Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Definitions for Characterization of non-local gates

Definition
QAlgorithms_Papers_Independent_Intakethird100_P99

by Gohan · Nov 1, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

The objects the formalized statements of Characterization of non-local gates (arXiv:quant-ph/0205100) are stated in. Each was checked against the paper by a blind read-back and a moderator.

  • OneMatrix: PDF p.2: operators on one qubit.
  • TwoMatrix: PDF p.2: operators on the ordered tensor product A then B.
  • OneGate: PDF pp.1–2: a bundled arbitrary one-qubit unitary control.
  • Triple: PDF p.3: three real interaction coefficients, indexed from zero in Lean.
  • OneState: PDF pp.8–9: one-qubit pure-state vectors; IsState supplies normalization.
  • IsGate: PDF p.2, II.A: a two-qubit gate is unitary.
  • localMatrix: PDF pp.1–2: instantaneous independent local gates on A and B.
  • TwoQubitLocalEquivalent: PDF p.2, II.A: local equivalence; U(2) factors also absorb the global phase.
  • canonicalPauliInteraction: PDF p.3, (3): the canonical interaction Hamiltonian, with no strength bound.
  • contentGate: PDF pp.2–3: U_beta = exp(-i H_beta).
  • SOrdered: PDF p.3, II.B: s-ordered coefficients (not ordinary decreasing order).
  • GateChamber: PDF p.2, (1): the gate chamber, as distinguished from Hamiltonian strength.
  • sOrder: PDF p.3, II.B: sort absolute values and give the least one the product's sign. If the product vanishes the least absolute value is zero, so the sign choice is harmless.
  • SMajorized: PDF p.3, (5): b is s-majorized by a; both sides are s-ordered first.
  • contentIntegerShift: PDF pp.4–6, (7): periodic shift by integer multiples of pi/2.
  • secondShift: PDF p.6, Result 2: the only nonzero shift needed for the minimum.
  • HasContent: PDF pp.2–3: an actual canonical-content witness for the given unitary.
  • HasCanonicalHamiltonian: PDF p.4, III.A: the nonlocal part of H has canonical coefficients a after local conjugation. K and L include arbitrary local terms and the scalar identity term.
  • PureInteraction: PDF p.6, III.C: a pure interaction has no scalar or one-body terms.
  • PulseProtocol: PDF p.4, (6): initial local gates followed by evolution/local-control pairs. Nonnegative times are built into the type; the pulse list counts actual H applications.
  • protocolTime: PDF p.4: only interaction intervals contribute to total duration.
  • protocolMatrix: PDF p.4, (6): the earliest step acts on the right. This is an operator on every input, not a protocol chosen after seeing an input state.
  • SamePhysicalGate: PDF pp.1–2: equality of physical gates disregards one scalar phase.
  • IdealSimulates: PDF p.4: ideal fast-LU simulation, explicitly including the limit of arbitrarily fine pulses needed to eliminate local drift. Every approximant has exactly duration t. An operator-norm approximation is uniform over all input states.
  • interactionCost: PDF pp.4,6: interaction cost as an extended nonnegative infimum, including impossible simulations with value infinity. Attainment is asserted in the result, not assumed here.
  • branchCost: PDF p.6: cost of one unshifted s-majorization branch.
  • MoreNonlocal: PDF pp.7–8, III.E: order by interaction cost for every Hermitian H, including zero interaction. This is a preorder on matrices; no antisymmetry on raw gates is claimed.
  • dcnot: PDF p.7, III.D.2: DCNOT sends |i,j> to |j,i xor j>.
  • reverseParties: PDF p.7: interchange Alice and Bob; first register remains the sender in the predicates.
  • CanCbit: PDF pp.8–10, (11): one perfect classical bit, with fixed orthogonal encoding and a fixed pure receiver input chosen before the bit. The sender's residual states may differ. All states are normalized; no ancillary or prior entangled state occurs in the definition.
  • CanTwoCbits: PDF pp.10–11, (16)–(18): two simultaneous perfect classical bits. Each receiving measurement may depend on that receiver's own sent bit, through the two orthogonality conditions. Inputs are fixed independent orthogonal encodings, not shared entanglement.
  • CanQubit: PDF pp.8,11: unknown-qubit transmission preserves every normalized superposition with one fixed encoding, pure receiver input, recovery, and residual sender state.
  • CanHybrid: PDF pp.11–12: a qubit A→B and a cbit B→A in the same use. Bob's recovery depends on his sent bit j, but the same recovery works for every unknown superposition.
  • CanTwoQubits: PDF p.12, IV.B.4: simultaneous exchange of arbitrary qubits with fixed free local encoding and decoding, uniformly over both unknown states.
  • taskCost: PDF pp.9–12: minimize interaction cost over every gate capable of the task P. The empty set has infimum infinity in ENNReal.
  • OptimalTask: PDF pp.10–12: an explicit attaining gate and time, together with the lower bound against every capable gate. Simulation is ideal fast control as on p.4.
  • optimalTransmissionContent: PDF p.11, Result 5: the minimizing interaction content (not necessarily unique).
  • ControlPhase: PDF p.13, Appendix A: determinant phase of the controlled one-qubit unitary is free on the control. b is an eigenphase half-angle, before folding into the gate chamber.
  • controlledContent: PDF p.13, Appendix A: fold b at pi/4 rather than selecting a principal argument.
  • ChamberClass: PDF p.12, IV.C: four disjoint chamber strata; both signed SWAP endpoints are included.
  • TransmissionProfile: PDF p.12, IV.C: operational profiles, including both orientations. "No capability" means no perfect full bit in this single-use model, not zero Shannon capacity.
  • interactionPhases: PDF p.3, (4): interaction eigenvalues in the exact magic-basis order (2).
  • magicBasis: PDF p.3, (2): columns are -i(|01>+|10>)/sqrt2, (|00>+|11>)/sqrt2, -i(|00>-|11>)/sqrt2, (|01>-|10>)/sqrt2, respectively.
  • magicMatrix: PDF p.13, Appendix A: express a computational-basis matrix in basis (2).
  • PhaseCompatibleContent: PDF p.13: a phase-compatible SU(2) local decomposition of the actual representative. This excludes assuming that an arbitrary projective phase has the same magic spectrum.
Definition code
import Mathlib
import Definitions.Def_QAlgorithms_Defs_Gates
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P10
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P32

/-!
Statements from Hammerer, Vidal and Cirac, arXiv:quant-ph/0205100v3.
Paper-specific definitions are confined to this paper; the accepted Pauli selector and
matrix evolution are imported from the pinned library. Source indices 1,2,3 are Lean indices 0,1,2.
Registers and Pauli gates use the frozen library: Bool for one qubit, Bool × Bool for A then B.
Local U(2) factors include physically irrelevant scalar phases. Appendix A instead explicitly
requires determinant-one local factors for the chosen representative.

General-H simulation is the ideal fast-control limit of finite protocols of exactly the stated
interaction duration (p.4). The norm used to take that limit is the frozen spectral norm.
Finite exact protocols are distinguished from this limit. Costs lie in ENNReal; an empty feasible
set has cost infinity. No gate-chamber bound is imposed on Hamiltonian strength.
-/

namespace QAlgorithms.Papers.Independent.Intakethird100.P99

open scoped Matrix Kronecker NNReal ENNReal

open QAlgorithms.Papers.Independent.Information100.P10 (pauli)
open QAlgorithms.Papers.Independent.Information100.P32 (evolution)

noncomputable section

/-- PDF p.2: operators on one qubit. -/
abbrev OneMatrix := Matrix Bool Bool ℂ

/-- PDF p.2: operators on the ordered tensor product A then B. -/
abbrev TwoMatrix := Matrix (Bool × Bool) (Bool × Bool) ℂ

/-- PDF pp.1–2: a bundled arbitrary one-qubit unitary control. -/
abbrev OneGate := Matrix.unitaryGroup Bool ℂ

/-- PDF p.3: three real interaction coefficients, indexed from zero in Lean. -/
abbrev Triple := Fin 3 → ℝ

/-- PDF pp.8–9: one-qubit pure-state vectors; IsState supplies normalization. -/
abbrev OneState := EuclideanSpace ℂ Bool

/-- PDF p.2, II.A: a two-qubit gate is unitary. -/
def IsGate (U : TwoMatrix) : Prop := U ∈ Matrix.unitaryGroup (Bool × Bool) ℂ

/-- PDF pp.1–2: instantaneous independent local gates on A and B. -/
def localMatrix (A B : OneGate) : TwoMatrix :=
  (A : OneMatrix) ⊗ₖ (B : OneMatrix)

/-- PDF p.2, II.A: local equivalence; U(2) factors also absorb the global phase. -/
def TwoQubitLocalEquivalent (U V : TwoMatrix) : Prop :=
  ∃ A B C D : OneGate, U = localMatrix A B * V * localMatrix C D

/-- PDF p.3, (3): the canonical interaction Hamiltonian, with no strength bound. -/
def canonicalPauliInteraction (a : Triple) : TwoMatrix :=
  ∑ i : Fin 3, (a i : ℂ) • (pauli i ⊗ₖ pauli i)

/-- PDF pp.2–3: U_beta = exp(-i H_beta). -/
def contentGate (b : Triple) : TwoMatrix := evolution (canonicalPauliInteraction b) 1

/-- PDF p.3, II.B: s-ordered coefficients (not ordinary decreasing order). -/
def SOrdered (a : Triple) : Prop := a 1 ≤ a 0 ∧ |a 2| ≤ a 1

/-- PDF p.2, (1): the gate chamber, as distinguished from Hamiltonian strength. -/
def GateChamber (b : Triple) : Prop := SOrdered b ∧ b 0 ≤ Real.pi / 4

