Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Definitions for The structure of degradable quantum channels

Definition
QAlgorithms_Papers_Independent_Gap20_P19

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

prove2quantum

The objects the formalized statements of The structure of degradable quantum channels (arXiv:0802.1360) are stated in. Each was checked against the paper by a blind read-back and a moderator.

  • PseudoDiagonal: PDF p.19, (51): arbitrary PSD Gram matrix and frame vectors. Trace preservation is imposed separately, not on each frame vector. The output unitary permits the arbitrary orthonormal basis in (51).
  • TwistedDiagonal: PDF pp.18–20: simultaneous diagonalization by invertible left/right factors.
  • RectangularUnitaryConjugate: PDF pp.12–15: independent input/output conjugations for rectangular maps. The accepted P60.UnitaryEquivalent covers only square Fin d channels.
  • canonicalKraus: PDF pp.14,28–29: corrected normalized pair for (85). Printed (39) interchanges the off-diagonal entries of A_minus.
  • rankTwoQubitMap: PDF p.28, (85), on every operator. This is a qubit channel, distinct from the accepted P69.canonical, which constructs a purification vector.
  • qutritToQubitKraus: PDF p.12, (27): the specific qutrit-to-qubit Kraus pair. The accepted Dependency50.P23.qutritKraus instead has qutrit output.
  • qutritCanonical: PDF p.12, (26).
  • paulis: PDF pp.15,28: sigma_0=I and ordered X,Y,Z.
  • pauliKraus: PDF pp.15,28: a_k are Kraus amplitudes.
  • qubitPauliMap: PDF p.28: qubit Kraus amplitudes in I,X,Y,Z order. The accepted Frontier18.P17.pauliChannel uses probability weights and Fin d Weyl labels.
  • pauliMultipliers: PDF p.28, (87): multipliers in X,Y,Z order.
  • multiplierChannel: PDF p.15, (41): linear extension of the unital diagonal Bloch map.
  • CPBounds: PDF pp.15,28,38: both CP inequalities and physical multiplier bounds.
  • antiExpression: PDF p.15, (42): cyclic indices enumerate the other two multipliers.
  • WeylProbabilities: PDF pp.20–22, (55): probabilities, not Kraus amplitudes.
  • weylChannel: PDF pp.20–22: positive-clock, forward-shift X^j Z^k mixture.
  • weylMultiplier: PDF p.21: coefficient computed from Phi with the omitted 1/d restored.
  • threeWeyl: PDF pp.22–23: three-term qutrit family.
  • flaggedIdentity: PDF pp.10–11, (19): identity/noise with orthogonal flags.
  • flaggedDilation: PDF p.10: normalized dilation of (19), none denotes the identity branch.
  • flaggedComplement: PDF p.10, (20): complement order scalar first, noise environment second.
  • flaggedDegrader: PDF p.11, (22): TP extension to every operator, discarding flag coherences. The scalar-first order agrees with flaggedComplement.
  • symmetricIsometry: PDF p.11, (23): an explicit orthonormal symmetric-subspace basis.
  • symmetricSubspaceMap: PDF p.11: the symmetric-subspace quantum channel from six to three dimensions; accepted P06.symmetricChannel is a classical transition matrix.
  • coFlags: PDF p.11, (25): weighted flags on a common input.
  • coFlagDilation: PDF p.11: normalized direct-sum dilation, with distinct environments.
  • coherentAt: PDF p.26, (4): coherent information at an input state.
  • feasibilityMatrix: PDF pp.32–33, (97): direct PSD matrix without a singular Schur inverse.
  • pauliAntidegrader: PDF p.31, (94): the factor 1/2 converts Pauli traces to actual output.
  • faceAmplitudes: PDF pp.34–35: face probability coordinates, a_0=0.
  • twoPauliAmplitudes: PDF p.35: parameter s=t^2, with its boundary s=1/2 retained.
  • depolarizingAmplitudes: PDF pp.35–36: fixed Pauli dilation for a depolarizing multiplier.
  • ExtremeIn: PDF pp.10,35: extremality inside a named convex class of channels.
  • twoPauliOrbit: PDF p.35: orbit of (2/3,2/3,1/3) under permutations and even sign changes.
  • wernerIsometry: PDF p.10: the normalized antisymmetric qutrit dilation of W_3. Its columns are the Levi-Civita tensor divided by sqrt2.
Definition code
import Mathlib
import Definitions.Def_QAlgorithms_Defs_QubitBlochMaps
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P67
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P60
import Definitions.Def_QAlgorithms_Papers_Independent_Claimed50_P48
import Definitions.Def_QAlgorithms_Papers_Independent_Claimed50_P24
import Definitions.Def_QAlgorithms_Papers_Independent_Gap48_P36
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P69
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P50
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P08

/-! Cubitt–Ruskai–Smith, arXiv:0802.1360v2. PDF pages throughout.
Source corrections: Theorem 3 uses rank rho; (39) has A_minus transposed to
recover (85); Theorem 12 uses output rank d_E, as computed in (54).
Compilation and independent moderation are pending the batch handoff. -/

noncomputable section

namespace QAlgorithms.Papers.Independent.Gap20.P19

open scoped BigOperators Matrix ComplexConjugate ComplexOrder Kronecker
open Filter Topology
open QAlgorithms.Papers.Independent.Information100.P67
  (IsIsometry receiverChannel complementaryChannel IsDegradable IsAntidegradable
   maximumCoherentInformation dilationPower)
open QAlgorithms.Papers.Independent.Claimed50.P48
  (MinimalDilation indexedKrausDilation arbitraryVectorComplement superopTensor tensorDilation)
open QAlgorithms.Papers.Independent.Claimed50.P24 (ChannelEB)
open QAlgorithms.Papers.Independent.Information100.P60 (wernerHolevo)
open QAlgorithms.Papers.Independent.Gap48.P36 (finWeyl WeylLabel)
open QAlgorithms.Papers.Independent.Information100.P69 (choi)
open QAlgorithms.Papers.Independent.Information100.P50 (matrixRange)
open QAlgorithms.Papers.Independent.Information100.P08 (IsPure)

attribute [local instance] Classical.propDecidable Classical.decEq

variable {A B C E F R K A2 B2 E2 : Type}
variable [Fintype A] [DecidableEq A] [Nonempty A]
variable [Fintype B] [DecidableEq B] [Nonempty B]
variable [Fintype C] [DecidableEq C] [Nonempty C]
variable [Fintype E] [DecidableEq E] [Nonempty E]
variable [Fintype F] [DecidableEq F] [Nonempty F]
variable [Fintype R] [DecidableEq R] [Nonempty R]
variable [Fintype K] [DecidableEq K]
variable [Fintype A2] [DecidableEq A2] [Nonempty A2]
variable [Fintype B2] [DecidableEq B2] [Nonempty B2]
variable [Fintype E2] [DecidableEq E2] [Nonempty E2]

/-- PDF p.19, (51): arbitrary PSD Gram matrix and frame vectors.
Trace preservation is imposed separately, not on each frame vector.
The output unitary permits the arbitrary orthonormal basis in (51). -/
def PseudoDiagonal (Phi : Superop A B) : Prop :=
  ∃ c : Matrix B B ℂ, c.PosSemidef ∧ ∃ w : B → EuclideanSpace ℂ A,
    ∃ V : Matrix.unitaryGroup B ℂ, ∀ rho,
      Phi rho = conjSuperop (V : Matrix B B ℂ) (arbitraryVectorComplement c w rho)

/-- PDF pp.18–20: simultaneous diagonalization by invertible left/right factors. -/
def TwistedDiagonal (Phi : Superop A A) : Prop :=
  ∃ m : ℕ, ∃ H : Fin m → Matrix A A ℂ, IsKrausFamily H ∧ Phi = krausApply H ∧
    ∃ X Y : Matrix A A ℂ, IsUnit X ∧ IsUnit Y ∧
      ∀ k, ∃ v : A → ℂ, Y * H k * X = Matrix.diagonal v

