Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Definitions for Optimal quantum source coding with quantum information at the encoder and decoder

Definition
QAlgorithms_Papers_Independent_Gap48_P38

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

prove2quantum

The objects the formalized statements of Optimal quantum source coding with quantum information at the encoder and decoder (arXiv:0706.2907) are stated in. Each was checked against the paper by a blind read-back and a moderator.

  • Restricted: Configurations of exactly the named subsystem, including the empty tensor product (p.2).
  • subsystemIndex: Assemble retained and traced-out computational indices; no coordinate is discarded twice (p.2).
  • subsystemReduction: Actual partial trace of the full operator over the complement of s (PDF p.2). The accepted traceRight is applied after regrouping retained/complementary indices; off-diagonal retained entries are preserved.
  • registerEntropy: Entropy in bits of an actual reduced operator, not independent entropy data (PDF p.2).
  • registerMutualInfo: I(S;T)=H(S)+H(T)-H(ST); the statements use disjoint subsystems (PDF p.2).
  • registerCMI: I(S;T|V)=H(SV)+H(TV)-H(V)-H(STV), in bits (PDF p.2).
  • cornerQ: The numerical corner qubit cost from (2), with A,B,C,D numbered 0,1,2,3 (PDF p.2). This definition is the entropy expression, not a definition of operational achievability.
  • cornerE: The signed numerical corner ebit cost from (2), consumption positive (PDF p.2).
  • iidGrouped: IID copies regrouped by subsystem, with exactly the original amplitudes (PDF pp.4--5).
  • productProjection: Product of the local projections, in the same regrouped tensor convention (PDF p.5).
  • TypicalBounds: All clauses of Theorem 5, including its continuation on PDF p.5. Ranks are Schatten zero quantities; frobNorm squared is the squared Schatten two norm; specNorm is the Schatten infinity norm. The six subsystems exclude ABC.
  • decouplingMarginal: The actual Bhat,E marginal of (WU tensor I_E)rho(WU tensor I_E)^dagger, tracing S only, with S,Bhat,E order fixed explicitly (PDF pp.6,11).
  • localAct: Apply an X operator, leaving the reference Y untouched (PDF p.7).
  • coherentDecoder: The explicitly constructed coherent decoder, with output X,K order (PDF p.7). The square root on Lambda is essential.
  • decoderOverlap: The overlap <j|<psi|(L tensor I_Y)|phi>, including both references (PDF p.7).
  • coherentMeasurementError: P in Lemma 5 is the average measurement error on U_j psi (PDF p.7).
  • decodingFidelity: F in Lemma 5 is the average SQUARED pure-state fidelity (PDF p.7).
  • assistedChannel: The actual channel D composed with (E tensor identity) on rho tensor tau. E acts on X,L; D acts on Y,Lprime; the untouched Lprime off-diagonal blocks are retained. No extra communication or classical message is available (PDF p.9).
Definition code
import Mathlib
import Definitions.Def_QAlgorithms_Defs_MatrixDistances
import Definitions.Def_QAlgorithms_Defs_SchurUnitaryEnsembles
import Definitions.Def_QAlgorithms_Defs_FiniteRegisters
import Definitions.Def_QAlgorithms_Defs_Channel
import Definitions.Def_QAlgorithms_Defs_Ensemble
import Definitions.Def_QAlgorithms_Defs_Measurement
import Definitions.Def_QAlgorithms_Defs_LinearSystems
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P08
import Definitions.Def_QAlgorithms_Papers_Independent_Next46_P39

/-!
Statements from Yard--Devetak, arXiv:0706.2907v2. PDF pages are used throughout.
All registers are finite complex registers with fixed bases; nonempty register
instances permit dimension one. Entropies are in bits, distances are FULL trace
norms, and pure-state fidelities are squared overlaps. Operational resource
inequalities without an audited simulation interface are explicitly deferred
in the unit's HARD.md rather than replaced by entropy inequalities.
-/

noncomputable section

open MeasureTheory
open scoped Matrix ComplexConjugate ComplexOrder Matrix.Norms.Elementwise

attribute [local instance] Classical.propDecidable Classical.decEq

namespace QAlgorithms.Papers.Independent.Gap48.P38

/-- Configurations of exactly the named subsystem, including the empty tensor product (p.2). -/
abbrev Restricted {m : ℕ} (I : Fin m → Type) (s : Finset (Fin m)) :=
  (i : {j : Fin m // j ∈ s}) → I i.val

/-- Assemble retained and traced-out computational indices; no coordinate is discarded twice (p.2). -/
def subsystemIndex {m : ℕ} (I : Fin m → Type) (s : Finset (Fin m))
    (x : Restricted I s) (z : Restricted I (Finset.univ \ s)) : (∀ i, I i) :=
  fun i => if h : i ∈ s then x ⟨i, h⟩
    else z ⟨i, Finset.mem_sdiff.mpr ⟨Finset.mem_univ i, h⟩⟩

/-- Actual partial trace of the full operator over the complement of s (PDF p.2).
The accepted traceRight is applied after regrouping retained/complementary
indices; off-diagonal retained entries are preserved. -/
def subsystemReduction {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)]
    (rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) (s : Finset (Fin m)) :
    Matrix (Restricted I s) (Restricted I s) ℂ :=
  traceRight (fun x y : Restricted I s × Restricted I (Finset.univ \ s) =>
    rho (subsystemIndex I s x.1 x.2) (subsystemIndex I s y.1 y.2))

/-- Entropy in bits of an actual reduced operator, not independent entropy data (PDF p.2). -/
def registerEntropy {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)]
    (rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) (s : Finset (Fin m)) : ℝ :=
  vnEntropy (subsystemReduction I rho s)

/-- I(S;T)=H(S)+H(T)-H(ST); the statements use disjoint subsystems (PDF p.2). -/
def registerMutualInfo {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)]
    (rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) (s t : Finset (Fin m)) : ℝ :=
  registerEntropy I rho s + registerEntropy I rho t - registerEntropy I rho (s ∪ t)