/-- PDF p.3, II.B: sort absolute values and give the least one the product's sign.
If the product vanishes the least absolute value is zero, so the sign choice is harmless. -/
def sOrder (a : Triple) : Triple :=
  let hi := max |a 0| (max |a 1| |a 2|)
  let lo := min |a 0| (min |a 1| |a 2|)
  ![hi, |a 0| + |a 1| + |a 2| - hi - lo,
    if a 0 * a 1 * a 2 < 0 then -lo else lo]

/-- PDF p.3, (5): b is s-majorized by a; both sides are s-ordered first. -/
def SMajorized (b a : Triple) : Prop :=
  let x := sOrder b
  let y := sOrder a
  x 0 ≤ y 0 ∧ x 0 + x 1 - x 2 ≤ y 0 + y 1 - y 2 ∧
    x 0 + x 1 + x 2 ≤ y 0 + y 1 + y 2

/-- PDF pp.4–6, (7): periodic shift by integer multiples of pi/2. -/
def contentIntegerShift (b : Triple) (n : Fin 3 → ℤ) : Triple :=
  fun i => b i + Real.pi / 2 * (n i : ℝ)

/-- PDF p.6, Result 2: the only nonzero shift needed for the minimum. -/
def secondShift (b : Triple) : Triple := contentIntegerShift b ![-1, 0, 0]

/-- PDF pp.2–3: an actual canonical-content witness for the given unitary. -/
def HasContent (U : TwoMatrix) (b : Triple) : Prop :=
  IsGate U ∧ GateChamber b ∧ TwoQubitLocalEquivalent U (contentGate b)

/-- PDF p.4, III.A: the nonlocal part of H has canonical coefficients a after local
conjugation. K and L include arbitrary local terms and the scalar identity term. -/
def HasCanonicalHamiltonian (H : TwoMatrix) (a : Triple) : Prop :=
  H.IsHermitian ∧ SOrdered a ∧
    ∃ A B : OneGate, ∃ K L : OneMatrix, K.IsHermitian ∧ L.IsHermitian ∧
      localMatrix A B * H * star (localMatrix A B) =
        canonicalPauliInteraction a + K ⊗ₖ (1 : OneMatrix) + (1 : OneMatrix) ⊗ₖ L

/-- PDF p.6, III.C: a pure interaction has no scalar or one-body terms. -/
def PureInteraction (H : TwoMatrix) : Prop :=
  ∃ c : Fin 3 → Fin 3 → ℝ,
    H = ∑ i : Fin 3, ∑ j : Fin 3, (c i j : ℂ) • (pauli i ⊗ₖ pauli j)

/-- PDF p.4, (6): initial local gates followed by evolution/local-control pairs.
Nonnegative times are built into the type; the pulse list counts actual H applications. -/
structure PulseProtocol where
  initialA : OneGate
  initialB : OneGate
  pulses : List (ℝ≥0 × (OneGate × OneGate))

/-- PDF p.4: only interaction intervals contribute to total duration. -/
def protocolTime (p : PulseProtocol) : ℝ≥0 := (p.pulses.map Prod.fst).sum

/-- PDF p.4, (6): the earliest step acts on the right. This is an operator on every input,
not a protocol chosen after seeing an input state. -/
def protocolMatrix (H : TwoMatrix) (p : PulseProtocol) : TwoMatrix :=
  p.pulses.foldl (fun U s => localMatrix s.2.1 s.2.2 * evolution H (s.1 : ℝ) * U)
    (localMatrix p.initialA p.initialB)

/-- PDF pp.1–2: equality of physical gates disregards one scalar phase. -/
def SamePhysicalGate (U V : TwoMatrix) : Prop :=
  ∃ z : ℂ, ‖z‖ = 1 ∧ U = z • V

/-- PDF p.4: ideal fast-LU simulation, explicitly including the limit of arbitrarily fine
pulses needed to eliminate local drift. Every approximant has exactly duration t.
An operator-norm approximation is uniform over all input states. -/
def IdealSimulates (H U : TwoMatrix) (t : ℝ) : Prop :=
  H.IsHermitian ∧ IsGate U ∧ 0 ≤ t ∧
    ∀ e : ℝ, 0 < e → ∃ p : PulseProtocol, (protocolTime p : ℝ) = t ∧
      ∃ z : ℂ, ‖z‖ = 1 ∧ specNorm (protocolMatrix H p - z • U) < e

/-- PDF pp.4,6: interaction cost as an extended nonnegative infimum, including impossible
simulations with value infinity. Attainment is asserted in the result, not assumed here. -/
def interactionCost (H U : TwoMatrix) : ℝ≥0∞ :=
  sInf ((fun t : ℝ≥0 => (t : ℝ≥0∞)) '' {t | IdealSimulates H U (t : ℝ)})

/-- PDF p.6: cost of one unshifted s-majorization branch. -/
def branchCost (a b : Triple) : ℝ≥0∞ :=
  sInf ((fun t : ℝ≥0 => (t : ℝ≥0∞)) '' {t | SMajorized b ((t : ℝ) • a)})

/-- PDF pp.7–8, III.E: order by interaction cost for every Hermitian H, including zero
interaction. This is a preorder on matrices; no antisymmetry on raw gates is claimed. -/
def MoreNonlocal (U V : TwoMatrix) : Prop :=
  ∀ H : TwoMatrix, H.IsHermitian → interactionCost H V ≤ interactionCost H U

/-- PDF p.7, III.D.2: DCNOT sends |i,j> to |j,i xor j>. -/
def dcnot : TwoMatrix :=
  Matrix.of fun p q => if p = (q.2, xor q.1 q.2) then 1 else 0

/-- PDF p.7: interchange Alice and Bob; first register remains the sender in the predicates. -/
def reverseParties (U : TwoMatrix) : TwoMatrix := swapGate * U * swapGate

/-- PDF pp.8–10, (11): one perfect classical bit, with fixed orthogonal encoding and a
fixed pure receiver input chosen before the bit. The sender's residual states may differ.
All states are normalized; no ancillary or prior entangled state occurs in the definition. -/
def CanCbit (U : TwoMatrix) : Prop :=
  IsGate U ∧ ∃ E D : OneGate, ∃ v : OneState, ∃ w : Bool → OneState,
    IsState v ∧ (∀ i, IsState (w i)) ∧
      ∀ i : Bool, act U (tensorVec (act (E : OneMatrix) (ket i)) v) =
        tensorVec (w i) (act (D : OneMatrix) (ket i))

/-- PDF pp.10–11, (16)–(18): two simultaneous perfect classical bits. Each receiving
measurement may depend on that receiver's own sent bit, through the two orthogonality
conditions. Inputs are fixed independent orthogonal encodings, not shared entanglement. -/
def CanTwoCbits (U : TwoMatrix) : Prop :=
  IsGate U ∧ ∃ E F : OneGate, ∃ a b : Bool → Bool → OneState,
    (∀ i j, IsState (a i j) ∧ IsState (b i j)) ∧
    (∀ i, inner ℂ (a i false) (a i true) = 0) ∧
    (∀ j, inner ℂ (b false j) (b true j) = 0) ∧
    ∀ i j : Bool, act U (tensorVec (act (E : OneMatrix) (ket i))
      (act (F : OneMatrix) (ket j))) = tensorVec (a i j) (b i j)

/-- PDF pp.8,11: unknown-qubit transmission preserves every normalized superposition
with one fixed encoding, pure receiver input, recovery, and residual sender state. -/
def CanQubit (U : TwoMatrix) : Prop :=
  IsGate U ∧ ∃ E R : OneGate, ∃ v w : OneState,
    IsState v ∧ IsState w ∧
    ∀ q : OneState, IsState q →
      act (((1 : OneMatrix) ⊗ₖ (R : OneMatrix)) * U)
        (tensorVec (act (E : OneMatrix) q) v) = tensorVec w q

/-- PDF pp.11–12: a qubit A→B and a cbit B→A in the same use. Bob's recovery depends
on his sent bit j, but the same recovery works for every unknown superposition. -/
def CanHybrid (U : TwoMatrix) : Prop :=
  IsGate U ∧ ∃ E F : OneGate, ∃ R : Bool → OneGate, ∃ a : Bool → OneState,
    (∀ j, IsState (a j)) ∧ inner ℂ (a false) (a true) = 0 ∧
    ∀ j : Bool, ∀ q : OneState, IsState q →
      act (((1 : OneMatrix) ⊗ₖ (R j : OneMatrix)) * U)
        (tensorVec (act (E : OneMatrix) q) (act (F : OneMatrix) (ket j))) =
          tensorVec (a j) q

/-- PDF p.12, IV.B.4: simultaneous exchange of arbitrary qubits with fixed free local
encoding and decoding, uniformly over both unknown states. -/
def CanTwoQubits (U : TwoMatrix) : Prop :=
  IsGate U ∧ ∃ E F R S : OneGate,
    ∀ q r : OneState, IsState q → IsState r →
      act (localMatrix R S * U)
        (tensorVec (act (E : OneMatrix) q) (act (F : OneMatrix) r)) = tensorVec r q

/-- PDF pp.9–12: minimize interaction cost over every gate capable of the task P.
The empty set has infimum infinity in ENNReal. -/
def taskCost (H : TwoMatrix) (P : TwoMatrix → Prop) : ℝ≥0∞ :=
  sInf (interactionCost H '' {U | P U})