/-- PDF pp.12–15: independent input/output conjugations for rectangular maps.
The accepted P60.UnitaryEquivalent covers only square Fin d channels. -/
def RectangularUnitaryConjugate (Phi Psi : Superop A B) : Prop :=
  ∃ U : Matrix.unitaryGroup A ℂ, ∃ V : Matrix.unitaryGroup B ℂ,
    ∀ rho, Phi rho = conjSuperop (V : Matrix B B ℂ)
      (Psi (conjSuperop (U : Matrix A A ℂ) rho))

/-- PDF pp.14,28–29: corrected normalized pair for (85).
Printed (39) interchanges the off-diagonal entries of A_minus. -/
def canonicalKraus (u v : ℝ) : Bool → Matrix Bool Bool ℂ :=
  fun k b a => if k then
    if b = a then 0 else if b then (Real.sin ((v-u)/2) : ℂ)
      else (Real.sin ((u+v)/2) : ℂ)
  else if b = a then
    if b then (Real.cos ((u+v)/2) : ℂ) else (Real.cos ((v-u)/2) : ℂ)
  else 0

/-- PDF p.28, (85), on every operator. This is a qubit channel, distinct
from the accepted P69.canonical, which constructs a purification vector. -/
def rankTwoQubitMap (u v : ℝ) : Superop Bool Bool := krausApply (canonicalKraus u v)

/-- PDF p.12, (27): the specific qutrit-to-qubit Kraus pair.
The accepted Dependency50.P23.qutritKraus instead has qutrit output. -/
def qutritToQubitKraus : Bool → Matrix Bool (Fin 3) ℂ :=
  fun k b a => if k then
    if b then ![0, 0, 1] a else ![0, ((Real.sqrt 2)⁻¹ : ℂ), 0] a
  else if b then ![0, ((Real.sqrt 2)⁻¹ : ℂ), 0] a else ![1, 0, 0] a

/-- PDF p.12, (26). -/
def qutritCanonical : Superop (Fin 3) Bool := krausApply qutritToQubitKraus

/-- PDF pp.15,28: sigma_0=I and ordered X,Y,Z. -/
def paulis : Fin 4 → Matrix Bool Bool ℂ := ![1, pauliX, pauliY, pauliZ]

/-- PDF pp.15,28: a_k are Kraus amplitudes. -/
def pauliKraus (a : Fin 4 → ℝ) : Fin 4 → Matrix Bool Bool ℂ :=
  fun k => (a k : ℂ) • paulis k

/-- PDF p.28: qubit Kraus amplitudes in I,X,Y,Z order. The accepted
Frontier18.P17.pauliChannel uses probability weights and Fin d Weyl labels. -/
def qubitPauliMap (a : Fin 4 → ℝ) : Superop Bool Bool := krausApply (pauliKraus a)

/-- PDF p.28, (87): multipliers in X,Y,Z order. -/
def pauliMultipliers (a : Fin 4 → ℝ) : Fin 3 → ℝ :=
  ![a 0^2+a 1^2-a 2^2-a 3^2, a 0^2-a 1^2+a 2^2-a 3^2,
    a 0^2-a 1^2-a 2^2+a 3^2]

/-- PDF p.15, (41): linear extension of the unital diagonal Bloch map. -/
def multiplierChannel (lam : Fin 3 → ℝ) : Superop Bool Bool :=
  QubitChannels.blochAffineMap 0 (Matrix.diagonal lam)

/-- PDF pp.15,28,38: both CP inequalities and physical multiplier bounds. -/
def CPBounds (lam : Fin 3 → ℝ) : Prop :=
  (∀ k, |lam k| ≤ 1) ∧ ∀ i j k : Fin 3,
    i ≠ j → i ≠ k → j ≠ k →
    (lam i+lam j)^2 ≤ (1+lam k)^2 ∧ (lam i-lam j)^2 ≤ (1-lam k)^2

/-- PDF p.15, (42): cyclic indices enumerate the other two multipliers. -/
def antiExpression (lam : Fin 3 → ℝ) : ℝ :=
  ∑ k : Fin 3, (1-|lam k| + Real.sqrt
    ((1-|lam k|)^2-(|lam (k+1)|-|lam (k+2)|)^2))

/-- PDF pp.20–22, (55): probabilities, not Kraus amplitudes. -/
def WeylProbabilities {d : ℕ} [NeZero d] (a : WeylLabel d → ℝ) : Prop :=
  (∀ j, 0 ≤ a j) ∧ ∑ j, a j = 1

/-- PDF pp.20–22: positive-clock, forward-shift X^j Z^k mixture. -/
def weylChannel (d : ℕ) [NeZero d] (a : WeylLabel d → ℝ) : Superop (Fin d) (Fin d) :=
  fun rho => ∑ k, (a k : ℂ) • conjSuperop (finWeyl d k) rho

/-- PDF p.21: coefficient computed from Phi with the omitted 1/d restored. -/
def weylMultiplier (d : ℕ) [NeZero d] (a : WeylLabel d → ℝ) (m : WeylLabel d) : ℂ :=
  (d : ℂ)⁻¹ * ((finWeyl d m)ᴴ * weylChannel d a (finWeyl d m)).trace

/-- PDF pp.22–23: three-term qutrit family. -/
def threeWeyl (a0 a1 a2 : ℝ) : Superop (Fin 3) (Fin 3) :=
  fun rho => (a0 : ℂ) • rho + (a1 : ℂ) • conjSuperop (finWeyl 3 (1,0)) rho +
    (a2 : ℂ) • conjSuperop (finWeyl 3 (0,1)) rho

/-- PDF pp.10–11, (19): identity/noise with orthogonal flags. -/
def flaggedIdentity (x : ℝ) (N : Superop A A) : Superop A (Bool × A) :=
  fun rho p q => if p.1 = q.1 then
    if p.1 then ((1-x : ℝ) : ℂ)*N rho p.2 q.2 else (x : ℂ)*rho p.2 q.2
  else 0

/-- PDF p.10: normalized dilation of (19), none denotes the identity branch. -/
def flaggedDilation (x : ℝ) (U : Matrix (A × E) A ℂ) :
    Matrix ((Bool × A) × Option E) A ℂ :=
  fun be a => match be.2 with
  | none => if be.1 = (false,a) then (Real.sqrt x : ℂ) else 0
  | some e => if be.1.1 then (Real.sqrt (1-x) : ℂ)*U (be.1.2,e) a else 0

/-- PDF p.10, (20): complement order scalar first, noise environment second. -/
def flaggedComplement (x : ℝ) (U : Matrix (A × E) A ℂ) : Superop A (Option E) :=
  fun rho p q => match p,q with
  | none,none => (x : ℂ)*rho.trace
  | some e,some f => ((1-x : ℝ) : ℂ)*complementaryChannel U rho e f
  | _,_ => 0

/-- PDF p.11, (22): TP extension to every operator, discarding flag coherences.
The scalar-first order agrees with flaggedComplement. -/
def flaggedDegrader (x : ℝ) (U : Matrix (A × E) A ℂ) : Superop (Bool × A) (Option E) :=
  fun rho p q =>
    let r0 : Matrix A A ℂ := fun a b => rho (false,a) (false,b)
    let r1 : Matrix A A ℂ := fun a b => rho (true,a) (true,b)
    match p,q with
    | none,none => (((2*x-1)/x : ℝ) : ℂ)*r0.trace + r1.trace
    | some e,some f => (((1-x)/x : ℝ) : ℂ)*complementaryChannel U r0 e f
    | _,_ => 0

/-- PDF p.11, (23): an explicit orthonormal symmetric-subspace basis. -/
def symmetricIsometry : Matrix (Fin 3 × Fin 3) (Fin 6) ℂ :=
  fun p k => if h : k.val < 3 then
    if p.1.val = k.val ∧ p.2.val = k.val then 1 else 0
  else
    let ij : Fin 3 × Fin 3 := ![(0,1),(0,2),(1,2)] ⟨k.val-3, by omega⟩
    if p = ij ∨ p = (ij.2,ij.1) then ((Real.sqrt 2)⁻¹ : ℂ) else 0