/-- I(S;T|V)=H(SV)+H(TV)-H(V)-H(STV), in bits (PDF p.2). -/
def registerCMI {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)]
    (rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) (s t v : Finset (Fin m)) : ℝ :=
  registerEntropy I rho (s ∪ v) + registerEntropy I rho (t ∪ v) -
    registerEntropy I rho v - registerEntropy I rho (s ∪ t ∪ v)

/-- The numerical corner qubit cost from (2), with A,B,C,D numbered 0,1,2,3 (PDF p.2).
This definition is the entropy expression, not a definition of operational achievability. -/
def cornerQ (I : Fin 4 → Type) [∀ i, Fintype (I i)]
    (rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) : ℝ := registerCMI I rho {2} {3} {1} / 2

/-- The signed numerical corner ebit cost from (2), consumption positive (PDF p.2). -/
def cornerE (I : Fin 4 → Type) [∀ i, Fintype (I i)]
    (rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) : ℝ :=
  registerMutualInfo I rho {0} {2} / 2 - registerMutualInfo I rho {1} {2} / 2

example {m : ℕ} (I : Fin m → Type) (s : Finset (Fin m))
    (x : Restricted I s) (z : Restricted I (Finset.univ \ s))
    (i : Fin m) (hi : i ∈ s) : subsystemIndex I s x z i = x ⟨i, hi⟩ := by
  simp [subsystemIndex, hi]

example {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)] (s : Finset (Fin m)) :
    subsystemReduction I (0 : Matrix (∀ i, I i) (∀ i, I i) ℂ) s = 0 := by
  ext x y
  simp [subsystemReduction, traceRight]

example {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)] (s : Finset (Fin m)) :
    registerEntropy I (0 : Matrix (∀ i, I i) (∀ i, I i) ℂ) s = 0 := by
  have hz : subsystemReduction I (0 : Matrix (∀ i, I i) (∀ i, I i) ℂ) s = 0 := by
    ext x y
    simp [subsystemReduction, traceRight]
  rw [registerEntropy, hz]
  simp [vnEntropy]

/-- IID copies regrouped by subsystem, with exactly the original amplitudes (PDF pp.4--5). -/
def iidGrouped {m : ℕ} (I : Fin m → Type) (n : ℕ)
    (psi : EuclideanSpace ℂ (∀ i, I i)) : EuclideanSpace ℂ (∀ i, Fin n → I i) :=
  WithLp.toLp 2 fun x => ∏ t : Fin n, psi (fun i => x i t)

/-- Product of the local projections, in the same regrouped tensor convention (PDF p.5). -/
def productProjection (I : Fin 3 → Type) [∀ i, Fintype (I i)] (n : ℕ)
    (P : (i : Fin 3) → Matrix (Restricted (fun j => Fin n → I j) {i})
      (Restricted (fun j => Fin n → I j) {i}) ℂ) :
    Matrix (∀ i, Fin n → I i) (∀ i, Fin n → I i) ℂ :=
  fun x y => ∏ i : Fin 3, P i (fun j => x j.val) (fun j => y j.val)

/-- All clauses of Theorem 5, including its continuation on PDF p.5.
Ranks are Schatten zero quantities; frobNorm squared is the squared Schatten two norm;
specNorm is the Schatten infinity norm. The six subsystems exclude ABC. -/
def TypicalBounds (I : Fin 3 → Type) [∀ i, Fintype (I i)]
    (psi : EuclideanSpace ℂ (∀ i, I i)) (n : ℕ) (delta epsilon : ℝ)
    (P : (i : Fin 3) → Matrix (Restricted (fun j => Fin n → I j) {i})
      (Restricted (fun j => Fin n → I j) {i}) ℂ) : Prop :=
  let J := fun i => Fin n → I i
  let rho := pureDensity psi
  let source := iidGrouped I n psi
  let v := act (productProjection I n P) source
  let vC := act (productProjection I n (fun i => if i = 2 then P i else 1)) source
  let phi := pureDensity (normalizedVec v)
  (∀ i : Fin 3, (P i).IsHermitian ∧ P i * P i = P i ∧
    effectProb (P i) (subsystemReduction J (pureDensity source) {i}) ≥ 1 - epsilon ∧
    Real.rpow 2 ((n : ℝ) * registerEntropy I rho {i} - n * delta) ≤ (P i).trace.re ∧
    (P i).trace.re ≤ Real.rpow 2 ((n : ℝ) * registerEntropy I rho {i} + n * delta)) ∧
  0 < ‖v‖ ∧ traceNorm (phi - pureDensity source) ≤ epsilon ∧
  (∀ s : Finset (Fin 3), s ∈ ({ {0}, {1}, {2}, {0,1}, {1,2}, {0,2} } :
      Finset (Finset (Fin 3))) →
    let h := registerEntropy I rho s
    let tau := subsystemReduction J phi s
    Real.rpow 2 ((n : ℝ) * h - n * delta) ≤ (Matrix.rank tau : ℝ) ∧
    (Matrix.rank tau : ℝ) ≤ Real.rpow 2 ((n : ℝ) * h + n * delta) ∧
    Real.rpow 2 (-(n : ℝ) * h - n * delta) ≤ frobNorm tau ^ 2 ∧
    frobNorm tau ^ 2 ≤ Real.rpow 2 (-(n : ℝ) * h + n * delta) ∧
    Real.rpow 2 (-(n : ℝ) * h - n * delta) ≤ specNorm tau ∧
    specNorm tau ≤ Real.rpow 2 (-(n : ℝ) * h + n * delta)) ∧
  0 < ‖vC‖ ∧ traceNorm (pureDensity (normalizedVec vC) - pureDensity source) ≤ epsilon

example {m : ℕ} (I : Fin m → Type) (psi : EuclideanSpace ℂ (∀ i, I i))
    (x : (∀ i, Fin 0 → I i)) : iidGrouped I 0 psi x = 1 := by
  simp [iidGrouped]

example {J : Type} [Fintype J] : normalizedVec (0 : EuclideanSpace ℂ J) = 0 := by
  simp [normalizedVec]

example (I : Fin 3 → Type) [∀ i, Fintype (I i)] (n : ℕ) :
    productProjection I n (fun _ => 0) = 0 := by
  ext x y
  simp [productProjection]

