Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Definitions for Quantum Illumination with Gaussian States

Definition
QAlgorithms_Papers_Independent_Intakenext100_P85

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

prove2quantum

The objects the formalized statements of Quantum Illumination with Gaussian States (arXiv:0810.0534) are stated in. Each was checked against the paper by a blind read-back and a moderator.

  • Occ: Full occupation strings, source pp.2–3; n counts modes, not a cutoff.
  • Array: Coefficient arrays on a complete number basis.
  • IlluminationOperator: Generic bounded operators on that completed space.
  • basis: The complete occupation basis, with conjugate-linear first inner argument.
  • HasArray: All coefficients, not merely diagonal elements, pin an operator.
  • realizeOccupationArray: Unique bounded realization when it exists. Every constructed-state target concludes HasArray and density, ruling out the zero fallback.
  • JointDecisionEffect: A bounded binary effect decides H1. Both E and 1-E are positive, including self-adjointness. Source p.2, optimum joint quantum reception.
  • optimumError: Equal-prior error, independently optimized over every joint binary effect. The extended nonnegative infimum never uses a real empty-infimum default.
  • signalS: Eq.(2): S=2 NS+1.
  • correlationQ: Eq.(2): Cq=2 sqrt(NS(NS+1)).
  • correlationC: Classical positive-P correlation limit, p.2.
  • bathB: Eq.(3): B=2 NB+1.
  • returnA: Eq.(4): received variance A; the received background is NB.
  • phaseSensitiveWignerCov: Standard-form interleaved covariance, with source vacuum variance 1/4.
  • spdcCov: Eq.(2), constructed covariance matrix.
  • returnCov: Eqs.(3)–(4), the two explicit returned covariance matrices.
  • SourceGaussian: Actual Gaussian operator characteristic, in source normalization. The accepted predicate includes normalized positive trace class and symmetry.
  • CenteredMoments: Symmetric, finite-energy second moments of arbitrary states, not just Gaussians. With accepted Weyl coordinates x=2(q,p), the directional second characteristic derivative is -4 xi^T Lambda xi. All trace sums converge. Symmetry is explicit, repairing the known defect of frozen P47 moments.
  • spdcCoefficient: Eq.(1), all SPDC number coefficients, with NS=0 giving |0,0>.
  • spdcVector: The constructed full SPDC vector. Its realization and unit norm are conclusions, rather than Gaussianity or covariance fields.
  • spdcArray: Pure density from the constructed vector (1).
  • marginalPhotonNumber: Marginal mean photon number; every use concludes summability.
  • coherentVector: Full coherent vector, with coefficient and norm conclusions below.
  • coherentArray: Exact coherent projector coefficients, independent of a fallback vector.
  • PositiveP: Genuine positive P probability measure, including singular distributions at the classicality threshold. It is stronger than separability.
  • twoOcc: Relabel the two occupation numbers without changing the Hilbert space.
  • pairArray: Signal/idler pair arrays for invoking accepted complete separability.
  • signalChannel: The actual single signal channel. H0 discards input and inserts theta(NB); H1 uses the thermal beam splitter with bath occupation NB/(1-k). No characteristic-function or output-moment law is assumed here.
  • coordinateChannel: One named signal coordinate is sent through the channel; off-diagonal matrix slices are retained. Idler registers are never sent through this map.
  • signalBlock: Independent channel uses on every signal mode, with no IID assumption on the input. The empty list is the identity; result nodes require M>=1.
  • targetBlock: The signal block is transmitted and the entire idler is perfectly retained.
  • onePair: Explicit relabeling of a two-mode density as a one-signal/one-idler pair.
  • fromOnePair: Inverse relabeling; no physical operation is involved.
  • qiReturn: Actual returned SPDC pair, applying the physical channel to (1).
  • coherentReturn: Exact returned coherent states for any multimode amplitude vector.
  • constantAmplitude: Source coherent energy NS in each transmitted mode.
  • coherentExponent: Eq.(11): exact Chernoff exponent coefficient, not an optimal error definition.
  • coherentLower: Eq.(13) at total transmitted energy E, exact at every finite parameter.
  • returnDiscriminant: Eq.(16): positive square-root discriminant on the physical domain.
  • williamsonX: Eq.(16): x+ and x-, with positive and negative choices respectively.
  • williamsonT: Eq.(15): X+=diag(x+,x+), X-=diag(x-,-x-).
  • returnSpectrum: Eq.(17), k_source=1,2 corresponding to Fin 2 values 0,1.
  • ClassicalDensity: Eq.(19): genuine Lebesgue P density with ensemble average signal energy. There is no IID, peak-energy, per-realization, or idler-energy constraint.
  • classicalArray: The actual correlated coherent-state mixture in (19). Coefficient integrability and density are conclusions of its target, not hypotheses.
  • genieIntegrand: Genie error when the complete signal amplitude is revealed; this is independent all-effect optimal reception conditioned on that classical datum.
  • genieError: Ensemble mean of the genie experiment, p.4.
  • BlockRealized: Guard all single-mode bounded realizations used by the physical block, including off-diagonal slices and intermediate arrays. These are conclusions.
  • TargetRealized: Every retained-idler matrix block has its actual channel realization.
Definition code
import Mathlib
import Definitions.Def_QAlgorithms_Papers_Independent_Claimed50_P30
import Definitions.Def_QAlgorithms_Papers_Independent_Claimed50_P39
import Definitions.Def_QAlgorithms_Papers_Independent_Dependency50_P33
import Definitions.Def_QAlgorithms_Papers_Independent_Claimed50_P19
import Definitions.Def_QAlgorithms_Papers_Independent_Renew46_P41

/-!
Statements from Tan et al., arXiv:0810.0534v1, PDF pp.1–4.
All oscillator number spaces are completed, without photon cutoff. Source
quadratures are half the accepted P30 quadratures: means and covariances are
converted by u=2d and V=4 Lambda. No defective P47 moment predicate is used.
All local definitions have bodies. Five source-ambiguity nodes are in HARD.md.
Compilation and independent moderation are pending the batch handoff.
-/

noncomputable section

open MeasureTheory Filter
open scoped BigOperators ENNReal Topology

attribute [local instance] Classical.propDecidable Classical.decEq

namespace QAlgorithms.Papers.Independent.Intakenext100.P85

open QAlgorithms.Papers.Independent.Claimed50.P19 (Cov)
open QAlgorithms.Papers.Independent.Renew46.P41 (coherentProductCoefficient)

namespace G
export QAlgorithms.Papers.Independent.Claimed50.P30
  (ModeHilbert ModeOperator Quadrature PhaseVector numberKet numberBasis
   IsModeDensity GaussianMoments FockWeylSystem TraceSummable modeWeylTrace
   canonicalOmega Symplectic repeatedDiagonal chernoffTrace quantumChernoff)
end G

namespace B
export QAlgorithms.Papers.Independent.Dependency50.P39
  (coherentCoefficient fockKet HasFockMatrix operatorOfMatrix thermalMatrix)
export QAlgorithms.Papers.Independent.Claimed50.P39
  (thermalOutputMatrix thermalProbability beamAmplitude)
end B

namespace T
export QAlgorithms.Papers.Independent.Dependency50.P33
  (IsDensity IsSeparable IsBarycenter rankOne tensorApply)
end T

/-- Full occupation strings, source pp.2–3; n counts modes, not a cutoff. -/
abbrev Occ (n : ℕ) := Fin n → ℕ

/-- Coefficient arrays on a complete number basis. -/
abbrev Array (I : Type) := Matrix I I ℂ

/-- Generic bounded operators on that completed space. -/
abbrev IlluminationOperator (I : Type) := QAlgorithms.RealRegisters.Hilbert I →L[ℂ] QAlgorithms.RealRegisters.Hilbert I

/-- The complete occupation basis, with conjugate-linear first inner argument. -/
def basis (I : Type) : HilbertBasis I ℂ (QAlgorithms.RealRegisters.Hilbert I) :=
  ⟨LinearIsometryEquiv.refl ℂ (QAlgorithms.RealRegisters.Hilbert I)⟩

/-- All coefficients, not merely diagonal elements, pin an operator. -/
def HasArray {I : Type} (A : IlluminationOperator I) (X : Array I) : Prop :=
  ∀ i j, A (basis I j) i = X i j

/-- Unique bounded realization when it exists. Every constructed-state target
concludes HasArray and density, ruling out the zero fallback. -/
def realizeOccupationArray {I : Type} (X : Array I) : IlluminationOperator I :=
  if h : ∃ A : IlluminationOperator I, HasArray A X then Classical.choose h else 0

/-- A bounded binary effect decides H1. Both E and 1-E are positive,
including self-adjointness. Source p.2, optimum joint quantum reception. -/
def JointDecisionEffect {I : Type} (E : IlluminationOperator I) : Prop :=
  QAlgorithms.Papers.Independent.Dependency50.P13.IsPositive E ∧
    QAlgorithms.Papers.Independent.Dependency50.P13.IsPositive (1 - E)

/-- Equal-prior error, independently optimized over every joint binary effect.
The extended nonnegative infimum never uses a real empty-infimum default. -/
def optimumError {I : Type} (X0 X1 : Array I) : ℝ≥0∞ :=
  ⨅ E : {E : IlluminationOperator I // JointDecisionEffect E}, ENNReal.ofReal
    (((QAlgorithms.Papers.Independent.Dependency50.P13.operatorTrace (basis I) (E.val.comp (realizeOccupationArray X0))).re +
      (QAlgorithms.Papers.Independent.Dependency50.P13.operatorTrace (basis I) ((1 - E.val).comp (realizeOccupationArray X1))).re) / 2)

/-- Eq.(2): S=2 NS+1. -/
def signalS (NS : ℝ) : ℝ := 2 * NS + 1

/-- Eq.(2): Cq=2 sqrt(NS(NS+1)). -/
def correlationQ (NS : ℝ) : ℝ := 2 * Real.sqrt (NS * (NS + 1))

/-- Classical positive-P correlation limit, p.2. -/
def correlationC (NS : ℝ) : ℝ := 2 * NS

/-- Eq.(3): B=2 NB+1. -/
def bathB (NB : ℝ) : ℝ := 2 * NB + 1

/-- Eq.(4): received variance A; the received background is NB. -/
def returnA (k NS NB : ℝ) : ℝ := 2 * k * NS + bathB NB

/-- Standard-form interleaved covariance, with source vacuum variance 1/4. -/
def phaseSensitiveWignerCov (a b c : ℝ) : Cov 2 := fun i j =>
  (if i = j then (if i.1 = 0 then a else b)
   else if i.1 ≠ j.1 ∧ i.2 = j.2 then (if i.2 = 0 then c else -c)
   else 0) / 4

/-- Eq.(2), constructed covariance matrix. -/
def spdcCov (NS : ℝ) : Cov 2 :=
  phaseSensitiveWignerCov (signalS NS) (signalS NS) (correlationQ NS)

/-- Eqs.(3)–(4), the two explicit returned covariance matrices. -/
def returnCov (present : Bool) (k NS NB : ℝ) : Cov 2 :=
  if present then phaseSensitiveWignerCov (returnA k NS NB) (signalS NS)
    (Real.sqrt k * correlationQ NS)
  else phaseSensitiveWignerCov (bathB NB) (signalS NS) 0

/-- Actual Gaussian operator characteristic, in source normalization.
The accepted predicate includes normalized positive trace class and symmetry. -/
def SourceGaussian {n : ℕ} (W : G.FockWeylSystem n) (X : Array (Occ n))
    (d : G.PhaseVector n) (V : Cov n) : Prop :=
  HasArray (realizeOccupationArray X) X ∧ G.GaussianMoments W (realizeOccupationArray X) (2 • d) (4 • V)

/-- Symmetric, finite-energy second moments of arbitrary states, not just
Gaussians. With accepted Weyl coordinates x=2(q,p), the directional second
characteristic derivative is -4 xi^T Lambda xi. All trace sums converge.
Symmetry is explicit, repairing the known defect of frozen P47 moments. -/
def CenteredMoments {n : ℕ} (W : G.FockWeylSystem n) (X : Array (Occ n))
    (V : Cov n) : Prop :=
  HasArray (realizeOccupationArray X) X ∧ G.IsModeDensity (realizeOccupationArray X) ∧ V.transpose = V ∧
  Summable (fun j : Occ n => (∑ k, (j k : ℝ)) * (X j j).re) ∧
  (∀ xi, G.TraceSummable ((realizeOccupationArray X).comp (W.operator xi))) ∧
  ∀ xi : G.PhaseVector n,
    HasDerivAt (fun t : ℝ => G.modeWeylTrace W (realizeOccupationArray X) (t • xi)) 0 0 ∧
    HasDerivAt (deriv (fun t : ℝ => G.modeWeylTrace W (realizeOccupationArray X) (t • xi)))
      (((-4 * ∑ i, ∑ j, xi i * V i j * xi j : ℝ) : ℂ)) 0

/-- Eq.(1), all SPDC number coefficients, with NS=0 giving |0,0>. -/
def spdcCoefficient (NS : ℝ) (j : Occ 2) : ℂ :=
  if j 0 = j 1 then (Real.sqrt (NS ^ (j 0) / (NS + 1) ^ (j 0 + 1)) : ℂ)
  else 0

/-- The constructed full SPDC vector. Its realization and unit norm are
conclusions, rather than Gaussianity or covariance fields. -/
def spdcVector (NS : ℝ) : G.ModeHilbert 2 :=
  if h : ∃ v : G.ModeHilbert 2, ∀ j, v j = spdcCoefficient NS j then
    Classical.choose h else 0

/-- Pure density from the constructed vector (1). -/
def spdcArray (NS : ℝ) : Array (Occ 2) := T.rankOne (spdcVector NS)