/-- PDF p.11: the symmetric-subspace quantum channel from six to three
dimensions; accepted P06.symmetricChannel is a classical transition matrix. -/
def symmetricSubspaceMap : Superop (Fin 6) (Fin 3) := receiverChannel symmetricIsometry

/-- PDF p.11, (25): weighted flags on a common input. -/
def coFlags (M N : Superop A B) : Superop A (Bool × B) :=
  fun rho p q => if p.1 = q.1 then
    (1/2 : ℂ)*(if p.1 then N rho p.2 q.2 else M rho p.2 q.2) else 0

/-- PDF p.11: normalized direct-sum dilation, with distinct environments. -/
def coFlagDilation (U : Matrix (B × E) A ℂ) (V : Matrix (B × F) A ℂ) :
    Matrix ((Bool × B) × (E ⊕ F)) A ℂ :=
  fun be a => match be.2 with
  | .inl e => if be.1.1 then 0 else ((Real.sqrt 2)⁻¹ : ℂ)*U (be.1.2,e) a
  | .inr f => if be.1.1 then ((Real.sqrt 2)⁻¹ : ℂ)*V (be.1.2,f) a else 0

/-- PDF p.26, (4): coherent information at an input state. -/
def coherentAt (U : Matrix (B × E) A ℂ) (rho : Matrix A A ℂ) : ℝ :=
  vnEntropy (receiverChannel U rho) - vnEntropy (complementaryChannel U rho)

/-- PDF pp.32–33, (97): direct PSD matrix without a singular Schur inverse. -/
def feasibilityMatrix (x y : Fin 3 → ℝ) : Matrix (Fin 4) (Fin 4) ℂ :=
  !![1, (x 2 : ℂ), (x 0 : ℂ), (x 1 : ℂ);
     (x 2 : ℂ), 1, (y 1 : ℂ), (y 0 : ℂ);
     (x 0 : ℂ), (y 1 : ℂ), 1, (y 2 : ℂ);
     (x 1 : ℂ), (y 0 : ℂ), (y 2 : ℂ), 1]

/-- PDF p.31, (94): the factor 1/2 converts Pauli traces to actual output. -/
def pauliAntidegrader (x y : Fin 3 → ℝ) : Superop (Fin 4) Bool :=
  fun r => (1/2 : ℂ) • (r.trace • 1 +
    ((x 0 : ℂ)*(r 0 1+r 1 0)+Complex.I*(y 0 : ℂ)*(r 2 3-r 3 2)) • pauliX +
    ((x 1 : ℂ)*(r 0 2+r 2 0)+Complex.I*(y 1 : ℂ)*(r 3 1-r 1 3)) • pauliY +
    ((x 2 : ℂ)*(r 0 3+r 3 0)+Complex.I*(y 2 : ℂ)*(r 1 2-r 2 1)) • pauliZ)

/-- PDF pp.34–35: face probability coordinates, a_0=0. -/
def faceAmplitudes (p : Fin 3 → ℝ) : Fin 4 → ℝ :=
  ![0, Real.sqrt (p 0), Real.sqrt (p 1), Real.sqrt (p 2)]

/-- PDF p.35: parameter s=t^2, with its boundary s=1/2 retained. -/
def twoPauliAmplitudes (s : ℝ) : Fin 4 → ℝ :=
  ![Real.sqrt (1-2*s), Real.sqrt s, Real.sqrt s, 0]

/-- PDF pp.35–36: fixed Pauli dilation for a depolarizing multiplier. -/
def depolarizingAmplitudes (lam : ℝ) : Fin 4 → ℝ :=
  ![Real.sqrt ((1+3*lam)/4), Real.sqrt ((1-lam)/4),
    Real.sqrt ((1-lam)/4), Real.sqrt ((1-lam)/4)]

/-- PDF pp.10,35: extremality inside a named convex class of channels. -/
def ExtremeIn (P : Superop A B → Prop) (Phi : Superop A B) : Prop :=
  P Phi ∧ ∀ (N M : Superop A B) (p : ℝ), P N → P M → 0 < p → p < 1 →
    Phi = (fun r => (p : ℂ) • N r + ((1-p : ℝ) : ℂ) • M r) → N = Phi ∧ M = Phi

/-- PDF p.35: orbit of (2/3,2/3,1/3) under permutations and even sign changes. -/
def twoPauliOrbit : Set (Fin 3 → ℝ) :=
  {lam | ∃ perm : Equiv.Perm (Fin 3), ∃ sign : Fin 3 → ℝ,
    (∀ k, sign k = 1 ∨ sign k = -1) ∧ (∏ k, sign k) = 1 ∧
    ∀ k, lam k = sign k * (![2/3,2/3,1/3] : Fin 3 → ℝ) (perm k)}

/-! Definition sanity examples; no source theorem is proved here. -/

example (u v : ℝ) : canonicalKraus u v false false false = Real.cos ((v-u)/2) := by
  simp [canonicalKraus]

example : canonicalKraus 0 0 true = 0 := by
  ext b a
  simp [canonicalKraus]

example : paulis 0 = 1 := rfl

example (x : ℝ) (N : Superop A A) (rho : Matrix A A ℂ) (a b : A) :
    flaggedIdentity x N rho (false,a) (false,b) = (x : ℂ)*rho a b := by
  simp [flaggedIdentity]

example (x : ℝ) (U : Matrix (A × E) A ℂ) (rho : Matrix A A ℂ) :
    flaggedComplement x U rho none none = (x : ℂ)*rho.trace := rfl

example (x y : Fin 3 → ℝ) : feasibilityMatrix x y 0 0 = 1 := rfl

example (p : Fin 3 → ℝ) : faceAmplitudes p 0 = 0 := rfl

example (Phi : Superop A B) (a c : A) (b d : B) :
    choi Phi (a,b) (c,d) = Phi (Matrix.single a c 1) b d := rfl

/-- PDF p.10: the normalized antisymmetric qutrit dilation of W_3.
Its columns are the Levi-Civita tensor divided by sqrt2. -/
def wernerIsometry : Matrix (Fin 3 × Fin 3) (Fin 3) ℂ :=
  fun p k => if p.1=p.2 ∨ p.1=k ∨ p.2=k then 0
    else if (p.1.val+1)%3=p.2.val then ((Real.sqrt 2)⁻¹ : ℂ)
    else -((Real.sqrt 2)⁻¹ : ℂ)

example (p : Fin 3 × Fin 3) : wernerIsometry p p.1 = 0 := by
  simp [wernerIsometry]

end QAlgorithms.Papers.Independent.Gap20.P19
Source
The structure of degradable quantum channels, arXiv:0802.1360v2, https://arxiv.org/abs/0802.1360v2
Read-back

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

QAlgorithms.Papers.Independent.Gap20.P19.CPBounds

For any real triple λ=(λ0,λ1,λ2)\lambda=(\lambda_0,\lambda_1,\lambda_2)λ=(λ0​,λ1​,λ2​), this predicate means that ∣λk∣≤1|\lambda_k|\le 1∣λk​∣≤1 for every k∈{0,1,2}k\in\{0,1,2\}k∈{0,1,2} and that, for every ordered triple of pairwise distinct indices i,j,k∈{0,1,2}i,j,k\in\{0,1,2\}i,j,k∈{0,1,2}, both inequalities hold:

(λi+λj)2≤(1+λk)2,(λi−λj)2≤(1−λk)2.(\lambda_i+\lambda_j)^2\le(1+\lambda_k)^2, \qquad (\lambda_i-\lambda_j)^2\le(1-\lambda_k)^2.(λi​+λj​)2≤(1+λk​)2,(λi​−λj​)2≤(1−λk​)2.

