫U(d)​U†Q(UXU†)UdU,Pm​(Q)(X)=Δ(m!1​π∑​Pπ†​Q(Pπ​Δ(X)Pπ†​)Pπ​),

where dUdUdU is Haar measure normalized to total mass 111, the sum ranges over all permutations, and Pπ∣j⟩=∣π(j)⟩P_\pi|j\rangle=|\pi(j)\ranglePπ​∣j⟩=∣π(j)⟩. Put

Td(M)=Hd ⁣(Ld(Pd2(M),Td)),Dd(N)=Pd2 ⁣(Ld2(Hd(N),Sd)).\mathcal T_d(M)=\mathcal H_d\!\left(\mathcal L_d(\mathcal P_{d^2}(M),T_d)\right), \qquad \mathcal D_d(N)=\mathcal P_{d^2}\!\left(\mathcal L_{d^2}(\mathcal H_d(N),S_d)\right).Td​(M)=Hd​(Ld​(Pd2​(M),Td​)),Dd​(N)=Pd2​(Ld2​(Hd​(N),Sd​)).

For m>1m>1m>1, let Sm,λ\mathcal S_{m,\lambda}Sm,λ​ have zero off-diagonal output entries and diagonal entries

Sm,λ(X)ii=λXii+1−λm−1∑j≠iXjj;\mathcal S_{m,\lambda}(X)_{ii} =\lambda X_{ii}+\frac{1-\lambda}{m-1}\sum_{j\ne i}X_{jj};Sm,λ​(X)ii​=λXii​+m−11−λ​j=i∑​Xjj​;

set S1,λ=Δ\mathcal S_{1,\lambda}=\DeltaS1,λ​=Δ. For d>1d>1d>1, set

Fd,λ(X)=aX+1−adtr⁡(X)I,a=d2λ−1d2−1,\mathcal F_{d,\lambda}(X) =aX+\frac{1-a}{d}\operatorname{tr}(X)I, \qquad a=\frac{d^2\lambda-1}{d^2-1},Fd,λ​(X)=aX+d1−a​tr(X)I,a=d2−1d2λ−1​,

and set F1,λ\mathcal F_{1,\lambda}F1,λ​ to the identity map. Then TdT_dTd​ and SdS_dSd​ are allowed codes with message dimensions ddd and d2d^2d2, respectively. For every λ∈[0,1]\lambda\in[0,1]λ∈[0,1], additionally requiring λ=1\lambda=1λ=1 when d=1d=1d=1,

Td(Sd2,λ)=Fd,λ,Dd(Fd,λ)=Sd2,λ.\mathcal T_d(\mathcal S_{d^2,\lambda})=\mathcal F_{d,\lambda}, \qquad \mathcal D_d(\mathcal F_{d,\lambda})=\mathcal S_{d^2,\lambda}.Td​(Sd2,λ​)=Fd,λ​,Dd​(Fd,λ​)=Sd2,λ​.

For every channel MMM on d2d^2d2-dimensional matrices, Td(M)\mathcal T_d(M)Td​(M) is a channel,

Td(M)=Fd,ud2(M),Fd(Td(M))=ud2(M),\mathcal T_d(M)=\mathcal F_{d,u_{d^2}(M)},\qquad F_d(\mathcal T_d(M))=u_{d^2}(M),Td​(M)=Fd,ud2​(M)​,Fd​(Td​(M))=ud2​(M),

and there exists an allowed code Z:d×d2→d2×dZ:d\times d^2\to d^2\times dZ:d×d2→d2×d with Ld(M,Z)=Td(M)\mathcal L_d(M,Z)=\mathcal T_d(M)Ld​(M,Z)=Td​(M). For every channel NNN on ddd-dimensional matrices, Dd(N)\mathcal D_d(N)Dd​(N) is a channel,

Dd(N)=Sd2,Fd(N),ud2(Dd(N))=Fd(N),\mathcal D_d(N)=\mathcal S_{d^2,F_d(N)},\qquad u_{d^2}(\mathcal D_d(N))=F_d(N),Dd​(N)=Sd2,Fd​(N)​,ud2​(Dd​(N))=Fd​(N),

and there exists an allowed code Z:d2×d→d×d2Z:d^2\times d\to d\times d^2Z:d2×d→d×d2 with Ld2(N,Z)=Dd(N)\mathcal L_{d^2}(N,Z)=\mathcal D_d(N)Ld2​(N,Z)=Dd​(N); each existential code may depend on the channel being converted.

Human review
  • Endorsed by Shuze Chen · Oct 24, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 24, 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