/-- Marginal mean photon number; every use concludes summability. -/
def marginalPhotonNumber {n : ℕ} (X : Array (Occ n)) (k : Fin n) : ℝ :=
  ∑' j : Occ n, (j k : ℝ) * (X j j).re

/-- Full coherent vector, with coefficient and norm conclusions below. -/
def coherentVector {n : ℕ} (alpha : Fin n → ℂ) : G.ModeHilbert n :=
  if h : ∃ v : G.ModeHilbert n, ∀ j, v j = coherentProductCoefficient alpha j then
    Classical.choose h else 0

/-- Exact coherent projector coefficients, independent of a fallback vector. -/
def coherentArray {n : ℕ} (alpha : Fin n → ℂ) : Array (Occ n) :=
  fun i j => coherentProductCoefficient alpha i * star (coherentProductCoefficient alpha j)

/-- Genuine positive P probability measure, including singular distributions
at the classicality threshold. It is stronger than separability. -/
def PositiveP {n : ℕ} (X : Array (Occ n)) : Prop :=
  T.IsDensity X ∧ ∃ mu : Measure (Fin n → ℂ), IsProbabilityMeasure mu ∧
    T.IsBarycenter mu (fun alpha => coherentArray alpha) X

/-- Relabel the two occupation numbers without changing the Hilbert space. -/
def twoOcc (a b : ℕ) : Occ 2 := fun k => if k = 0 then a else b

/-- Signal/idler pair arrays for invoking accepted complete separability. -/
def pairArray (X : Array (Occ 2)) : Array (ℕ × ℕ) :=
  fun p q => X (twoOcc p.1 p.2) (twoOcc q.1 q.2)

/-- The actual single signal channel. H0 discards input and inserts theta(NB);
H1 uses the thermal beam splitter with bath occupation NB/(1-k).
No characteristic-function or output-moment law is assumed here. -/
def signalChannel (present : Bool) (k NB : ℝ) (X : Array ℕ) : Array ℕ :=
  if present then B.thermalOutputMatrix k (NB / (1 - k)) (B.operatorOfMatrix X)
  else fun i j => B.thermalMatrix (bathB NB) i j * (∑' m : ℕ, X m m)

/-- One named signal coordinate is sent through the channel; off-diagonal
matrix slices are retained. Idler registers are never sent through this map. -/
def coordinateChannel {M : ℕ} (present : Bool) (k NB : ℝ) (r : Fin M)
    (X : Array (Occ M)) : Array (Occ M) := fun i j =>
  signalChannel present k NB
    (fun a b => X (Function.update i r a) (Function.update j r b)) (i r) (j r)

/-- Independent channel uses on every signal mode, with no IID assumption
on the input. The empty list is the identity; result nodes require M>=1. -/
def signalBlock (M : ℕ) (present : Bool) (k NB : ℝ)
    (X : Array (Occ M)) : Array (Occ M) :=
  (List.ofFn (fun r : Fin M => r)).foldl
    (fun Y r => coordinateChannel present k NB r Y) X

/-- The signal block is transmitted and the entire idler is perfectly retained. -/
def targetBlock (M : ℕ) (present : Bool) (k NB : ℝ)
    (X : Array (Occ M × Occ M)) : Array (Occ M × Occ M) :=
  T.tensorApply (signalBlock M present k NB) X

/-- Explicit relabeling of a two-mode density as a one-signal/one-idler pair. -/
def onePair (X : Array (Occ 2)) : Array (Occ 1 × Occ 1) :=
  fun p q => X (twoOcc (p.1 0) (p.2 0)) (twoOcc (q.1 0) (q.2 0))

/-- Inverse relabeling; no physical operation is involved. -/
def fromOnePair (X : Array (Occ 1 × Occ 1)) : Array (Occ 2) :=
  fun p q => X ((fun _ => p 0), (fun _ => p 1))
    ((fun _ => q 0), (fun _ => q 1))

/-- Actual returned SPDC pair, applying the physical channel to (1). -/
def qiReturn (present : Bool) (k NS NB : ℝ) : Array (Occ 2) :=
  fromOnePair (targetBlock 1 present k NB (onePair (spdcArray NS)))

/-- Exact returned coherent states for any multimode amplitude vector. -/
def coherentReturn {M : ℕ} (present : Bool) (k NB : ℝ)
    (alpha : Fin M → ℂ) : Array (Occ M) :=
  signalBlock M present k NB (coherentArray alpha)

/-- Source coherent energy NS in each transmitted mode. -/
def constantAmplitude (M : ℕ) (NS : ℝ) : Fin M → ℂ :=
  fun _ => (Real.sqrt NS : ℂ)

/-- Eq.(11): exact Chernoff exponent coefficient, not an optimal error definition. -/
def coherentExponent (k NS NB : ℝ) : ℝ :=
  k * NS * (Real.sqrt (NB + 1) - Real.sqrt NB) ^ 2

/-- Eq.(13) at total transmitted energy E, exact at every finite parameter. -/
def coherentLower (k E NB : ℝ) : ℝ :=
  (1 - Real.sqrt (1 - Real.exp (-2 * coherentExponent k E NB))) / 2

/-- Eq.(16): positive square-root discriminant on the physical domain. -/
def returnDiscriminant (k NS NB : ℝ) : ℝ :=
  Real.sqrt ((returnA k NS NB + signalS NS) ^ 2 - 4 * k * correlationQ NS ^ 2)

/-- Eq.(16): x+ and x-, with positive and negative choices respectively. -/
def williamsonX (plus : Bool) (k NS NB : ℝ) : ℝ :=
  Real.sqrt ((returnA k NS NB + signalS NS +
    (if plus then returnDiscriminant k NS NB else -returnDiscriminant k NS NB)) / (2 * returnDiscriminant k NS NB))

/-- Eq.(15): X+=diag(x+,x+), X-=diag(x-,-x-). -/
def williamsonT (k NS NB : ℝ) : Cov 2 := fun i j =>
  if i.2 ≠ j.2 then 0 else if i.1 = j.1 then williamsonX true k NS NB
  else if i.2 = 0 then williamsonX false k NS NB else -williamsonX false k NS NB

/-- Eq.(17), k_source=1,2 corresponding to Fin 2 values 0,1. -/
def returnSpectrum (present : Bool) (k NS NB : ℝ) : Fin 2 → ℝ := fun r =>
  if present then (((-1 : ℝ) ^ (r.val + 1)) * (signalS NS - returnA k NS NB) +
    returnDiscriminant k NS NB) / 8
  else if r = 0 then bathB NB / 4 else signalS NS / 4

/-- Eq.(19): genuine Lebesgue P density with ensemble average signal energy.
There is no IID, peak-energy, per-realization, or idler-energy constraint. -/
def ClassicalDensity (M : ℕ) (NS : ℝ) (P : ((Fin M → ℂ) × (Fin M → ℂ)) → ℝ) : Prop :=
  Measurable P ∧ (∀ a, 0 ≤ P a) ∧ Integrable P ∧ (∫ a, P a) = 1 ∧
  Integrable (fun a => P a * ∑ r, Complex.normSq (a.1 r)) ∧
    (∫ a, P a * ∑ r, Complex.normSq (a.1 r)) = (M : ℝ) * NS

/-- The actual correlated coherent-state mixture in (19). Coefficient
integrability and density are conclusions of its target, not hypotheses. -/
def classicalArray (M : ℕ) (P : ((Fin M → ℂ) × (Fin M → ℂ)) → ℝ) :
    Array (Occ M × Occ M) := fun i j => ∫ a,
  (P a : ℂ) * coherentArray a.1 i.1 j.1 * coherentArray a.2 i.2 j.2

/-- Genie error when the complete signal amplitude is revealed; this is
independent all-effect optimal reception conditioned on that classical datum. -/
def genieIntegrand (M : ℕ) (k NB : ℝ)
    (P : ((Fin M → ℂ) × (Fin M → ℂ)) → ℝ)
    (a : (Fin M → ℂ) × (Fin M → ℂ)) : ℝ :=
  P a * (optimumError (coherentReturn false k NB a.1)
    (coherentReturn true k NB a.1)).toReal

/-- Ensemble mean of the genie experiment, p.4. -/
def genieError (M : ℕ) (k NB : ℝ)
    (P : ((Fin M → ℂ) × (Fin M → ℂ)) → ℝ) : ℝ :=
  ∫ a, genieIntegrand M k NB P a

/-! Sanity examples pinning normalization, signs, endpoints and register maps. -/

example : signalS 0 = 1 ∧ correlationQ 0 = 0 ∧ correlationC 0 = 0 := by
  norm_num [signalS, correlationQ, correlationC]

example (NS : ℝ) : spdcCoefficient NS (twoOcc 0 1) = 0 := by
  simp [spdcCoefficient, twoOcc]

example (a b : ℕ) : twoOcc a b 0 = a ∧ twoOcc a b 1 = b := by
  simp [twoOcc]

example (NS : ℝ) : spdcCov NS (0, 0) (1, 0) = correlationQ NS / 4 := by
  simp [spdcCov, phaseSensitiveWignerCov]

example (NS : ℝ) : spdcCov NS (0, 1) (1, 1) = -correlationQ NS / 4 := by
  simp [spdcCov, phaseSensitiveWignerCov]

example {n : ℕ} (alpha : Fin n → ℂ) (i j : Occ n) :
    coherentArray alpha i j = coherentProductCoefficient alpha i * star (coherentProductCoefficient alpha j) := rfl

example (k NB : ℝ) (X : Array (Occ 0)) : signalBlock 0 true k NB X = X := by
  simp [signalBlock]

example (k NB : ℝ) : coherentExponent k 0 NB = 0 := by
  simp [coherentExponent]

example (k NB : ℝ) : coherentLower k 0 NB = 1 / 2 := by
  norm_num [coherentLower, coherentExponent]

example (k NS NB : ℝ) : returnSpectrum false k NS NB 0 = bathB NB / 4 := by
  simp [returnSpectrum]

example (k NS NB : ℝ) : williamsonT k NS NB (0, 1) (1, 1) =
    -williamsonX false k NS NB := by simp [williamsonT]

/-- Guard all single-mode bounded realizations used by the physical block,
including off-diagonal slices and intermediate arrays. These are conclusions. -/
def BlockRealized (M : ℕ) (present : Bool) (k NB : ℝ) (X : Array (Occ M)) : Prop :=
  ∀ t : Fin (M + 1),
    let Y := ((List.ofFn (fun r : Fin M => r)).take t.val).foldl
      (fun Z r => coordinateChannel present k NB r Z) X
    ∀ (r : Fin M) (i j : Occ M),
      let slice := fun a b => Y (Function.update i r a) (Function.update j r b)
      B.HasFockMatrix (B.operatorOfMatrix slice) slice

/-- Every retained-idler matrix block has its actual channel realization. -/
def TargetRealized (M : ℕ) (present : Bool) (k NB : ℝ)
    (X : Array (Occ M × Occ M)) : Prop :=
  ∀ a b : Occ M, BlockRealized M present k NB (fun i j => X (i, a) (j, b))

end QAlgorithms.Papers.Independent.Intakenext100.P85
Source
Quantum Illumination with Gaussian States, arXiv:0810.0534v1, https://arxiv.org/abs/0810.0534v1
Read-back

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

QAlgorithms.Papers.Independent.Intakenext100.P85.Array

For every type III, Array⁡(I)\operatorname{Array}(I)Array(I) is the type of complex matrices indexed by III in both coordinates, equivalently all functions I×I→CI\times I\to\mathbb CI×I→C. The index type may be empty or infinite; no boundedness, summability, positivity, or normalization is required.

QAlgorithms.Papers.Independent.Intakenext100.P85.BlockRealized

For every natural number MMM, Boolean present\mathrm{present}present, real numbers k,NBk,N_Bk,NB​, and arbitrary complex array XXX indexed by NM×NM\mathbb N^M\times\mathbb N^MNM×NM, this predicate imposes the following condition. Define a coordinate update Cr\mathcal C_rCr​ by

(CrY)(i,j)=Spresent((Y(i[r←b],j[r←c]))b,c∈N)(ir,jr),(\mathcal C_rY)(i,j)= \mathcal S_{\mathrm{present}} \bigl((Y(i[r\leftarrow b],j[r\leftarrow c]))_{b,c\in\mathbb N}\bigr)(i_r,j_r),(Cr​Y)(i,j)=Spresent​((Y(i[r←b],j[r←c]))b,c∈N​)(ir​,jr​),

and set Y0=XY_0=XY0​=X and Yt+1=CtYtY_{t+1}=\mathcal C_tY_tYt+1​=Ct​Yt​ for 0≤t<M0\le t<M0≤t<M. For every t∈{0,…,M}t\in\{0,\ldots,M\}t∈{0,…,M}, every r∈{0,…,M−1}r\in\{0,\ldots,M-1\}r∈{0,…,M−1}, every i,j∈NMi,j\in\mathbb N^Mi,j∈NM, and all b,c∈Nb,c\in\mathbb Nb,c∈N, the selected realization of the slice

Zbc=Yt(i[r←b],j[r←c])Z_{bc}=Y_t(i[r\leftarrow b],j[r\leftarrow c])Zbc​=Yt​(i[r←b],j[r←c])

must have exactly those matrix entries. More explicitly, for any array ZZZ on N×N\mathbb N\times\mathbb NN×N, its selected realization R(Z)R(Z)R(Z) is a chosen bounded operator on ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C) satisfying (R(Z)ec)b=Zbc(R(Z)e_c)_b=Z_{bc}(R(Z)ec​)b​=Zbc​ whenever such an operator exists, and is zero otherwise; the required condition is (R(Z)ec)b=Zbc(R(Z)e_c)_b=Z_{bc}(R(Z)ec​)b​=Zbc​ for every b,cb,cb,c. The transformations used to form the intermediate arrays are