QAlgorithms.Papers.Independent.Gap20.P19.ExtremeIn

For index types A,BA,BA,B, a predicate PPP on functions from complex A×AA\times AA×A matrices to complex B×BB\times BB×B matrices, and such a function Φ\PhiΦ, this defines the proposition that P(Φ)P(\Phi)P(Φ) holds and that, for every pair of such functions N,MN,MN,M and every real ppp with 0<p<10<p<10<p<1, if P(N)P(N)P(N), P(M)P(M)P(M), and

Φ(ρ)=pN(ρ)+(1−p)M(ρ)for every complex A×A matrix ρ,\Phi(\rho)=pN(\rho)+(1-p)M(\rho) \quad\text{for every complex }A\times A\text{ matrix }\rho,Φ(ρ)=pN(ρ)+(1−p)M(ρ)for every complex A×A matrix ρ,

then N=ΦN=\PhiN=Φ and M=ΦM=\PhiM=Φ as functions. No convexity assumption on PPP, or linearity or channel assumption on these functions, is required.

QAlgorithms.Papers.Independent.Gap20.P19.PseudoDiagonal

For finite types A,BA,BA,B with decidable equality and a function Φ:CA×A→CB×B\Phi:\mathbb C^{A\times A}\to\mathbb C^{B\times B}Φ:CA×A→CB×B, this property means that there exist a positive semidefinite matrix c∈CB×Bc\in\mathbb C^{B\times B}c∈CB×B, vectors wb∈CAw_b\in\mathbb C^Awb​∈CA for every b∈Bb\in Bb∈B, and a unitary matrix V∈CB×BV\in\mathbb C^{B\times B}V∈CB×B such that, for every matrix ρ∈CA×A\rho\in\mathbb C^{A\times A}ρ∈CA×A, Φ(ρ)=VQ(ρ)V†\Phi(\rho)=VQ(\rho)V^\daggerΦ(ρ)=VQ(ρ)V†, where

Q(ρ)ab=cab∑i,j∈Awa(i)‾ ρij wb(j)(a,b∈B).Q(\rho)_{ab}=c_{ab}\sum_{i,j\in A}\overline{w_a(i)}\,\rho_{ij}\,w_b(j) \qquad(a,b\in B).Q(ρ)ab​=cab​i,j∈A∑​wa​(i)​ρij​wb​(j)(a,b∈B).

The vectors need not be normalized, and ρ\rhoρ ranges over all matrices, without a positivity or trace assumption. Empty index types are allowed, with empty sums equal to zero.

QAlgorithms.Papers.Independent.Gap20.P19.RectangularUnitaryConjugate

For finite types A,BA,BA,B with decidable equality and functions Φ,Ψ:CA×A→CB×B\Phi,\Psi:\mathbb C^{A\times A}\to\mathbb C^{B\times B}Φ,Ψ:CA×A→CB×B, this property means that there exist unitary matrices U∈CA×AU\in\mathbb C^{A\times A}U∈CA×A and V∈CB×BV\in\mathbb C^{B\times B}V∈CB×B such that

Φ(ρ)=V Ψ(UρU†) V†\Phi(\rho)=V\,\Psi(U\rho U^\dagger)\,V^\daggerΦ(ρ)=VΨ(UρU†)V†

for every matrix ρ∈CA×A\rho\in\mathbb C^{A\times A}ρ∈CA×A. No channel, linearity, positivity, or normalization hypothesis is imposed on the given functions. Empty index types are included.

QAlgorithms.Papers.Independent.Gap20.P19.TwistedDiagonal

For a finite type AAA with decidable equality and a function Φ:CA×A→CA×A\Phi:\mathbb C^{A\times A}\to\mathbb C^{A\times A}Φ:CA×A→CA×A, this property means that there exist a natural number mmm, matrices H0,…,Hm−1∈CA×AH_0,\ldots,H_{m-1}\in\mathbb C^{A\times A}H0​,…,Hm−1​∈CA×A, and invertible matrices X,Y∈CA×AX,Y\in\mathbb C^{A\times A}X,Y∈CA×A such that

∑k=0m−1Hk†Hk=I,Φ(ρ)=∑k=0m−1HkρHk†for every ρ,\sum_{k=0}^{m-1}H_k^\dagger H_k=I,\qquad \Phi(\rho)=\sum_{k=0}^{m-1}H_k\rho H_k^\dagger \quad\text{for every }\rho,k=0∑m−1​Hk†​Hk​=I,Φ(ρ)=k=0∑m−1​Hk​ρHk†​for every ρ,

and, for each kkk, there exists a function vk:A→Cv_k:A\to\mathbb Cvk​:A→C with YHkX=diag⁡(vk)YH_kX=\operatorname{diag}(v_k)YHk​X=diag(vk​). Invertibility of X,YX,YX,Y is required, without requiring unitarity. The case m=0m=0m=0 is included: its sums are zero and its diagonalization condition is vacuous, while completeness requires 0=I0=I0=I, which is impossible when AAA is nonempty.

QAlgorithms.Papers.Independent.Gap20.P19.WeylProbabilities

For any natural number d≠0d\ne0d=0 and any real-valued function aaa on (Z/dZ)2(\mathbb Z/d\mathbb Z)^2(Z/dZ)2, this predicate means precisely that

aj≥0for every j∈(Z/dZ)2,∑j∈(Z/dZ)2aj=1.a_j\ge0\quad\text{for every }j\in(\mathbb Z/d\mathbb Z)^2, \qquad \sum_{j\in(\mathbb Z/d\mathbb Z)^2}a_j=1.aj​≥0for every j∈(Z/dZ)2,j∈(Z/dZ)2∑​aj​=1.

The case d=1d=1d=1 is included.

QAlgorithms.Papers.Independent.Gap20.P19.antiExpression

For any real triple λ=(λ0,λ1,λ2)\lambda=(\lambda_0,\lambda_1,\lambda_2)λ=(λ0​,λ1​,λ2​), this defines the real number

∑k=02[1−∣λk∣+(1−∣λk∣)2−(∣λk+1∣−∣λk+2∣)2],\sum_{k=0}^{2}\left[ 1-|\lambda_k|+ \sqrt{(1-|\lambda_k|)^2- \bigl(|\lambda_{k+1}|-|\lambda_{k+2}|\bigr)^2} \right],k=0∑2​[1−∣λk​∣+(1−∣λk​∣)2−(∣λk+1​∣−∣λk+2​∣)2​],

where indices are added modulo 333. No bounds on λ\lambdaλ or nonnegativity of the radicands are assumed; the real square root is defined to be 000 on negative arguments.

QAlgorithms.Papers.Independent.Gap20.P19.canonicalKraus

For arbitrary real numbers u,vu,vu,v, this defines two complex matrices indexed by Boolean row and column labels. In the order (false,true)(\mathrm{false},\mathrm{true})(false,true), they are

Kfalse=(cos⁡v−u200cos⁡u+v2),Ktrue=(0sin⁡u+v2sin⁡v−u20).K_{\mathrm{false}}= \begin{pmatrix} \cos\frac{v-u}{2}&0\\ 0&\cos\frac{u+v}{2} \end{pmatrix}, \qquad K_{\mathrm{true}}= \begin{pmatrix} 0&\sin\frac{u+v}{2}\\ \sin\frac{v-u}{2}&0 \end{pmatrix}.Kfalse​=(cos2v−u​0​0cos2u+v​​),Ktrue​=(0sin2v−u​​sin2u+v​0​).

All displayed real entries are regarded as complex numbers; there are no restrictions on the angles.

