Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Definitions for Lower bound on the relative error of mixed-state cloning and related operations

Definition
QAlgorithms_Papers_Independent_Intakenext100_P97

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

prove2quantum

The objects the formalized statements of Lower bound on the relative error of mixed-state cloning and related operations (arXiv:quant-ph/0208159) are stated in. Each was checked against the paper by a blind read-back and a moderator.

  • dilationInput: Product initial state in Eq. (12), PDF p.3, regrouped as (A B) E. The ancillary state on B E may be correlated or mixed.
  • relativeError: Relative error of two joint outputs against two ideal joint states, Eq. (14), PDF p.3. The numerator is an unweighted sum. At identical ideals Lean division returns zero; no theorem below uses this extraneous extension.
  • oneToTwoError: Relative error of the 1 → 2 dilation in Eqs. (12)--(14), PDF p.3. The same V is applied for both input labels.
  • copiesError: Computed N → N+M relative error, Eq. (23), PDF p.4. The ideal (N+M)-copy state is grouped as its first N and last M copies.
  • cloningBound: Scalar lower bound, Eq. (23), PDF p.4; L=N+M avoids natural subtraction. The denominator divides only the second term. N=M=1 gives Eq. (15).
  • standardBound: Standard cloning lower bound, Eq. (22), PDF p.4.
Definition code
import Mathlib
import Definitions.Def_QAlgorithms_Defs_QuantumInformation
import Definitions.Def_QAlgorithms_Defs_Measurement
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P67
import Definitions.Def_QAlgorithms_Papers_Independent_Dependency50_P48

/-!
Statements from A. E. Rastegin, arXiv:quant-ph/0208159v2, PDF pp.2--5.
The paper's fidelity is squared Uhlmann fidelity. All dilation outputs are
computed from product initial states and a single common unitary. Relative
error is used only for distinct input states, so its denominator is positive.
The environment is arbitrary finite dimensional, with no dimension bound.
Squared fidelity, sine distance, computed receiver evolution and tensor powers
reuse the pinned accepted library; only the cloning-specific expressions are local.
-/

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)

/-- Product initial state in Eq. (12), PDF p.3, regrouped as `(A B) E`.
The ancillary state on `B E` may be correlated or mixed. -/
def dilationInput {A B E : Type*} (rho : Matrix A A ℂ)
    (tau : Matrix (B × E) (B × E) ℂ) : Matrix ((A × B) × E) ((A × B) × E) ℂ :=
  fun x y => rho x.1.1 y.1.1 * tau (x.1.2, x.2) (y.1.2, y.2)

/-- Relative error of two joint outputs against two ideal joint states,
Eq. (14), PDF p.3. The numerator is an unweighted sum. At identical ideals
Lean division returns zero; no theorem below uses this extraneous extension. -/
def relativeError {A : Type*} [Fintype A] [DecidableEq A]
    (out1 out2 ideal1 ideal2 : Matrix A A ℂ) : ℝ :=
  (sineDistance out1 ideal1 + sineDistance out2 ideal2) / sineDistance ideal1 ideal2

/-- Relative error of the `1 → 2` dilation in Eqs. (12)--(14), PDF p.3.
The same `V` is applied for both input labels. -/
def oneToTwoError {d : ℕ} {E : Type*} [Fintype E] [DecidableEq E]
    (rho1 rho2 : Matrix (Fin d) (Fin d) ℂ)
    (tau1 tau2 : Matrix (Fin d × E) (Fin d × E) ℂ)
    (V : Matrix.unitaryGroup ((Fin d × Fin d) × E) ℂ) : ℝ :=
  relativeError (receiverChannel (V : Matrix ((Fin d × Fin d) × E) ((Fin d × Fin d) × E) ℂ) (dilationInput rho1 tau1))
    (receiverChannel (V : Matrix ((Fin d × Fin d) × E) ((Fin d × Fin d) × E) ℂ) (dilationInput rho2 tau2))
    (Matrix.kronecker rho1 rho1) (Matrix.kronecker rho2 rho2)

/-- Computed `N → N+M` relative error, Eq. (23), PDF p.4.
The ideal `(N+M)`-copy state is grouped as its first `N` and last `M` copies. -/
def copiesError {d : ℕ} (N M : ℕ) {E : Type*} [Fintype E] [DecidableEq 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) : ℝ :=
  relativeError (receiverChannel (V : Matrix (((Fin N → Fin d) × (Fin M → Fin d)) × E) (((Fin N → Fin d) × (Fin M → Fin d)) × E) ℂ) (dilationInput (densityPower rho1 N) tau1))
    (receiverChannel (V : Matrix (((Fin N → Fin d) × (Fin M → Fin d)) × E) (((Fin N → Fin d) × (Fin M → Fin d)) × E) ℂ) (dilationInput (densityPower rho2 N) tau2))
    (Matrix.kronecker (densityPower rho1 N) (densityPower rho1 M))
    (Matrix.kronecker (densityPower rho2 N) (densityPower rho2 M))