Sfalse(Z)(m,n)={22NB+2(2NB2NB+2)n∑b′Zbb,m=n,0,m≠n,\mathcal S_{\mathrm{false}}(Z)(m,n)= \begin{cases} \displaystyle\frac{2}{2N_B+2} \left(\frac{2N_B}{2N_B+2}\right)^n\sum'_b Z_{bb},&m=n,\\ 0,&m\ne n, \end{cases}Sfalse​(Z)(m,n)=⎩⎨⎧​2NB​+22​(2NB​+22NB​​)nb∑′​Zbb​,0,​m=n,m=n,​

and, writing ν=NB/(1−k)\nu=N_B/(1-k)ν=NB​/(1−k),

Strue(Z)(m,n)=∑(h,ℓ)∈N2′{(ν/(ν+1))hν+1 bk(m,ℓ,m+ℓ−h,h) bk(n,ℓ,n+ℓ−h,h) (R(Z)en+ℓ−h)m+ℓ−h,h≤m+ℓ, h≤n+ℓ,0,otherwise,\mathcal S_{\mathrm{true}}(Z)(m,n)= \sum'_{(h,\ell)\in\mathbb N^2} \begin{cases} \displaystyle \frac{(\nu/(\nu+1))^h}{\nu+1}\, b_k(m,\ell,m+\ell-h,h)\, b_k(n,\ell,n+\ell-h,h)\, (R(Z)e_{n+\ell-h})_{m+\ell-h}, &h\le m+\ell,\ h\le n+\ell,\\ 0,&\text{otherwise}, \end{cases}Strue​(Z)(m,n)=(h,ℓ)∈N2∑′​⎩⎨⎧​ν+1(ν/(ν+1))h​bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)(R(Z)en+ℓ−h​)m+ℓ−h​,0,​h≤m+ℓ, h≤n+ℓ,otherwise,​

where

bη(m,ℓ,p,q)={m!ℓ!p!q!∑u=0p{(pu)(qm−u)(η)u+q−(m−u)(−1−η)p−u(1−η)m−u,u≤m, m−u≤q,0,otherwise,m+ℓ=p+q,0,otherwise.b_\eta(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{u=0}^{p} \begin{cases} \displaystyle \binom pu\binom q{m-u} (\sqrt\eta)^{u+q-(m-u)} (-\sqrt{1-\eta})^{p-u} (\sqrt{1-\eta})^{m-u}, &u\le m,\ m-u\le q,\\ 0,&\text{otherwise}, \end{cases} &m+\ell=p+q,\\ 0,&\text{otherwise}. \end{cases}bη​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​u=0∑p​⎩⎨⎧​(up​)(m−uq​)(η​)u+q−(m−u)(−1−η​)p−u(1−η​)m−u,0,​u≤m, m−u≤q,otherwise,​0,​m+ℓ=p+q,otherwise.​

The condition includes slices of the initial array and of the array after all MMM updates. It requires no density, positivity, trace, or parameter-range assumptions. Natural-number subtraction is truncated at zero, division by zero and square roots of negative real numbers give zero, and nonsummable infinite sums give zero. When M=0M=0M=0, there is no coordinate rrr, so the predicate holds vacuously for every X,k,NB,presentX,k,N_B,\mathrm{present}X,k,NB​,present.

QAlgorithms.Papers.Independent.Intakenext100.P85.CenteredMoments

For every natural number nnn, including zero, let J=N{0,…,n−1}J=\mathbb N^{\{0,\ldots,n-1\}}J=N{0,…,n−1}, H=ℓ2(J;C)\mathcal H=\ell^2(J;\mathbb C)H=ℓ2(J;C), and Q={0,…,n−1}×{0,1}Q=\{0,\ldots,n-1\}\times\{0,1\}Q={0,…,n−1}×{0,1}, with coordinate basis eje_jej​. The arguments are a supplied family of bounded complex-linear unitary operators W(ξ)W(\xi)W(ξ) on H\mathcal HH, a complex array XXX indexed by JJJ, and a real matrix VVV indexed by QQQ. The supplied family has coefficients (W(ξ)ej)i=∏kD(−ξk,1+iξk,0;ik,jk)(W(\xi)e_j)_i=\prod_k D(-\xi_{k,1}+\mathrm i\xi_{k,0};i_k,j_k)(W(ξ)ej​)i​=∏k​D(−ξk,1​+iξk,0​;ik​,jk​), where

D(z;m,l)=e−∣z∣2/2m! l!∑h=0min⁡(m,l)zm−h(−z‾)l−hh!(m−h)!(l−h)!.D(z;m,l)=e^{-|z|^2/2}\sqrt{m!\,l!} \sum_{h=0}^{\min(m,l)} \frac{z^{m-h}(-\overline z)^{l-h}}{h!(m-h)!(l-h)!}.D(z;m,l)=e−∣z∣2/2m!l!​h=0∑min(m,l)​h!(m−h)!(l−h)!zm−h(−z)l−h​.

Let RXR_XRX​ be a chosen bounded complex-linear operator with coefficients XXX, if one exists, and zero otherwise, and write tr⁡(T)=∑j∈J⟨ej,Tej⟩\operatorname{tr}(T)=\sum_{j\in J}\langle e_j,Te_j\rangletr(T)=∑j∈J​⟨ej​,Tej​⟩. This predicate requires (RXej)i=Xij(R_Xe_j)_i=X_{ij}(RX​ej​)i​=Xij​ for all i,ji,ji,j; sequences ur,vr∈Hu_r,v_r\in\mathcal Hur​,vr​∈H with ∑r∥ur∥∥vr∥<∞\sum_r\|u_r\|\|v_r\|<\infty∑r​∥ur​∥∥vr​∥<∞ and RXx=∑r⟨vr,x⟩urR_Xx=\sum_r\langle v_r,x\rangle u_rRX​x=∑r​⟨vr​,x⟩ur​ for every xxx; self-adjointness of RXR_XRX​ and Re⁡⟨x,RXx⟩≥0\operatorname{Re}\langle x,R_Xx\rangle\ge0Re⟨x,RX​x⟩≥0 for every xxx; summability of its diagonal trace series and tr⁡(RX)=1\operatorname{tr}(R_X)=1tr(RX​)=1; symmetry VT=VV^\mathsf T=VVT=V; summability of

j⟼(∑k=0n−1jk)Re⁡Xjj;j\longmapsto\left(\sum_{k=0}^{n-1}j_k\right)\operatorname{Re}X_{jj};j⟼(k=0∑n−1​jk​)ReXjj​;

and summability of the diagonal trace series of RXW(ξ)R_XW(\xi)RX​W(ξ) for every ξ∈RQ\xi\in\mathbb R^Qξ∈RQ. Finally, for each such ξ\xiξ, the complex-valued real function fξ(t)=tr⁡(RXW(tξ))f_\xi(t)=\operatorname{tr}(R_XW(t\xi))fξ​(t)=tr(RX​W(tξ)) must have derivative zero at t=0t=0t=0, and its derivative function must have derivative

−4∑a,b∈QξaVabξb-4\sum_{a,b\in Q}\xi_aV_{ab}\xi_b−4a,b∈Q∑​ξa​Vab​ξb​

at t=0t=0t=0, viewed as a complex number. Here the derivative function is the total derivative operation, which returns zero at points where a derivative does not exist; the two stated derivative-existence conditions are nevertheless required. For n=0n=0n=0, the finite sums are empty.

QAlgorithms.Papers.Independent.Intakenext100.P85.ClassicalDensity

For every natural number MMM, real number NSN_SNS​, and real-valued function PPP on CM×CM\mathbb C^M\times\mathbb C^MCM×CM, this predicate means that PPP is measurable, is nonnegative at every point, is integrable with respect to Lebesgue measure on that product space, has integral one, and has the following integrable weighted first-register energy with the specified integral:

∫CM×CMP(α,β) d(α,β)=1,∫CM×CMP(α,β)∑r=0M−1∣αr∣2 d(α,β)=MNS.\int_{\mathbb C^M\times\mathbb C^M}P(\alpha,\beta)\,d(\alpha,\beta)=1,\qquad \int_{\mathbb C^M\times\mathbb C^M} P(\alpha,\beta)\sum_{r=0}^{M-1}|\alpha_r|^2\,d(\alpha,\beta) =MN_S.∫CM×CM​P(α,β)d(α,β)=1,∫CM×CM​P(α,β)r=0∑M−1​∣αr​∣2d(α,β)=MNS​.

The weighted function in the second integral must itself be integrable. No analogous energy condition is imposed on β\betaβ. The definition permits M=0M=0M=0: then the domain is a singleton, both energy sums are zero, and the predicate reduces to PPP having value one at that point, with no restriction on NSN_SNS​. For M>0M>0M>0, a negative NSN_SNS​ cannot satisfy the stated nonnegative-energy equality.

QAlgorithms.Papers.Independent.Intakenext100.P85.HasArray

For any type III, continuous complex-linear operator AAA on ℓ2(I;C)\ell^2(I;\mathbb C)ℓ2(I;C), and complex array X:I×I→CX:I\times I\to\mathbb CX:I×I→C, this predicate means that (Aej)(i)=X(i,j)(Ae_j)(i)=X(i,j)(Aej​)(i)=X(i,j) for every i,j∈Ii,j\in Ii,j∈I, where eje_jej​ is the coordinate vector supported at jjj with value 111. For empty III, the condition is vacuous.

QAlgorithms.Papers.Independent.Intakenext100.P85.IlluminationOperator

For every type III, an illumination operator is a continuous complex-linear map from ℓ2(I;C)\ell^2(I;\mathbb C)ℓ2(I;C) to itself, where ℓ2(I;C)\ell^2(I;\mathbb C)ℓ2(I;C) is the Hilbert space of square-summable complex families indexed by III. Empty and infinite index types are allowed.

QAlgorithms.Papers.Independent.Intakenext100.P85.JointDecisionEffect

For any type III and continuous complex-linear operator EEE on ℓ2(I;C)\ell^2(I;\mathbb C)ℓ2(I;C), this predicate means that both EEE and 1−E1-E1−E, with 111 the identity operator, are self-adjoint and have nonnegative real quadratic forms: E∗=EE^*=EE∗=E, (1−E)∗=1−E(1-E)^*=1-E(1−E)∗=1−E, and for every vector xxx, Re⁡⟨x,Ex⟩≥0\operatorname{Re}\langle x,Ex\rangle\ge0Re⟨x,Ex⟩≥0 and Re⁡⟨x,(1−E)x⟩≥0\operatorname{Re}\langle x,(1-E)x\rangle\ge0Re⟨x,(1−E)x⟩≥0.

QAlgorithms.Papers.Independent.Intakenext100.P85.Occ

For every natural number nnn, Occ⁡(n)\operatorname{Occ}(n)Occ(n) is the type of functions {0,…,n−1}→N\{0,\ldots,n-1\}\to\mathbb N{0,…,n−1}→N, assigning a natural number to each of nnn positions. The case n=0n=0n=0 is included and has exactly one such function, the empty function.

QAlgorithms.Papers.Independent.Intakenext100.P85.PositiveP

For every natural number nnn, including n=0n=0n=0, and complex array XXX indexed by occupation vectors i,j:{0,…,n−1}→Ni,j:\{0,\ldots,n-1\}\to\mathbb Ni,j:{0,…,n−1}→N, this predicate requires that XXX be a density array and that there exist a probability measure μ\muμ on Cn\mathbb C^nCn representing XXX entry by entry as a mixture of coherent arrays. Here being a density array means that there are sequences ur,vr∈ℓ2(Nn;C)u_r,v_r\in\ell^2(\mathbb N^n;\mathbb C)ur​,vr​∈ℓ2(Nn;C) with ∑r∥ur∥∥vr∥<∞\sum_r\|u_r\|\|v_r\|<\infty∑r​∥ur​∥∥vr​∥<∞ and Xij=∑rur(i)vr(j)‾X_{ij}=\sum_r u_r(i)\overline{v_r(j)}Xij​=∑r​ur​(i)vr​(j)​; every finite quadratic form ∑i,j∈szi‾Xijzj\sum_{i,j\in s}\overline{z_i}X_{ij}z_j∑i,j∈s​zi​​Xij​zj​, for every finite set sss of occupation vectors and every complex function zzz, is real and nonnegative; and the real diagonal entries form a summable family with sum 111. The required probability measure satisfies, for every i,ji,ji,j, integrability of the following integrand and the equality

Xij=∫Cncα(i)cα(j)‾ dμ(α),cα(i)=∏r=0n−1e−∣αr∣2/2αr irir!.X_{ij}=\int_{\mathbb C^n}c_\alpha(i)\overline{c_\alpha(j)}\,d\mu(\alpha), \qquad c_\alpha(i)=\prod_{r=0}^{n-1} \frac{e^{-|\alpha_r|^2/2}\alpha_r^{\,i_r}}{\sqrt{i_r!}}.Xij​=∫Cn​cα​(i)cα​(j)​dμ(α),cα​(i)=r=0∏n−1​ir​!​e−∣αr​∣2/2αrir​​​.

When n=0n=0n=0, the occupation index and amplitude spaces each have one element, and the empty product defining cαc_\alphacα​ equals 111.

