Definitions for Optimal quantum source coding with quantum information at the encoder and decoder
DefinitionQAlgorithms_Papers_Independent_Gap48_P38The objects the formalized statements of Optimal quantum source coding with quantum information at the encoder and decoder (arXiv:0706.2907) are stated in. Each was checked against the paper by a blind read-back and a moderator.
Restricted: Configurations of exactly the named subsystem, including the empty tensor product (p.2).subsystemIndex: Assemble retained and traced-out computational indices; no coordinate is discarded twice (p.2).subsystemReduction: Actual partial trace of the full operator over the complement of s (PDF p.2). The accepted traceRight is applied after regrouping retained/complementary indices; off-diagonal retained entries are preserved.registerEntropy: Entropy in bits of an actual reduced operator, not independent entropy data (PDF p.2).registerMutualInfo: I(S;T)=H(S)+H(T)-H(ST); the statements use disjoint subsystems (PDF p.2).registerCMI: I(S;T|V)=H(SV)+H(TV)-H(V)-H(STV), in bits (PDF p.2).cornerQ: The numerical corner qubit cost from (2), with A,B,C,D numbered 0,1,2,3 (PDF p.2). This definition is the entropy expression, not a definition of operational achievability.cornerE: The signed numerical corner ebit cost from (2), consumption positive (PDF p.2).iidGrouped: IID copies regrouped by subsystem, with exactly the original amplitudes (PDF pp.4--5).productProjection: Product of the local projections, in the same regrouped tensor convention (PDF p.5).TypicalBounds: All clauses of Theorem 5, including its continuation on PDF p.5. Ranks are Schatten zero quantities; frobNorm squared is the squared Schatten two norm; specNorm is the Schatten infinity norm. The six subsystems exclude ABC.decouplingMarginal: The actual Bhat,E marginal of (WU tensor I_E)rho(WU tensor I_E)^dagger, tracing S only, with S,Bhat,E order fixed explicitly (PDF pp.6,11).localAct: Apply an X operator, leaving the reference Y untouched (PDF p.7).coherentDecoder: The explicitly constructed coherent decoder, with output X,K order (PDF p.7). The square root on Lambda is essential.decoderOverlap: The overlap <j|<psi|(L tensor I_Y)|phi>, including both references (PDF p.7).coherentMeasurementError: P in Lemma 5 is the average measurement error on U_j psi (PDF p.7).decodingFidelity: F in Lemma 5 is the average SQUARED pure-state fidelity (PDF p.7).assistedChannel: The actual channel D composed with (E tensor identity) on rho tensor tau. E acts on X,L; D acts on Y,Lprime; the untouched Lprime off-diagonal blocks are retained. No extra communication or classical message is available (PDF p.9).
import Mathlib
import Definitions.Def_QAlgorithms_Defs_MatrixDistances
import Definitions.Def_QAlgorithms_Defs_SchurUnitaryEnsembles
import Definitions.Def_QAlgorithms_Defs_FiniteRegisters
import Definitions.Def_QAlgorithms_Defs_Channel
import Definitions.Def_QAlgorithms_Defs_Ensemble
import Definitions.Def_QAlgorithms_Defs_Measurement
import Definitions.Def_QAlgorithms_Defs_LinearSystems
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P08
import Definitions.Def_QAlgorithms_Papers_Independent_Next46_P39
/-!
Statements from Yard--Devetak, arXiv:0706.2907v2. PDF pages are used throughout.
All registers are finite complex registers with fixed bases; nonempty register
instances permit dimension one. Entropies are in bits, distances are FULL trace
norms, and pure-state fidelities are squared overlaps. Operational resource
inequalities without an audited simulation interface are explicitly deferred
in the unit's HARD.md rather than replaced by entropy inequalities.
-/
noncomputable section
open MeasureTheory
open scoped Matrix ComplexConjugate ComplexOrder Matrix.Norms.Elementwise
attribute [local instance] Classical.propDecidable Classical.decEq
namespace QAlgorithms.Papers.Independent.Gap48.P38
/-- Configurations of exactly the named subsystem, including the empty tensor product (p.2). -/
abbrev Restricted {m : ℕ} (I : Fin m → Type) (s : Finset (Fin m)) :=
(i : {j : Fin m // j ∈ s}) → I i.val
/-- Assemble retained and traced-out computational indices; no coordinate is discarded twice (p.2). -/
def subsystemIndex {m : ℕ} (I : Fin m → Type) (s : Finset (Fin m))
(x : Restricted I s) (z : Restricted I (Finset.univ \ s)) : (∀ i, I i) :=
fun i => if h : i ∈ s then x ⟨i, h⟩
else z ⟨i, Finset.mem_sdiff.mpr ⟨Finset.mem_univ i, h⟩⟩
/-- Actual partial trace of the full operator over the complement of s (PDF p.2).