/-- The actual Bhat,E marginal of (WU tensor I_E)rho(WU tensor I_E)^dagger,
tracing S only, with S,Bhat,E order fixed explicitly (PDF pp.6,11). -/
def decouplingMarginal {c s b e : ℕ}
    (W : Matrix (Fin s × Fin b) (Fin c) ℂ)
    (rho : Matrix (Fin c × Fin e) (Fin c × Fin e) ℂ)
    (U : Matrix.unitaryGroup (Fin c) ℂ) :
    Matrix (Fin b × Fin e) (Fin b × Fin e) ℂ :=
  let V := Matrix.kronecker (W * (U : Matrix (Fin c) (Fin c) ℂ))
    (1 : Matrix (Fin e) (Fin e) ℂ)
  let tau := V * rho * Vᴴ
  fun x y => ∑ a : Fin s, tau ((a, x.1), x.2) ((a, y.1), y.2)

example : QAlgorithms.Papers.Independent.Information100.P08.maximallyMixed 1 = (1 : Matrix (Fin 1) (Fin 1) ℂ) := by
  simp [QAlgorithms.Papers.Independent.Information100.P08.maximallyMixed]

example {c s b e : ℕ} (W : Matrix (Fin s × Fin b) (Fin c) ℂ)
    (U : Matrix.unitaryGroup (Fin c) ℂ) :
    decouplingMarginal (e := e) W 0 U = 0 := by
  ext x y
  simp [decouplingMarginal]

example {J : Type} [Fintype J] [DecidableEq J] [Nonempty J] :
    QAlgorithms.Papers.Independent.Next46.P39.inverseSqrt (0 : Matrix J J ℂ) = 0 := by
  simp [QAlgorithms.Papers.Independent.Next46.P39.inverseSqrt, cfc_zero]

/-- Apply an X operator, leaving the reference Y untouched (PDF p.7). -/
def localAct {X Y : Type} [Fintype X] [Fintype Y] [DecidableEq Y]
    (M : Matrix X X ℂ) (psi : EuclideanSpace ℂ (X × Y)) :
    EuclideanSpace ℂ (X × Y) := act (Matrix.kronecker M (1 : Matrix Y Y ℂ)) psi

/-- The explicitly constructed coherent decoder, with output X,K order (PDF p.7).
The square root on Lambda is essential. -/
def coherentDecoder {X : Type} [Fintype X] [DecidableEq X] {k : ℕ}
    (alpha : Fin k → ℂ) (U : Fin k → Matrix.unitaryGroup X ℂ)
    (Lambda : Fin k → Matrix X X ℂ) : Matrix (X × Fin k) X ℂ :=
  fun x y => alpha x.2 * (((U x.2 : Matrix X X ℂ)ᴴ * cfc Real.sqrt (Lambda x.2)) x.1 y)

/-- The overlap <j|<psi|(L tensor I_Y)|phi>, including both references (PDF p.7). -/
def decoderOverlap {X Y : Type} [Fintype X] [Fintype Y] {k : ℕ}
    (psi : EuclideanSpace ℂ (X × Y)) (L : Matrix (X × Fin k) X ℂ)
    (phi : EuclideanSpace ℂ (X × Y)) (j : Fin k) : ℂ :=
  ∑ x : X, ∑ y : Y, conj (psi (x,y)) * ∑ z : X, L (x,j) z * phi (z,y)

/-- P in Lemma 5 is the average measurement error on U_j psi (PDF p.7). -/
def coherentMeasurementError {X Y : Type} [Fintype X] [Fintype Y]
    [DecidableEq X] [DecidableEq Y] {k : ℕ}
    (psi : EuclideanSpace ℂ (X × Y)) (U : Fin k → Matrix.unitaryGroup X ℂ)
    (Lambda : Fin k → Matrix X X ℂ) : ℝ :=
  1 - (k : ℝ)⁻¹ * ∑ j, effectProb (Lambda j)
    (traceRight (pureDensity (localAct (U j : Matrix X X ℂ) psi)))

/-- F in Lemma 5 is the average SQUARED pure-state fidelity (PDF p.7). -/
def decodingFidelity {X Y : Type} [Fintype X] [Fintype Y]
    [DecidableEq X] [DecidableEq Y] {k : ℕ}
    (psi : EuclideanSpace ℂ (X × Y)) (U : Fin k → Matrix.unitaryGroup X ℂ)
    (psiPrime : Fin k → EuclideanSpace ℂ (X × Y)) : ℝ :=
  (k : ℝ)⁻¹ * ∑ j, Complex.normSq (inner ℂ (localAct (U j : Matrix X X ℂ) psi) (psiPrime j))

example {X Y : Type} [Fintype X] [Fintype Y] [DecidableEq Y]
    (psi : EuclideanSpace ℂ (X × Y)) : localAct (0 : Matrix X X ℂ) psi = 0 := by
  simp [localAct, act, Matrix.mulVec]

example {X : Type} [Fintype X] [DecidableEq X] {k : ℕ}
    (U : Fin k → Matrix.unitaryGroup X ℂ) (Lambda : Fin k → Matrix X X ℂ) :
    coherentDecoder (fun _ => 0) U Lambda = 0 := by
  ext x y
  simp [coherentDecoder]

example {X Y : Type} [Fintype X] [Fintype Y] {k : ℕ}
    (psi phi : EuclideanSpace ℂ (X × Y)) (j : Fin k) :
    decoderOverlap psi (0 : Matrix (X × Fin k) X ℂ) phi j = 0 := by
  simp [decoderOverlap]