/-- Scalar lower bound, Eq. (23), PDF p.4; `L=N+M` avoids natural subtraction.
The denominator divides only the second term. `N=M=1` gives Eq. (15). -/
def cloningBound (N M : ℕ) (f phi : ℝ) : ℝ :=
  f ^ N * phi - f ^ (N + M) * Real.sqrt (1 - f ^ (2 * N) * phi ^ 2) /
    Real.sqrt (1 - f ^ (2 * (N + M)))

/-- Standard cloning lower bound, Eq. (22), PDF p.4. -/
def standardBound (f : ℝ) : ℝ :=
  f - f ^ 2 / Real.sqrt (1 + f ^ 2)

/-! Sanity examples pin square convention, register order, tensor powers,
identity evolution, the unweighted ratio, and the orthogonal endpoint. -/

example : squaredFidelity (1 : Matrix (Fin 1) (Fin 1) ℂ) 1 = 1 := by
  simp [squaredFidelity, rootFidelity, cfc_one]

example : sineDistance (1 : Matrix (Fin 1) (Fin 1) ℂ) 1 = 0 := by
  simp [sineDistance, squaredFidelity, rootFidelity, cfc_one]

example {A B E : Type*} (rho : Matrix A A ℂ)
    (tau : Matrix (B × E) (B × E) ℂ) (a a' : A) (b b' : B) (e e' : E) :
    dilationInput rho tau ((a,b),e) ((a',b'),e') = rho a a' * tau (b,e) (b',e') := by
  rfl

example {A B E : Type*} [Fintype A] [Fintype B] [Fintype E]
    [DecidableEq A] [DecidableEq B] [DecidableEq E]
    (rho : Matrix A A ℂ) (tau : Matrix (B × E) (B × E) ℂ) :
    receiverChannel (1 : Matrix ((A × B) × E) ((A × B) × E) ℂ)
      (dilationInput rho tau) = traceRight (dilationInput rho tau) := by
  simp [receiverChannel]

example {A : Type*} [Fintype A] [DecidableEq A] (o1 o2 t1 t2 : Matrix A A ℂ) :
    relativeError o1 o2 t1 t2 =
      (sineDistance o1 t1 + sineDistance o2 t2) / sineDistance t1 t2 := by
  rfl

example {d : ℕ} {E : Type*} [Fintype E] [DecidableEq E]
    (r1 r2 : Matrix (Fin d) (Fin d) ℂ)
    (t1 t2 : Matrix (Fin d × E) (Fin d × E) ℂ)
    (V : Matrix.unitaryGroup ((Fin d × Fin d) × E) ℂ) :
    oneToTwoError r1 r2 t1 t2 V =
      relativeError (receiverChannel (V : Matrix ((Fin d × Fin d) × E) ((Fin d × Fin d) × E) ℂ) (dilationInput r1 t1))
        (receiverChannel (V : Matrix ((Fin d × Fin d) × E) ((Fin d × Fin d) × E) ℂ) (dilationInput r2 t2))
        (Matrix.kronecker r1 r1) (Matrix.kronecker r2 r2) := by
  rfl

example {d : ℕ} (rho : Matrix (Fin d) (Fin d) ℂ) (x y : Fin 1 → Fin d) :
    densityPower rho 1 x y = rho (x 0) (y 0) := by
  simp [densityPower]

example {d : ℕ} (rho : Matrix (Fin d) (Fin d) ℂ) (x y : Fin 0 → Fin d) :
    densityPower rho 0 x y = 1 := by
  simp [densityPower]

example {d N M : ℕ} {E : Type*} [Fintype E] [DecidableEq E]
    (r1 r2 : Matrix (Fin d) (Fin d) ℂ)
    (t1 t2 : 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) :
    copiesError N M r1 r2 t1 t2 V =
      relativeError (receiverChannel (V : Matrix (((Fin N → Fin d) × (Fin M → Fin d)) × E) (((Fin N → Fin d) × (Fin M → Fin d)) × E) ℂ) (dilationInput (densityPower r1 N) t1))
        (receiverChannel (V : Matrix (((Fin N → Fin d) × (Fin M → Fin d)) × E) (((Fin N → Fin d) × (Fin M → Fin d)) × E) ℂ) (dilationInput (densityPower r2 N) t2))
        (Matrix.kronecker (densityPower r1 N) (densityPower r1 M))
        (Matrix.kronecker (densityPower r2 N) (densityPower r2 M)) := by
  rfl

example (phi : ℝ) : cloningBound 1 1 0 phi = 0 := by
  norm_num [cloningBound]

example : standardBound 0 = 0 := by
  norm_num [standardBound]

end QAlgorithms.Papers.Independent.Intakenext100.P97
Source
Lower bound on the relative error of mixed-state cloning and related operations, arXiv:quant-ph/0208159v2, https://arxiv.org/abs/quant-ph/0208159v2
Read-back

