Definitions for Asymptotic state transformations of continuous variable resources
DefinitionQAlgorithms_Papers_Independent_Intakethird100_P88The objects the formalized statements of Asymptotic state transformations of continuous variable resources (arXiv:2010.00044) are stated in. Each was checked against the paper by a blind read-back and a moderator.
BoundedSelfAdjoint: PDF p.3: bounded self-adjoint operators on the entire Hilbert space.weakPositiveTopology: PDF pp.3–4: the weak topology on the positive trace-class cone, tested against all bounded self-adjoint operators, rather than just compact ones.normPositiveTopology: PDF p.4: restriction of the accepted nuclear trace-norm topology to the cone.weakOperatorPositiveTopology: PDF p.3: WOT on the self-adjoint positive cone is generated by all quadratic forms.weakStateTopology: PDF pp.3–4: weak topology restricted to trace-one positive states.weakOperatorStateTopology: PDF pp.3–4: WOT restricted to trace-one positive states.measuredKLSummand: PDF p.5: a single classical KL summand, extended on support failure. No trace-correction term is added for an unnormalized second argument.FinitePOVM: PDF pp.3,5: exactly finite-outcome POVMs of positive bounded effects.traceClassMeasuredEntropyBits: PDF p.5, (10), extended to the positive trace-class second argument of Lemma 20, p.11. EReal retains negative finite values when Tr sigma exceeds one. The input rho is always a positive trace-one state. The accepted Bridge50.P11 measuredEntropy is a natural-log variational program on algebra functionals, not this finite-POVM definition with an unnormalized trace-class reference. Source correction: printed (10), p.5, repeats P_rho in both arguments. Changed: the second distribution is P_sigma. Why: the surrounding two-state definition and the explicit Tr(sigma E_x) calculation on pp.19-20 require it; the literal printed expression would be identically zero on normalized rho.StrictlyPositiveQuadraticForm: PDF pp.11,18: strict positivity means a positive quadratic form on every nonzero vector, without a uniform spectral lower bound. The accepted Dependency50.P37.StrictlyPositive is instead a predicate on CP maps.UniformlyPositive: PDF p.18: L has some positive lower spectral bound depending on L. The proof explicitly uses L >= delta I. This is stronger than zero kernel.twoPower: PDF pp.11,18: the bounded positive exponential 2^h, via real continuous functional calculus on the bounded spectrum of a self-adjoint h.logarithmicExpectation: PDF pp.18–19: extended logarithmic expectation. The continuous bounded-below spectral cutoffs log2(max(x,1/(n+1))) decrease to log2(x). The infimum is in EReal, so nonintegrable negative logarithmic tails give minus infinity. Only positive L and trace-one rho are used; this is not cfc Real.log at zero.exponentialProgram: PDF p.11, (23)–(24): optimization over all bounded self-adjoint h. The Bool selects the linear penalty (true) or logarithmic penalty (false). The false branch is used only with sigma nonzero.positiveProgram: PDF p.11, (25)–(28): optimization over bounded uniformly positive L (uniform=true) or bounded strictly positive L (uniform=false), with either penalty. The logarithmic penalty is used only with sigma nonzero.zeroPositive: PDF p.20, Remark 32: the zero positive trace-class operator, explicitly without a normalization or nonzero requirement.
import Mathlib
import Definitions.Def_QAlgorithms_Papers_Independent_Dependency50_P13
import Definitions.Def_QAlgorithms_Papers_Independent_Dependency50_P39
import Definitions.Def_QAlgorithms_Papers_Independent_Information100_P62
/-!
Ferrari, Lami, Theurer and Plenio, arXiv:2010.00044v3.
Drafted subset: Corollary 3, Remark 4, Lemma 20, Remark 32.
All remaining planned nodes are explicitly deferred in HARD.md and GAPS.json.
The independent-paper override permits the inspected accepted-library imports.
Only definition bodies are reused. Compilation and independent review are pending.