/-- PDF pp.10–12: an explicit attaining gate and time, together with the lower bound
against every capable gate. Simulation is ideal fast control as on p.4. -/
def OptimalTask (H : TwoMatrix) (P : TwoMatrix → Prop) (U : TwoMatrix) (t : ℝ) : Prop :=
  P U ∧ IdealSimulates H U t ∧ interactionCost H U = ENNReal.ofReal t ∧
    taskCost H P = ENNReal.ofReal t ∧
    ∀ V : TwoMatrix, P V → interactionCost H U ≤ interactionCost H V

/-- PDF p.11, Result 5: the minimizing interaction content (not necessarily unique). -/
def optimalTransmissionContent (a : Triple) : Triple :=
  ![Real.pi / 4, Real.pi / 4, Real.pi / 4 * (2 * a 2 / (a 0 + a 1))]

/-- PDF p.13, Appendix A: determinant phase of the controlled one-qubit unitary is
free on the control. b is an eigenphase half-angle, before folding into the gate chamber. -/
def ControlPhase (W : OneMatrix) (b : ℝ) : Prop :=
  0 ≤ b ∧ b ≤ Real.pi / 2 ∧ ∃ z : ℂ, ‖z‖ = 1 ∧
    (z • W).charpoly =
      (Polynomial.X - Polynomial.C (Complex.exp (2 * Complex.I * (b : ℂ)))) *
      (Polynomial.X - Polynomial.C (Complex.exp (-2 * Complex.I * (b : ℂ))))

/-- PDF p.13, Appendix A: fold b at pi/4 rather than selecting a principal argument. -/
def controlledContent (b : ℝ) : Triple := ![min b (Real.pi / 2 - b), 0, 0]

/-- PDF p.12, IV.C: four disjoint chamber strata; both signed SWAP endpoints are included. -/
def ChamberClass (b : Triple) (k : Fin 4) : Prop :=
  if k = 0 then b 0 < Real.pi / 4
  else if k = 1 then b 0 = Real.pi / 4 ∧ b 1 < Real.pi / 4
  else if k = 2 then b 0 = Real.pi / 4 ∧ b 1 = Real.pi / 4 ∧ |b 2| < Real.pi / 4
  else b 0 = Real.pi / 4 ∧ b 1 = Real.pi / 4 ∧ |b 2| = Real.pi / 4

/-- PDF p.12, IV.C: operational profiles, including both orientations. "No capability"
means no perfect full bit in this single-use model, not zero Shannon capacity. -/
def TransmissionProfile (U : TwoMatrix) (k : Fin 4) : Prop :=
  if k = 0 then
    ¬ CanCbit U ∧ ¬ CanCbit (reverseParties U) ∧ ¬ CanTwoCbits U ∧
      ¬ CanQubit U ∧ ¬ CanQubit (reverseParties U) ∧
      ¬ CanHybrid U ∧ ¬ CanHybrid (reverseParties U) ∧ ¬ CanTwoQubits U
  else if k = 1 then
    CanCbit U ∧ CanCbit (reverseParties U) ∧ ¬ CanTwoCbits U ∧
      ¬ CanQubit U ∧ ¬ CanQubit (reverseParties U) ∧
      ¬ CanHybrid U ∧ ¬ CanHybrid (reverseParties U) ∧ ¬ CanTwoQubits U
  else if k = 2 then
    CanCbit U ∧ CanCbit (reverseParties U) ∧ CanTwoCbits U ∧
      CanQubit U ∧ CanQubit (reverseParties U) ∧
      CanHybrid U ∧ CanHybrid (reverseParties U) ∧ ¬ CanTwoQubits U
  else
    CanCbit U ∧ CanCbit (reverseParties U) ∧ CanTwoCbits U ∧
      CanQubit U ∧ CanQubit (reverseParties U) ∧
      CanHybrid U ∧ CanHybrid (reverseParties U) ∧ CanTwoQubits U

/-- PDF p.3, (4): interaction eigenvalues in the exact magic-basis order (2). -/
def interactionPhases (b : Triple) : Fin 4 → ℝ :=
  ![b 0 + b 1 - b 2, b 0 - b 1 + b 2, -b 0 + b 1 + b 2, -b 0 - b 1 - b 2]

/-- PDF p.3, (2): columns are -i(|01>+|10>)/sqrt2, (|00>+|11>)/sqrt2,
-i(|00>-|11>)/sqrt2, (|01>-|10>)/sqrt2, respectively. -/
def magicBasis : Matrix (Bool × Bool) (Fin 4) ℂ :=
  fun p j =>
    ![if p.1 ≠ p.2 then -Complex.I / (Real.sqrt 2 : ℂ) else 0,
      if p.1 = p.2 then 1 / (Real.sqrt 2 : ℂ) else 0,
      if p.1 = p.2 then (if p.1 then Complex.I else -Complex.I) / (Real.sqrt 2 : ℂ)
        else 0,
      if p.1 ≠ p.2 then (if p.1 then -1 else 1) / (Real.sqrt 2 : ℂ) else 0] j

/-- PDF p.13, Appendix A: express a computational-basis matrix in basis (2). -/
def magicMatrix (U : TwoMatrix) : Matrix (Fin 4) (Fin 4) ℂ :=
  magicBasis.conjTranspose * U * magicBasis

/-- PDF p.13: a phase-compatible SU(2) local decomposition of the actual representative.
This excludes assuming that an arbitrary projective phase has the same magic spectrum. -/
def PhaseCompatibleContent (U : TwoMatrix) (b : Triple) : Prop :=
  GateChamber b ∧ ∃ A B C D : OneGate,
    (A : OneMatrix).det = 1 ∧ (B : OneMatrix).det = 1 ∧
    (C : OneMatrix).det = 1 ∧ (D : OneMatrix).det = 1 ∧
    U = localMatrix A B * contentGate b * localMatrix C D

/-! Definition sanity checks; no paper theorem is proved here. -/

example : canonicalPauliInteraction (0 : Triple) = 0 := by
  simp [canonicalPauliInteraction]

example (H : TwoMatrix) : evolution H 0 = 1 := by
  simp [evolution]

example : GateChamber (0 : Triple) := by
  refine ⟨?_, ?_⟩
  · norm_num [SOrdered]
  · exact div_nonneg Real.pi_pos.le (by norm_num)

example : sOrder (0 : Triple) = 0 := by
  funext i
  fin_cases i <;> norm_num [sOrder]

example : SMajorized (0 : Triple) 0 := by
  norm_num [SMajorized, sOrder]

example (b : Triple) : contentIntegerShift b 0 = b := by
  funext i
  simp [contentIntegerShift]

example (A B : OneGate) : protocolTime ⟨A, B, []⟩ = 0 := rfl

example (A B : OneGate) (H : TwoMatrix) :
    protocolMatrix H ⟨A, B, []⟩ = localMatrix A B := rfl

example : dcnot (true, true) (false, true) = 1 := by
  simp [dcnot]

example (H : TwoMatrix) : taskCost H (fun _ => False) = ⊤ := by
  simp [taskCost]

example (U : TwoMatrix) (h : CanCbit U) : IsGate U := h.1

example (U : TwoMatrix) (h : CanTwoCbits U) : IsGate U := h.1

example (U : TwoMatrix) (h : CanQubit U) : IsGate U := h.1

example (U : TwoMatrix) (h : CanHybrid U) : IsGate U := h.1

example (U : TwoMatrix) (h : CanTwoQubits U) : IsGate U := h.1

example (b : Triple) : interactionPhases b 3 = -b 0 - b 1 - b 2 := rfl

example : magicBasis (false, false) 0 = 0 := by simp [magicBasis]

example : magicBasis (false, true) 3 = 1 / (Real.sqrt 2 : ℂ) := by simp [magicBasis]

end

end QAlgorithms.Papers.Independent.Intakethird100.P99
Source
Characterization of non-local gates, arXiv:quant-ph/0205100v3, https://arxiv.org/abs/quant-ph/0205100v3
Read-back

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

QAlgorithms.Papers.Independent.Intakethird100.P99.CanCbit

For a complex two-qubit matrix UUU, this predicate says that UUU is unitary and there exist single-qubit unitaries E,DE,DE,D, a vector v∈C{false,true}v\in\mathbb C^{\{\mathrm{false},\mathrm{true}\}}v∈C{false,true}, and vectors wiw_iwi​ in the same space for each Boolean iii, such that ∥v∥=1\|v\|=1∥v∥=1, ∥wi∥=1\|w_i\|=1∥wi​∥=1 for both iii, and

U((E∣i⟩)⊗v)=wi⊗(D∣i⟩)for every i∈{false,true}.U\bigl((E|i\rangle)\otimes v\bigr)=w_i\otimes(D|i\rangle) \qquad\text{for every }i\in\{\mathrm{false},\mathrm{true}\}.U((E∣i⟩)⊗v)=wi​⊗(D∣i⟩)for every i∈{false,true}.

Here ∣i⟩|i\rangle∣i⟩ is the coordinate vector with value one at iii and zero at the other Boolean, matrix action is ordinary matrix-vector multiplication, and (x⊗y)(a,b)=xayb(x\otimes y)_{(a,b)}=x_a y_b(x⊗y)(a,b)​=xa​yb​. All norms are Euclidean. The same E,D,vE,D,vE,D,v must work for both values of iii; wiw_iwi​ may depend on iii.