The accepted traceRight is applied after regrouping retained/complementary
indices; off-diagonal retained entries are preserved. -/
def subsystemReduction {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)]
(rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) (s : Finset (Fin m)) :
Matrix (Restricted I s) (Restricted I s) ℂ :=
traceRight (fun x y : Restricted I s × Restricted I (Finset.univ \ s) =>
rho (subsystemIndex I s x.1 x.2) (subsystemIndex I s y.1 y.2))
/-- Entropy in bits of an actual reduced operator, not independent entropy data (PDF p.2). -/
def registerEntropy {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)]
(rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) (s : Finset (Fin m)) : ℝ :=
vnEntropy (subsystemReduction I rho s)
/-- I(S;T)=H(S)+H(T)-H(ST); the statements use disjoint subsystems (PDF p.2). -/
def registerMutualInfo {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)]
(rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) (s t : Finset (Fin m)) : ℝ :=
registerEntropy I rho s + registerEntropy I rho t - registerEntropy I rho (s ∪ t)
/-- I(S;T|V)=H(SV)+H(TV)-H(V)-H(STV), in bits (PDF p.2). -/
def registerCMI {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)]
(rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) (s t v : Finset (Fin m)) : ℝ :=
registerEntropy I rho (s ∪ v) + registerEntropy I rho (t ∪ v) -
registerEntropy I rho v - registerEntropy I rho (s ∪ t ∪ v)
/-- The numerical corner qubit cost from (2), with A,B,C,D numbered 0,1,2,3 (PDF p.2).
This definition is the entropy expression, not a definition of operational achievability. -/
def cornerQ (I : Fin 4 → Type) [∀ i, Fintype (I i)]
(rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) : ℝ := registerCMI I rho {2} {3} {1} / 2
/-- The signed numerical corner ebit cost from (2), consumption positive (PDF p.2). -/
def cornerE (I : Fin 4 → Type) [∀ i, Fintype (I i)]
(rho : Matrix (∀ i, I i) (∀ i, I i) ℂ) : ℝ :=
registerMutualInfo I rho {0} {2} / 2 - registerMutualInfo I rho {1} {2} / 2
example {m : ℕ} (I : Fin m → Type) (s : Finset (Fin m))
(x : Restricted I s) (z : Restricted I (Finset.univ \ s))
(i : Fin m) (hi : i ∈ s) : subsystemIndex I s x z i = x ⟨i, hi⟩ := by
simp [subsystemIndex, hi]
example {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)] (s : Finset (Fin m)) :
subsystemReduction I (0 : Matrix (∀ i, I i) (∀ i, I i) ℂ) s = 0 := by
ext x y
simp [subsystemReduction, traceRight]
example {m : ℕ} (I : Fin m → Type) [∀ i, Fintype (I i)] (s : Finset (Fin m)) :
registerEntropy I (0 : Matrix (∀ i, I i) (∀ i, I i) ℂ) s = 0 := by
have hz : subsystemReduction I (0 : Matrix (∀ i, I i) (∀ i, I i) ℂ) s = 0 := by
ext x y
simp [subsystemReduction, traceRight]
rw [registerEntropy, hz]
simp [vnEntropy]
/-- IID copies regrouped by subsystem, with exactly the original amplitudes (PDF pp.4--5). -/
def iidGrouped {m : ℕ} (I : Fin m → Type) (n : ℕ)
(psi : EuclideanSpace ℂ (∀ i, I i)) : EuclideanSpace ℂ (∀ i, Fin n → I i) :=
WithLp.toLp 2 fun x => ∏ t : Fin n, psi (fun i => x i t)
/-- Product of the local projections, in the same regrouped tensor convention (PDF p.5). -/
def productProjection (I : Fin 3 → Type) [∀ i, Fintype (I i)] (n : ℕ)
(P : (i : Fin 3) → Matrix (Restricted (fun j => Fin n → I j) {i})
(Restricted (fun j => Fin n → I j) {i}) ℂ) :
Matrix (∀ i, Fin n → I i) (∀ i, Fin n → I i) ℂ :=
fun x y => ∏ i : Fin 3, P i (fun j => x j.val) (fun j => y j.val)
/-- All clauses of Theorem 5, including its continuation on PDF p.5.