-/
noncomputable section
namespace QAlgorithms.Papers.Independent.Intakethird100.P88
open Filter
open QAlgorithms.Papers.Independent.Dependency50.P13
open QAlgorithms.Papers.Independent.Dependency50.P39 (Fock fockKet fockDyad)
open QAlgorithms.Papers.Independent.Information100.P62 (logTwo)
attribute [local instance] Classical.propDecidable
variable {H : Type*} [NormedAddCommGroup H] [InnerProductSpace ℂ H] [CompleteSpace H]
/-- PDF p.3: bounded self-adjoint operators on the entire Hilbert space. -/
abbrev BoundedSelfAdjoint (H : Type*) [NormedAddCommGroup H] [InnerProductSpace ℂ H]
[CompleteSpace H] := {A : H →L[ℂ] H // A.adjoint = A}
/-- PDF pp.3–4: the weak topology on the positive trace-class cone, tested
against all bounded self-adjoint operators, rather than just compact ones. -/
def weakPositiveTopology {I : Type*} (b : HilbertBasis I ℂ H) :
TopologicalSpace (PositiveTraceClass H) :=
TopologicalSpace.induced
(fun T : PositiveTraceClass H => fun A : BoundedSelfAdjoint H =>
(operatorTrace b (T.val.val.comp A.val)).re) inferInstance
/-- PDF p.4: restriction of the accepted nuclear trace-norm topology to the cone. -/
def normPositiveTopology : TopologicalSpace (PositiveTraceClass H) :=
TopologicalSpace.induced (fun T : PositiveTraceClass H => T.val) traceNormTopology
/-- PDF p.3: WOT on the self-adjoint positive cone is generated by all quadratic forms. -/
def weakOperatorPositiveTopology : TopologicalSpace (PositiveTraceClass H) :=
TopologicalSpace.induced
(fun T : PositiveTraceClass H => fun x : H => (inner ℂ x (T.val.val x)).re)
inferInstance
/-- PDF pp.3–4: weak topology restricted to trace-one positive states. -/
def weakStateTopology {I : Type*} (b : HilbertBasis I ℂ H) : TopologicalSpace (State b) :=
TopologicalSpace.induced
(fun rho : State b => fun A : BoundedSelfAdjoint H =>
(operatorTrace b (rho.val.val.comp A.val)).re) inferInstance
/-- PDF pp.3–4: WOT restricted to trace-one positive states. -/
def weakOperatorStateTopology {I : Type*} (b : HilbertBasis I ℂ H) :
TopologicalSpace (State b) :=
TopologicalSpace.induced
(fun rho : State b => fun x : H => (inner ℂ x (rho.val.val x)).re) inferInstance
/-- PDF p.5: a single classical KL summand, extended on support failure.
No trace-correction term is added for an unnormalized second argument. -/
def measuredKLSummand (p q : ℝ) : EReal :=
if p = 0 then 0 else if q = 0 then ⊤ else (p * (logTwo p - logTwo q) : ℝ)
/-- PDF pp.3,5: exactly finite-outcome POVMs of positive bounded effects. -/
def FinitePOVM {n : ℕ} (M : Fin n → H →L[ℂ] H) : Prop :=
(∀ i, IsPositive (M i)) ∧ ∑ i, M i = 1
/-- PDF p.5, (10), extended to the positive trace-class second argument of
Lemma 20, p.11. EReal retains negative finite values when Tr sigma exceeds one.
The input rho is always a positive trace-one state. The accepted Bridge50.P11
measuredEntropy is a natural-log variational program on algebra functionals,
not this finite-POVM definition with an unnormalized trace-class reference.
Source correction: printed (10), p.5, repeats P_rho in both arguments.
Changed: the second distribution is P_sigma. Why: the surrounding two-state
definition and the explicit Tr(sigma E_x) calculation on pp.19-20 require it;
the literal printed expression would be identically zero on normalized rho. -/
def traceClassMeasuredEntropyBits {I : Type*} (b : HilbertBasis I ℂ H) (rho : State b)
(sigma : PositiveTraceClass H) : EReal :=
⨆ n : ℕ, ⨆ M : {M : Fin n → H →L[ℂ] H // FinitePOVM M},
∑ i, measuredKLSummand
(operatorTrace b (rho.val.val.comp (M.val i))).re
(operatorTrace b (sigma.val.val.comp (M.val i))).re
/-- PDF pp.11,18: strict positivity means a positive quadratic form on every
nonzero vector, without a uniform spectral lower bound. The accepted
Dependency50.P37.StrictlyPositive is instead a predicate on CP maps. -/
def StrictlyPositiveQuadraticForm (L : H →L[ℂ] H) : Prop :=
IsPositive L ∧ ∀ x : H, x ≠ 0 → 0 < (inner ℂ x (L x)).re
/-- PDF p.18: L has some positive lower spectral bound depending on L.