QAlgorithms.Papers.Independent.Intakethird100.P99.CanHybrid

For a complex two-qubit matrix UUU, this predicate says that UUU is unitary and there exist fixed single-qubit unitaries E,FE,FE,F, a single-qubit unitary RjR_jRj​ for each Boolean jjj, and single-qubit vectors aja_jaj​ such that ∥aj∥=1\|a_j\|=1∥aj​∥=1 for both jjj, ⟨afalse,atrue⟩=0\langle a_{\mathrm{false}},a_{\mathrm{true}}\rangle=0⟨afalse​,atrue​⟩=0, and

((I2⊗Rj)U)((Eq)⊗(F∣j⟩))=aj⊗qfor every Boolean j and every q with ∥q∥=1.\bigl((I_2\otimes R_j)U\bigr) \bigl((Eq)\otimes(F|j\rangle)\bigr)=a_j\otimes q \qquad\text{for every Boolean }j\text{ and every }q\text{ with }\|q\|=1.((I2​⊗Rj​)U)((Eq)⊗(F∣j⟩))=aj​⊗qfor every Boolean j and every q with ∥q∥=1.

The vectors lie in C{false,true}\mathbb C^{\{\mathrm{false},\mathrm{true}\}}C{false,true}, norms are Euclidean, the inner product is the standard complex inner product conjugate-linear in its first argument, and ∣j⟩|j\rangle∣j⟩ is the coordinate basis vector. Matrix actions are ordinary multiplication and tensor-vector coordinates are products. Both RjR_jRj​ and aja_jaj​ may depend on jjj, but none of the chosen objects may depend on qqq.

QAlgorithms.Papers.Independent.Intakethird100.P99.CanQubit

For a complex two-qubit matrix UUU, this predicate says that UUU is unitary and there exist fixed single-qubit unitaries E,RE,RE,R and fixed single-qubit vectors v,wv,wv,w of Euclidean norm one such that, for every single-qubit vector qqq with ∥q∥=1\|q\|=1∥q∥=1,

((I2⊗R)U)((Eq)⊗v)=w⊗q.\bigl((I_2\otimes R)U\bigr)\bigl((Eq)\otimes v\bigr)=w\otimes q.((I2​⊗R)U)((Eq)⊗v)=w⊗q.

All single-qubit vectors belong to C{false,true}\mathbb C^{\{\mathrm{false},\mathrm{true}\}}C{false,true}, I2I_2I2​ is its identity matrix, and tensor-vector coordinates are products of coordinates. The equality is exact, and the same choices E,R,v,wE,R,v,wE,R,v,w must work for every unit vector qqq.

QAlgorithms.Papers.Independent.Intakethird100.P99.CanTwoCbits

For a complex two-qubit matrix UUU, this predicate says that UUU is unitary and there exist single-qubit unitaries E,FE,FE,F and single-qubit vectors aij,bija_{ij},b_{ij}aij​,bij​ for every pair of Booleans i,ji,ji,j, such that

∥aij∥=∥bij∥=1,⟨ai,false,ai,true⟩=0for every i,⟨bfalse,j,btrue,j⟩=0for every j,\|a_{ij}\|=\|b_{ij}\|=1,\qquad \langle a_{i,\mathrm{false}},a_{i,\mathrm{true}}\rangle=0 \quad\text{for every }i,\qquad \langle b_{\mathrm{false},j},b_{\mathrm{true},j}\rangle=0 \quad\text{for every }j,∥aij​∥=∥bij​∥=1,⟨ai,false​,ai,true​⟩=0for every i,⟨bfalse,j​,btrue,j​⟩=0for every j,

and

U((E∣i⟩)⊗(F∣j⟩))=aij⊗bijfor every i,j.U\bigl((E|i\rangle)\otimes(F|j\rangle)\bigr)=a_{ij}\otimes b_{ij} \qquad\text{for every }i,j.U((E∣i⟩)⊗(F∣j⟩))=aij​⊗bij​for every i,j.

The vectors lie in C{false,true}\mathbb C^{\{\mathrm{false},\mathrm{true}\}}C{false,true}, with the Euclidean norm and the standard complex inner product, conjugate-linear in its first argument. The ket ∣i⟩|i\rangle∣i⟩ is the corresponding coordinate vector, matrix action is ordinary multiplication, and tensor-vector coordinates are products of coordinates. The unitaries E,FE,FE,F are fixed across all four inputs.

QAlgorithms.Papers.Independent.Intakethird100.P99.CanTwoQubits

For a complex two-qubit matrix UUU, this predicate says that UUU is unitary and there exist fixed single-qubit unitaries E,F,R,SE,F,R,SE,F,R,S such that

((R⊗S)U)((Eq)⊗(Fr))=r⊗qfor all single-qubit vectors q,r with ∥q∥=∥r∥=1.\bigl((R\otimes S)U\bigr)\bigl((Eq)\otimes(Fr)\bigr)=r\otimes q \qquad\text{for all single-qubit vectors }q,r\text{ with }\|q\|=\|r\|=1.((R⊗S)U)((Eq)⊗(Fr))=r⊗qfor all single-qubit vectors q,r with ∥q∥=∥r∥=1.

The single-qubit space is C{false,true}\mathbb C^{\{\mathrm{false},\mathrm{true}\}}C{false,true}, the norms are Euclidean, matrix action is ordinary multiplication, and tensor-vector coordinates are products of coordinates. The equality is exact and the four unitaries must work simultaneously for every such pair q,rq,rq,r.

QAlgorithms.Papers.Independent.Intakethird100.P99.ChamberClass

For every real triple b=(b0,b1,b2)b=(b_0,b_1,b_2)b=(b0​,b1​,b2​) and every index k∈{0,1,2,3}k\in\{0,1,2,3\}k∈{0,1,2,3}, this predicate is defined by

{b0<π/4,k=0,b0=π/4 and b1<π/4,k=1,b0=π/4, b1=π/4, and ∣b2∣<π/4,k=2,b0=π/4, b1=π/4, and ∣b2∣=π/4,k=3.\begin{cases} b_0<\pi/4, & k=0,\\ b_0=\pi/4\ \text{and}\ b_1<\pi/4, & k=1,\\ b_0=\pi/4,\ b_1=\pi/4,\ \text{and}\ |b_2|<\pi/4, & k=2,\\ b_0=\pi/4,\ b_1=\pi/4,\ \text{and}\ |b_2|=\pi/4, & k=3. \end{cases}⎩⎨⎧​b0​<π/4,b0​=π/4 and b1​<π/4,b0​=π/4, b1​=π/4, and ∣b2​∣<π/4,b0​=π/4, b1​=π/4, and ∣b2​∣=π/4,​k=0,k=1,k=2,k=3.​

No additional inequalities on the coordinates are part of this predicate; in particular, it does not separately require b1≤b0b_1\le b_0b1​≤b0​ or ∣b2∣≤b1|b_2|\le b_1∣b2​∣≤b1​.

QAlgorithms.Papers.Independent.Intakethird100.P99.ControlPhase

For every complex matrix WWW whose rows and columns are indexed by {false,true}\{\mathrm{false},\mathrm{true}\}{false,true}, and every real number bbb, this predicate means that 0≤b≤π/20\le b\le\pi/20≤b≤π/2 and there exists a complex number zzz with ∣z∣=1|z|=1∣z∣=1 such that the characteristic polynomial of the scalar multiple zWzWzW is

χzW(x)=(x−e2ib)(x−e−2ib).\chi_{zW}(x)=(x-e^{2ib})(x-e^{-2ib}).χzW​(x)=(x−e2ib)(x−e−2ib).

The endpoints b=0b=0b=0 and b=π/2b=\pi/2b=π/2 are included. The predicate imposes no separate unitarity or Hermiticity condition on WWW.

QAlgorithms.Papers.Independent.Intakethird100.P99.GateChamber

For a real triple b=(b0,b1,b2)b=(b_0,b_1,b_2)b=(b0​,b1​,b2​), this predicate means precisely that b1≤b0b_1\le b_0b1​≤b0​, ∣b2∣≤b1|b_2|\le b_1∣b2​∣≤b1​, and b0≤π/4b_0\le\pi/4b0​≤π/4. All boundary equalities are allowed.

QAlgorithms.Papers.Independent.Intakethird100.P99.HasCanonicalHamiltonian

For a complex 4×44\times44×4 matrix HHH, indexed by pairs of Boolean values, and a real triple a=(a0,a1,a2)a=(a_0,a_1,a_2)a=(a0​,a1​,a2​), this predicate holds precisely when HHH is Hermitian, a1≤a0a_1\le a_0a1​≤a0​ and ∣a2∣≤a1|a_2|\le a_1∣a2​∣≤a1​, and there exist complex 2×22\times22×2 unitary matrices A,BA,BA,B and complex 2×22\times22×2 Hermitian matrices K,LK,LK,L satisfying

(A⊗B)H(A⊗B)†=∑j=02aj σj⊗σj+K⊗I2+I2⊗L,(A\otimes B)H(A\otimes B)^\dagger =\sum_{j=0}^{2}a_j\,\sigma_j\otimes\sigma_j+K\otimes I_2+I_2\otimes L,(A⊗B)H(A⊗B)†=j=0∑2​aj​σj​⊗σj​+K⊗I2​+I2​⊗L,

where (σ0,σ1,σ2)=(X,Y,Z)(\sigma_0,\sigma_1,\sigma_2)=(X,Y,Z)(σ0​,σ1​,σ2​)=(X,Y,Z) are the usual Pauli matrices, I2I_2I2​ is the identity matrix, and †\dagger† denotes conjugate transpose. No upper bound on a0a_0a0​, determinant-one restriction on A,BA,BA,B, or trace restriction on K,LK,LK,L is imposed.