Ranks are Schatten zero quantities; frobNorm squared is the squared Schatten two norm;
specNorm is the Schatten infinity norm. The six subsystems exclude ABC. -/
def TypicalBounds (I : Fin 3 → Type) [∀ i, Fintype (I i)]
(psi : EuclideanSpace ℂ (∀ i, I i)) (n : ℕ) (delta epsilon : ℝ)
(P : (i : Fin 3) → Matrix (Restricted (fun j => Fin n → I j) {i})
(Restricted (fun j => Fin n → I j) {i}) ℂ) : Prop :=
let J := fun i => Fin n → I i
let rho := pureDensity psi
let source := iidGrouped I n psi
let v := act (productProjection I n P) source
let vC := act (productProjection I n (fun i => if i = 2 then P i else 1)) source
let phi := pureDensity (normalizedVec v)
(∀ i : Fin 3, (P i).IsHermitian ∧ P i * P i = P i ∧
effectProb (P i) (subsystemReduction J (pureDensity source) {i}) ≥ 1 - epsilon ∧
Real.rpow 2 ((n : ℝ) * registerEntropy I rho {i} - n * delta) ≤ (P i).trace.re ∧
(P i).trace.re ≤ Real.rpow 2 ((n : ℝ) * registerEntropy I rho {i} + n * delta)) ∧
0 < ‖v‖ ∧ traceNorm (phi - pureDensity source) ≤ epsilon ∧
(∀ s : Finset (Fin 3), s ∈ ({ {0}, {1}, {2}, {0,1}, {1,2}, {0,2} } :
Finset (Finset (Fin 3))) →
let h := registerEntropy I rho s
let tau := subsystemReduction J phi s
Real.rpow 2 ((n : ℝ) * h - n * delta) ≤ (Matrix.rank tau : ℝ) ∧
(Matrix.rank tau : ℝ) ≤ Real.rpow 2 ((n : ℝ) * h + n * delta) ∧
Real.rpow 2 (-(n : ℝ) * h - n * delta) ≤ frobNorm tau ^ 2 ∧
frobNorm tau ^ 2 ≤ Real.rpow 2 (-(n : ℝ) * h + n * delta) ∧
Real.rpow 2 (-(n : ℝ) * h - n * delta) ≤ specNorm tau ∧
specNorm tau ≤ Real.rpow 2 (-(n : ℝ) * h + n * delta)) ∧
0 < ‖vC‖ ∧ traceNorm (pureDensity (normalizedVec vC) - pureDensity source) ≤ epsilon
example {m : ℕ} (I : Fin m → Type) (psi : EuclideanSpace ℂ (∀ i, I i))
(x : (∀ i, Fin 0 → I i)) : iidGrouped I 0 psi x = 1 := by
simp [iidGrouped]
example {J : Type} [Fintype J] : normalizedVec (0 : EuclideanSpace ℂ J) = 0 := by
simp [normalizedVec]
example (I : Fin 3 → Type) [∀ i, Fintype (I i)] (n : ℕ) :
productProjection I n (fun _ => 0) = 0 := by
ext x y
simp [productProjection]
/-- The actual Bhat,E marginal of (WU tensor I_E)rho(WU tensor I_E)^dagger,
tracing S only, with S,Bhat,E order fixed explicitly (PDF pp.6,11). -/
def decouplingMarginal {c s b e : ℕ}
(W : Matrix (Fin s × Fin b) (Fin c) ℂ)
(rho : Matrix (Fin c × Fin e) (Fin c × Fin e) ℂ)
(U : Matrix.unitaryGroup (Fin c) ℂ) :
Matrix (Fin b × Fin e) (Fin b × Fin e) ℂ :=
let V := Matrix.kronecker (W * (U : Matrix (Fin c) (Fin c) ℂ))
(1 : Matrix (Fin e) (Fin e) ℂ)
let tau := V * rho * Vᴴ
fun x y => ∑ a : Fin s, tau ((a, x.1), x.2) ((a, y.1), y.2)
example : QAlgorithms.Papers.Independent.Information100.P08.maximallyMixed 1 = (1 : Matrix (Fin 1) (Fin 1) ℂ) := by
simp [QAlgorithms.Papers.Independent.Information100.P08.maximallyMixed]
example {c s b e : ℕ} (W : Matrix (Fin s × Fin b) (Fin c) ℂ)
(U : Matrix.unitaryGroup (Fin c) ℂ) :
decouplingMarginal (e := e) W 0 U = 0 := by
ext x y
simp [decouplingMarginal]
example {J : Type} [Fintype J] [DecidableEq J] [Nonempty J] :
QAlgorithms.Papers.Independent.Next46.P39.inverseSqrt (0 : Matrix J J ℂ) = 0 := by
simp [QAlgorithms.Papers.Independent.Next46.P39.inverseSqrt, cfc_zero]
/-- Apply an X operator, leaving the reference Y untouched (PDF p.7). -/
def localAct {X Y : Type} [Fintype X] [Fintype Y] [DecidableEq Y]
(M : Matrix X X ℂ) (psi : EuclideanSpace ℂ (X × Y)) :
EuclideanSpace ℂ (X × Y) := act (Matrix.kronecker M (1 : Matrix Y Y ℂ)) psi
/-- The explicitly constructed coherent decoder, with output X,K order (PDF p.7).