QAlgorithms.Papers.Independent.Gap20.P19.coFlagDilation

For index types A,B,E,FA,B,E,FA,B,E,F and arbitrary complex matrices UUU with rows indexed by B×EB\times EB×E and columns by AAA, and VVV with rows indexed by B×FB\times FB×F and columns by AAA, this defines a matrix with rows indexed by ({false,true}×B)×(E⊔F)(\{\mathrm{false},\mathrm{true}\}\times B)\times(E\sqcup F)({false,true}×B)×(E⊔F) and columns by AAA. Writing ιE,ιF\iota_E,\iota_FιE​,ιF​ for the two injections into the disjoint union, its entries are

W((b0,b),ιE(e)),a={U(b,e),a/2,b0=false,0,b0=true,W((b0,b),ιF(f)),a={0,b0=false,V(b,f),a/2,b0=true.W_{((b_0,b),\iota_E(e)),a}= \begin{cases} U_{(b,e),a}/\sqrt2,&b_0=\mathrm{false},\\ 0,&b_0=\mathrm{true}, \end{cases} \qquad W_{((b_0,b),\iota_F(f)),a}= \begin{cases} 0,&b_0=\mathrm{false},\\ V_{(b,f),a}/\sqrt2,&b_0=\mathrm{true}. \end{cases}W((b0​,b),ιE​(e)),a​={U(b,e),a​/2​,0,​b0​=false,b0​=true,​W((b0​,b),ιF​(f)),a​={0,V(b,f),a​/2​,​b0​=false,b0​=true.​

No isometry assumption is imposed on either input matrix.

QAlgorithms.Papers.Independent.Gap20.P19.coFlags

For index types A,BA,BA,B and arbitrary functions M,NM,NM,N from complex A×AA\times AA×A matrices to complex B×BB\times BB×B matrices, this defines a function from complex A×AA\times AA×A matrices to matrices indexed by {false,true}×B\{\mathrm{false},\mathrm{true}\}\times B{false,true}×B. For every input matrix ρ\rhoρ, its output is block diagonal in the Boolean index:

coFlags⁡(M,N)(ρ)=12(M(ρ)00N(ρ)),\operatorname{coFlags}(M,N)(\rho)=\frac12 \begin{pmatrix} M(\rho)&0\\ 0&N(\rho) \end{pmatrix},coFlags(M,N)(ρ)=21​(M(ρ)0​0N(ρ)​),

where the blocks are ordered false,true\mathrm{false},\mathrm{true}false,true. No linearity, positivity, or trace preservation is required of M,NM,NM,N.

QAlgorithms.Papers.Independent.Gap20.P19.coherentAt

For finite index types A,B,EA,B,EA,B,E with decidable equality, an arbitrary complex matrix U:(B×E)×AU:(B\times E)\times AU:(B×E)×A, and an arbitrary complex A×AA\times AA×A matrix ρ\rhoρ, this defines

coherentAt⁡(U,ρ)=S(ρB)−S(ρE),(ρB)b,b′=∑e(UρU†)(b,e),(b′,e),(ρE)e,e′=∑b(UρU†)(b,e),(b,e′),\operatorname{coherentAt}(U,\rho)=S(\rho_B)-S(\rho_E), \qquad (\rho_B)_{b,b'}=\sum_e(U\rho U^\dagger)_{(b,e),(b',e)}, \qquad (\rho_E)_{e,e'}=\sum_b(U\rho U^\dagger)_{(b,e),(b,e')},coherentAt(U,ρ)=S(ρB​)−S(ρE​),(ρB​)b,b′​=e∑​(UρU†)(b,e),(b′,e)​,(ρE​)e,e′​=b∑​(UρU†)(b,e),(b,e′)​,

where S(H)=−Re⁡tr⁡(f(H))S(H)=-\operatorname{Re}\operatorname{tr}(f(H))S(H)=−Retr(f(H)), with f(t)=tlog⁡2tf(t)=t\log_2tf(t)=tlog2​t applied through the continuous functional calculus. The definition requires neither that UUU be an isometry nor that ρ\rhoρ be positive semidefinite or have trace one.

QAlgorithms.Papers.Independent.Gap20.P19.depolarizingAmplitudes

For every real number λ\lambdaλ, this defines the real four-component vector

depolarizingAmplitudes⁡(λ)=(1+3λ4,1−λ4,1−λ4,1−λ4).\operatorname{depolarizingAmplitudes}(\lambda)= \left( \sqrt{\frac{1+3\lambda}{4}}, \sqrt{\frac{1-\lambda}{4}}, \sqrt{\frac{1-\lambda}{4}}, \sqrt{\frac{1-\lambda}{4}} \right).depolarizingAmplitudes(λ)=(41+3λ​​,41−λ​​,41−λ​​,41−λ​​).

No interval restriction on λ\lambdaλ is imposed; the real square-root function returns zero on negative arguments.

QAlgorithms.Papers.Independent.Gap20.P19.faceAmplitudes

For every real triple p=(p0,p1,p2)p=(p_0,p_1,p_2)p=(p0​,p1​,p2​), this defines the real four-component vector

faceAmplitudes⁡(p)=(0,p0,p1,p2).\operatorname{faceAmplitudes}(p)=(0,\sqrt{p_0},\sqrt{p_1},\sqrt{p_2}).faceAmplitudes(p)=(0,p0​​,p1​​,p2​​).

There is no nonnegativity or sum condition on ppp; each square root is the total real square-root function, which returns zero for a negative argument.

QAlgorithms.Papers.Independent.Gap20.P19.feasibilityMatrix

For arbitrary real triples x=(x0,x1,x2)x=(x_0,x_1,x_2)x=(x0​,x1​,x2​) and y=(y0,y1,y2)y=(y_0,y_1,y_2)y=(y0​,y1​,y2​), this defines the following complex matrix, with rows and columns ordered 0,1,2,30,1,2,30,1,2,3:

(1x2x0x1x21y1y0x0y11y2x1y0y21).\begin{pmatrix} 1&x_2&x_0&x_1\\ x_2&1&y_1&y_0\\ x_0&y_1&1&y_2\\ x_1&y_0&y_2&1 \end{pmatrix}.​1x2​x0​x1​​x2​1y1​y0​​x0​y1​1y2​​x1​y0​y2​1​​.

All displayed real entries are embedded into C\mathbb CC; there are no restrictions on x,yx,yx,y, and no positivity condition is part of the definition.

QAlgorithms.Papers.Independent.Gap20.P19.flaggedComplement

For any finite type AAA, any type EEE, any real number xxx, and any complex matrix UUU with rows indexed by A×EA\times EA×E and columns by AAA, this defines a function on arbitrary complex matrices ρ\rhoρ indexed by AAA, whose output is indexed by {none}⊔E\{\mathrm{none}\}\sqcup E{none}⊔E and has entries

Fx(ρ)none,none=x tr⁡ρ,Fx(ρ)some(e),some(f)=(1−x)∑c∈A∑a,b∈AU(c,e),aρa,bU(c,f),b‾,F_x(\rho)_{\mathrm{none},\mathrm{none}}=x\,\operatorname{tr}\rho, \qquad F_x(\rho)_{\mathrm{some}(e),\mathrm{some}(f)} =(1-x)\sum_{c\in A}\sum_{a,b\in A} U_{(c,e),a}\rho_{a,b}\overline{U_{(c,f),b}},Fx​(ρ)none,none​=xtrρ,Fx​(ρ)some(e),some(f)​=(1−x)c∈A∑​a,b∈A∑​U(c,e),a​ρa,b​U(c,f),b​​,

with both kinds of mixed none\mathrm{none}none–some\mathrm{some}some entries equal to 000. No isometry, channel, density-matrix, or interval assumptions are imposed; empty index types are included, with empty sums equal to 000.