QAlgorithms.Papers.Independent.Intakethird100.P99.HasContent

For a complex 4×44\times44×4 matrix UUU, indexed by pairs of Boolean values, and a real triple b=(b0,b1,b2)b=(b_0,b_1,b_2)b=(b0​,b1​,b2​), this predicate holds precisely when UUU is unitary,

b1≤b0,∣b2∣≤b1,b0≤π4,b_1\le b_0,\qquad |b_2|\le b_1,\qquad b_0\le\frac{\pi}{4},b1​≤b0​,∣b2​∣≤b1​,b0​≤4π​,

and there exist four complex 2×22\times22×2 unitary matrices A,B,C,DA,B,C,DA,B,C,D such that

U=(A⊗B)exp⁡ ⁣(−i∑j=02bj σj⊗σj)(C⊗D),U=(A\otimes B)\exp\!\left(-\mathrm{i}\sum_{j=0}^{2}b_j\,\sigma_j\otimes\sigma_j\right)(C\otimes D),U=(A⊗B)exp(−ij=0∑2​bj​σj​⊗σj​)(C⊗D),

where (σ0,σ1,σ2)=(X,Y,Z)(\sigma_0,\sigma_1,\sigma_2)=(X,Y,Z)(σ0​,σ1​,σ2​)=(X,Y,Z) are the usual Pauli matrices and ⊗\otimes⊗ denotes the Kronecker product. The equality is exact, and the four local unitary matrices have no determinant-one requirement.

QAlgorithms.Papers.Independent.Intakethird100.P99.IdealSimulates

For complex 4×44\times44×4 matrices H,UH,UH,U, indexed by pairs of Boolean values, and a real number ttt, this predicate holds precisely when HHH is Hermitian, UUU is unitary, t≥0t\ge0t≥0, and for every real ε>0\varepsilon>0ε>0 there exist a finite list (tj,Aj,Bj)j=1m(t_j,A_j,B_j)_{j=1}^{m}(tj​,Aj​,Bj​)j=1m​, initial matrices A0,B0A_0,B_0A0​,B0​, and a complex scalar zzz, such that all Aj,BjA_j,B_jAj​,Bj​, including the initial pair, are complex 2×22\times22×2 unitaries, every tj≥0t_j\ge0tj​≥0, ∑j=1mtj=t\sum_{j=1}^{m}t_j=t∑j=1m​tj​=t, ∣z∣=1|z|=1∣z∣=1, and

∥Vm−zU∥op<ε,V0=A0⊗B0,Vj=(Aj⊗Bj)exp⁡(−itjH)Vj−1.\|V_m-zU\|_{\mathrm{op}}<\varepsilon,\qquad V_0=A_0\otimes B_0,\qquad V_j=(A_j\otimes B_j)\exp(-\mathrm{i}t_jH)V_{j-1}.∥Vm​−zU∥op​<ε,V0​=A0​⊗B0​,Vj​=(Aj​⊗Bj​)exp(−itj​H)Vj−1​.

Here ∥⋅∥op\|\cdot\|_{\mathrm{op}}∥⋅∥op​ is the operator norm induced by the Euclidean norm. The list and phase may depend on ε\varepsilonε; empty lists, zero durations, and total time t=0t=0t=0 are included.

QAlgorithms.Papers.Independent.Intakethird100.P99.IsGate

For any 4×44\times44×4 complex matrix UUU indexed by pairs of Boolean values, this predicate means that UUU is unitary: U†U=UU†=I4U^\dagger U=UU^\dagger=I_4U†U=UU†=I4​.

QAlgorithms.Papers.Independent.Intakethird100.P99.MoreNonlocal

For arbitrary complex two-qubit matrices U,VU,VU,V, this predicate says that for every Hermitian complex two-qubit matrix HHH, CH(V)≤CH(U)C_H(V)\le C_H(U)CH​(V)≤CH​(U). Here CH(W)∈[0,∞]C_H(W)\in[0,\infty]CH​(W)∈[0,∞] is the infimum of all nonnegative real times ttt such that WWW is unitary and, for every ε>0\varepsilon>0ε>0, there exist a finite list of nonnegative durations t1,…,tmt_1,\ldots,t_mt1​,…,tm​ summing exactly to ttt, single-qubit unitaries A0,B0,…,Am,BmA_0,B_0,\ldots,A_m,B_mA0​,B0​,…,Am​,Bm​, and z∈Cz\in\mathbb Cz∈C with ∣z∣=1|z|=1∣z∣=1, for which

∥(Am⊗Bm)e−itmH⋯(A1⊗B1)e−it1H(A0⊗B0)−zW∥op<ε.\left\|(A_m\otimes B_m)e^{-it_mH}\cdots (A_1\otimes B_1)e^{-it_1H}(A_0\otimes B_0)-zW\right\|_{\mathrm{op}}<\varepsilon.​(Am​⊗Bm​)e−itm​H⋯(A1​⊗B1​)e−it1​H(A0​⊗B0​)−zW​op​<ε.

The norm is the Euclidean operator norm. The list may be empty, giving duration zero and matrix A0⊗B0A_0\otimes B_0A0​⊗B0​; the protocol and phase may depend on ε\varepsilonε. The infimum of an empty set is +∞+\infty+∞. Neither UUU nor VVV is assumed unitary in the predicate itself, and the comparison includes infinite costs, with +∞≤+∞+\infty\le+\infty+∞≤+∞ allowed.

QAlgorithms.Papers.Independent.Intakethird100.P99.OneGate

This is the type of unitary 2×22\times22×2 complex matrices indexed by Boolean values: matrices AAA satisfying A†A=AA†=I2A^\dagger A=AA^\dagger=I_2A†A=AA†=I2​. Their determinants are not required to equal 111.

QAlgorithms.Papers.Independent.Intakethird100.P99.OneMatrix

This is the type of all complex matrices whose rows and columns are indexed by {false,true}\{\mathrm{false},\mathrm{true}\}{false,true}, hence all 2×22\times22×2 complex matrices, with no further restrictions.

QAlgorithms.Papers.Independent.Intakethird100.P99.OneState

This is the complex Euclidean space of vectors indexed by Boolean values, namely C2\mathbb C^2C2 with its Euclidean inner product. Vectors are not required to have norm 111; the zero vector is included.

QAlgorithms.Papers.Independent.Intakethird100.P99.OptimalTask

For arbitrary complex two-qubit matrices H,UH,UH,U, a predicate PPP on such matrices, and a real number ttt, this predicate asserts all of the following simultaneously: P(U)P(U)P(U) holds; HHH is Hermitian, UUU is unitary, and t≥0t\ge0t≥0; for every ε>0\varepsilon>0ε>0 there exist a finite list of nonnegative durations t1,…,tmt_1,\ldots,t_mt1​,…,tm​ summing exactly to ttt, single-qubit unitaries A0,B0,…,Am,BmA_0,B_0,\ldots,A_m,B_mA0​,B0​,…,Am​,Bm​, and z∈Cz\in\mathbb Cz∈C with ∣z∣=1|z|=1∣z∣=1, such that

∥(Am⊗Bm)e−itmH⋯(A1⊗B1)e−it1H(A0⊗B0)−zU∥op<ε;\left\|(A_m\otimes B_m)e^{-it_mH}\cdots (A_1\otimes B_1)e^{-it_1H}(A_0\otimes B_0)-zU\right\|_{\mathrm{op}}<\varepsilon;​(Am​⊗Bm​)e−itm​H⋯(A1​⊗B1​)e−it1​H(A0​⊗B0​)−zU​op​<ε;

moreover,

CH(U)=ofReal⁡(t),inf⁡V: P(V)CH(V)=ofReal⁡(t),∀V, P(V)⟹CH(U)≤CH(V).C_H(U)=\operatorname{ofReal}(t),\qquad \inf_{V:\,P(V)}C_H(V)=\operatorname{ofReal}(t),\qquad \forall V,\ P(V)\Longrightarrow C_H(U)\le C_H(V).CH​(U)=ofReal(t),V:P(V)inf​CH​(V)=ofReal(t),∀V, P(V)⟹CH​(U)≤CH​(V).

Here CH(W)C_H(W)CH​(W) is the infimum in [0,∞][0,\infty][0,∞] of all nonnegative times admitting the same approximation condition with target WWW, including the requirements that HHH be Hermitian and WWW unitary; an empty admissible set has cost +∞+\infty+∞. The operator norm is induced by the Euclidean norm. Empty pulse lists are allowed and give duration zero and matrix A0⊗B0A_0\otimes B_0A0​⊗B0​, while protocols and phases may vary with ε\varepsilonε. Finally, ofReal⁡(t)\operatorname{ofReal}(t)ofReal(t) is max⁡(t,0)\max(t,0)max(t,0) regarded as an extended nonnegative real, which here equals ttt because nonnegativity is required.

QAlgorithms.Papers.Independent.Intakethird100.P99.PhaseCompatibleContent

For every complex matrix UUU on two qubits and every real triple b=(b0,b1,b2)b=(b_0,b_1,b_2)b=(b0​,b1​,b2​), this predicate means that

b1≤b0,∣b2∣≤b1,b0≤π4,b_1\le b_0,\qquad |b_2|\le b_1,\qquad b_0\le\frac{\pi}{4},b1​≤b0​,∣b2​∣≤b1​,b0​≤4π​,

and there exist four complex 2×22\times22×2 unitary matrices A,B,C,DA,B,C,DA,B,C,D, each with determinant 111, such that

U=(A⊗B)exp⁡ ⁣(−i∑j=02bj(Pj⊗Pj))(C⊗D),U=(A\otimes B)\exp\!\left(-i\sum_{j=0}^{2}b_j(P_j\otimes P_j)\right)(C\otimes D),U=(A⊗B)exp(−ij=0∑2​bj​(Pj​⊗Pj​))(C⊗D),

where P0=X,P1=Y,P2=ZP_0=X,P_1=Y,P_2=ZP0​=X,P1​=Y,P2​=Z are the Pauli matrices

X=(0110),Y=(0−ii0),Z=(100−1)X=\begin{pmatrix}0&1\\1&0\end{pmatrix},\qquad Y=\begin{pmatrix}0&-i\\i&0\end{pmatrix},\qquad Z=\begin{pmatrix}1&0\\0&-1\end{pmatrix}X=(01​10​),Y=(0i​−i0​),Z=(10​0−1​)

in Boolean basis order false,true\mathrm{false},\mathrm{true}false,true. The displayed equality is exact and contains no additional overall phase factor.