QAlgorithms.Papers.Independent.Intakenext100.P85.SourceGaussian

For every natural number nnn, including zero, put J=N{0,…,n−1}J=\mathbb N^{\{0,\ldots,n-1\}}J=N{0,…,n−1}, H=ℓ2(J;C)\mathcal H=\ell^2(J;\mathbb C)H=ℓ2(J;C), and Q={0,…,n−1}×{0,1}Q=\{0,\ldots,n-1\}\times\{0,1\}Q={0,…,n−1}×{0,1}, with coordinate basis vectors eje_jej​. The arguments are a supplied family W(ξ)W(\xi)W(ξ) of bounded complex-linear unitary operators on H\mathcal HH, a complex array X=(Xij)i,j∈JX=(X_{ij})_{i,j\in J}X=(Xij​)i,j∈J​, a real vector d∈RQd\in\mathbb R^Qd∈RQ, and a real matrix VVV indexed by QQQ. The family WWW comes equipped with the coefficient identities (W(ξ)ej)i=∏kD(−ξk,1+iξk,0;ik,jk)(W(\xi)e_j)_i=\prod_k D(-\xi_{k,1}+\mathrm i\xi_{k,0};i_k,j_k)(W(ξ)ej​)i​=∏k​D(−ξk,1​+iξk,0​;ik​,jk​), where

D(z;m,l)=e−∣z∣2/2m! l!∑h=0min⁡(m,l)zm−h(−z‾)l−hh!(m−h)!(l−h)!.D(z;m,l)=e^{-|z|^2/2}\sqrt{m!\,l!} \sum_{h=0}^{\min(m,l)} \frac{z^{m-h}(-\overline z)^{l-h}}{h!(m-h)!(l-h)!}.D(z;m,l)=e−∣z∣2/2m!l!​h=0∑min(m,l)​h!(m−h)!(l−h)!zm−h(−z)l−h​.

Let RXR_XRX​ be a chosen bounded complex-linear operator satisfying (RXej)i=Xij(R_Xe_j)_i=X_{ij}(RX​ej​)i​=Xij​, if one exists, and the zero operator otherwise; define tr⁡(T)=∑j∈J⟨ej,Tej⟩\operatorname{tr}(T)=\sum_{j\in J}\langle e_j,Te_j\rangletr(T)=∑j∈J​⟨ej​,Tej​⟩. This predicate means that RXR_XRX​ actually satisfies all these coefficient identities; that there exist sequences ur,vr∈Hu_r,v_r\in\mathcal Hur​,vr​∈H with ∑r∥ur∥∥vr∥<∞\sum_r\|u_r\|\|v_r\|<\infty∑r​∥ur​∥∥vr​∥<∞ and RXx=∑r⟨vr,x⟩urR_Xx=\sum_r\langle v_r,x\rangle u_rRX​x=∑r​⟨vr​,x⟩ur​ for every x∈Hx\in\mathcal Hx∈H; that RX∗=RXR_X^*=R_XRX∗​=RX​, Re⁡⟨x,RXx⟩≥0\operatorname{Re}\langle x,R_Xx\rangle\ge0Re⟨x,RX​x⟩≥0 for every xxx, its diagonal trace series is summable, and tr⁡(RX)=1\operatorname{tr}(R_X)=1tr(RX​)=1; that 4V4V4V is symmetric and positive definite and its complexification plus iΩ\mathrm i\OmegaiΩ is positive semidefinite, where Ω(k,0),(k,1)=1\Omega_{(k,0),(k,1)}=1Ω(k,0),(k,1)​=1, Ω(k,1),(k,0)=−1\Omega_{(k,1),(k,0)}=-1Ω(k,1),(k,0)​=−1, and all other entries are zero; and that, for every ξ∈RQ\xi\in\mathbb R^Qξ∈RQ, the diagonal trace series of RXW(ξ)R_XW(\xi)RX​W(ξ) is summable and

tr⁡(RXW(ξ))=exp⁡ ⁣(−2∑a,b∈QξaVabξb+2i∑a∈Qdaξa).\operatorname{tr}(R_XW(\xi)) =\exp\!\left(-2\sum_{a,b\in Q}\xi_aV_{ab}\xi_b +2\mathrm i\sum_{a\in Q}d_a\xi_a\right).tr(RX​W(ξ))=exp​−2a,b∈Q∑​ξa​Vab​ξb​+2ia∈Q∑​da​ξa​​.

All conditions remain part of the definition when n=0n=0n=0, with empty finite sums and products interpreted as zero and one.

QAlgorithms.Papers.Independent.Intakenext100.P85.TargetRealized

For every natural number MMM, Boolean present\mathrm{present}present, real numbers k,NBk,N_Bk,NB​, and complex array XXX indexed by (NM×NM)×(NM×NM)(\mathbb N^M\times\mathbb N^M)\times(\mathbb N^M\times\mathbb N^M)(NM×NM)×(NM×NM), this predicate requires the following for every pair of fixed occupation vectors a,b∈NMa,b\in\mathbb N^Ma,b∈NM. Begin with the array Y0(i,j)=X((i,a),(j,b))Y_0(i,j)=X((i,a),(j,b))Y0​(i,j)=X((i,a),(j,b)), and for t=0,…,M−1t=0,\ldots,M-1t=0,…,M−1 define

Yt+1(i,j)=Spresent((Yt(i[t←u],j[t←v]))u,v∈N)(it,jt).Y_{t+1}(i,j)= \mathcal S_{\mathrm{present}} \bigl((Y_t(i[t\leftarrow u],j[t\leftarrow v]))_{u,v\in\mathbb N}\bigr)(i_t,j_t).Yt+1​(i,j)=Spresent​((Yt​(i[t←u],j[t←v]))u,v∈N​)(it​,jt​).

For every t∈{0,…,M}t\in\{0,\ldots,M\}t∈{0,…,M}, every coordinate r<Mr<Mr<M, and every i,j∈NMi,j\in\mathbb N^Mi,j∈NM, the slice Zuv=Yt(i[r←u],j[r←v])Z_{uv}=Y_t(i[r\leftarrow u],j[r\leftarrow v])Zuv​=Yt​(i[r←u],j[r←v]) must satisfy (R(Z)ev)u=Zuv(R(Z)e_v)_u=Z_{uv}(R(Z)ev​)u​=Zuv​ for all u,v∈Nu,v\in\mathbb Nu,v∈N. Here R(Z)R(Z)R(Z) is a chosen bounded operator on ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C) having all matrix entries ZuvZ_{uv}Zuv​ if such an operator exists, and otherwise is zero. The transformations in the recursion are

Sfalse(Z)(m,n)={22NB+2(2NB2NB+2)n∑u′Zuu,m=n,0,m≠n,\mathcal S_{\mathrm{false}}(Z)(m,n)= \begin{cases} \displaystyle\frac{2}{2N_B+2} \left(\frac{2N_B}{2N_B+2}\right)^n\sum'_u Z_{uu},&m=n,\\ 0,&m\ne n, \end{cases}Sfalse​(Z)(m,n)=⎩⎨⎧​2NB​+22​(2NB​+22NB​​)nu∑′​Zuu​,0,​m=n,m=n,​

and, setting ν=NB/(1−k)\nu=N_B/(1-k)ν=NB​/(1−k),

Strue(Z)(m,n)=∑(h,ℓ)∈N2′{(ν/(ν+1))hν+1 bk(m,ℓ,m+ℓ−h,h) bk(n,ℓ,n+ℓ−h,h) (R(Z)en+ℓ−h)m+ℓ−h,h≤m+ℓ, h≤n+ℓ,0,otherwise,\mathcal S_{\mathrm{true}}(Z)(m,n)= \sum'_{(h,\ell)\in\mathbb N^2} \begin{cases} \displaystyle \frac{(\nu/(\nu+1))^h}{\nu+1}\, b_k(m,\ell,m+\ell-h,h)\, b_k(n,\ell,n+\ell-h,h)\, (R(Z)e_{n+\ell-h})_{m+\ell-h}, &h\le m+\ell,\ h\le n+\ell,\\ 0,&\text{otherwise}, \end{cases}Strue​(Z)(m,n)=(h,ℓ)∈N2∑′​⎩⎨⎧​ν+1(ν/(ν+1))h​bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)(R(Z)en+ℓ−h​)m+ℓ−h​,0,​h≤m+ℓ, h≤n+ℓ,otherwise,​

with

bη(m,ℓ,p,q)={m!ℓ!p!q!∑u=0p{(pu)(qm−u)(η)u+q−(m−u)(−1−η)p−u(1−η)m−u,u≤m, m−u≤q,0,otherwise,m+ℓ=p+q,0,otherwise.b_\eta(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{u=0}^{p} \begin{cases} \displaystyle \binom pu\binom q{m-u} (\sqrt\eta)^{u+q-(m-u)} (-\sqrt{1-\eta})^{p-u} (\sqrt{1-\eta})^{m-u}, &u\le m,\ m-u\le q,\\ 0,&\text{otherwise}, \end{cases} &m+\ell=p+q,\\ 0,&\text{otherwise}. \end{cases}bη​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​u=0∑p​⎩⎨⎧​(up​)(m−uq​)(η​)u+q−(m−u)(−1−η​)p−u(1−η​)m−u,0,​u≤m, m−u≤q,otherwise,​0,​m+ℓ=p+q,otherwise.​

Thus the requirement ranges over every fixed second-register row and column, including unequal a,ba,ba,b, and every indicated first-register slice before, during, and after the updates. No density, positivity, summability, or parameter-range assumptions are imposed separately. Natural-number subtraction is truncated at zero; division by zero, square roots of negative real numbers, and nonsummable infinite sums use the total conventions giving zero. For M=0M=0M=0, the absence of a coordinate rrr makes the predicate vacuously true for every array XXX.

QAlgorithms.Papers.Independent.Intakenext100.P85.basis

For every type III, this defines the canonical Hilbert basis (ej)j∈I(e_j)_{j\in I}(ej​)j∈I​ of ℓ2(I;C)\ell^2(I;\mathbb C)ℓ2(I;C), whose coordinate representation is the identity: ej(i)=1e_j(i)=1ej​(i)=1 when i=ji=ji=j and 000 otherwise. For empty III, this is the empty Hilbert basis of the zero space.

QAlgorithms.Papers.Independent.Intakenext100.P85.bathB

For every real number NBN_BNB​, this defines B(NB)=2NB+1B(N_B)=2N_B+1B(NB​)=2NB​+1, with no nonnegativity restriction on NBN_BNB​.

QAlgorithms.Papers.Independent.Intakenext100.P85.classicalArray

For every natural number MMM and arbitrary real-valued function PPP on CM×CM\mathbb C^M\times\mathbb C^MCM×CM, this defines a complex matrix whose row and column indices are pairs of occupation vectors in NM×NM\mathbb N^M\times\mathbb N^MNM×NM. For α∈CM\alpha\in\mathbb C^Mα∈CM and u∈NMu\in\mathbb N^Mu∈NM, write

cα(u)=∏r=0M−1e−∣αr∣2/2αrurur!.c_\alpha(u)=\prod_{r=0}^{M-1} \frac{e^{-|\alpha_r|^2/2}\alpha_r^{u_r}}{\sqrt{u_r!}}.cα​(u)=r=0∏M−1​ur​!​e−∣αr​∣2/2αrur​​​.

The entry at row (u,v)(u,v)(u,v) and column (u′,v′)(u',v')(u′,v′) is

∫CM×CMP(α,β) cα(u)cα(u′)‾ cβ(v)cβ(v′)‾ d(α,β).\int_{\mathbb C^M\times\mathbb C^M} P(\alpha,\beta)\, c_\alpha(u)\overline{c_\alpha(u')}\, c_\beta(v)\overline{c_\beta(v')}\,d(\alpha,\beta).∫CM×CM​P(α,β)cα​(u)cα​(u′)​cβ​(v)cβ​(v′)​d(α,β).

The integral is the complex Bochner integral with respect to Lebesgue measure. No measurability, integrability, nonnegativity, normalization, or energy condition on PPP is assumed; a nonintegrable entry integrand has integral zero under the total integral convention. When M=0M=0M=0, all products are empty products equal to one, and the matrix has a single entry, namely ∫P\int P∫P.

QAlgorithms.Papers.Independent.Intakenext100.P85.coherentArray

For every natural number nnn, including zero, and every function α:{0,…,n−1}→C\alpha:\{0,\ldots,n-1\}\to\mathbb Cα:{0,…,n−1}→C, this defines a complex array indexed by occupation functions i,j:{0,…,n−1}→Ni,j:\{0,\ldots,n-1\}\to\mathbb Ni,j:{0,…,n−1}→N through

Xij=(∏k=0n−1e−∣αk∣2/2αkikik!)(∏k=0n−1e−∣αk∣2/2αkjkjk!)‾.X_{ij} = \left(\prod_{k=0}^{n-1} \frac{e^{-|\alpha_k|^2/2}\alpha_k^{i_k}}{\sqrt{i_k!}}\right) \overline{\left(\prod_{k=0}^{n-1} \frac{e^{-|\alpha_k|^2/2}\alpha_k^{j_k}}{\sqrt{j_k!}}\right)}.Xij​=(k=0∏n−1​ik​!​e−∣αk​∣2/2αkik​​​)(k=0∏n−1​jk​!​e−∣αk​∣2/2αkjk​​​)​.

The definition imposes no additional hypotheses and makes no normalization assertion. When n=0n=0n=0, both products are empty and the array has its sole entry equal to one.

QAlgorithms.Papers.Independent.Intakenext100.P85.coherentExponent

For arbitrary real numbers k,NS,NBk,N_S,N_Bk,NS​,NB​, this defines the real number

kNS(NB+1−NB)2.kN_S\bigl(\sqrt{N_B+1}-\sqrt{N_B}\bigr)^2.kNS​(NB​+1​−NB​​)2.

There are no sign or range restrictions on these arguments. Every real square root here is defined to be zero when its argument is negative; in particular, this expression is zero whenever NB≤−1N_B\le -1NB​≤−1, or whenever kNS=0kN_S=0kNS​=0.

QAlgorithms.Papers.Independent.Intakenext100.P85.coherentLower

For arbitrary real numbers k,E,NBk,E,N_Bk,E,NB​, this defines

1− 1−exp⁡ ⁣(−2kE(NB+1−NB)2)2.\frac{1-\sqrt{\,1-\exp\!\left(-2kE\bigl(\sqrt{N_B+1}-\sqrt{N_B}\bigr)^2\right)}}{2}.21−1−exp(−2kE(NB​+1​−NB​​)2)​​.