QAlgorithms.Papers.Independent.Gap20.P19.flaggedDegrader

For any finite type AAA, any type EEE, any real number xxx, and any complex matrix UUU with rows indexed by A×EA\times EA×E and columns by AAA, this defines a function on arbitrary complex matrices ρ\rhoρ indexed by {false,true}×A\{\mathrm{false},\mathrm{true}\}\times A{false,true}×A. Write (r0)a,b=ρ(false,a),(false,b)(r_0)_{a,b}=\rho_{(\mathrm{false},a),(\mathrm{false},b)}(r0​)a,b​=ρ(false,a),(false,b)​ and (r1)a,b=ρ(true,a),(true,b)(r_1)_{a,b}=\rho_{(\mathrm{true},a),(\mathrm{true},b)}(r1​)a,b​=ρ(true,a),(true,b)​. Its output, indexed by {none}⊔E\{\mathrm{none}\}\sqcup E{none}⊔E, is

Dx(ρ)none,none=2x−1xtr⁡r0+tr⁡r1,Dx(ρ)some(e),some(f)=1−xx∑c∈A∑a,b∈AU(c,e),a(r0)a,bU(c,f),b‾,D_x(\rho)_{\mathrm{none},\mathrm{none}} =\frac{2x-1}{x}\operatorname{tr}r_0+\operatorname{tr}r_1, \qquad D_x(\rho)_{\mathrm{some}(e),\mathrm{some}(f)} =\frac{1-x}{x}\sum_{c\in A}\sum_{a,b\in A} U_{(c,e),a}(r_0)_{a,b}\overline{U_{(c,f),b}},Dx​(ρ)none,none​=x2x−1​trr0​+trr1​,Dx​(ρ)some(e),some(f)​=x1−x​c∈A∑​a,b∈A∑​U(c,e),a​(r0​)a,b​U(c,f),b​​,

and its mixed none\mathrm{none}none–some\mathrm{some}some entries are 000. The off-diagonal flag blocks of ρ\rhoρ do not enter these formulas. No restrictions on xxx or UUU are assumed; division by 000 is defined as 000, so at x=0x=0x=0 only the entry tr⁡r1\operatorname{tr}r_1trr1​ at (none,none)(\mathrm{none},\mathrm{none})(none,none) can remain nonzero. Empty index types are allowed.

QAlgorithms.Papers.Independent.Gap20.P19.flaggedDilation

For any type AAA with decidable equality, any type EEE, any real number xxx, and any complex matrix UUU with rows indexed by A×EA\times EA×E and columns by AAA, this defines a matrix VVV with rows indexed by ({false,true}×A)×({none}⊔E)(\{\mathrm{false},\mathrm{true}\}\times A)\times(\{\mathrm{none}\}\sqcup E)({false,true}×A)×({none}⊔E) and columns by AAA, by

V((b,c),none),a={x,b=false and c=a,0,otherwise,V((b,c),some(e)),a={1−x U(c,e),a,b=true,0,b=false.V_{((b,c),\mathrm{none}),a}= \begin{cases}\sqrt{x},&b=\mathrm{false}\text{ and }c=a,\\0,&\text{otherwise},\end{cases} \qquad V_{((b,c),\mathrm{some}(e)),a}= \begin{cases}\sqrt{1-x}\,U_{(c,e),a},&b=\mathrm{true},\\0,&b=\mathrm{false}.\end{cases}V((b,c),none),a​={x​,0,​b=false and c=a,otherwise,​V((b,c),some(e)),a​={1−x​U(c,e),a​,0,​b=true,b=false.​

No isometry assumption on UUU or restriction on xxx is imposed; real square roots of negative numbers are 000, and either index type may be empty.

QAlgorithms.Papers.Independent.Gap20.P19.flaggedIdentity

For any type AAA, any real number xxx, and any function NNN from complex matrices indexed by AAA to complex matrices indexed by AAA, this defines a function sending such a matrix ρ\rhoρ to a matrix indexed by {false,true}×A\{\mathrm{false},\mathrm{true}\}\times A{false,true}×A, with entries

Fx(ρ)(b,a),(c,a′)={x ρa,a′,b=c=false,(1−x) N(ρ)a,a′,b=c=true,0,b≠c.F_x(\rho)_{(b,a),(c,a')}= \begin{cases} x\,\rho_{a,a'},&b=c=\mathrm{false},\\ (1-x)\,N(\rho)_{a,a'},&b=c=\mathrm{true},\\ 0,&b\ne c. \end{cases}Fx​(ρ)(b,a),(c,a′)​=⎩⎨⎧​xρa,a′​,(1−x)N(ρ)a,a′​,0,​b=c=false,b=c=true,b=c.​

There are no restrictions on xxx, no channel or linearity assumption on NNN, and no density-matrix assumption on ρ\rhoρ; AAA may be empty.

QAlgorithms.Papers.Independent.Gap20.P19.multiplierChannel

For an arbitrary real triple λ=(λ0,λ1,λ2)\lambda=(\lambda_0,\lambda_1,\lambda_2)λ=(λ0​,λ1​,λ2​), this defines a function on complex 2×22\times22×2 matrices by

ρ⟼Tr⁡ρ2I2+12∑j=02λjTr⁡(Pjρ) Pj,\rho\longmapsto \frac{\operatorname{Tr}\rho}{2}I_2+ \frac12\sum_{j=0}^{2}\lambda_j\operatorname{Tr}(P_j\rho)\,P_j,ρ⟼2Trρ​I2​+21​j=0∑2​λj​Tr(Pj​ρ)Pj​,

where, in Boolean basis order (false,true)(\mathrm{false},\mathrm{true})(false,true),

P0=(0110),P1=(0−ii0),P2=(100−1).P_0=\begin{pmatrix}0&1\\1&0\end{pmatrix},\qquad P_1=\begin{pmatrix}0&-i\\i&0\end{pmatrix},\qquad P_2=\begin{pmatrix}1&0\\0&-1\end{pmatrix}.P0​=(01​10​),P1​=(0i​−i0​),P2​=(10​0−1​).

The input ρ\rhoρ is arbitrary, and no restrictions on λ\lambdaλ ensuring positivity or complete positivity are part of this definition.

QAlgorithms.Papers.Independent.Gap20.P19.pauliAntidegrader

For arbitrary real triples x,yx,yx,y, this defines a function on every complex 4×44\times44×4 matrix rrr by

r⟼12(tr⁡(r)I2+c0X+c1Y+c2Z),r\longmapsto\frac12\left(\operatorname{tr}(r)I_2+c_0X+c_1Y+c_2Z\right),r⟼21​(tr(r)I2​+c0​X+c1​Y+c2​Z),

where

c0=x0(r01+r10)+iy0(r23−r32),c1=x1(r02+r20)+iy1(r31−r13),c2=x2(r03+r30)+iy2(r12−r21),X=(0110),Y=(0−ii0),Z=(100−1).\begin{aligned} c_0&=x_0(r_{01}+r_{10})+\mathrm{i}y_0(r_{23}-r_{32}),\\ c_1&=x_1(r_{02}+r_{20})+\mathrm{i}y_1(r_{31}-r_{13}),\\ c_2&=x_2(r_{03}+r_{30})+\mathrm{i}y_2(r_{12}-r_{21}), \end{aligned} \qquad X=\begin{pmatrix}0&1\\1&0\end{pmatrix},\quad Y=\begin{pmatrix}0&-\mathrm{i}\\\mathrm{i}&0\end{pmatrix},\quad Z=\begin{pmatrix}1&0\\0&-1\end{pmatrix}.c0​c1​c2​​=x0​(r01​+r10​)+iy0​(r23​−r32​),=x1​(r02​+r20​)+iy1​(r31​−r13​),=x2​(r03​+r30​)+iy2​(r12​−r21​),​X=(01​10​),Y=(0i​−i0​),Z=(10​0−1​).

The output indices are Boolean, ordered false,true\mathrm{false},\mathrm{true}false,true. No positivity or normalization restrictions on r,x,yr,x,yr,x,y, or channel property of the resulting function, are included.