QAlgorithms.Papers.Independent.Intakethird100.P99.PulseProtocol

A pulse protocol consists of two initial complex 2×22\times22×2 unitary matrices A0,B0A_0,B_0A0​,B0​ and a finite ordered list of triples (tj,Aj,Bj)(t_j,A_j,B_j)(tj​,Aj​,Bj​), where every tjt_jtj​ is a nonnegative real number and every Aj,BjA_j,B_jAj​,Bj​ is a complex 2×22\times22×2 unitary matrix. The list may be empty, durations may be zero, and none of the unitary matrices is required to have determinant one.

QAlgorithms.Papers.Independent.Intakethird100.P99.PureInteraction

For a complex 4×44\times44×4 matrix HHH, indexed by pairs of Boolean values, this predicate holds precisely when there exist nine real coefficients cijc_{ij}cij​, for i,j∈{0,1,2}i,j\in\{0,1,2\}i,j∈{0,1,2}, such that

H=∑i=02∑j=02cij σi⊗σj,H=\sum_{i=0}^{2}\sum_{j=0}^{2}c_{ij}\,\sigma_i\otimes\sigma_j,H=i=0∑2​j=0∑2​cij​σi​⊗σj​,

where (σ0,σ1,σ2)=(X,Y,Z)(\sigma_0,\sigma_1,\sigma_2)=(X,Y,Z)(σ0​,σ1​,σ2​)=(X,Y,Z) are the usual Pauli matrices and ⊗\otimes⊗ is the Kronecker product. The coefficients are unrestricted real numbers, including zero, so the zero matrix is included.

QAlgorithms.Papers.Independent.Intakethird100.P99.SMajorized

For arbitrary real triples bbb and aaa, this predicate holds precisely when the following three inequalities hold. For each u∈{a,b}u\in\{a,b\}u∈{a,b}, put hu=max⁡i∣ui∣h_u=\max_i|u_i|hu​=maxi​∣ui​∣, ℓu=min⁡i∣ui∣\ell_u=\min_i|u_i|ℓu​=mini​∣ui​∣, mu=∣u0∣+∣u1∣+∣u2∣−hu−ℓum_u=|u_0|+|u_1|+|u_2|-h_u-\ell_umu​=∣u0​∣+∣u1​∣+∣u2​∣−hu​−ℓu​, and su=−ℓus_u=-\ell_usu​=−ℓu​ if u0u1u2<0u_0u_1u_2<0u0​u1​u2​<0, otherwise su=ℓus_u=\ell_usu​=ℓu​. The required inequalities are

hb≤ha,hb+mb−sb≤ha+ma−sa,hb+mb+sb≤ha+ma+sa.h_b\le h_a,\qquad h_b+m_b-s_b\le h_a+m_a-s_a,\qquad h_b+m_b+s_b\le h_a+m_a+s_a.hb​≤ha​,hb​+mb​−sb​≤ha​+ma​−sa​,hb​+mb​+sb​≤ha​+ma​+sa​.

No ordering or nonzero-coordinate assumptions are imposed on either input triple.

QAlgorithms.Papers.Independent.Intakethird100.P99.SOrdered

For a real triple a=(a0,a1,a2)a=(a_0,a_1,a_2)a=(a0​,a1​,a2​), this predicate means precisely that a1≤a0a_1\le a_0a1​≤a0​ and ∣a2∣≤a1|a_2|\le a_1∣a2​∣≤a1​. Both inequalities allow equality, including the zero triple.

QAlgorithms.Papers.Independent.Intakethird100.P99.SamePhysicalGate

For any two complex 4×44\times44×4 matrices U,VU,VU,V, indexed by pairs of Boolean values, this predicate holds precisely when there exists a complex scalar zzz with ∣z∣=1|z|=1∣z∣=1 such that

U=zV.U=zV.U=zV.

Neither matrix is assumed unitary; in particular, both may be zero.

QAlgorithms.Papers.Independent.Intakethird100.P99.TransmissionProfile

For every complex matrix UUU on C2⊗C2\mathbb C^2\otimes\mathbb C^2C2⊗C2 and every k∈{0,1,2,3}k\in\{0,1,2,3\}k∈{0,1,2,3}, define Urev=SUSU^{\mathrm{rev}}=SUSUrev=SUS, where S(x⊗y)=y⊗xS(x\otimes y)=y\otimes xS(x⊗y)=y⊗x; use the standard basis vectors e0,e1e_0,e_1e0​,e1​, with 0=false0=\mathrm{false}0=false and 1=true1=\mathrm{true}1=true, and tensor coordinates (x⊗y)ij=xiyj(x\otimes y)_{ij}=x_i y_j(x⊗y)ij​=xi​yj​. To state the predicate explicitly, let C(B)C(B)C(B) mean that BBB is unitary and there exist single-qubit unitary matrices E,DE,DE,D, a unit vector vvv, and unit vectors wiw_iwi​ for i∈{0,1}i\in\{0,1\}i∈{0,1}, such that B(Eei⊗v)=wi⊗DeiB(Ee_i\otimes v)=w_i\otimes De_iB(Eei​⊗v)=wi​⊗Dei​ for every iii. Let D2(B)D_2(B)D2​(B) mean that BBB is unitary and there exist single-qubit unitary matrices E,FE,FE,F and unit vectors aij,bija_{ij},b_{ij}aij​,bij​ for every i,j∈{0,1}i,j\in\{0,1\}i,j∈{0,1}, such that ⟨ai0,ai1⟩=0\langle a_{i0},a_{i1}\rangle=0⟨ai0​,ai1​⟩=0 for every iii, ⟨b0j,b1j⟩=0\langle b_{0j},b_{1j}\rangle=0⟨b0j​,b1j​⟩=0 for every jjj, and B(Eei⊗Fej)=aij⊗bijB(Ee_i\otimes Fe_j)=a_{ij}\otimes b_{ij}B(Eei​⊗Fej​)=aij​⊗bij​ for every i,ji,ji,j. Let Q(B)Q(B)Q(B) mean that BBB is unitary and there exist single-qubit unitary matrices E,RE,RE,R and unit vectors v,wv,wv,w such that (I2⊗R)B(Eq⊗v)=w⊗q(I_2\otimes R)B(Eq\otimes v)=w\otimes q(I2​⊗R)B(Eq⊗v)=w⊗q for every unit vector q∈C2q\in\mathbb C^2q∈C2. Let H(B)H(B)H(B) mean that BBB is unitary and there exist single-qubit unitary matrices E,FE,FE,F, single-qubit unitary matrices RjR_jRj​ for j∈{0,1}j\in\{0,1\}j∈{0,1}, and orthogonal unit vectors a0,a1a_0,a_1a0​,a1​, such that (I2⊗Rj)B(Eq⊗Fej)=aj⊗q(I_2\otimes R_j)B(Eq\otimes Fe_j)=a_j\otimes q(I2​⊗Rj​)B(Eq⊗Fej​)=aj​⊗q for every jjj and every unit vector qqq. Finally, let Q2(B)Q_2(B)Q2​(B) mean that BBB is unitary and there exist single-qubit unitary matrices E,F,R,TE,F,R,TE,F,R,T such that (R⊗T)B(Eq⊗Fr)=r⊗q(R\otimes T)B(Eq\otimes Fr)=r\otimes q(R⊗T)B(Eq⊗Fr)=r⊗q for every pair of unit vectors q,r∈C2q,r\in\mathbb C^2q,r∈C2. The defined profile requires exactly the following truth values, where a false entry means the negation of the corresponding existential predicate:

kC(U)C(Urev)D2(U)Q(U)Q(Urev)H(U)H(Urev)Q2(U)0FFFFFFFF1TTFFFFFF2TTTTTTTF3TTTTTTTT\begin{array}{c|cccccccc} k&C(U)&C(U^{\mathrm{rev}})&D_2(U)&Q(U)&Q(U^{\mathrm{rev}})&H(U)&H(U^{\mathrm{rev}})&Q_2(U)\\ \hline 0&\mathrm F&\mathrm F&\mathrm F&\mathrm F&\mathrm F&\mathrm F&\mathrm F&\mathrm F\\ 1&\mathrm T&\mathrm T&\mathrm F&\mathrm F&\mathrm F&\mathrm F&\mathrm F&\mathrm F\\ 2&\mathrm T&\mathrm T&\mathrm T&\mathrm T&\mathrm T&\mathrm T&\mathrm T&\mathrm F\\ 3&\mathrm T&\mathrm T&\mathrm T&\mathrm T&\mathrm T&\mathrm T&\mathrm T&\mathrm T \end{array}k0123​C(U)FTTT​C(Urev)FTTT​D2​(U)FFTT​Q(U)FFTT​Q(Urev)FFTT​H(U)FFTT​H(Urev)FFTT​Q2​(U)FFFT​​

There is no separate unitarity assumption on UUU; in particular, every nonunitary UUU satisfies the k=0k=0k=0 branch.

QAlgorithms.Papers.Independent.Intakethird100.P99.Triple

This is the type of real triples a=(a0,a1,a2)a=(a_0,a_1,a_2)a=(a0​,a1​,a2​), with no restrictions on their entries.