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

QAlgorithms.Papers.Independent.Intakenext100.P97.cloningBound

For arbitrary natural numbers N,MN,MN,M and arbitrary real numbers f,ϕf,\phif,ϕ, this defines

fNϕ−fN+M1−f2Nϕ21−f2(N+M).f^{N}\phi- \frac{f^{N+M}\sqrt{1-f^{2N}\phi^{2}}} {\sqrt{1-f^{2(N+M)}}}.fNϕ−1−f2(N+M)​fN+M1−f2Nϕ2​​.

There are no restrictions on f,ϕ,N,Mf,\phi,N,Mf,ϕ,N,M, so zero exponents and negative square-root arguments are included. Zeroth powers, including 000^000, equal 111; the real square root of a nonpositive number is zero; and a quotient with zero denominator is zero.

QAlgorithms.Papers.Independent.Intakenext100.P97.copiesError

For arbitrary natural numbers d,N,Md,N,Md,N,M, a finite type EEE with decidable equality, arbitrary complex matrices ρ1,ρ2\rho_1,\rho_2ρ1​,ρ2​ on {0,…,d−1}\{0,\ldots,d-1\}{0,…,d−1}, arbitrary complex matrices τ1,τ2\tau_1,\tau_2τ1​,τ2​ on {0,…,d−1}M×E\{0,\ldots,d-1\}^{M}\times E{0,…,d−1}M×E, and a unitary VVV 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, write

(ρi⊗K)a,a′=∏j=0K−1(ρi)aj,aj′,(Di)((a,b),e),((a′,b′),e′)=(ρi⊗N)a,a′(τi)(b,e),(b′,e′),Oi=Tr⁡E(VDiV†).(\rho_i^{\otimes K})_{a,a'}=\prod_{j=0}^{K-1}(\rho_i)_{a_j,a'_j}, \quad (D_i)_{((a,b),e),((a',b'),e')} =(\rho_i^{\otimes N})_{a,a'}(\tau_i)_{(b,e),(b',e')}, \quad O_i=\operatorname{Tr}_{E}(VD_iV^\dagger).(ρi⊗K​)a,a′​=j=0∏K−1​(ρi​)aj​,aj′​​,(Di​)((a,b),e),((a′,b′),e′)​=(ρi⊗N​)a,a′​(τi​)(b,e),(b′,e′)​,Oi​=TrE​(VDi​V†).

Here partial trace means summing matrix entries over equal row and column coordinates in EEE. With Ii=ρi⊗N⊗ρi⊗MI_i=\rho_i^{\otimes N}\otimes\rho_i^{\otimes M}Ii​=ρi⊗N​⊗ρi⊗M​, the defined error is

s(O1,I1)+s(O2,I2)s(I1,I2),s(X,Y)=1−[Re⁡tr⁡S ⁣(S(X)YS(X))]2,\frac{s(O_1,I_1)+s(O_2,I_2)}{s(I_1,I_2)}, \qquad s(X,Y)=\sqrt{1-\left[\operatorname{Re}\operatorname{tr} \mathcal S\!\bigl(\mathcal S(X)Y\mathcal S(X)\bigr)\right]^2},s(I1​,I2​)s(O1​,I1​)+s(O2​,I2​)​,s(X,Y)=1−[RetrS(S(X)YS(X))]2​,

where S\mathcal SS is real continuous functional calculus for the real square-root function. No positivity, normalization, or nonemptiness assumptions are imposed. In particular, N=0N=0N=0 or M=0M=0M=0 is allowed, and a zeroth tensor power has its single entry equal to the empty product 111, including when d=0d=0d=0. Real square roots of nonpositive numbers and division by zero return zero.

QAlgorithms.Papers.Independent.Intakenext100.P97.dilationInput

For arbitrary types A,B,EA,B,EA,B,E, a complex matrix ρ\rhoρ indexed by A×AA\times AA×A, and a complex matrix τ\tauτ indexed by (B×E)×(B×E)(B\times E)\times(B\times E)(B×E)×(B×E), this defines the matrix D(ρ,τ)D(\rho,\tau)D(ρ,τ) on (A×B)×E(A\times B)\times E(A×B)×E by

D(ρ,τ)((a,b),e),((a′,b′),e′)=ρa,a′ τ(b,e),(b′,e′).D(\rho,\tau)_{((a,b),e),((a',b'),e')} =\rho_{a,a'}\,\tau_{(b,e),(b',e')}.D(ρ,τ)((a,b),e),((a′,b′),e′)​=ρa,a′​τ(b,e),(b′,e′)​.