/-- The actual channel D composed with (E tensor identity) on rho tensor tau.
E acts on X,L; D acts on Y,Lprime; the untouched Lprime off-diagonal blocks
are retained. No extra communication or classical message is available (PDF p.9). -/
def assistedChannel {X Y L Lprime : Type} [Fintype X] [Fintype Y] [Fintype L]
    (E : Superop (X × L) Y) (D : Superop (Y × Lprime) X)
    (tau : Matrix (L × Lprime) (L × Lprime) ℂ) : Superop X X :=
  fun rho => D (fun a b => E
    (fun x y => rho x.1 y.1 * tau (x.2,a.2) (y.2,b.2)) a.1 b.1)

example : assistedChannel
    (fun M : Matrix (Unit × Unit) (Unit × Unit) ℂ =>
      (fun _ _ : Unit => M ((),()) ((),())))
    (fun M : Matrix (Unit × Unit) (Unit × Unit) ℂ =>
      (fun _ _ : Unit => M ((),()) ((),())))
    (1 : Matrix (Unit × Unit) (Unit × Unit) ℂ)
    (1 : Matrix Unit Unit ℂ) () () = 1 := by
  simp [assistedChannel]

end QAlgorithms.Papers.Independent.Gap48.P38
Source
Optimal quantum source coding with quantum information at the encoder and decoder, arXiv:0706.2907v2, https://arxiv.org/abs/0706.2907v2
Read-back

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

QAlgorithms.Papers.Independent.Gap48.P38.Restricted

For any natural number mmm, family of types (Ii)i∈{0,…,m−1}(I_i)_{i\in\{0,\ldots,m-1\}}(Ii​)i∈{0,…,m−1}​, and subset s⊆{0,…,m−1}s\subseteq\{0,\ldots,m-1\}s⊆{0,…,m−1}, this denotes the type of assignments xxx that choose an element xi∈Iix_i\in I_ixi​∈Ii​ for every i∈si\in si∈s. No finiteness or nonemptiness of the types IiI_iIi​ is assumed. If s=∅s=\varnothings=∅, there is exactly one such assignment, including when m=0m=0m=0; if some IiI_iIi​ with i∈si\in si∈s is empty, there are no such assignments.

QAlgorithms.Papers.Independent.Gap48.P38.TypicalBounds

For three finite types I0,I1,I2I_0,I_1,I_2I0​,I1​,I2​ with specified finite enumerations, an arbitrary vector ψ∈C∏iIi\psi\in\mathbb C^{\prod_iI_i}ψ∈C∏i​Ii​, a natural number nnn, real numbers δ,ϵ\delta,\epsilonδ,ϵ, and three arbitrary complex matrices PiP_iPi​ indexed by assignments of nnn values to coordinate iii, this defines the following conjunction. Write Ji=Ii{0,…,n−1}J_i=I_i^{\{0,\ldots,n-1\}}Ji​=Ii{0,…,n−1}​; for any vector www, write D(w)x,y=wxwy‾D(w)_{x,y}=w_x\overline{w_y}D(w)x,y​=wx​wy​​; for a matrix MMM indexed by a product of coordinate types, write (Ms)x,y=∑zMx⊔sz,y⊔sz(M_s)_{x,y}=\sum_zM_{x\sqcup_s z,y\sqcup_s z}(Ms​)x,y​=∑z​Mx⊔s​z,y⊔s​z​, summing over assignments on the complementary coordinates, and write HM(s)=−Re⁡tr⁡f(Ms)H_M(s)=-\operatorname{Re}\operatorname{tr}f(M_s)HM​(s)=−Retrf(Ms​) for f(x)=xlog⁡2xf(x)=x\log_2xf(x)=xlog2​x evaluated by continuous functional calculus using the total real logarithm. Set ρ=D(ψ)\rho=D(\psi)ρ=D(ψ), Ψ(x)=∏t<nψ((xi(t))i=02)\Psi(x)=\prod_{t<n}\psi((x_i(t))_{i=0}^2)Ψ(x)=∏t<n​ψ((xi​(t))i=02​), Qx,y=∏i=02Pi(x∣{i},y∣{i})Q_{x,y}=\prod_{i=0}^2P_i(x|_{\{i\}},y|_{\{i\}})Qx,y​=∏i=02​Pi​(x∣{i}​,y∣{i}​), v=QΨv=Q\Psiv=QΨ, and vC=QCΨv_C=Q_C\PsivC​=QC​Ψ, where QCQ_CQC​ is the same product matrix with P0,P1P_0,P_1P0​,P1​ replaced by identity matrices and P2P_2P2​ retained. Set ϕ=D(∥v∥−1v)\phi=D(\|v\|^{-1}v)ϕ=D(∥v∥−1v), where normalization is defined to give zero when v=0v=0v=0. The required conjunction is

∀i∈{0,1,2}:Pi†=Pi,Pi2=Pi,Re⁡tr⁡ ⁣(PiD(Ψ){i})≥1−ϵ,2 nHρ({i})−nδ≤Re⁡tr⁡Pi≤2 nHρ({i})+nδ;∥v∥>0,∥ϕ−D(Ψ)∥tr≤ϵ;∀s∈{{0},{1},{2},{0,1},{1,2},{0,2}},h=Hρ(s),τ=ϕs:2nh−nδ≤rank⁡τ≤2nh+nδ,2−nh−nδ≤∥τ∥F 2≤2−nh+nδ,2−nh−nδ≤∥τ∥op≤2−nh+nδ;∥vC∥>0,∥D(∥vC∥−1vC)−D(Ψ)∥tr≤ϵ.\begin{gathered} \forall i\in\{0,1,2\}:\quad P_i^\dagger=P_i,\qquad P_i^2=P_i,\qquad \operatorname{Re}\operatorname{tr}\!\left(P_iD(\Psi)_{\{i\}}\right)\ge1-\epsilon,\\ 2^{\,nH_\rho(\{i\})-n\delta} \le\operatorname{Re}\operatorname{tr}P_i \le2^{\,nH_\rho(\{i\})+n\delta};\\ \|v\|>0,\qquad \|\phi-D(\Psi)\|_{\mathrm{tr}}\le\epsilon;\\ \forall s\in \bigl\{\{0\},\{1\},\{2\},\{0,1\},\{1,2\},\{0,2\}\bigr\}, \quad h=H_\rho(s),\quad \tau=\phi_s:\\ 2^{nh-n\delta}\le\operatorname{rank}\tau\le2^{nh+n\delta},\\ 2^{-nh-n\delta}\le\|\tau\|_{\mathrm F}^{\,2}\le2^{-nh+n\delta}, \qquad 2^{-nh-n\delta}\le\|\tau\|_{\mathrm{op}}\le2^{-nh+n\delta};\\ \|v_C\|>0,\qquad \bigl\|D(\|v_C\|^{-1}v_C)-D(\Psi)\bigr\|_{\mathrm{tr}}\le\epsilon. \end{gathered}∀i∈{0,1,2}:Pi†​=Pi​,Pi2​=Pi​,Retr(Pi​D(Ψ){i}​)≥1−ϵ,2nHρ​({i})−nδ≤RetrPi​≤2nHρ​({i})+nδ;∥v∥>0,∥ϕ−D(Ψ)∥tr​≤ϵ;∀s∈{{0},{1},{2},{0,1},{1,2},{0,2}},h=Hρ​(s),τ=ϕs​:2nh−nδ≤rankτ≤2nh+nδ,2−nh−nδ≤∥τ∥F2​≤2−nh+nδ,2−nh−nδ≤∥τ∥op​≤2−nh+nδ;∥vC​∥>0,​D(∥vC​∥−1vC​)−D(Ψ)​tr​≤ϵ.​