QAlgorithms.Papers.Independent.Intakethird100.P99.TwoMatrix

This is the type of all complex matrices whose rows and columns are indexed by ordered pairs of Boolean values, hence all 4×44\times44×4 complex matrices, with no further restrictions.

QAlgorithms.Papers.Independent.Intakethird100.P99.TwoQubitLocalEquivalent

For arbitrary 4×44\times44×4 complex matrices UUU and VVV indexed by Boolean pairs, this predicate means that there exist four unitary 2×22\times22×2 complex matrices A,B,C,DA,B,C,DA,B,C,D such that U=(A⊗B)V(C⊗D)U=(A\otimes B)V(C\otimes D)U=(A⊗B)V(C⊗D). Neither UUU nor VVV is separately required to be unitary, and the equality is exact.

QAlgorithms.Papers.Independent.Intakethird100.P99.branchCost

For arbitrary real triples a=(a0,a1,a2)a=(a_0,a_1,a_2)a=(a0​,a1​,a2​) and b=(b0,b1,b2)b=(b_0,b_1,b_2)b=(b0​,b1​,b2​), define, for each real triple vvv,

hv=max⁡(∣v0∣,∣v1∣,∣v2∣),ℓv=min⁡(∣v0∣,∣v1∣,∣v2∣),s(v)=(hv, ∣v0∣+∣v1∣+∣v2∣−hv−ℓv, {−ℓvv0v1v2<0,ℓvv0v1v2≥0.).h_v=\max(|v_0|,|v_1|,|v_2|),\quad \ell_v=\min(|v_0|,|v_1|,|v_2|),\quad s(v)=\left(h_v,\ |v_0|+|v_1|+|v_2|-h_v-\ell_v,\ \begin{cases}-\ell_v&v_0v_1v_2<0,\\ \ell_v&v_0v_1v_2\ge0.\end{cases}\right).hv​=max(∣v0​∣,∣v1​∣,∣v2​∣),ℓv​=min(∣v0​∣,∣v1​∣,∣v2​∣),s(v)=(hv​, ∣v0​∣+∣v1​∣+∣v2​∣−hv​−ℓv​, {−ℓv​ℓv​​v0​v1​v2​<0,v0​v1​v2​≥0.​).

The branch cost is the infimum in [0,∞][0,\infty][0,∞] of the nonnegative real numbers ttt for which, writing x=s(b)x=s(b)x=s(b) and y=s(ta)y=s(ta)y=s(ta) with (ta)i=tai(ta)_i=ta_i(ta)i​=tai​,

x0≤y0,x0+x1−x2≤y0+y1−y2,x0+x1+x2≤y0+y1+y2.x_0\le y_0,\qquad x_0+x_1-x_2\le y_0+y_1-y_2,\qquad x_0+x_1+x_2\le y_0+y_1+y_2.x0​≤y0​,x0​+x1​−x2​≤y0​+y1​−y2​,x0​+x1​+x2​≤y0​+y1​+y2​.

No ordering or nonzero assumptions are imposed on either triple, and t=0t=0t=0 is included among the candidate times. An empty set of candidate times has infimum +∞+\infty+∞; an infimum is not required to be attained.

QAlgorithms.Papers.Independent.Intakethird100.P99.canonicalPauliInteraction

For every real triple a=(a0,a1,a2)a=(a_0,a_1,a_2)a=(a0​,a1​,a2​), this defines the 4×44\times44×4 complex matrix a0(X⊗X)+a1(Y⊗Y)+a2(Z⊗Z)a_0(X\otimes X)+a_1(Y\otimes Y)+a_2(Z\otimes Z)a0​(X⊗X)+a1​(Y⊗Y)+a2​(Z⊗Z), where, in Boolean order (false,true)(\mathrm{false},\mathrm{true})(false,true), X=(0110)X=\begin{pmatrix}0&1\\1&0\end{pmatrix}X=(01​10​), Y=(0−ii0)Y=\begin{pmatrix}0&-\mathrm i\\ \mathrm i&0\end{pmatrix}Y=(0i​−i0​), and Z=(100−1)Z=\begin{pmatrix}1&0\\0&-1\end{pmatrix}Z=(10​0−1​).

QAlgorithms.Papers.Independent.Intakethird100.P99.contentGate

For every real triple b=(b0,b1,b2)b=(b_0,b_1,b_2)b=(b0​,b1​,b2​), this defines the matrix exponential exp⁡ ⁣(−i[b0(X⊗X)+b1(Y⊗Y)+b2(Z⊗Z)])\exp\!\bigl(-\mathrm i[b_0(X\otimes X)+b_1(Y\otimes Y)+b_2(Z\otimes Z)]\bigr)exp(−i[b0​(X⊗X)+b1​(Y⊗Y)+b2​(Z⊗Z)]), where X=(0110)X=\begin{pmatrix}0&1\\1&0\end{pmatrix}X=(01​10​), Y=(0−ii0)Y=\begin{pmatrix}0&-\mathrm i\\ \mathrm i&0\end{pmatrix}Y=(0i​−i0​), and Z=(100−1)Z=\begin{pmatrix}1&0\\0&-1\end{pmatrix}Z=(10​0−1​) in Boolean order (false,true)(\mathrm{false},\mathrm{true})(false,true). No ordering or size restrictions are imposed on bbb.

QAlgorithms.Papers.Independent.Intakethird100.P99.contentIntegerShift

For any real triple b=(b0,b1,b2)b=(b_0,b_1,b_2)b=(b0​,b1​,b2​) and any integer triple n=(n0,n1,n2)n=(n_0,n_1,n_2)n=(n0​,n1​,n2​), this function returns the real triple whose coordinates are

bi+π2ni,i=0,1,2.b_i+\frac{\pi}{2}n_i,\qquad i=0,1,2.bi​+2π​ni​,i=0,1,2.

Each integer coordinate may be positive, zero, or negative; no reduction modulo any period is performed.

QAlgorithms.Papers.Independent.Intakethird100.P99.controlledContent

For every real number bbb, this defines the real triple

(min⁡{b,π2−b},0,0).\left(\min\left\{b,\frac{\pi}{2}-b\right\},0,0\right).(min{b,2π​−b},0,0).

There is no restriction on bbb; in particular, the first coordinate can be negative when bbb lies outside [0,π/2][0,\pi/2][0,π/2].

QAlgorithms.Papers.Independent.Intakethird100.P99.dcnot

This defines a complex matrix DDD whose rows and columns are indexed by pairs of Booleans: for every row p=(p1,p2)p=(p_1,p_2)p=(p1​,p2​) and column q=(q1,q2)q=(q_1,q_2)q=(q1​,q2​),

Dp,q={1,p=(q2,q1⊕q2),0,otherwise,D_{p,q}= \begin{cases} 1,&p=(q_2,q_1\mathbin{\oplus}q_2),\\ 0,&\text{otherwise}, \end{cases}Dp,q​={1,0,​p=(q2​,q1​⊕q2​),otherwise,​

where ⊕\oplus⊕ is exclusive-or. Equivalently, on coordinate basis vectors it sends ∣q1,q2⟩|q_1,q_2\rangle∣q1​,q2​⟩ to ∣q2,q1⊕q2⟩|q_2,q_1\oplus q_2\rangle∣q2​,q1​⊕q2​⟩.

QAlgorithms.Papers.Independent.Intakethird100.P99.interactionCost

For arbitrary complex matrices H,UH,UH,U acting on C{false,true}2\mathbb C^{\{\mathrm{false},\mathrm{true}\}^2}C{false,true}2, the interaction cost is the infimum, in [0,∞][0,\infty][0,∞], of all nonnegative real times ttt satisfying the following condition: HHH is Hermitian, UUU is unitary, and for every ε>0\varepsilon>0ε>0 there exist a finite list of nonnegative durations t1,…,tmt_1,\ldots,t_mt1​,…,tm​, single-qubit unitaries A0,B0,…,Am,BmA_0,B_0,\ldots,A_m,B_mA0​,B0​,…,Am​,Bm​, and a complex number zzz with ∣z∣=1|z|=1∣z∣=1, such that

∑j=1mtj=t,∥(Am⊗Bm)e−itmH⋯(A1⊗B1)e−it1H(A0⊗B0)−zU∥op<ε.\sum_{j=1}^m t_j=t,\qquad \left\|(A_m\otimes B_m)e^{-it_mH}\cdots (A_1\otimes B_1)e^{-it_1H}(A_0\otimes B_0)-zU\right\|_{\mathrm{op}}<\varepsilon.j=1∑m​tj​=t,​(Am​⊗Bm​)e−itm​H⋯(A1​⊗B1​)e−it1​H(A0​⊗B0​)−zU​op​<ε.

Here the norm is the operator norm for the Euclidean norm, and the tensor products are ordinary Kronecker products. The list may be empty, in which case its total duration is zero and its matrix is A0⊗B0A_0\otimes B_0A0​⊗B0​. The protocols and zzz may depend on ε\varepsilonε. The infimum need not be attained; if there are no admissible times, the cost is +∞+\infty+∞, including whenever HHH is not Hermitian or UUU is not unitary.

QAlgorithms.Papers.Independent.Intakethird100.P99.interactionPhases

For every real triple b=(b0,b1,b2)b=(b_0,b_1,b_2)b=(b0​,b1​,b2​), this defines a function on {0,1,2,3}\{0,1,2,3\}{0,1,2,3} with values, in that order,

(b0+b1−b2,b0−b1+b2,−b0+b1+b2,−b0−b1−b2).\left( b_0+b_1-b_2,\quad b_0-b_1+b_2,\quad -b_0+b_1+b_2,\quad -b_0-b_1-b_2 \right).(b0​+b1​−b2​,b0​−b1​+b2​,−b0​+b1​+b2​,−b0​−b1​−b2​).