No positivity, normalization, or parameter-range hypotheses are imposed. Real square roots of negative numbers are zero, so the definition remains meaningful even when the argument of the outer square root is negative; in that case its value is 1/21/21/2.

QAlgorithms.Papers.Independent.Intakenext100.P85.coherentReturn

For every natural number MMM, Boolean present\mathrm{present}present, real numbers k,NBk,N_Bk,NB​, and amplitude vector α∈CM\alpha\in\mathbb C^Mα∈CM, start with the complex array

Xij(0)=cα(i)cα(j)‾,cα(i)=∏r=0M−1e−∣αr∣2/2αr irir!.X^{(0)}_{ij}=c_\alpha(i)\overline{c_\alpha(j)}, \qquad c_\alpha(i)=\prod_{r=0}^{M-1} \frac{e^{-|\alpha_r|^2/2}\alpha_r^{\,i_r}}{\sqrt{i_r!}}.Xij(0)​=cα​(i)cα​(j)​,cα​(i)=r=0∏M−1​ir​!​e−∣αr​∣2/2αrir​​​.

This definition successively transforms coordinates r=0,…,M−1r=0,\ldots,M-1r=0,…,M−1: for each i,ji,ji,j, form Yab=Xi[r:=a],j[r:=b](r)Y_{ab}=X^{(r)}_{i[r:=a],j[r:=b]}Yab​=Xi[r:=a],j[r:=b](r)​, and set Xij(r+1)=C(Y)irjrX^{(r+1)}_{ij}=\mathcal C(Y)_{i_rj_r}Xij(r+1)​=C(Y)ir​jr​​; the result is X(M)X^{(M)}X(M). Explicitly, let tL(h)=(L/(L+1))h/(L+1)t_L(h)=(L/(L+1))^h/(L+1)tL​(h)=(L/(L+1))h/(L+1), let Y^=Y\widehat Y=YY=Y if it admits a bounded complex-linear operator realization on ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C) and zero otherwise, and define

bk(m,ℓ,p,q)={m!ℓ!p!q!∑0≤s≤ps≤m, m−s≤q(ps)(qm−s)(k)s+q−(m−s)(−1−k)p−s(1−k)m−s,m+ℓ=p+q,0,otherwise.b_k(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{\substack{0\le s\le p\\s\le m,\ m-s\le q}} \binom ps\binom q{m-s} (\sqrt{k})^{s+q-(m-s)} (-\sqrt{1-k})^{p-s} (\sqrt{1-k})^{m-s}, &m+\ell=p+q,\\ 0,&\text{otherwise}. \end{cases}bk​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​0≤s≤ps≤m, m−s≤q​∑​(sp​)(m−sq​)(k​)s+q−(m−s)(−1−k​)p−s(1−k​)m−s,0,​m+ℓ=p+q,otherwise.​

For true present\mathrm{present}present,

C(Y)mn=∑(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ′tNB/(1−k)(h)bk(m,ℓ,m+ℓ−h,h)bk(n,ℓ,n+ℓ−h,h)Y^m+ℓ−h,n+ℓ−h;\mathcal C(Y)_{mn}= \sum_{\substack{(h,\ell)\in\mathbb N^2\\h\le m+\ell,\ h\le n+\ell}}^{\prime} t_{N_B/(1-k)}(h) b_k(m,\ell,m+\ell-h,h)b_k(n,\ell,n+\ell-h,h) \widehat Y_{m+\ell-h,n+\ell-h};C(Y)mn​=(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ​∑′​tNB​/(1−k)​(h)bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)Ym+ℓ−h,n+ℓ−h​;

for false present\mathrm{present}present, C(Y)mn=1m=ntNB(n)∑a′Yaa\mathcal C(Y)_{mn}=\mathbf1_{m=n}t_{N_B}(n)\sum_a^{\prime}Y_{aa}C(Y)mn​=1m=n​tNB​​(n)∑a′​Yaa​. No bounds on k,NBk,N_Bk,NB​ are assumed; division by zero, negative-argument real square roots, and nonsummable complex sums give zero. When M=0M=0M=0, there are no transformations and the resulting one-entry array has entry 111.

QAlgorithms.Papers.Independent.Intakenext100.P85.coherentVector

For every natural number nnn, including zero, and every function α:{0,…,n−1}→C\alpha:\{0,\ldots,n-1\}\to\mathbb Cα:{0,…,n−1}→C, this defines a vector in ℓ2(N{0,…,n−1};C)\ell^2(\mathbb N^{\{0,\ldots,n-1\}};\mathbb C)ℓ2(N{0,…,n−1};C) by choosing a vector with coefficients

vj=∏k=0n−1e−∣αk∣2/2αkjkjk!v_j=\prod_{k=0}^{n-1} \frac{e^{-|\alpha_k|^2/2}\alpha_k^{j_k}}{\sqrt{j_k!}}vj​=k=0∏n−1​jk​!​e−∣αk​∣2/2αkjk​​​

for every occupation function jjj, if such a square-summable vector exists, and returning zero otherwise. This definition itself asserts neither existence nor normalization. When n=0n=0n=0, there is one occupation function and its prescribed coefficient is the empty product, one.

QAlgorithms.Papers.Independent.Intakenext100.P85.constantAmplitude

For every natural number MMM and every real number NSN_SNS​, this defines the complex amplitude vector of length MMM with every coordinate equal to the real square root of NSN_SNS​, regarded as a complex number:

αr=NS(0≤r<M).\alpha_r=\sqrt{N_S}\qquad(0\le r<M).αr​=NS​​(0≤r<M).

When M=0M=0M=0, this is the unique empty vector; when NS≤0N_S\le0NS​≤0, every coordinate is zero under the total real-square-root convention.

QAlgorithms.Papers.Independent.Intakenext100.P85.coordinateChannel

For every natural number MMM, Boolean present\mathrm{present}present, real numbers k,NBk,N_Bk,NB​, coordinate r∈{0,…,M−1}r\in\{0,\ldots,M-1\}r∈{0,…,M−1}, and complex array XXX indexed by MMM-coordinate occupation vectors, the output entry at i,ji,ji,j is obtained by forming the one-coordinate slice Yab=Xi[r:=a],j[r:=b]Y_{ab}=X_{i[r:=a],j[r:=b]}Yab​=Xi[r:=a],j[r:=b]​, applying the following rule to YYY, and selecting entry (ir,jr)(i_r,j_r)(ir​,jr​). Let Y^=Y\widehat Y=YY=Y if YYY is the matrix of some bounded complex-linear operator on ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C), and Y^=0\widehat Y=0Y=0 otherwise; set tL(h)=(L/(L+1))h/(L+1)t_L(h)=(L/(L+1))^h/(L+1)tL​(h)=(L/(L+1))h/(L+1), and define

bk(m,ℓ,p,q)={m!ℓ!p!q!∑0≤s≤ps≤m, m−s≤q(ps)(qm−s)(k)s+q−(m−s)(−1−k)p−s(1−k)m−s,m+ℓ=p+q,0,otherwise.b_k(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{\substack{0\le s\le p\\s\le m,\ m-s\le q}} \binom ps\binom q{m-s} (\sqrt{k})^{s+q-(m-s)} (-\sqrt{1-k})^{p-s} (\sqrt{1-k})^{m-s}, &m+\ell=p+q,\\ 0,&\text{otherwise}. \end{cases}bk​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​0≤s≤ps≤m, m−s≤q​∑​(sp​)(m−sq​)(k​)s+q−(m−s)(−1−k​)p−s(1−k​)m−s,0,​m+ℓ=p+q,otherwise.​

For true present\mathrm{present}present, the rule gives

C(Y)mn=∑(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ′tNB/(1−k)(h) bk(m,ℓ,m+ℓ−h,h) bk(n,ℓ,n+ℓ−h,h) Y^m+ℓ−h,n+ℓ−h;\mathcal C(Y)_{mn}= \sum_{\substack{(h,\ell)\in\mathbb N^2\\h\le m+\ell,\ h\le n+\ell}}^{\prime} t_{N_B/(1-k)}(h)\, b_k(m,\ell,m+\ell-h,h)\, b_k(n,\ell,n+\ell-h,h)\, \widehat Y_{m+\ell-h,n+\ell-h};C(Y)mn​=(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ​∑′​tNB​/(1−k)​(h)bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)Ym+ℓ−h,n+ℓ−h​;

for false present\mathrm{present}present, it gives C(Y)mn=1m=ntNB(n)∑a′Yaa\mathcal C(Y)_{mn}=\mathbf1_{m=n}t_{N_B}(n)\sum_a^{\prime}Y_{aa}C(Y)mn​=1m=n​tNB​​(n)∑a′​Yaa​. Thus the defined entry is C(Y)irjr\mathcal C(Y)_{i_rj_r}C(Y)ir​jr​​. There are no assumptions on X,k,NBX,k,N_BX,k,NB​; zero denominators and nonsummable complex sums are assigned zero, and negative real square-root arguments give zero. The coordinate argument cannot be supplied when M=0M=0M=0.

QAlgorithms.Papers.Independent.Intakenext100.P85.correlationC

For every real number NSN_SNS​, this defines C(NS)=2NSC(N_S)=2N_SC(NS​)=2NS​, with no nonnegativity restriction on NSN_SNS​.

QAlgorithms.Papers.Independent.Intakenext100.P85.correlationQ

For every real number NSN_SNS​, this defines Q(NS)=2NS(NS+1)Q(N_S)=2\sqrt{N_S(N_S+1)}Q(NS​)=2NS​(NS​+1)​, using the real square root, which equals 000 for negative arguments. In particular, the value is 000 for −1≤NS≤0-1\le N_S\le0−1≤NS​≤0; no restriction on NSN_SNS​ is assumed.

QAlgorithms.Papers.Independent.Intakenext100.P85.fromOnePair

For any complex array XXX indexed by pairs of one-coordinate occupation vectors, this defines an array indexed by two-coordinate occupation vectors p,qp,qp,q by

fromOnePair⁡(X)p,q=X(p0‾,p1‾),(q0‾,q1‾),\operatorname{fromOnePair}(X)_{p,q} =X_{(\underline{p_0},\underline{p_1}),(\underline{q_0},\underline{q_1})},fromOnePair(X)p,q​=X(p0​​,p1​​),(q0​​,q1​​)​,

where a‾\underline aa​ denotes the one-coordinate vector whose sole value is aaa. No condition is imposed on XXX.

QAlgorithms.Papers.Independent.Intakenext100.P85.genieError

For every natural number MMM, arbitrary real numbers k,NBk,N_Bk,NB​, and arbitrary real-valued function PPP on CM×CM\mathbb C^M\times\mathbb C^MCM×CM, this defines the Lebesgue integral

∫CM×CMP(α,β) e(Z0α,Z1α) d(α,β),\int_{\mathbb C^M\times\mathbb C^M}P(\alpha,\beta)\,e(Z_0^\alpha,Z_1^\alpha)\,d(\alpha,\beta),∫CM×CM​P(α,β)e(Z0α​,Z1α​)d(α,β),

with the following explicit construction. For a countable index set III, let RI(Z)R_I(Z)RI​(Z) be a chosen bounded operator on ℓ2(I;C)\ell^2(I;\mathbb C)ℓ2(I;C) with matrix entries Zij=(RI(Z)ej)iZ_{ij}=(R_I(Z)e_j)_iZij​=(RI​(Z)ej​)i​, if one exists, and zero otherwise, and put τI(T)=∑i′⟨ei,Tei⟩\tau_I(T)=\sum'_i\langle e_i,Te_i\rangleτI​(T)=∑i′​⟨ei​,Tei​⟩. Define e(Z0,Z1)e(Z_0,Z_1)e(Z0​,Z1​) by taking the infimum, in the extended nonnegative reals, over bounded operators EEE for which both EEE and I−EI-EI−E are self-adjoint and have nonnegative real quadratic forms, of

max⁡{0, Re⁡τI(ERI(Z0))+Re⁡τI((I−E)RI(Z1))2},\max\left\{0,\, \frac{\operatorname{Re}\tau_I(ER_I(Z_0))+ \operatorname{Re}\tau_I((I-E)R_I(Z_1))}{2}\right\},max{0,2ReτI​(ERI​(Z0​))+ReτI​((I−E)RI​(Z1​))​},

and then converting that infimum to a real number, with +∞+\infty+∞ converted to zero. Here I=NMI=\mathbb N^MI=NM. Start both arrays ZhαZ_h^\alphaZhα​ with

Cα(i,j)=∏r=0M−1e−∣αr∣2/2αririr! e−∣αr∣2/2αrjrjr!‾,C_\alpha(i,j)= \prod_{r=0}^{M-1} \frac{e^{-|\alpha_r|^2/2}\alpha_r^{i_r}}{\sqrt{i_r!}}\, \overline{\frac{e^{-|\alpha_r|^2/2}\alpha_r^{j_r}}{\sqrt{j_r!}}},Cα​(i,j)=r=0∏M−1​ir​!​e−∣αr​∣2/2αrir​​​jr​!​e−∣αr​∣2/2αrjr​​​​,

and, successively for r=0,…,M−1r=0,\ldots,M-1r=0,…,M−1, replace an array YYY by the array with entry Sh(Z)(ir,jr)\mathcal S_h(Z)(i_r,j_r)Sh​(Z)(ir​,jr​), where Zbc=Y(i[r←b],j[r←c])Z_{bc}=Y(i[r\leftarrow b],j[r\leftarrow c])Zbc​=Y(i[r←b],j[r←c]). These one-coordinate transformations are