QAlgorithms.Papers.Independent.Gap20.P19.pauliKraus

For any function a:{0,1,2,3}→Ra:\{0,1,2,3\}\to\mathbb Ra:{0,1,2,3}→R, this defines four complex matrices Kk=akPkK_k=a_kP_kKk​=ak​Pk​, where, in Boolean basis order (false,true)(\mathrm{false},\mathrm{true})(false,true),

P0=I2,P1=(0110),P2=(0−ii0),P3=(100−1).P_0=I_2,\qquad P_1=\begin{pmatrix}0&1\\1&0\end{pmatrix},\qquad P_2=\begin{pmatrix}0&-i\\i&0\end{pmatrix},\qquad P_3=\begin{pmatrix}1&0\\0&-1\end{pmatrix}.P0​=I2​,P1​=(01​10​),P2​=(0i​−i0​),P3​=(10​0−1​).

The coefficients are arbitrary real numbers, including negative and zero values; no normalization is required.

QAlgorithms.Papers.Independent.Gap20.P19.pauliMultipliers

For any function a:{0,1,2,3}→Ra:\{0,1,2,3\}\to\mathbb Ra:{0,1,2,3}→R, this defines the real triple

(a02+a12−a22−a32,  a02−a12+a22−a32,  a02−a12−a22+a32).\left( a_0^2+a_1^2-a_2^2-a_3^2,\; a_0^2-a_1^2+a_2^2-a_3^2,\; a_0^2-a_1^2-a_2^2+a_3^2 \right).(a02​+a12​−a22​−a32​,a02​−a12​+a22​−a32​,a02​−a12​−a22​+a32​).

The coefficients have no sign or normalization restrictions.

QAlgorithms.Papers.Independent.Gap20.P19.paulis

This defines a family of four complex matrices indexed by k∈{0,1,2,3}k\in\{0,1,2,3\}k∈{0,1,2,3}, in Boolean basis order (false,true)(\mathrm{false},\mathrm{true})(false,true):

P0=(1001),P1=(0110),P2=(0−ii0),P3=(100−1).P_0=\begin{pmatrix}1&0\\0&1\end{pmatrix},\quad P_1=\begin{pmatrix}0&1\\1&0\end{pmatrix},\quad P_2=\begin{pmatrix}0&-i\\i&0\end{pmatrix},\quad P_3=\begin{pmatrix}1&0\\0&-1\end{pmatrix}.P0​=(10​01​),P1​=(01​10​),P2​=(0i​−i0​),P3​=(10​0−1​).

QAlgorithms.Papers.Independent.Gap20.P19.qubitPauliMap

For any function a:{0,1,2,3}→Ra:\{0,1,2,3\}\to\mathbb Ra:{0,1,2,3}→R, this defines a function on all complex 2×22\times22×2 matrices by

ρ⟼∑k=03(akPk)ρ(akPk)†=∑k=03ak2PkρPk†,\rho\longmapsto\sum_{k=0}^{3}(a_kP_k)\rho(a_kP_k)^\dagger =\sum_{k=0}^{3}a_k^2P_k\rho P_k^\dagger,ρ⟼k=0∑3​(ak​Pk​)ρ(ak​Pk​)†=k=0∑3​ak2​Pk​ρPk†​,

where, in Boolean basis order (false,true)(\mathrm{false},\mathrm{true})(false,true), P0=I2P_0=I_2P0​=I2​, P1=(0110)P_1=\begin{pmatrix}0&1\\1&0\end{pmatrix}P1​=(01​10​), P2=(0−ii0)P_2=\begin{pmatrix}0&-i\\i&0\end{pmatrix}P2​=(0i​−i0​), and P3=(100−1)P_3=\begin{pmatrix}1&0\\0&-1\end{pmatrix}P3​=(10​0−1​). No condition ∑kak2=1\sum_k a_k^2=1∑k​ak2​=1 is imposed, and the definition includes a=0a=0a=0, which gives the zero map.

QAlgorithms.Papers.Independent.Gap20.P19.qutritCanonical

This defines the function from complex 3×33\times33×3 matrices to complex 2×22\times22×2 matrices given, for every input matrix ρ\rhoρ, by

ρ⟼K0ρK0†+K1ρK1†,K0=(10001/20),K1=(01/20001).\rho\longmapsto K_0\rho K_0^\dagger+K_1\rho K_1^\dagger, \qquad K_0=\begin{pmatrix}1&0&0\\0&1/\sqrt2&0\end{pmatrix}, \quad K_1=\begin{pmatrix}0&1/\sqrt2&0\\0&0&1\end{pmatrix}.ρ⟼K0​ρK0†​+K1​ρK1†​,K0​=(10​01/2​​00​),K1​=(00​1/2​0​01​).

The output basis order is (false,true)(\mathrm{false},\mathrm{true})(false,true), and the input basis order is (0,1,2)(0,1,2)(0,1,2); no positivity or trace restriction is placed on ρ\rhoρ.

QAlgorithms.Papers.Independent.Gap20.P19.qutritToQubitKraus

This defines two complex 2×32\times32×3 matrices, with Boolean rows ordered (false,true)(\mathrm{false},\mathrm{true})(false,true) and columns ordered (0,1,2)(0,1,2)(0,1,2):

Kfalse=(10001/20),Ktrue=(01/20001).K_{\mathrm{false}}= \begin{pmatrix} 1&0&0\\ 0&1/\sqrt2&0 \end{pmatrix}, \qquad K_{\mathrm{true}}= \begin{pmatrix} 0&1/\sqrt2&0\\ 0&0&1 \end{pmatrix}.Kfalse​=(10​01/2​​00​),Ktrue​=(00​1/2​0​01​).

QAlgorithms.Papers.Independent.Gap20.P19.rankTwoQubitMap

For arbitrary real numbers u,vu,vu,v, this defines a function on complex 2×22\times22×2 matrices by

ρ⟼K0ρK0†+K1ρK1†,K0=(cos⁡v−u200cos⁡u+v2),K1=(0sin⁡u+v2sin⁡v−u20),\rho\longmapsto K_0\rho K_0^\dagger+K_1\rho K_1^\dagger,\qquad K_0=\begin{pmatrix}\cos\frac{v-u}{2}&0\\0&\cos\frac{u+v}{2}\end{pmatrix}, \quad K_1=\begin{pmatrix}0&\sin\frac{u+v}{2}\\\sin\frac{v-u}{2}&0\end{pmatrix},ρ⟼K0​ρK0†​+K1​ρK1†​,K0​=(cos2v−u​0​0cos2u+v​​),K1​=(0sin2v−u​​sin2u+v​0​),

using Boolean basis order (false,true)(\mathrm{false},\mathrm{true})(false,true). The input ρ\rhoρ is arbitrary, and the definition imposes no condition that the map or its Kraus representation have rank exactly two; either displayed matrix may vanish for some angles.

QAlgorithms.Papers.Independent.Gap20.P19.symmetricIsometry

This defines a complex matrix VVV with rows indexed by {0,1,2}2\{0,1,2\}^2{0,1,2}2 and columns indexed by {0,…,5}\{0,\ldots,5\}{0,…,5} through its six columns:

Ve0=e(0,0),Ve1=e(1,1),Ve2=e(2,2),Ve3=e(0,1)+e(1,0)2,Ve4=e(0,2)+e(2,0)2,Ve5=e(1,2)+e(2,1)2.Ve_0=e_{(0,0)},\quad Ve_1=e_{(1,1)},\quad Ve_2=e_{(2,2)},\quad Ve_3=\frac{e_{(0,1)}+e_{(1,0)}}{\sqrt2},\quad Ve_4=\frac{e_{(0,2)}+e_{(2,0)}}{\sqrt2},\quad Ve_5=\frac{e_{(1,2)}+e_{(2,1)}}{\sqrt2}.Ve0​=e(0,0)​,Ve1​=e(1,1)​,Ve2​=e(2,2)​,Ve3​=2​e(0,1)​+e(1,0)​​,Ve4​=2​e(0,2)​+e(2,0)​​,Ve5​=2​e(1,2)​+e(2,1)​​.

Here the eee's are standard coordinate vectors; all entries not specified by these columns are 000.