Here ∥M∥tr=Re⁡tr⁡M†M\|M\|_{\mathrm{tr}}=\operatorname{Re}\operatorname{tr}\sqrt{M^\dagger M}∥M∥tr​=RetrM†M​, with the square root given by continuous functional calculus, ∥⋅∥F\|\cdot\|_{\mathrm F}∥⋅∥F​ is the Frobenius norm, and ∥⋅∥op\|\cdot\|_{\mathrm{op}}∥⋅∥op​ is the Euclidean operator norm; all displayed powers of 222 are real powers. No normalization of ψ\psiψ, positivity of δ\deltaδ or ϵ\epsilonϵ, positivity of nnn, or nonemptiness of IiI_iIi​ is assumed in defining this predicate. Its positive-norm requirements exclude zero projected vectors; in particular, if any IiI_iIi​ is empty and n>0n>0n>0, those requirements cannot hold. Empty coordinate products and n=0n=0n=0 otherwise use their usual unique assignments and empty products.

QAlgorithms.Papers.Independent.Gap48.P38.assistedChannel

For arbitrary types X,Y,L,L′X,Y,L,L'X,Y,L,L′, with finite structures on X,Y,LX,Y,LX,Y,L, let EEE be any function from complex matrices on X×LX\times LX×L to complex matrices on YYY, let DDD be any function from complex matrices on Y×L′Y\times L'Y×L′ to complex matrices on XXX, and let τ\tauτ be any complex matrix on L×L′L\times L'L×L′. This defines a function A\mathcal AA from matrices on XXX to matrices on XXX: for each input ρ\rhoρ, form the matrix ZZZ on Y×L′Y\times L'Y×L′ with

Z(y,ℓ′),(z,m′)=E ⁣([ρx,x′τ(ℓ,ℓ′),(m,m′)](x,ℓ),(x′,m))y,z,A(ρ)=D(Z).Z_{(y,\ell'),(z,m')} = E\!\left( \bigl[\rho_{x,x'}\tau_{(\ell,\ell'),(m,m')}\bigr]_{(x,\ell),(x',m)} \right)_{y,z}, \qquad \mathcal A(\rho)=D(Z).Z(y,ℓ′),(z,m′)​=E([ρx,x′​τ(ℓ,ℓ′),(m,m′)​](x,ℓ),(x′,m)​)y,z​,A(ρ)=D(Z).

No linearity, channel condition, positivity, or normalization is imposed on these functions or matrices. The finite types may be empty, and no finiteness assumption is imposed on L′L'L′.

QAlgorithms.Papers.Independent.Gap48.P38.coherentDecoder

For a finite type XXX with decidable equality, a natural number kkk, arbitrary complex coefficients αj\alpha_jαj​, unitary matrices UjU_jUj​ indexed by XXX, and arbitrary complex matrices Λj\Lambda_jΛj​ indexed by XXX, for 0≤j<k0\le j<k0≤j<k, this defines a rectangular matrix L:CX→CX×Fin(k)L:\mathbb C^X\to\mathbb C^{X\times\mathrm{Fin}(k)}L:CX→CX×Fin(k) by

L(x,j),y=αj(Uj† cfc⁡(⋅,Λj))x,y.L_{(x,j),y} =\alpha_j\bigl(U_j^\dagger\,\operatorname{cfc}(\sqrt{\cdot},\Lambda_j)\bigr)_{x,y}.L(x,j),y​=αj​(Uj†​cfc(⋅​,Λj​))x,y​.

The square root is the real continuous functional calculus used in the code; no positivity or Hermitian condition on Λj\Lambda_jΛj​, and no modulus condition on αj\alpha_jαj​, is assumed by this definition. The type XXX may be empty, and k=0k=0k=0 gives a matrix with no rows.

QAlgorithms.Papers.Independent.Gap48.P38.coherentMeasurementError

For finite types X,YX,YX,Y with decidable equality, a natural number kkk, an arbitrary vector ψ∈CX×Y\psi\in\mathbb C^{X\times Y}ψ∈CX×Y, unitary matrices UjU_jUj​ on CX\mathbb C^XCX, and arbitrary complex matrices Λj\Lambda_jΛj​ on CX\mathbb C^XCX, indexed by 0≤j<k0\le j<k0≤j<k, set

vj(x,y)=∑z∈X(Uj)x,zψ(z,y),(Rj)x,z=∑y∈Yvj(x,y)vj(z,y)‾.v_j(x,y)=\sum_{z\in X}(U_j)_{x,z}\psi(z,y), \qquad (R_j)_{x,z}=\sum_{y\in Y}v_j(x,y)\overline{v_j(z,y)}.vj​(x,y)=z∈X∑​(Uj​)x,z​ψ(z,y),(Rj​)x,z​=y∈Y∑​vj​(x,y)vj​(z,y)​.

This definition returns

1−1k∑j=0k−1Re⁡tr⁡(ΛjRj).1-\frac1k\sum_{j=0}^{k-1}\operatorname{Re}\operatorname{tr}(\Lambda_jR_j).1−k1​j=0∑k−1​Retr(Λj​Rj​).

No normalization of ψ\psiψ, positivity of the Λj\Lambda_jΛj​, or condition that their sum is the identity is assumed. The types may be empty. At k=0k=0k=0, the sum and the defined inverse of kkk are both zero, so the value is 111.