There are no assumptions on bbb, and no reduction of these real values modulo a period is performed.

QAlgorithms.Papers.Independent.Intakethird100.P99.localMatrix

For any unitary 2×22\times22×2 complex matrices AAA and BBB indexed by Boolean values, this defines their Kronecker product A⊗BA\otimes BA⊗B, whose entry at row (x,y)(x,y)(x,y) and column (x′,y′)(x',y')(x′,y′) is Ax,x′By,y′A_{x,x'}B_{y,y'}Ax,x′​By,y′​.

QAlgorithms.Papers.Independent.Intakethird100.P99.magicBasis

This defines the complex matrix with rows indexed by Boolean pairs and columns indexed by 0,1,2,30,1,2,30,1,2,3 which, in row order (false,false),(false,true),(true,false),(true,true)(\mathrm{false},\mathrm{false}),(\mathrm{false},\mathrm{true}),(\mathrm{true},\mathrm{false}),(\mathrm{true},\mathrm{true})(false,false),(false,true),(true,false),(true,true), is

B=12(01−i0−i001−i00−101i0).B=\frac{1}{\sqrt2} \begin{pmatrix} 0&1&-i&0\\ -i&0&0&1\\ -i&0&0&-1\\ 0&1&i&0 \end{pmatrix}.B=2​1​​0−i−i0​1001​−i00i​01−10​​.

QAlgorithms.Papers.Independent.Intakethird100.P99.magicMatrix

For every complex matrix UUU whose rows and columns are indexed by Boolean pairs, this defines the complex matrix on indices 0,1,2,30,1,2,30,1,2,3 by

B†UB,B=12(01−i0−i001−i00−101i0),B^\dagger U B, \qquad B=\frac{1}{\sqrt2} \begin{pmatrix} 0&1&-i&0\\ -i&0&0&1\\ -i&0&0&-1\\ 0&1&i&0 \end{pmatrix},B†UB,B=2​1​​0−i−i0​1001​−i00i​01−10​​,

where the rows of BBB are ordered as (false,false),(false,true),(true,false),(true,true)(\mathrm{false},\mathrm{false}),(\mathrm{false},\mathrm{true}),(\mathrm{true},\mathrm{false}),(\mathrm{true},\mathrm{true})(false,false),(false,true),(true,false),(true,true), its columns are ordered 0,1,2,30,1,2,30,1,2,3, and B†B^\daggerB† is its conjugate transpose. No unitarity assumption on UUU is required.

QAlgorithms.Papers.Independent.Intakethird100.P99.optimalTransmissionContent

For every real triple a=(a0,a1,a2)a=(a_0,a_1,a_2)a=(a0​,a1​,a2​), this defines the real triple

(π4,π4,π42a2a0+a1).\left(\frac{\pi}{4},\frac{\pi}{4}, \frac{\pi}{4}\frac{2a_2}{a_0+a_1}\right).(4π​,4π​,4π​a0​+a1​2a2​​).

There are no ordering or nonzero-denominator assumptions on aaa. Division is total: if a0+a1=0a_0+a_1=0a0​+a1​=0, the third coordinate is 000.

QAlgorithms.Papers.Independent.Intakethird100.P99.protocolMatrix

For any complex 4×44\times44×4 matrix HHH, indexed by pairs of Boolean values, and any pulse protocol with initial complex 2×22\times22×2 unitaries A0,B0A_0,B_0A0​,B0​ and ordered list (tj,Aj,Bj)j=1m(t_j,A_j,B_j)_{j=1}^{m}(tj​,Aj​,Bj​)j=1m​, where tj≥0t_j\ge0tj​≥0 and Aj,BjA_j,B_jAj​,Bj​ are complex 2×22\times22×2 unitaries, define

V0=A0⊗B0,Vj=(Aj⊗Bj)exp⁡(−itjH)Vj−1(1≤j≤m).V_0=A_0\otimes B_0,\qquad V_j=(A_j\otimes B_j)\exp(-\mathrm{i}t_jH)V_{j-1}\quad(1\le j\le m).V0​=A0​⊗B0​,Vj​=(Aj​⊗Bj​)exp(−itj​H)Vj−1​(1≤j≤m).

The protocol matrix is VmV_mVm​. For an empty list it is A0⊗B0A_0\otimes B_0A0​⊗B0​; zero durations are allowed, and HHH is not assumed Hermitian.

QAlgorithms.Papers.Independent.Intakethird100.P99.protocolTime

For any pulse protocol ppp, consisting of initial complex 2×22\times22×2 unitaries and a finite ordered list (tj,Aj,Bj)j=1m(t_j,A_j,B_j)_{j=1}^{m}(tj​,Aj​,Bj​)j=1m​ of nonnegative real durations paired with two complex 2×22\times22×2 unitaries, its protocol time is the nonnegative real number

∑j=1mtj.\sum_{j=1}^{m}t_j.j=1∑m​tj​.

The initial unitary matrices do not contribute to this sum, and an empty list has protocol time zero.

QAlgorithms.Papers.Independent.Intakethird100.P99.reverseParties

For every complex matrix UUU indexed by Boolean pairs, this defines SUSSUSSUS, where SSS is the matrix with entry Sp,q=1S_{p,q}=1Sp,q​=1 when p=(q2,q1)p=(q_2,q_1)p=(q2​,q1​) and entry zero otherwise. Thus the resulting entries are

(reverseParties⁡(U))(a,b),(c,d)=U(b,a),(d,c)\bigl(\operatorname{reverseParties}(U)\bigr)_{(a,b),(c,d)} =U_{(b,a),(d,c)}(reverseParties(U))(a,b),(c,d)​=U(b,a),(d,c)​

for all Booleans a,b,c,da,b,c,da,b,c,d. There is no unitarity assumption on UUU.

QAlgorithms.Papers.Independent.Intakethird100.P99.sOrder

For any real triple a=(a0,a1,a2)a=(a_0,a_1,a_2)a=(a0​,a1​,a2​), define h=max⁡{∣a0∣,∣a1∣,∣a2∣}h=\max\{|a_0|,|a_1|,|a_2|\}h=max{∣a0​∣,∣a1​∣,∣a2​∣} and ℓ=min⁡{∣a0∣,∣a1∣,∣a2∣}\ell=\min\{|a_0|,|a_1|,|a_2|\}ℓ=min{∣a0​∣,∣a1​∣,∣a2​∣}. The function returns the real triple

(h, ∣a0∣+∣a1∣+∣a2∣−h−ℓ, {−ℓ,a0a1a2<0,ℓ,a0a1a2≥0.).\left(h,\ |a_0|+|a_1|+|a_2|-h-\ell,\ \begin{cases} -\ell,&a_0a_1a_2<0,\\ \ell,&a_0a_1a_2\ge0. \end{cases}\right).(h, ∣a0​∣+∣a1​∣+∣a2​∣−h−ℓ, {−ℓ,ℓ,​a0​a1​a2​<0,a0​a1​a2​≥0.​).

Thus the first two coordinates are the largest and middle absolute values; the third is the smallest absolute value with the indicated sign. Ties and zero coordinates are included, and a zero product takes the nonnegative branch.

QAlgorithms.Papers.Independent.Intakethird100.P99.secondShift

For any real triple b=(b0,b1,b2)b=(b_0,b_1,b_2)b=(b0​,b1​,b2​), this function returns

(b0−π2, b1, b2).\left(b_0-\frac{\pi}{2},\,b_1,\,b_2\right).(b0​−2π​,b1​,b2​).

There are no ordering or range restrictions on bbb.

QAlgorithms.Papers.Independent.Intakethird100.P99.taskCost

For an arbitrary complex two-qubit matrix HHH and an arbitrary predicate PPP on complex two-qubit matrices, this defines inf⁡{CH(U):P(U)}\inf\{C_H(U):P(U)\}inf{CH​(U):P(U)} in [0,∞][0,\infty][0,∞]. Here CH(U)C_H(U)CH​(U) is the infimum of the nonnegative real times ttt for which HHH is Hermitian, UUU is unitary, and for every ε>0\varepsilon>0ε>0 there exist nonnegative durations t1,…,tmt_1,\ldots,t_mt1​,…,tm​ summing exactly to ttt, single-qubit unitaries A0,B0,…,Am,BmA_0,B_0,\ldots,A_m,B_mA0​,B0​,…,Am​,Bm​, and a complex number zzz with ∣z∣=1|z|=1∣z∣=1, such that

∥(Am⊗Bm)e−itmH⋯(A1⊗B1)e−it1H(A0⊗B0)−zU∥op<ε.\left\|(A_m\otimes B_m)e^{-it_mH}\cdots (A_1\otimes B_1)e^{-it_1H}(A_0\otimes B_0)-zU\right\|_{\mathrm{op}}<\varepsilon.​(Am​⊗Bm​)e−itm​H⋯(A1​⊗B1​)e−it1​H(A0​⊗B0​)−zU​op​<ε.

The norm is the Euclidean operator norm; the list is finite and may be empty, yielding duration zero and matrix A0⊗B0A_0\otimes B_0A0​⊗B0​. The protocol and phase may depend on ε\varepsilonε. Both infima use the convention inf⁡∅=+∞\inf\varnothing=+\inftyinf∅=+∞, and neither infimum is required to be attained. There is no separate restriction on HHH or on the matrices satisfying PPP.

Human review
  • Endorsed by Shuze Chen · Nov 1, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Nov 1, 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 paper
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me