S0(Z)(m,n)={22NB+2(2NB2NB+2)n∑b′Zbb,m=n,0,m≠n,\mathcal S_0(Z)(m,n)= \begin{cases} \displaystyle\frac{2}{2N_B+2} \left(\frac{2N_B}{2N_B+2}\right)^n\sum'_b Z_{bb},&m=n,\\ 0,&m\ne n, \end{cases}S0​(Z)(m,n)=⎩⎨⎧​2NB​+22​(2NB​+22NB​​)nb∑′​Zbb​,0,​m=n,m=n,​

and, with ν=NB/(1−k)\nu=N_B/(1-k)ν=NB​/(1−k) and A=RN(Z)A=R_{\mathbb N}(Z)A=RN​(Z),

S1(Z)(m,n)=∑(h,ℓ)∈N2′{(ν/(ν+1))hν+1 bk(m,ℓ,m+ℓ−h,h) bk(n,ℓ,n+ℓ−h,h) (Aen+ℓ−h)m+ℓ−h,h≤m+ℓ, h≤n+ℓ,0,otherwise,\mathcal S_1(Z)(m,n)= \sum'_{(h,\ell)\in\mathbb N^2} \begin{cases} \displaystyle \frac{(\nu/(\nu+1))^h}{\nu+1}\, b_k(m,\ell,m+\ell-h,h)\, b_k(n,\ell,n+\ell-h,h)\, (Ae_{n+\ell-h})_{m+\ell-h}, &h\le m+\ell,\ h\le n+\ell,\\ 0,&\text{otherwise}, \end{cases}S1​(Z)(m,n)=(h,ℓ)∈N2∑′​⎩⎨⎧​ν+1(ν/(ν+1))h​bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)(Aen+ℓ−h​)m+ℓ−h​,0,​h≤m+ℓ, h≤n+ℓ,otherwise,​

where

bη(m,ℓ,p,q)={m!ℓ!p!q!∑t=0p{(pt)(qm−t)(η)t+q−(m−t)(−1−η)p−t(1−η)m−t,t≤m, m−t≤q,0,otherwise,m+ℓ=p+q,0,otherwise.b_\eta(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{t=0}^{p} \begin{cases} \displaystyle \binom pt\binom q{m-t} (\sqrt\eta)^{t+q-(m-t)} (-\sqrt{1-\eta})^{p-t} (\sqrt{1-\eta})^{m-t}, &t\le m,\ m-t\le q,\\ 0,&\text{otherwise}, \end{cases} &m+\ell=p+q,\\ 0,&\text{otherwise}. \end{cases}bη​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​t=0∑p​⎩⎨⎧​(tp​)(m−tq​)(η​)t+q−(m−t)(−1−η​)p−t(1−η​)m−t,0,​t≤m, m−t≤q,otherwise,​0,​m+ℓ=p+q,otherwise.​

Natural-number subtraction is truncated at zero, division by zero is zero, and real square roots of negative arguments are zero. Infinite sums use their total convention, giving zero for nonsummable families. No assumptions on PPP, the parameters, or the arrays are required, and the outer real Bochner integral is zero if its integrand is not integrable. The vector β\betaβ affects the integrand only through PPP. When M=0M=0M=0, no coordinate updates occur.

QAlgorithms.Papers.Independent.Intakenext100.P85.genieIntegrand

For every natural number MMM, arbitrary real numbers k,NBk,N_Bk,NB​, arbitrary real-valued function PPP on CM×CM\mathbb C^M\times\mathbb C^MCM×CM, and point a=(α,β)a=(\alpha,\beta)a=(α,β) of that space, this defines P(α,β)e(Z0,Z1)P(\alpha,\beta)e(Z_0,Z_1)P(α,β)e(Z0​,Z1​), with the following meaning of every quantity. For any countable index set III, let RI(Z)R_I(Z)RI​(Z) be a chosen bounded complex-linear operator on ℓ2(I;C)\ell^2(I;\mathbb C)ℓ2(I;C) satisfying (RI(Z)ej)i=Zij(R_I(Z)e_j)_i=Z_{ij}(RI​(Z)ej​)i​=Zij​ for every standard basis vector eje_jej​, if such an operator exists, and otherwise let it be zero. For operators on this space put τI(T)=∑i′⟨ei,Tei⟩\tau_I(T)=\sum'_i\langle e_i,Te_i\rangleτI​(T)=∑i′​⟨ei​,Tei​⟩, using the total infinite-sum convention, which gives zero if the family is not summable, and define

e(Z0,Z1)=[inf⁡E=E∗,I−E=(I−E)∗Re⁡⟨x,Ex⟩≥0,Re⁡⟨x,(I−E)x⟩≥0 ∀x[Re⁡τI(ERI(Z0))+Re⁡τI((I−E)RI(Z1))2]+]R,e(Z_0,Z_1)= \left[ \inf_{\substack{E=E^*,\,I-E=(I-E)^*\\ \operatorname{Re}\langle x,Ex\rangle\ge0,\, \operatorname{Re}\langle x,(I-E)x\rangle\ge0\ \forall x}} \left[ \frac{\operatorname{Re}\tau_I(ER_I(Z_0))+ \operatorname{Re}\tau_I((I-E)R_I(Z_1))}{2} \right]_+ \right]_{\mathbb R},e(Z0​,Z1​)=​E=E∗,I−E=(I−E)∗Re⟨x,Ex⟩≥0,Re⟨x,(I−E)x⟩≥0 ∀x​inf​[2ReτI​(ERI​(Z0​))+ReτI​((I−E)RI​(Z1​))​]+​​R​,

where [r]+=max⁡(r,0)[r]_+=\max(r,0)[r]+​=max(r,0) is regarded as an extended nonnegative real, the infimum is taken there, and the final conversion to a real number sends +∞+\infty+∞ to zero. Here I=NMI=\mathbb N^MI=NM, and ZhZ_hZh​, for h∈{0,1}h\in\{0,1\}h∈{0,1}, is obtained by starting with

Cα(i,j)=∏r=0M−1e−∣αr∣2/2αririr! e−∣αr∣2/2αrjrjr!‾C_\alpha(i,j)= \prod_{r=0}^{M-1} \frac{e^{-|\alpha_r|^2/2}\alpha_r^{i_r}}{\sqrt{i_r!}}\, \overline{\frac{e^{-|\alpha_r|^2/2}\alpha_r^{j_r}}{\sqrt{j_r!}}}Cα​(i,j)=r=0∏M−1​ir​!​e−∣αr​∣2/2αrir​​​jr​!​e−∣αr​∣2/2αrjr​​​​

and updating coordinates r=0,…,M−1r=0,\ldots,M-1r=0,…,M−1 in that order: an array YYY is replaced by the array whose (i,j)(i,j)(i,j)-entry is Sh(Z)(ir,jr)\mathcal S_h(Z)(i_r,j_r)Sh​(Z)(ir​,jr​), where Zbc=Y(i[r←b],j[r←c])Z_{bc}=Y(i[r\leftarrow b],j[r\leftarrow c])Zbc​=Y(i[r←b],j[r←c]). The one-coordinate transformations are

S0(Z)(m,n)={22NB+2(2NB2NB+2)n∑b′Zbb,m=n,0,m≠n,\mathcal S_0(Z)(m,n)= \begin{cases} \displaystyle \frac{2}{2N_B+2}\left(\frac{2N_B}{2N_B+2}\right)^n \sum'_b Z_{bb},&m=n,\\ 0,&m\ne n, \end{cases}S0​(Z)(m,n)=⎩⎨⎧​2NB​+22​(2NB​+22NB​​)nb∑′​Zbb​,0,​m=n,m=n,​

and, writing ν=NB/(1−k)\nu=N_B/(1-k)ν=NB​/(1−k) and A=RN(Z)A=R_{\mathbb N}(Z)A=RN​(Z),

S1(Z)(m,n)=∑(h,ℓ)∈N2′{(ν/(ν+1))hν+1 bk(m,ℓ,m+ℓ−h,h) bk(n,ℓ,n+ℓ−h,h) (Aen+ℓ−h)m+ℓ−h,h≤m+ℓ, h≤n+ℓ,0,otherwise.\mathcal S_1(Z)(m,n)= \sum'_{(h,\ell)\in\mathbb N^2} \begin{cases} \displaystyle \frac{(\nu/(\nu+1))^h}{\nu+1}\, b_k(m,\ell,m+\ell-h,h)\, b_k(n,\ell,n+\ell-h,h)\, (Ae_{n+\ell-h})_{m+\ell-h}, &h\le m+\ell,\ h\le n+\ell,\\ 0,&\text{otherwise}. \end{cases}S1​(Z)(m,n)=(h,ℓ)∈N2∑′​⎩⎨⎧​ν+1(ν/(ν+1))h​bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)(Aen+ℓ−h​)m+ℓ−h​,0,​h≤m+ℓ, h≤n+ℓ,otherwise.​

The real coefficient used here is

bη(m,ℓ,p,q)={m!ℓ!p!q!∑t=0p{(pt)(qm−t)(η)t+q−(m−t)(−1−η)p−t(1−η)m−t,t≤m, m−t≤q,0,otherwise,m+ℓ=p+q,0,otherwise.b_\eta(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{t=0}^{p} \begin{cases} \displaystyle \binom pt\binom q{m-t} (\sqrt\eta)^{t+q-(m-t)} (-\sqrt{1-\eta})^{p-t} (\sqrt{1-\eta})^{m-t}, &t\le m,\ m-t\le q,\\ 0,&\text{otherwise}, \end{cases} &m+\ell=p+q,\\ 0,&\text{otherwise}. \end{cases}bη​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​t=0∑p​⎩⎨⎧​(tp​)(m−tq​)(η​)t+q−(m−t)(−1−η​)p−t(1−η​)m−t,0,​t≤m, m−t≤q,otherwise,​0,​m+ℓ=p+q,otherwise.​

All subtractions between natural-number indices are truncated at zero; real square roots of negative arguments and division by zero return zero, so these definitions also cover k=1k=1k=1 and unrestricted signed parameters. No density or integrability hypotheses are imposed. The second amplitude vector β\betaβ enters only through P(α,β)P(\alpha,\beta)P(α,β); for M=0M=0M=0, the coordinate-update list is empty.

QAlgorithms.Papers.Independent.Intakenext100.P85.marginalPhotonNumber

For every natural number nnn, every complex array XXX indexed by occupation functions j:{0,…,n−1}→Nj:\{0,\ldots,n-1\}\to\mathbb Nj:{0,…,n−1}→N, and every index k∈{0,…,n−1}k\in\{0,\ldots,n-1\}k∈{0,…,n−1}, this defines the real number

∑j∈N{0,…,n−1}jk Re⁡Xjj.\sum_{j\in\mathbb N^{\{0,\ldots,n-1\}}}j_k\,\operatorname{Re}X_{jj}.j∈N{0,…,n−1}∑​jk​ReXjj​.

No positivity, normalization, or summability assumption is imposed on XXX; the total infinite-sum operation returns zero if this family is not summable. When n=0n=0n=0, there is no possible argument kkk.

QAlgorithms.Papers.Independent.Intakenext100.P85.onePair

For any complex array XXX indexed by two-coordinate occupation vectors, this defines an array indexed by pairs of one-coordinate occupation vectors. For row index (u,v)(u,v)(u,v) and column index (u′,v′)(u',v')(u′,v′), its entry is

onePair⁡(X)(u,v),(u′,v′)=X(u0,v0),(u0′,v0′),\operatorname{onePair}(X)_{(u,v),(u',v')} =X_{(u_0,v_0),(u'_0,v'_0)},onePair(X)(u,v),(u′,v′)​=X(u0​,v0​),(u0′​,v0′​)​,

where each pair on the right denotes the two-coordinate occupation vector with the displayed coordinates. No condition is imposed on XXX.

QAlgorithms.Papers.Independent.Intakenext100.P85.optimumError

For any type III and arbitrary complex arrays X0,X1:I×I→CX_0,X_1:I\times I\to\mathbb CX0​,X1​:I×I→C, let RaR_aRa​, for a=0,1a=0,1a=0,1, be a chosen continuous complex-linear operator on ℓ2(I;C)\ell^2(I;\mathbb C)ℓ2(I;C) whose matrix in the canonical coordinate basis (ej)(e_j)(ej​) is XaX_aXa​, if such an operator exists, and the zero operator otherwise. This defines the following value in [0,∞][0,\infty][0,∞]:

inf⁡E∗=E, (1−E)∗=1−E∀x, Re⁡⟨x,Ex⟩≥0, Re⁡⟨x,(1−E)x⟩≥0max⁡ ⁣{0,Re⁡ ⁣(∑j∈I⟨ej,E(R0ej)⟩)+Re⁡ ⁣(∑j∈I⟨ej,(1−E)(R1ej)⟩)2},\inf_{\substack{E^*=E,\ (1-E)^*=1-E\\ \forall x,\ \operatorname{Re}\langle x,Ex\rangle\ge0,\ \operatorname{Re}\langle x,(1-E)x\rangle\ge0}} \max\!\left\{0,\frac{ \operatorname{Re}\!\left(\sum_{j\in I}\langle e_j,E(R_0e_j)\rangle\right) +\operatorname{Re}\!\left(\sum_{j\in I}\langle e_j,(1-E)(R_1e_j)\rangle\right)} {2}\right\},E∗=E, (1−E)∗=1−E∀x, Re⟨x,Ex⟩≥0, Re⟨x,(1−E)x⟩≥0​inf​max⎩⎨⎧​0,2Re(∑j∈I​⟨ej​,E(R0​ej​)⟩)+Re(∑j∈I​⟨ej​,(1−E)(R1​ej​)⟩)​⎭⎬⎫​,

where EEE ranges over continuous complex-linear operators and 111 is the identity. Each sum is the total Hilbert-space-indexed sum used in the definition, taking value 000 if its complex summands are not summable. No density, trace-class, realization, or trace-summability assumptions are imposed on the arrays. The effect set includes the zero operator, and empty III is allowed, in which case the value is 000.