The square root on Lambda is essential. -/
def coherentDecoder {X : Type} [Fintype X] [DecidableEq X] {k : ℕ}
(alpha : Fin k → ℂ) (U : Fin k → Matrix.unitaryGroup X ℂ)
(Lambda : Fin k → Matrix X X ℂ) : Matrix (X × Fin k) X ℂ :=
fun x y => alpha x.2 * (((U x.2 : Matrix X X ℂ)ᴴ * cfc Real.sqrt (Lambda x.2)) x.1 y)
/-- The overlap <j|<psi|(L tensor I_Y)|phi>, including both references (PDF p.7). -/
def decoderOverlap {X Y : Type} [Fintype X] [Fintype Y] {k : ℕ}
(psi : EuclideanSpace ℂ (X × Y)) (L : Matrix (X × Fin k) X ℂ)
(phi : EuclideanSpace ℂ (X × Y)) (j : Fin k) : ℂ :=
∑ x : X, ∑ y : Y, conj (psi (x,y)) * ∑ z : X, L (x,j) z * phi (z,y)
/-- P in Lemma 5 is the average measurement error on U_j psi (PDF p.7). -/
def coherentMeasurementError {X Y : Type} [Fintype X] [Fintype Y]
[DecidableEq X] [DecidableEq Y] {k : ℕ}
(psi : EuclideanSpace ℂ (X × Y)) (U : Fin k → Matrix.unitaryGroup X ℂ)
(Lambda : Fin k → Matrix X X ℂ) : ℝ :=
1 - (k : ℝ)⁻¹ * ∑ j, effectProb (Lambda j)
(traceRight (pureDensity (localAct (U j : Matrix X X ℂ) psi)))
/-- F in Lemma 5 is the average SQUARED pure-state fidelity (PDF p.7). -/
def decodingFidelity {X Y : Type} [Fintype X] [Fintype Y]
[DecidableEq X] [DecidableEq Y] {k : ℕ}
(psi : EuclideanSpace ℂ (X × Y)) (U : Fin k → Matrix.unitaryGroup X ℂ)
(psiPrime : Fin k → EuclideanSpace ℂ (X × Y)) : ℝ :=
(k : ℝ)⁻¹ * ∑ j, Complex.normSq (inner ℂ (localAct (U j : Matrix X X ℂ) psi) (psiPrime j))
example {X Y : Type} [Fintype X] [Fintype Y] [DecidableEq Y]
(psi : EuclideanSpace ℂ (X × Y)) : localAct (0 : Matrix X X ℂ) psi = 0 := by
simp [localAct, act, Matrix.mulVec]
example {X : Type} [Fintype X] [DecidableEq X] {k : ℕ}
(U : Fin k → Matrix.unitaryGroup X ℂ) (Lambda : Fin k → Matrix X X ℂ) :
coherentDecoder (fun _ => 0) U Lambda = 0 := by
ext x y
simp [coherentDecoder]
example {X Y : Type} [Fintype X] [Fintype Y] {k : ℕ}
(psi phi : EuclideanSpace ℂ (X × Y)) (j : Fin k) :
decoderOverlap psi (0 : Matrix (X × Fin k) X ℂ) phi j = 0 := by
simp [decoderOverlap]
/-- The actual channel D composed with (E tensor identity) on rho tensor tau.