QAlgorithms.Papers.Independent.Gap48.P38.cornerE

For four finite types I0,I1,I2,I3I_0,I_1,I_2,I_3I0​,I1​,I2​,I3​ with specified finite enumerations and any complex matrix ρ\rhoρ indexed by their full assignments, this defines

12(Hρ({0})+Hρ({2})−Hρ({0,2}))−12(Hρ({1})+Hρ({2})−Hρ({1,2})).\frac12\Bigl(H_\rho(\{0\})+H_\rho(\{2\})-H_\rho(\{0,2\})\Bigr) -\frac12\Bigl(H_\rho(\{1\})+H_\rho(\{2\})-H_\rho(\{1,2\})\Bigr).21​(Hρ​({0})+Hρ​({2})−Hρ​({0,2}))−21​(Hρ​({1})+Hρ​({2})−Hρ​({1,2})).

Here (ρa)x,y=∑z∈∏i∉aIiρx⊔az,y⊔az(\rho_a)_{x,y}=\sum_{z\in\prod_{i\notin a}I_i}\rho_{x\sqcup_a z,y\sqcup_a z}(ρa​)x,y​=∑z∈∏i∈/a​Ii​​ρx⊔a​z,y⊔a​z​, with ⊔a\sqcup_a⊔a​ combining assignments, and Hρ(a)=−Re⁡tr⁡f(ρa)H_\rho(a)=-\operatorname{Re}\operatorname{tr}f(\rho_a)Hρ​(a)=−Retrf(ρa​) for f(x)=xlog⁡2xf(x)=x\log_2xf(x)=xlog2​x evaluated by continuous functional calculus using the total real logarithm. No positivity, normalization, or nonemptiness is assumed; the formula applies also when coordinate types are empty.

QAlgorithms.Papers.Independent.Gap48.P38.cornerQ

For four finite types I0,I1,I2,I3I_0,I_1,I_2,I_3I0​,I1​,I2​,I3​ with specified finite enumerations and any complex matrix ρ\rhoρ indexed by I0×I1×I2×I3I_0\times I_1\times I_2\times I_3I0​×I1​×I2​×I3​, this defines

12(Hρ({1,2})+Hρ({1,3})−Hρ({1})−Hρ({1,2,3})).\frac12\Bigl( H_\rho(\{1,2\})+H_\rho(\{1,3\}) -H_\rho(\{1\})-H_\rho(\{1,2,3\}) \Bigr).21​(Hρ​({1,2})+Hρ​({1,3})−Hρ​({1})−Hρ​({1,2,3})).

For each coordinate subset aaa, (ρa)x,y=∑z∈∏i∉aIiρx⊔az,y⊔az(\rho_a)_{x,y}=\sum_{z\in\prod_{i\notin a}I_i}\rho_{x\sqcup_a z,y\sqcup_a z}(ρa​)x,y​=∑z∈∏i∈/a​Ii​​ρx⊔a​z,y⊔a​z​, where ⊔a\sqcup_a⊔a​ combines assignments, and Hρ(a)=−Re⁡tr⁡f(ρa)H_\rho(a)=-\operatorname{Re}\operatorname{tr}f(\rho_a)Hρ​(a)=−Retrf(ρa​) with f(x)=xlog⁡2xf(x)=x\log_2xf(x)=xlog2​x evaluated by continuous functional calculus using the total real logarithm. The coordinate labels are 0,1,2,30,1,2,30,1,2,3. No density-matrix or nonemptiness hypothesis is imposed, so empty coordinate types are included.

QAlgorithms.Papers.Independent.Gap48.P38.decoderOverlap

For arbitrary finite types X,YX,YX,Y, a natural number kkk, vectors ψ,ϕ∈CX×Y\psi,\phi\in\mathbb C^{X\times Y}ψ,ϕ∈CX×Y, a complex matrix LLL with rows indexed by X×Fin(k)X\times\mathrm{Fin}(k)X×Fin(k) and columns indexed by XXX, and an index j∈Fin(k)j\in\mathrm{Fin}(k)j∈Fin(k), this defines the complex number

∑x∈X∑y∈Yψ(x,y)‾∑z∈XL(x,j),zϕ(z,y).\sum_{x\in X}\sum_{y\in Y} \overline{\psi(x,y)} \sum_{z\in X}L_{(x,j),z}\phi(z,y).x∈X∑​y∈Y∑​ψ(x,y)​z∈X∑​L(x,j),z​ϕ(z,y).

No normalization or isometry assumptions are imposed. If XXX or YYY is empty, the value is zero. Supplying jjj requires k>0k>0k>0, so there is no instance of the definition with such an argument when k=0k=0k=0.

QAlgorithms.Papers.Independent.Gap48.P38.decodingFidelity

For finite types X,YX,YX,Y with decidable equality, a natural number kkk, an arbitrary vector ψ∈CX×Y\psi\in\mathbb C^{X\times Y}ψ∈CX×Y, unitary matrices UjU_jUj​ on CX\mathbb C^XCX, and arbitrary vectors ψj′∈CX×Y\psi'_j\in\mathbb C^{X\times Y}ψj′​∈CX×Y, indexed by 0≤j<k0\le j<k0≤j<k, this defines

1k∑j=0k−1∣∑x∈X∑y∈Y(∑z∈X(Uj)x,zψ(z,y))‾ ψj′(x,y)∣2.\frac1k\sum_{j=0}^{k-1} \left| \sum_{x\in X}\sum_{y\in Y} \overline{\left(\sum_{z\in X}(U_j)_{x,z}\psi(z,y)\right)} \,\psi'_j(x,y) \right|^2.k1​j=0∑k−1​​x∈X∑​y∈Y∑​(z∈X∑​(Uj​)x,z​ψ(z,y))​ψj′​(x,y)​2.

Neither ψ\psiψ nor the ψj′\psi'_jψj′​ are required to be normalized. Either type may be empty, in which case the inner sums are zero. At k=0k=0k=0, the empty sum and the defined inverse of zero give value zero.