QAlgorithms.Papers.Independent.Intakenext100.P85.pairArray

For any complex array XXX whose row and column indices are two-coordinate occupation vectors, this defines an array indexed by pairs of natural numbers by

pairArray⁡(X)(a,b),(c,d)=X(a,b),(c,d),\operatorname{pairArray}(X)_{(a,b),(c,d)}=X_{(a,b),(c,d)},pairArray(X)(a,b),(c,d)​=X(a,b),(c,d)​,

where each pair on the right is understood as the occupation vector taking those values at coordinates 000 and 111, respectively. No positivity, normalization, boundedness, or summability condition is imposed on XXX.

QAlgorithms.Papers.Independent.Intakenext100.P85.phaseSensitiveWignerCov

For arbitrary real numbers a,b,ca,b,ca,b,c, this defines a real matrix indexed by pairs (r,p)∈{0,1}×{0,1}(r,p)\in\{0,1\}\times\{0,1\}(r,p)∈{0,1}×{0,1}. In the index order (0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1), it is

14(a0c00a0−cc0b00−c0b).\frac14 \begin{pmatrix} a&0&c&0\\ 0&a&0&-c\\ c&0&b&0\\ 0&-c&0&b \end{pmatrix}.41​​a0c0​0a0−c​c0b0​0−c0b​​.

No positivity or other restrictions on the matrix are imposed.

QAlgorithms.Papers.Independent.Intakenext100.P85.qiReturn

For a Boolean present\mathrm{present}present and arbitrary real numbers k,NS,NBk,N_S,N_Bk,NS​,NB​, this defines a complex array on two-coordinate occupation vectors as follows. Let ψ∈ℓ2(N2;C)\psi\in\ell^2(\mathbb N^2;\mathbb C)ψ∈ℓ2(N2;C) have coefficients

ψa,b={NSa(NS+1)a+1,a=b,0,a≠b,\psi_{a,b}= \begin{cases} \displaystyle\sqrt{\frac{N_S^a}{(N_S+1)^{a+1}}},&a=b,\\ 0,&a\ne b, \end{cases}ψa,b​=⎩⎨⎧​(NS​+1)a+1NSa​​​,0,​a=b,a=b,​

if a vector with exactly these coefficients exists, and let ψ=0\psi=0ψ=0 otherwise. For output indices i,ji,ji,j, form Yab=ψa,i1ψb,j1‾Y_{ab}=\psi_{a,i_1}\overline{\psi_{b,j_1}}Yab​=ψa,i1​​ψb,j1​​​; the defined entry is C(Y)i0j0\mathcal C(Y)_{i_0j_0}C(Y)i0​j0​​. Here tL(h)=(L/(L+1))h/(L+1)t_L(h)=(L/(L+1))^h/(L+1)tL​(h)=(L/(L+1))h/(L+1), Y^=Y\widehat Y=YY=Y if it is the matrix of a bounded complex-linear operator on ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C) and zero otherwise, and

bk(m,ℓ,p,q)={m!ℓ!p!q!∑0≤s≤ps≤m, m−s≤q(ps)(qm−s)(k)s+q−(m−s)(−1−k)p−s(1−k)m−s,m+ℓ=p+q,0,otherwise.b_k(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{\substack{0\le s\le p\\s\le m,\ m-s\le q}} \binom ps\binom q{m-s} (\sqrt{k})^{s+q-(m-s)} (-\sqrt{1-k})^{p-s} (\sqrt{1-k})^{m-s}, &m+\ell=p+q,\\ 0,&\text{otherwise}. \end{cases}bk​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​0≤s≤ps≤m, m−s≤q​∑​(sp​)(m−sq​)(k​)s+q−(m−s)(−1−k​)p−s(1−k​)m−s,0,​m+ℓ=p+q,otherwise.​

For true present\mathrm{present}present,

C(Y)mn=∑(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ′tNB/(1−k)(h)bk(m,ℓ,m+ℓ−h,h)bk(n,ℓ,n+ℓ−h,h)Y^m+ℓ−h,n+ℓ−h;\mathcal C(Y)_{mn}= \sum_{\substack{(h,\ell)\in\mathbb N^2\\h\le m+\ell,\ h\le n+\ell}}^{\prime} t_{N_B/(1-k)}(h) b_k(m,\ell,m+\ell-h,h)b_k(n,\ell,n+\ell-h,h) \widehat Y_{m+\ell-h,n+\ell-h};C(Y)mn​=(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ​∑′​tNB​/(1−k)​(h)bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)Ym+ℓ−h,n+ℓ−h​;

for false present\mathrm{present}present, C(Y)mn=1m=ntNB(n)∑a′Yaa\mathcal C(Y)_{mn}=\mathbf1_{m=n}t_{N_B}(n)\sum_a^{\prime}Y_{aa}C(Y)mn​=1m=n​tNB​​(n)∑a′​Yaa​. No positivity or normalization conclusion is part of this definition. Division by zero and nonsummable complex sums give zero, and real square roots of negative arguments give zero.

QAlgorithms.Papers.Independent.Intakenext100.P85.realizeOccupationArray

For any type III and complex array X:I×I→CX:I\times I\to\mathbb CX:I×I→C, this defines a continuous complex-linear operator on ℓ2(I;C)\ell^2(I;\mathbb C)ℓ2(I;C) by choosing an operator AAA satisfying (Aej)(i)=X(i,j)(Ae_j)(i)=X(i,j)(Aej​)(i)=X(i,j) for all i,j∈Ii,j\in Ii,j∈I, if such an operator exists, where eje_jej​ is the canonical coordinate vector. If no such operator exists, the value is the zero operator. No hypothesis guarantees that the supplied array has an operator realization.

QAlgorithms.Papers.Independent.Intakenext100.P85.returnA

For arbitrary real numbers k,NS,NBk,N_S,N_Bk,NS​,NB​, this defines the real number

A(k,NS,NB)=2kNS+2NB+1.A(k,N_S,N_B)=2kN_S+2N_B+1.A(k,NS​,NB​)=2kNS​+2NB​+1.

There are no sign or range restrictions on these parameters.

QAlgorithms.Papers.Independent.Intakenext100.P85.returnCov

For any Boolean present\mathrm{present}present and arbitrary real numbers k,NS,NBk,N_S,N_Bk,NS​,NB​, this defines a real matrix indexed by (r,p)∈{0,1}2(r,p)\in\{0,1\}^2(r,p)∈{0,1}2. In the order (0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1), its value is

{14(2kNS+2NB+10c002kNS+2NB+10−cc02NS+100−c02NS+1),present=true,14diag⁡(2NB+1,2NB+1,2NS+1,2NS+1),present=false,c=2kNS(NS+1).\begin{cases} \displaystyle\frac14 \begin{pmatrix} 2kN_S+2N_B+1&0&c&0\\ 0&2kN_S+2N_B+1&0&-c\\ c&0&2N_S+1&0\\ 0&-c&0&2N_S+1 \end{pmatrix}, &\mathrm{present}=\mathrm{true},\\[1em] \displaystyle\frac14\operatorname{diag}(2N_B+1,2N_B+1,2N_S+1,2N_S+1), &\mathrm{present}=\mathrm{false}, \end{cases} \qquad c=2\sqrt{k}\sqrt{N_S(N_S+1)}.⎩⎨⎧​41​​2kNS​+2NB​+10c0​02kNS​+2NB​+10−c​c02NS​+10​0−c02NS​+1​​,41​diag(2NB​+1,2NB​+1,2NS​+1,2NS​+1),​present=true,present=false,​c=2k​NS​(NS​+1)​.

No parameter restrictions are imposed, and each real square root of a negative argument is zero.

QAlgorithms.Papers.Independent.Intakenext100.P85.returnDiscriminant

For arbitrary real numbers k,NS,NBk,N_S,N_Bk,NS​,NB​, this defines

D=(2kNS+2NB+1+2NS+1)2−4k(2NS(NS+1))2.D=\sqrt{\left(2kN_S+2N_B+1+2N_S+1\right)^2 -4k\left(2\sqrt{N_S(N_S+1)}\right)^2}.D=(2kNS​+2NB​+1+2NS​+1)2−4k(2NS​(NS​+1)​)2​.

There are no restrictions on the parameters. Both square roots are the total real square-root function, which returns zero on negative arguments; thus D=0D=0D=0 whenever the outer radicand is nonpositive.

QAlgorithms.Papers.Independent.Intakenext100.P85.returnSpectrum

For a Boolean value present\mathrm{present}present and arbitrary real numbers k,NS,NBk,N_S,N_Bk,NS​,NB​, put

A=2kNS+2NB+1,S=2NS+1,D=(A+S)2−4k(2NS(NS+1))2.A=2kN_S+2N_B+1,\qquad S=2N_S+1,\qquad D=\sqrt{(A+S)^2-4k\left(2\sqrt{N_S(N_S+1)}\right)^2}.A=2kNS​+2NB​+1,S=2NS​+1,D=(A+S)2−4k(2NS​(NS​+1)​)2​.

This defines a real-valued function on {0,1}\{0,1\}{0,1} whose ordered pair of values is

{(A−S+D8,S−A+D8),present=true,(2NB+14,2NS+14),present=false.\begin{cases} \displaystyle\left(\frac{A-S+D}{8},\frac{S-A+D}{8}\right), &\mathrm{present}=\mathrm{true},\\[7pt] \displaystyle\left(\frac{2N_B+1}{4},\frac{2N_S+1}{4}\right), &\mathrm{present}=\mathrm{false}. \end{cases}⎩⎨⎧​(8A−S+D​,8S−A+D​),(42NB​+1​,42NS​+1​),​present=true,present=false.​

There are no sign, ordering, or physicality hypotheses on these values. Each real square root is zero on negative arguments.

QAlgorithms.Papers.Independent.Intakenext100.P85.signalBlock

For every natural number MMM, Boolean present\mathrm{present}present, arbitrary real numbers k,NBk,N_Bk,NB​, and arbitrary complex array XXX indexed by MMM-coordinate occupation vectors, this defines a successive transformation of coordinates 0,1,…,M−10,1,\ldots,M-10,1,…,M−1, in that order. Start with X(0)=XX^{(0)}=XX(0)=X; at coordinate rrr, for each output index pair i,ji,ji,j, form Yab=Xi[r:=a],j[r:=b](r)Y_{ab}=X^{(r)}_{i[r:=a],j[r:=b]}Yab​=Xi[r:=a],j[r:=b](r)​ and set Xij(r+1)=C(Y)irjrX^{(r+1)}_{ij}=\mathcal C(Y)_{i_rj_r}Xij(r+1)​=C(Y)ir​jr​​. Here, writing tL(h)=(L/(L+1))h/(L+1)t_L(h)=(L/(L+1))^h/(L+1)tL​(h)=(L/(L+1))h/(L+1), the false-present\mathrm{present}present rule is C(Y)mn=1m=ntNB(n)∑a′Yaa\mathcal C(Y)_{mn}=\mathbf1_{m=n}t_{N_B}(n)\sum_a^{\prime}Y_{aa}C(Y)mn​=1m=n​tNB​​(n)∑a′​Yaa​, while the true-present\mathrm{present}present rule is

C(Y)mn=∑(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ′tNB/(1−k)(h) bk(m,ℓ,m+ℓ−h,h) bk(n,ℓ,n+ℓ−h,h) Y^m+ℓ−h,n+ℓ−h,\mathcal C(Y)_{mn}= \sum_{\substack{(h,\ell)\in\mathbb N^2\\h\le m+\ell,\ h\le n+\ell}}^{\prime} t_{N_B/(1-k)}(h)\, b_k(m,\ell,m+\ell-h,h)\, b_k(n,\ell,n+\ell-h,h)\, \widehat Y_{m+\ell-h,n+\ell-h},C(Y)mn​=(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ​∑′​tNB​/(1−k)​(h)bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)Ym+ℓ−h,n+ℓ−h​,

where Y^\widehat YY equals YYY when it has a bounded complex-linear realization on ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C), and is zero otherwise, and

bk(m,ℓ,p,q)={m!ℓ!p!q!∑0≤s≤ps≤m, m−s≤q(ps)(qm−s)(k)s+q−(m−s)(−1−k)p−s(1−k)m−s,m+ℓ=p+q,0,otherwise.b_k(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{\substack{0\le s\le p\\s\le m,\ m-s\le q}} \binom ps\binom q{m-s} (\sqrt{k})^{s+q-(m-s)} (-\sqrt{1-k})^{p-s} (\sqrt{1-k})^{m-s}, &m+\ell=p+q,\\ 0,&\text{otherwise}. \end{cases}bk​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​0≤s≤ps≤m, m−s≤q​∑​(sp​)(m−sq​)(k​)s+q−(m−s)(−1−k​)p−s(1−k​)m−s,0,​m+ℓ=p+q,otherwise.​

The result is X(M)X^{(M)}X(M); when M=0M=0M=0, it is XXX. All divisions and infinite sums are totalized: division by zero and nonsummable complex sums give zero, and real square roots of negative numbers give zero.