QAlgorithms.Papers.Independent.Gap20.P19.symmetricSubspaceMap

This defines a function from arbitrary complex 6×66\times66×6 matrices to complex 3×33\times33×3 matrices by

Φ(ρ)i,j=∑e=02(VρV†)(i,e),(j,e),i,j∈{0,1,2},\Phi(\rho)_{i,j}=\sum_{e=0}^{2}(V\rho V^\dagger)_{(i,e),(j,e)}, \qquad i,j\in\{0,1,2\},Φ(ρ)i,j​=e=0∑2​(VρV†)(i,e),(j,e)​,i,j∈{0,1,2},

where the columns of VVV are, in order,

e(0,0),e(1,1),e(2,2),e(0,1)+e(1,0)2,e(0,2)+e(2,0)2,e(1,2)+e(2,1)2.e_{(0,0)},\quad e_{(1,1)},\quad e_{(2,2)},\quad \frac{e_{(0,1)}+e_{(1,0)}}{\sqrt2},\quad \frac{e_{(0,2)}+e_{(2,0)}}{\sqrt2},\quad \frac{e_{(1,2)}+e_{(2,1)}}{\sqrt2}.e(0,0)​,e(1,1)​,e(2,2)​,2​e(0,1)​+e(1,0)​​,2​e(0,2)​+e(2,0)​​,2​e(1,2)​+e(2,1)​​.

The input ρ\rhoρ is not required to be positive or have trace 111.

QAlgorithms.Papers.Independent.Gap20.P19.threeWeyl

For arbitrary real numbers a0,a1,a2a_0,a_1,a_2a0​,a1​,a2​, this defines a function on complex 3×33\times33×3 matrices by

ρ⟼a0ρ+a1SρS†+a2ZρZ†,\rho\longmapsto a_0\rho+a_1S\rho S^\dagger+a_2Z\rho Z^\dagger,ρ⟼a0​ρ+a1​SρS†+a2​ZρZ†,

where, for j∈{0,1,2}j\in\{0,1,2\}j∈{0,1,2}, Sej=ej+1 mod 3Se_j=e_{j+1\bmod3}Sej​=ej+1mod3​ and Zej=e2πij/3ejZe_j=e^{2\pi i j/3}e_jZej​=e2πij/3ej​. No nonnegativity or normalization conditions on the coefficients, or density-matrix condition on ρ\rhoρ, are assumed.

QAlgorithms.Papers.Independent.Gap20.P19.twoPauliAmplitudes

For every real number sss, this defines the real four-component vector

twoPauliAmplitudes⁡(s)=(1−2s,s,s,0).\operatorname{twoPauliAmplitudes}(s)=(\sqrt{1-2s},\sqrt{s},\sqrt{s},0).twoPauliAmplitudes(s)=(1−2s​,s​,s​,0).

The parameter sss is unrestricted, and every negative square-root argument is assigned square root zero.

QAlgorithms.Papers.Independent.Gap20.P19.twoPauliOrbit

This defines the set of real triples λ\lambdaλ for which there exist a permutation π\piπ of {0,1,2}\{0,1,2\}{0,1,2} and signs ε0,ε1,ε2∈{1,−1}\varepsilon_0,\varepsilon_1,\varepsilon_2\in\{1,-1\}ε0​,ε1​,ε2​∈{1,−1} satisfying ε0ε1ε2=1\varepsilon_0\varepsilon_1\varepsilon_2=1ε0​ε1​ε2​=1, such that

λk=εkvπ(k)(k=0,1,2),v=(23,23,13).\lambda_k=\varepsilon_k v_{\pi(k)} \quad(k=0,1,2), \qquad v=\left(\frac23,\frac23,\frac13\right).λk​=εk​vπ(k)​(k=0,1,2),v=(32​,32​,31​).

Thus the signs have an even number of negative entries, and the permutation acts on the entries of the displayed fixed triple.

QAlgorithms.Papers.Independent.Gap20.P19.wernerIsometry

This defines the complex 9×39\times39×3 matrix VVV, with rows indexed by (i,j)∈{0,1,2}2(i,j)\in\{0,1,2\}^2(i,j)∈{0,1,2}2 and columns indexed by k∈{0,1,2}k\in\{0,1,2\}k∈{0,1,2}, by

V(i,j),k={0,i=j or i=k or j=k,1/2,otherwise and (i+1) mod 3=j,−1/2,otherwise.V_{(i,j),k}= \begin{cases} 0,&i=j\ \text{or}\ i=k\ \text{or}\ j=k,\\ 1/\sqrt2,&\text{otherwise and }(i+1)\bmod3=j,\\ -1/\sqrt2,&\text{otherwise}. \end{cases}V(i,j),k​=⎩⎨⎧​0,1/2​,−1/2​,​i=j or i=k or j=k,otherwise and (i+1)mod3=j,otherwise.​

Thus the nonzero case requires i,j,ki,j,ki,j,k to be pairwise distinct; the positive sign is selected by the indicated cyclic order of i,ji,ji,j.

QAlgorithms.Papers.Independent.Gap20.P19.weylChannel

For any natural number d≠0d\ne0d=0 and any real weights a:(Z/dZ)2→Ra:(\mathbb Z/d\mathbb Z)^2\to\mathbb Ra:(Z/dZ)2→R, this defines a function on arbitrary complex d×dd\times dd×d matrices by

Φa(ρ)=∑(n,m)∈(Z/dZ)2an,m Wn,mρWn,m†.\Phi_a(\rho)=\sum_{(n,m)\in(\mathbb Z/d\mathbb Z)^2} a_{n,m}\,W_{n,m}\rho W_{n,m}^{\dagger}.Φa​(ρ)=(n,m)∈(Z/dZ)2∑​an,m​Wn,m​ρWn,m†​.

Here Wn,m=SnˉZmˉW_{n,m}=S^{\bar n}Z^{\bar m}Wn,m​=SnˉZmˉ, where nˉ,mˉ∈{0,…,d−1}\bar n,\bar m\in\{0,\ldots,d-1\}nˉ,mˉ∈{0,…,d−1} are the residue representatives, Sej=ej+1 mod dSe_j=e_{j+1\bmod d}Sej​=ej+1modd​, and Zej=e2πij/dejZe_j=e^{2\pi i j/d}e_jZej​=e2πij/dej​ in the standard basis. The weights are not required to be nonnegative or sum to 111, and ρ\rhoρ is not required to be a density matrix; d=1d=1d=1 is allowed.

QAlgorithms.Papers.Independent.Gap20.P19.weylMultiplier

For any natural number d≠0d\ne0d=0, any real function a:(Z/dZ)2→Ra:(\mathbb Z/d\mathbb Z)^2\to\mathbb Ra:(Z/dZ)2→R, and any label m∈(Z/dZ)2m\in(\mathbb Z/d\mathbb Z)^2m∈(Z/dZ)2, this defines the complex number

1dtr⁡ ⁣(Wm†[∑k∈(Z/dZ)2akWkWmWk†]).\frac1d\operatorname{tr}\!\left( W_m^\dagger \left[\sum_{k\in(\mathbb Z/d\mathbb Z)^2}a_kW_kW_mW_k^\dagger\right] \right).d1​tr​Wm†​​k∈(Z/dZ)2∑​ak​Wk​Wm​Wk†​​​.

For a label (u,v)(u,v)(u,v), Wu,v=SuˉZvˉW_{u,v}=S^{\bar u}Z^{\bar v}Wu,v​=SuˉZvˉ, where Sej=ej+1 mod dSe_j=e_{j+1\bmod d}Sej​=ej+1modd​ and Zej=e2πij/dejZe_j=e^{2\pi i j/d}e_jZej​=e2πij/dej​, and bars denote representatives in {0,…,d−1}\{0,\ldots,d-1\}{0,…,d−1}. No probability conditions on aaa are imposed, the zero label mmm is included, and d=1d=1d=1 is allowed.

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