E acts on X,L; D acts on Y,Lprime; the untouched Lprime off-diagonal blocks
are retained. No extra communication or classical message is available (PDF p.9). -/
def assistedChannel {X Y L Lprime : Type} [Fintype X] [Fintype Y] [Fintype L]
(E : Superop (X × L) Y) (D : Superop (Y × Lprime) X)
(tau : Matrix (L × Lprime) (L × Lprime) ℂ) : Superop X X :=
fun rho => D (fun a b => E
(fun x y => rho x.1 y.1 * tau (x.2,a.2) (y.2,b.2)) a.1 b.1)
example : assistedChannel
(fun M : Matrix (Unit × Unit) (Unit × Unit) ℂ =>
(fun _ _ : Unit => M ((),()) ((),())))
(fun M : Matrix (Unit × Unit) (Unit × Unit) ℂ =>
(fun _ _ : Unit => M ((),()) ((),())))
(1 : Matrix (Unit × Unit) (Unit × Unit) ℂ)
(1 : Matrix Unit Unit ℂ) () () = 1 := by
simp [assistedChannel]
end QAlgorithms.Papers.Independent.Gap48.P38
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
QAlgorithms.Papers.Independent.Gap48.P38.Restricted
For any natural number , family of types , and subset , this denotes the type of assignments that choose an element for every . No finiteness or nonemptiness of the types is assumed. If , there is exactly one such assignment, including when ; if some with is empty, there are no such assignments.
QAlgorithms.Papers.Independent.Gap48.P38.TypicalBounds
For three finite types with specified finite enumerations, an arbitrary vector , a natural number , real numbers , and three arbitrary complex matrices indexed by assignments of values to coordinate , this defines the following conjunction. Write ; for any vector , write ; for a matrix indexed by a product of coordinate types, write , summing over assignments on the complementary coordinates, and write for evaluated by continuous functional calculus using the total real logarithm. Set , , , , and , where is the same product matrix with replaced by identity matrices and retained. Set , where normalization is defined to give zero when . The required conjunction is
Here , with the square root given by continuous functional calculus, is the Frobenius norm, and is the Euclidean operator norm; all displayed powers of are real powers. No normalization of , positivity of or , positivity of , or nonemptiness of is assumed in defining this predicate. Its positive-norm requirements exclude zero projected vectors; in particular, if any is empty and , those requirements cannot hold. Empty coordinate products and otherwise use their usual unique assignments and empty products.
QAlgorithms.Papers.Independent.Gap48.P38.assistedChannel
For arbitrary types , with finite structures on , let be any function from complex matrices on to complex matrices on , let be any function from complex matrices on to complex matrices on , and let be any complex matrix on . This defines a function from matrices on to matrices on : for each input , form the matrix on with
No linearity, channel condition, positivity, or normalization is imposed on these functions or matrices. The finite types may be empty, and no finiteness assumption is imposed on .
QAlgorithms.Papers.Independent.Gap48.P38.coherentDecoder
For a finite type with decidable equality, a natural number , arbitrary complex coefficients , unitary matrices indexed by , and arbitrary complex matrices indexed by , for , this defines a rectangular matrix by
The square root is the real continuous functional calculus used in the code; no positivity or Hermitian condition on , and no modulus condition on , is assumed by this definition. The type may be empty, and gives a matrix with no rows.
QAlgorithms.Papers.Independent.Gap48.P38.coherentMeasurementError
For finite types with decidable equality, a natural number , an arbitrary vector , unitary matrices on , and arbitrary complex matrices on , indexed by , set
This definition returns
No normalization of , positivity of the , or condition that their sum is the identity is assumed. The types may be empty. At , the sum and the defined inverse of are both zero, so the value is .
QAlgorithms.Papers.Independent.Gap48.P38.cornerE
For four finite types with specified finite enumerations and any complex matrix indexed by their full assignments, this defines
Here , with combining assignments, and for evaluated by continuous functional calculus using the total real logarithm. No positivity, normalization, or nonemptiness is assumed; the formula applies also when coordinate types are empty.
QAlgorithms.Papers.Independent.Gap48.P38.cornerQ
For four finite types with specified finite enumerations and any complex matrix indexed by , this defines
For each coordinate subset , , where combines assignments, and with evaluated by continuous functional calculus using the total real logarithm. The coordinate labels are . No density-matrix or nonemptiness hypothesis is imposed, so empty coordinate types are included.