QAlgorithms.Papers.Independent.Intakenext100.P85.signalChannel

For a Boolean present\mathrm{present}present, arbitrary real numbers k,NBk,N_Bk,NB​, and an arbitrary complex array XXX indexed by N\mathbb NN, this defines another such array as follows. Write tL(h)=(L/(L+1))h/(L+1)t_L(h)=(L/(L+1))^h/(L+1)tL​(h)=(L/(L+1))h/(L+1), and let X^=X\widehat X=XX=X if there exists a bounded complex-linear operator AAA on ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C) with (Aej)i=Xij(Ae_j)_i=X_{ij}(Aej​)i​=Xij​ for every i,ji,ji,j, and X^=0\widehat X=0X=0 otherwise. Define

bk(m,ℓ,p,q)={m!ℓ!p!q!∑0≤r≤pr≤m, m−r≤q(pr)(qm−r)(k)r+q−(m−r)(−1−k)p−r(1−k)m−r,m+ℓ=p+q,0,m+ℓ≠p+q.b_k(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{\substack{0\le r\le p\\r\le m,\ m-r\le q}} \binom pr\binom q{m-r} (\sqrt{k})^{r+q-(m-r)} (-\sqrt{1-k})^{p-r} (\sqrt{1-k})^{m-r}, &m+\ell=p+q,\\ 0,&m+\ell\ne p+q. \end{cases}bk​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​0≤r≤pr≤m, m−r≤q​∑​(rp​)(m−rq​)(k​)r+q−(m−r)(−1−k​)p−r(1−k​)m−r,0,​m+ℓ=p+q,m+ℓ=p+q.​

If present\mathrm{present}present is true, put L=NB/(1−k)L=N_B/(1-k)L=NB​/(1−k) and define the (m,n)(m,n)(m,n) entry to be

∑(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ′tL(h) bk(m,ℓ,m+ℓ−h,h) bk(n,ℓ,n+ℓ−h,h) X^m+ℓ−h,n+ℓ−h.\sum_{\substack{(h,\ell)\in\mathbb N^2\\h\le m+\ell,\ h\le n+\ell}}^{\prime} t_L(h)\, b_k(m,\ell,m+\ell-h,h)\, b_k(n,\ell,n+\ell-h,h)\, \widehat X_{m+\ell-h,n+\ell-h}.(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ​∑′​tL​(h)bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)Xm+ℓ−h,n+ℓ−h​.

If present\mathrm{present}present is false, the (m,n)(m,n)(m,n) entry is

1m=n tNB(n)∑j∈N′Xjj.\mathbf1_{m=n}\,t_{N_B}(n)\sum_{j\in\mathbb N}^{\prime}X_{jj}.1m=n​tNB​​(n)j∈N∑′​Xjj​.

The restricted sum denotes a sum over all pairs, with excluded terms set to zero. No parameter bounds or convergence hypotheses are required: division by zero is zero, real square roots of nonpositive arguments are zero, and a nonsummable complex family has totalized sum zero. In particular, the true branch at k=1k=1k=1 uses L=0L=0L=0.

QAlgorithms.Papers.Independent.Intakenext100.P85.signalS

For every real number NSN_SNS​, this defines S(NS)=2NS+1S(N_S)=2N_S+1S(NS​)=2NS​+1, with no nonnegativity restriction on NSN_SNS​.

QAlgorithms.Papers.Independent.Intakenext100.P85.spdcArray

For every real NSN_SNS​, this defines the complex array indexed by occupations i,j:{0,1}→Ni,j:\{0,1\}\to\mathbb Ni,j:{0,1}→N as Xij=vivj‾X_{ij}=v_i\overline{v_j}Xij​=vi​vj​​, where vvv is the square-summable vector with coefficients NSj0/(NS+1)j0+1\sqrt{N_S^{j_0}/(N_S+1)^{j_0+1}}NSj0​​/(NS​+1)j0​+1​ when j0=j1j_0=j_1j0​=j1​, and zero otherwise, if such a vector exists; if no such vector exists, vvv is zero and the entire array is zero. Real division by zero and square roots of negative arguments return zero. The definition imposes no restriction on NSN_SNS​ and asserts no normalization of this array.

QAlgorithms.Papers.Independent.Intakenext100.P85.spdcCoefficient

For any real NSN_SNS​ and any occupation function j:{0,1}→Nj:\{0,1\}\to\mathbb Nj:{0,1}→N, this defines the complex coefficient

cNS(j)={NSj0(NS+1)j0+1,j0=j1,0,j0≠j1,c_{N_S}(j)= \begin{cases} \displaystyle\sqrt{\frac{N_S^{j_0}}{(N_S+1)^{j_0+1}}},&j_0=j_1,\\ 0,&j_0\ne j_1, \end{cases}cNS​​(j)=⎩⎨⎧​(NS​+1)j0​+1NSj0​​​​,0,​j0​=j1​,j0​=j1​,​

where the displayed real value is embedded into C\mathbb CC. There is no restriction on NSN_SNS​; division by zero returns zero, a real square root of a negative number returns zero, and a zeroth power is one even when its base is zero.

QAlgorithms.Papers.Independent.Intakenext100.P85.spdcCov

For any real number NSN_SNS​, this defines the real matrix indexed by (r,p)∈{0,1}2(r,p)\in\{0,1\}^2(r,p)∈{0,1}2 which, in the order (0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1), is

14(2NS+102NS(NS+1)002NS+10−2NS(NS+1)2NS(NS+1)02NS+100−2NS(NS+1)02NS+1).\frac14 \begin{pmatrix} 2N_S+1&0&2\sqrt{N_S(N_S+1)}&0\\ 0&2N_S+1&0&-2\sqrt{N_S(N_S+1)}\\ 2\sqrt{N_S(N_S+1)}&0&2N_S+1&0\\ 0&-2\sqrt{N_S(N_S+1)}&0&2N_S+1 \end{pmatrix}.41​​2NS​+102NS​(NS​+1)​0​02NS​+10−2NS​(NS​+1)​​2NS​(NS​+1)​02NS​+10​0−2NS​(NS​+1)​02NS​+1​​.

The parameter need not be nonnegative; the real square root is defined to be zero when its argument is negative.

QAlgorithms.Papers.Independent.Intakenext100.P85.spdcVector

For every real NSN_SNS​, this defines a vector in ℓ2(N{0,1};C)\ell^2(\mathbb N^{\{0,1\}};\mathbb C)ℓ2(N{0,1};C) by choosing a vector whose coefficient at every occupation j:{0,1}→Nj:\{0,1\}\to\mathbb Nj:{0,1}→N is

{NSj0(NS+1)j0+1,j0=j1,0,j0≠j1,\begin{cases} \displaystyle\sqrt{\frac{N_S^{j_0}}{(N_S+1)^{j_0+1}}},&j_0=j_1,\\ 0,&j_0\ne j_1, \end{cases}⎩⎨⎧​(NS​+1)j0​+1NSj0​​​​,0,​j0​=j1​,j0​=j1​,​

if such a square-summable vector exists, and returning the zero vector otherwise. The real coefficient is embedded into C\mathbb CC, with division by zero and square roots of negative arguments returning zero. No parameter restriction, existence assertion, or normalization condition is included.

QAlgorithms.Papers.Independent.Intakenext100.P85.targetBlock

For every natural number MMM, Boolean present\mathrm{present}present, real numbers k,NBk,N_Bk,NB​, and arbitrary complex array XXX indexed by pairs of MMM-coordinate occupation vectors, this acts separately on each slice with the second row and column components fixed: for fixed a,ba,ba,b, begin with Yij(0)=X(i,a),(j,b)Y^{(0)}_{ij}=X_{(i,a),(j,b)}Yij(0)​=X(i,a),(j,b)​, and successively transform coordinates r=0,…,M−1r=0,\ldots,M-1r=0,…,M−1 by Yij(r+1)=C(Z)irjrY^{(r+1)}_{ij}=\mathcal C(Z)_{i_rj_r}Yij(r+1)​=C(Z)ir​jr​​, where Zuv=Yi[r:=u],j[r:=v](r)Z_{uv}=Y^{(r)}_{i[r:=u],j[r:=v]}Zuv​=Yi[r:=u],j[r:=v](r)​; the output at (i,a),(j,b)(i,a),(j,b)(i,a),(j,b) is Yij(M)Y^{(M)}_{ij}Yij(M)​. To specify this rule completely, set tL(h)=(L/(L+1))h/(L+1)t_L(h)=(L/(L+1))^h/(L+1)tL​(h)=(L/(L+1))h/(L+1), let Z^=Z\widehat Z=ZZ=Z if ZZZ is the matrix of a bounded complex-linear operator on ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C) and zero otherwise, and put

bk(m,ℓ,p,q)={m!ℓ!p!q!∑0≤s≤ps≤m, m−s≤q(ps)(qm−s)(k)s+q−(m−s)(−1−k)p−s(1−k)m−s,m+ℓ=p+q,0,otherwise.b_k(m,\ell,p,q)= \begin{cases} \displaystyle \sqrt{\frac{m!\ell!}{p!q!}} \sum_{\substack{0\le s\le p\\s\le m,\ m-s\le q}} \binom ps\binom q{m-s} (\sqrt{k})^{s+q-(m-s)} (-\sqrt{1-k})^{p-s} (\sqrt{1-k})^{m-s}, &m+\ell=p+q,\\ 0,&\text{otherwise}. \end{cases}bk​(m,ℓ,p,q)=⎩⎨⎧​p!q!m!ℓ!​​0≤s≤ps≤m, m−s≤q​∑​(sp​)(m−sq​)(k​)s+q−(m−s)(−1−k​)p−s(1−k​)m−s,0,​m+ℓ=p+q,otherwise.​

For true present\mathrm{present}present,

C(Z)mn=∑(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ′tNB/(1−k)(h)bk(m,ℓ,m+ℓ−h,h)bk(n,ℓ,n+ℓ−h,h)Z^m+ℓ−h,n+ℓ−h;\mathcal C(Z)_{mn}= \sum_{\substack{(h,\ell)\in\mathbb N^2\\h\le m+\ell,\ h\le n+\ell}}^{\prime} t_{N_B/(1-k)}(h) b_k(m,\ell,m+\ell-h,h)b_k(n,\ell,n+\ell-h,h) \widehat Z_{m+\ell-h,n+\ell-h};C(Z)mn​=(h,ℓ)∈N2h≤m+ℓ, h≤n+ℓ​∑′​tNB​/(1−k)​(h)bk​(m,ℓ,m+ℓ−h,h)bk​(n,ℓ,n+ℓ−h,h)Zm+ℓ−h,n+ℓ−h​;

for false present\mathrm{present}present, C(Z)mn=1m=ntNB(n)∑u′Zuu\mathcal C(Z)_{mn}=\mathbf1_{m=n}t_{N_B}(n)\sum_u^{\prime}Z_{uu}C(Z)mn​=1m=n​tNB​​(n)∑u′​Zuu​. No parameter or array hypotheses are imposed; division by zero, negative-argument real square roots, and nonsummable complex sums give zero. At M=0M=0M=0, the result is XXX.

QAlgorithms.Papers.Independent.Intakenext100.P85.twoOcc

For any natural numbers a,ba,ba,b, this defines the two-coordinate occupation vector whose coordinate 000 is aaa and whose coordinate 111 is bbb:

twoOcc⁡(a,b)=(a,b).\operatorname{twoOcc}(a,b)=(a,b).twoOcc(a,b)=(a,b).

Zero occupations are allowed in either coordinate.

QAlgorithms.Papers.Independent.Intakenext100.P85.williamsonT

For arbitrary real numbers k,NS,NBk,N_S,N_Bk,NS​,NB​, define

L=2kNS+2NB+1+2NS+1,D=L2−4k(2NS(NS+1))2,x±=L±D2D.L=2kN_S+2N_B+1+2N_S+1,\qquad D=\sqrt{L^2-4k\left(2\sqrt{N_S(N_S+1)}\right)^2},\qquad x_\pm=\sqrt{\frac{L\pm D}{2D}}.L=2kNS​+2NB​+1+2NS​+1,D=L2−4k(2NS​(NS​+1)​)2​,x±​=2DL±D​​.

This defines a real matrix indexed by pairs (a,p)(a,p)(a,p), where a,p∈{0,1}a,p\in\{0,1\}a,p∈{0,1}, through

T(a,p),(b,q)={0,p≠q,x+,p=q and a=b,x−,p=q=0 and a≠b,−x−,p=q=1 and a≠b.T_{(a,p),(b,q)}= \begin{cases} 0,&p\ne q,\\ x_+,&p=q\text{ and }a=b,\\ x_-,&p=q=0\text{ and }a\ne b,\\ -x_-,&p=q=1\text{ and }a\ne b. \end{cases}T(a,p),(b,q)​=⎩⎨⎧​0,x+​,x−​,−x−​,​p=q,p=q and a=b,p=q=0 and a=b,p=q=1 and a=b.​

No parameter restrictions or matrix properties are required by this definition. Real square roots of negative arguments and division by zero both give zero; in particular, D=0D=0D=0 makes the entire matrix zero.

QAlgorithms.Papers.Independent.Intakenext100.P85.williamsonX

For a Boolean value plus\mathrm{plus}plus and arbitrary real numbers k,NS,NBk,N_S,N_Bk,NS​,NB​, set

L=2kNS+2NB+1+2NS+1,D=L2−4k(2NS(NS+1))2.L=2kN_S+2N_B+1+2N_S+1,\qquad D=\sqrt{L^2-4k\left(2\sqrt{N_S(N_S+1)}\right)^2}.L=2kNS​+2NB​+1+2NS​+1,D=L2−4k(2NS​(NS​+1)​)2​.

This defines

{L+D2D,plus=true,L−D2D,plus=false.\begin{cases} \displaystyle\sqrt{\frac{L+D}{2D}},&\mathrm{plus}=\mathrm{true},\\[6pt] \displaystyle\sqrt{\frac{L-D}{2D}},&\mathrm{plus}=\mathrm{false}. \end{cases}⎩⎨⎧​2DL+D​​,2DL−D​​,​plus=true,plus=false.​

No parameter restrictions or nonzero-denominator hypotheses are imposed. Division by zero is defined to be zero, and a real square root of a negative number is zero; consequently, both branches are zero when D=0D=0D=0.

Human review
  • Endorsed by Shuze Chen · Oct 31, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 31, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me