QAlgorithms.Papers.Independent.Gap48.P38.decouplingMarginal

For arbitrary natural numbers c,s,b,ec,s,b,ec,s,b,e, a complex matrix W:Cc→Cs⊗CbW:\mathbb C^c\to\mathbb C^s\otimes\mathbb C^bW:Cc→Cs⊗Cb, an arbitrary complex matrix ρ\rhoρ indexed by Fin(c)×Fin(e)\mathrm{Fin}(c)\times\mathrm{Fin}(e)Fin(c)×Fin(e), and a unitary c×cc\times cc×c matrix UUU, this defines a matrix on Cb⊗Ce\mathbb C^b\otimes\mathbb C^eCb⊗Ce. Set V=(WU)⊗IeV=(WU)\otimes I_eV=(WU)⊗Ie​ and τ=VρV†\tau=V\rho V^\daggerτ=VρV†; the resulting matrix has entries

DW(ρ,U)(i,r),(j,t)=∑a=0s−1τ((a,i),r),((a,j),t).\mathcal D_W(\rho,U)_{(i,r),(j,t)} =\sum_{a=0}^{s-1}\tau_{((a,i),r),((a,j),t)}.DW​(ρ,U)(i,r),(j,t)​=a=0∑s−1​τ((a,i),r),((a,j),t)​.

Thus it sums over the first output factor of dimension sss. No positivity, normalization, or isometry condition is imposed on ρ\rhoρ or WWW. All four dimensions may be zero; in particular, when s=0s=0s=0, every entry is an empty sum and hence zero.

QAlgorithms.Papers.Independent.Gap48.P38.iidGrouped

For any natural numbers m,nm,nm,n, family of types (Ii)i<m(I_i)_{i<m}(Ii​)i<m​, and vector ψ\psiψ in the complex Euclidean space indexed by full assignments ∏i<mIi\prod_{i<m}I_i∏i<m​Ii​, this defines a vector Ψn\Psi_nΨn​ indexed by families x=(xi)i<mx=(x_i)_{i<m}x=(xi​)i<m​ with xi:{0,…,n−1}→Iix_i:\{0,\ldots,n-1\}\to I_ixi​:{0,…,n−1}→Ii​, by

Ψn(x)=∏t=0n−1ψ((xi(t))i<m).\Psi_n(x)=\prod_{t=0}^{n-1}\psi\bigl((x_i(t))_{i<m}\bigr).Ψn​(x)=t=0∏n−1​ψ((xi​(t))i<m​).

No finiteness or nonemptiness assumptions on the types, and no normalization assumption on ψ\psiψ, are included. When n=0n=0n=0, every coordinate sequence is empty and the product equals 111. When m=0m=0m=0, the full-assignment type has one element, and the value is the nnnth power of ψ\psiψ at that element.

QAlgorithms.Papers.Independent.Gap48.P38.localAct

For arbitrary finite types X,YX,YX,Y, with decidable equality on YYY, any complex matrix MMM indexed by XXX, and any vector ψ∈CX×Y\psi\in\mathbb C^{X\times Y}ψ∈CX×Y, this defines the vector obtained by matrix multiplication with M⊗IYM\otimes I_YM⊗IY​; explicitly,

localAct⁡(M,ψ)(x,y)=∑z∈XMx,zψ(z,y).\operatorname{localAct}(M,\psi)(x,y) =\sum_{z\in X}M_{x,z}\psi(z,y).localAct(M,ψ)(x,y)=z∈X∑​Mx,z​ψ(z,y).

There is no unitarity condition on MMM or normalization condition on ψ\psiψ. Either indexing type may be empty.

QAlgorithms.Papers.Independent.Gap48.P38.productProjection

For three finite types I0,I1,I2I_0,I_1,I_2I0​,I1​,I2​ with specified finite enumerations, any natural number nnn, and, for each i∈{0,1,2}i\in\{0,1,2\}i∈{0,1,2}, an arbitrary complex matrix PiP_iPi​ indexed by assignments of nnn values to the single coordinate iii, this defines a complex matrix QQQ on grouped assignments x=(xi)i=02x=(x_i)_{i=0}^2x=(xi​)i=02​, where each xi:{0,…,n−1}→Iix_i:\{0,\ldots,n-1\}\to I_ixi​:{0,…,n−1}→Ii​, by

Qx,y=∏i=02Pi(x∣{i},y∣{i}).Q_{x,y}=\prod_{i=0}^{2}P_i(x|_{\{i\}},y|_{\{i\}}).Qx,y​=i=0∏2​Pi​(x∣{i}​,y∣{i}​).

The restrictions retain only the indicated coordinate sequence. No Hermiticity, idempotence, positivity, normalization, or nonemptiness is assumed for these inputs. The case n=0n=0n=0 is included: each local assignment type then has one element, and QQQ is the one-entry matrix whose entry is the product of the three local entries.

QAlgorithms.Papers.Independent.Gap48.P38.registerCMI

For any natural number mmm, family of finite types (Ii)i<m(I_i)_{i<m}(Ii​)i<m​ with specified finite enumerations, arbitrary complex matrix ρ\rhoρ indexed by ∏i<mIi\prod_{i<m}I_i∏i<m​Ii​, and subsets s,t,v⊆{0,…,m−1}s,t,v\subseteq\{0,\ldots,m-1\}s,t,v⊆{0,…,m−1}, this defines

Hρ(s∪v)+Hρ(t∪v)−Hρ(v)−Hρ(s∪t∪v).H_\rho(s\cup v)+H_\rho(t\cup v)-H_\rho(v)-H_\rho(s\cup t\cup v).Hρ​(s∪v)+Hρ​(t∪v)−Hρ​(v)−Hρ​(s∪t∪v).