Thus it is the tensor product of ρ\rhoρ and τ\tauτ, with its coordinates grouped in the order (A×B)×E(A\times B)\times E(A×B)×E. There are no finiteness, nonemptiness, positivity, or normalization assumptions on the types or matrices.

QAlgorithms.Papers.Independent.Intakenext100.P97.oneToTwoError

For any natural number ddd, any finite type EEE with decidable equality, arbitrary complex matrices ρ1,ρ2\rho_1,\rho_2ρ1​,ρ2​ on {0,…,d−1}\{0,\ldots,d-1\}{0,…,d−1}, arbitrary complex matrices τ1,τ2\tau_1,\tau_2τ1​,τ2​ on {0,…,d−1}×E\{0,\ldots,d-1\}\times E{0,…,d−1}×E, and a unitary matrix VVV on ({0,…,d−1}×{0,…,d−1})×E(\{0,\ldots,d-1\}\times\{0,\ldots,d-1\})\times E({0,…,d−1}×{0,…,d−1})×E, form

(Di)((a,b),e),((a′,b′),e′)=(ρi)a,a′(τi)(b,e),(b′,e′),(Oi)(a,b),(a′,b′)=∑e∈E(VDiV†)((a,b),e),((a′,b′),e).(D_i)_{((a,b),e),((a',b'),e')} =(\rho_i)_{a,a'}(\tau_i)_{(b,e),(b',e')}, \qquad (O_i)_{(a,b),(a',b')} =\sum_{e\in E}(VD_iV^\dagger)_{((a,b),e),((a',b'),e)}.(Di​)((a,b),e),((a′,b′),e′)​=(ρi​)a,a′​(τi​)(b,e),(b′,e′)​,(Oi​)(a,b),(a′,b′)​=e∈E∑​(VDi​V†)((a,b),e),((a′,b′),e)​.

The defined error is

s(O1,ρ1⊗ρ1)+s(O2,ρ2⊗ρ2)s(ρ1⊗ρ1,ρ2⊗ρ2),s(X,Y)=1−[Re⁡tr⁡S ⁣(S(X)YS(X))]2,\frac{s(O_1,\rho_1\otimes\rho_1)+s(O_2,\rho_2\otimes\rho_2)} {s(\rho_1\otimes\rho_1,\rho_2\otimes\rho_2)}, \qquad s(X,Y)=\sqrt{1-\left[\operatorname{Re}\operatorname{tr} \mathcal S\!\bigl(\mathcal S(X)Y\mathcal S(X)\bigr)\right]^2},s(ρ1​⊗ρ1​,ρ2​⊗ρ2​)s(O1​,ρ1​⊗ρ1​)+s(O2​,ρ2​⊗ρ2​)​,s(X,Y)=1−[RetrS(S(X)YS(X))]2​,

where S\mathcal SS denotes real continuous functional calculus for the real square-root function. Neither positivity nor normalization of the input matrices is required; d=0d=0d=0 and empty EEE are included. Real square roots of nonpositive numbers and quotients with zero denominator are zero.

QAlgorithms.Papers.Independent.Intakenext100.P97.relativeError

For any finite type AAA with decidable equality and any four complex matrices O1,O2,I1,I2O_1,O_2,I_1,I_2O1​,O2​,I1​,I2​ indexed by A×AA\times AA×A, define

F(X,Y)=Re⁡tr⁡S ⁣(S(X)YS(X)),s(X,Y)=1−F(X,Y)2,F(X,Y)=\operatorname{Re}\operatorname{tr} \mathcal S\!\bigl(\mathcal S(X)Y\mathcal S(X)\bigr), \qquad s(X,Y)=\sqrt{1-F(X,Y)^2},F(X,Y)=RetrS(S(X)YS(X)),s(X,Y)=1−F(X,Y)2​,

where S\mathcal SS is the real continuous functional calculus applied to the real square-root function. The defined relative error is the unweighted ratio

s(O1,I1)+s(O2,I2)s(I1,I2).\frac{s(O_1,I_1)+s(O_2,I_2)}{s(I_1,I_2)}.s(I1​,I2​)s(O1​,I1​)+s(O2​,I2​)​.

No matrix is assumed to be a density matrix, and AAA may be empty. These are total operations: the real square root is zero on nonpositive arguments, and division by zero returns zero, so the relative error is zero whenever its denominator is zero.

QAlgorithms.Papers.Independent.Intakenext100.P97.standardBound

For every real number fff, this defines

f−f21+f2.f-\frac{f^{2}}{\sqrt{1+f^{2}}}.f−1+f2​f2​.

No restriction such as 0≤f≤10\le f\le10≤f≤1 is imposed. The denominator is strictly positive for every real fff.

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