QAlgorithms.Papers.Independent.Gap48.P38.decoderOverlap
For arbitrary finite types , a natural number , vectors , a complex matrix with rows indexed by and columns indexed by , and an index , this defines the complex number
No normalization or isometry assumptions are imposed. If or is empty, the value is zero. Supplying requires , so there is no instance of the definition with such an argument when .
QAlgorithms.Papers.Independent.Gap48.P38.decodingFidelity
For finite types with decidable equality, a natural number , an arbitrary vector , unitary matrices on , and arbitrary vectors , indexed by , this defines
Neither nor the are required to be normalized. Either type may be empty, in which case the inner sums are zero. At , the empty sum and the defined inverse of zero give value zero.
QAlgorithms.Papers.Independent.Gap48.P38.decouplingMarginal
For arbitrary natural numbers , a complex matrix , an arbitrary complex matrix indexed by , and a unitary matrix , this defines a matrix on . Set and ; the resulting matrix has entries
Thus it sums over the first output factor of dimension . No positivity, normalization, or isometry condition is imposed on or . All four dimensions may be zero; in particular, when , every entry is an empty sum and hence zero.
QAlgorithms.Papers.Independent.Gap48.P38.iidGrouped
For any natural numbers , family of types , and vector in the complex Euclidean space indexed by full assignments , this defines a vector indexed by families with , by
No finiteness or nonemptiness assumptions on the types, and no normalization assumption on , are included. When , every coordinate sequence is empty and the product equals . When , the full-assignment type has one element, and the value is the th power of at that element.
QAlgorithms.Papers.Independent.Gap48.P38.localAct
For arbitrary finite types , with decidable equality on , any complex matrix indexed by , and any vector , this defines the vector obtained by matrix multiplication with ; explicitly,
There is no unitarity condition on or normalization condition on . Either indexing type may be empty.
QAlgorithms.Papers.Independent.Gap48.P38.productProjection
For three finite types with specified finite enumerations, any natural number , and, for each , an arbitrary complex matrix indexed by assignments of values to the single coordinate , this defines a complex matrix on grouped assignments , where each , by
The restrictions retain only the indicated coordinate sequence. No Hermiticity, idempotence, positivity, normalization, or nonemptiness is assumed for these inputs. The case is included: each local assignment type then has one element, and is the one-entry matrix whose entry is the product of the three local entries.
QAlgorithms.Papers.Independent.Gap48.P38.registerCMI
For any natural number , family of finite types with specified finite enumerations, arbitrary complex matrix indexed by , and subsets , this defines
For every subset , the matrix has entries , where combines assignments, and for the continuous functional calculus expression associated with , using the total real logarithm. No disjointness among , density-matrix condition on , or nonemptiness of the coordinate types is assumed. Empty subsets and are included.
QAlgorithms.Papers.Independent.Gap48.P38.registerEntropy
For any natural number , family of finite types with specified finite enumerations, arbitrary complex matrix indexed by , and subset , this defines
where is the continuous functional calculus expression and , with combining assignments on and its complement. The logarithm is Lean’s total real base-two logarithm, with . No density-matrix or nonemptiness hypothesis is required: this is the stated expression for every matrix, including matrices on empty assignment types and the case .
QAlgorithms.Papers.Independent.Gap48.P38.registerMutualInfo
For any natural number , family of finite types with specified finite enumerations, arbitrary complex matrix indexed by , and subsets , this defines
Here, for each subset , , where combines the indicated assignments, and with evaluated by continuous functional calculus using the total real logarithm. The subsets need not be disjoint and may be empty. Neither nor the coordinate types are assumed to be normalized, positive, or nonempty, and is allowed.
QAlgorithms.Papers.Independent.Gap48.P38.subsystemIndex
For any natural number , family of types , subset , assignment on , and assignment on its complement , this defines the full assignment by when and otherwise. No finiteness or nonemptiness assumptions are imposed on the types. Empty subsets and are included; on an empty collection of coordinates the corresponding assignment is unique.
QAlgorithms.Papers.Independent.Gap48.P38.subsystemReduction
For any natural number , family of finite types with specified finite enumerations, arbitrary complex matrix indexed by full assignments in , and subset , this defines a matrix indexed by assignments on whose entries are
where uses on and on its complement. No positivity, Hermiticity, trace, or nonemptiness assumption is made. If the complementary assignment type is empty, the sum is zero; if the complement has no coordinates, it has one assignment and the sum has one term. The definition also includes and empty matrix index types.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.