For every subset aaa, the matrix ρa\rho_aρa​ has entries ∑z∈∏i∈acIiρx⊔az,y⊔az\sum_{z\in\prod_{i\in a^c}I_i}\rho_{x\sqcup_a z,y\sqcup_a z}∑z∈∏i∈ac​Ii​​ρx⊔a​z,y⊔a​z​, where ⊔a\sqcup_a⊔a​ combines assignments, and Hρ(a)=−Re⁡tr⁡f(ρa)H_\rho(a)=-\operatorname{Re}\operatorname{tr}f(\rho_a)Hρ​(a)=−Retrf(ρa​) for the continuous functional calculus expression associated with f(x)=xlog⁡2xf(x)=x\log_2xf(x)=xlog2​x, using the total real logarithm. No disjointness among s,t,vs,t,vs,t,v, density-matrix condition on ρ\rhoρ, or nonemptiness of the coordinate types is assumed. Empty subsets and m=0m=0m=0 are included.

QAlgorithms.Papers.Independent.Gap48.P38.registerEntropy

For any natural number mmm, family of finite types (Ii)i<m(I_i)_{i<m}(Ii​)i<m​ with specified finite enumerations, arbitrary complex matrix ρ\rhoρ indexed by ∏i<mIi\prod_{i<m}I_i∏i<m​Ii​, and subset s⊆{0,…,m−1}s\subseteq\{0,\ldots,m-1\}s⊆{0,…,m−1}, this defines

Hρ(s)=−Re⁡tr⁡(f(ρs)),f(x)=xlog⁡2x,H_\rho(s)=-\operatorname{Re}\operatorname{tr}\bigl(f(\rho_s)\bigr), \qquad f(x)=x\log_2 x,Hρ​(s)=−Retr(f(ρs​)),f(x)=xlog2​x,

where f(ρs)f(\rho_s)f(ρs​) is the continuous functional calculus expression and (ρs)u,v=∑z∈∏i∈scIiρu⊔sz,v⊔sz(\rho_s)_{u,v}=\sum_{z\in\prod_{i\in s^c}I_i}\rho_{u\sqcup_s z,v\sqcup_s z}(ρs​)u,v​=∑z∈∏i∈sc​Ii​​ρu⊔s​z,v⊔s​z​, with ⊔s\sqcup_s⊔s​ combining assignments on sss and its complement. The logarithm is Lean’s total real base-two logarithm, with log⁡20=0\log_2 0=0log2​0=0. No density-matrix or nonemptiness hypothesis is required: this is the stated expression for every matrix, including matrices on empty assignment types and the case m=0m=0m=0.

QAlgorithms.Papers.Independent.Gap48.P38.registerMutualInfo

For any natural number mmm, family of finite types (Ii)i<m(I_i)_{i<m}(Ii​)i<m​ with specified finite enumerations, arbitrary complex matrix ρ\rhoρ indexed by ∏i<mIi\prod_{i<m}I_i∏i<m​Ii​, and subsets s,t⊆{0,…,m−1}s,t\subseteq\{0,\ldots,m-1\}s,t⊆{0,…,m−1}, this defines

Hρ(s)+Hρ(t)−Hρ(s∪t).H_\rho(s)+H_\rho(t)-H_\rho(s\cup t).Hρ​(s)+Hρ​(t)−Hρ​(s∪t).

Here, for each subset aaa, (ρa)x,y=∑z∈∏i∈acIiρx⊔az,y⊔az(\rho_a)_{x,y}=\sum_{z\in\prod_{i\in a^c}I_i}\rho_{x\sqcup_a z,y\sqcup_a z}(ρa​)x,y​=∑z∈∏i∈ac​Ii​​ρx⊔a​z,y⊔a​z​, where ⊔a\sqcup_a⊔a​ combines the indicated assignments, and Hρ(a)=−Re⁡tr⁡f(ρa)H_\rho(a)=-\operatorname{Re}\operatorname{tr}f(\rho_a)Hρ​(a)=−Retrf(ρa​) with f(x)=xlog⁡2xf(x)=x\log_2xf(x)=xlog2​x evaluated by continuous functional calculus using the total real logarithm. The subsets need not be disjoint and may be empty. Neither ρ\rhoρ nor the coordinate types are assumed to be normalized, positive, or nonempty, and m=0m=0m=0 is allowed.

QAlgorithms.Papers.Independent.Gap48.P38.subsystemIndex

For any natural number mmm, family of types (Ii)i<m(I_i)_{i<m}(Ii​)i<m​, subset s⊆{0,…,m−1}s\subseteq\{0,\ldots,m-1\}s⊆{0,…,m−1}, assignment xi∈Iix_i\in I_ixi​∈Ii​ on sss, and assignment zi∈Iiz_i\in I_izi​∈Ii​ on its complement scs^csc, this defines the full assignment x⊔szx\sqcup_s zx⊔s​z by (x⊔sz)i=xi(x\sqcup_s z)_i=x_i(x⊔s​z)i​=xi​ when i∈si\in si∈s and (x⊔sz)i=zi(x\sqcup_s z)_i=z_i(x⊔s​z)i​=zi​ otherwise. No finiteness or nonemptiness assumptions are imposed on the types. Empty subsets and m=0m=0m=0 are included; on an empty collection of coordinates the corresponding assignment is unique.

QAlgorithms.Papers.Independent.Gap48.P38.subsystemReduction

For any natural number mmm, family of finite types (Ii)i<m(I_i)_{i<m}(Ii​)i<m​ with specified finite enumerations, arbitrary complex matrix ρ\rhoρ indexed by full assignments in ∏i<mIi\prod_{i<m}I_i∏i<m​Ii​, and subset s⊆{0,…,m−1}s\subseteq\{0,\ldots,m-1\}s⊆{0,…,m−1}, this defines a matrix indexed by assignments on sss whose entries are

(ρs)x,y=∑z∈∏i∈scIiρ x⊔sz,  y⊔sz,(\rho_s)_{x,y} =\sum_{z\in\prod_{i\in s^c}I_i} \rho_{\,x\sqcup_s z,\;y\sqcup_s z},(ρs​)x,y​=z∈∏i∈sc​Ii​∑​ρx⊔s​z,y⊔s​z​,

where x⊔szx\sqcup_s zx⊔s​z uses xxx on sss and zzz on its complement. No positivity, Hermiticity, trace, or nonemptiness assumption is made. If the complementary assignment type is empty, the sum is zero; if the complement has no coordinates, it has one assignment and the sum has one term. The definition also includes m=0m=0m=0 and empty matrix index types.

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).

  • Endorsed by Gohan · Oct 29, 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