The proof explicitly uses L >= delta I. This is stronger than zero kernel. -/
def UniformlyPositive (L : H →L[ℂ] H) : Prop :=
IsPositive L ∧ ∃ delta : ℝ, 0 < delta ∧
∀ x : H, delta * ‖x‖ ^ 2 ≤ (inner ℂ x (L x)).re
/-- PDF pp.11,18: the bounded positive exponential 2^h, via real continuous
functional calculus on the bounded spectrum of a self-adjoint h. -/
def twoPower (h : BoundedSelfAdjoint H) : H →L[ℂ] H :=
cfc (fun x : ℝ => Real.rpow 2 x) h.val
/-- PDF pp.18–19: extended logarithmic expectation. The continuous bounded-below
spectral cutoffs log2(max(x,1/(n+1))) decrease to log2(x). The infimum is in
EReal, so nonintegrable negative logarithmic tails give minus infinity.
Only positive L and trace-one rho are used; this is not cfc Real.log at zero. -/
def logarithmicExpectation {I : Type*} (b : HilbertBasis I ℂ H) (rho : State b)
(L : H →L[ℂ] H) : EReal :=
⨅ n : ℕ, ((operatorTrace b
(rho.val.val.comp (cfc
(fun x : ℝ => logTwo (max x ((n + 1 : ℝ)⁻¹))) L))).re : EReal)
/-- PDF p.11, (23)–(24): optimization over all bounded self-adjoint h.
The Bool selects the linear penalty (true) or logarithmic penalty (false).
The false branch is used only with sigma nonzero. -/
def exponentialProgram {I : Type*} (b : HilbertBasis I ℂ H) (rho : State b)
(sigma : PositiveTraceClass H) (linear : Bool) : EReal :=
⨆ h : BoundedSelfAdjoint H,
let t := (operatorTrace b (sigma.val.val.comp (twoPower h))).re
(((operatorTrace b (rho.val.val.comp h.val)).re +
if linear then (1 - t) / Real.log 2 else -logTwo t : ℝ) : EReal)
/-- PDF p.11, (25)–(28): optimization over bounded uniformly positive L
(uniform=true) or bounded strictly positive L (uniform=false), with either
penalty. The logarithmic penalty is used only with sigma nonzero. -/
def positiveProgram {I : Type*} (b : HilbertBasis I ℂ H) (rho : State b)
(sigma : PositiveTraceClass H) (uniform linear : Bool) : EReal :=
⨆ L : {L : H →L[ℂ] H // if uniform then UniformlyPositive L else StrictlyPositiveQuadraticForm L},
let t := (operatorTrace b (sigma.val.val.comp L.val)).re
logarithmicExpectation b rho L.val +
((if linear then (1 - t) / Real.log 2 else -logTwo t : ℝ) : EReal)
/-- PDF p.20, Remark 32: the zero positive trace-class operator, explicitly
without a normalization or nonzero requirement. -/
def zeroPositive : PositiveTraceClass H :=
⟨⟨0, by
unfold IsTraceClass NuclearRepresentation
refine ⟨fun _ => 0, fun _ => 0, ?_, ?_⟩
· simp
· intro x
simp⟩, by
constructor
· simp
· intro x
simp⟩
example : logTwo 1 = 0 := by simp [logTwo]
example (q : ℝ) : measuredKLSummand 0 q = 0 := by simp [measuredKLSummand]
example (p : ℝ) (hp : p ≠ 0) : measuredKLSummand p 0 = ⊤ := by
simp [measuredKLSummand, hp]
example : measuredKLSummand 1 1 = 0 := by simp [measuredKLSummand, logTwo]
example : measuredKLSummand 1 2 = (-1 : EReal) := by
have hlog : Real.log 2 ≠ 0 := Real.log_ne_zero_of_pos_of_ne_one (by norm_num) (by norm_num)
simp [measuredKLSummand, logTwo, hlog]
example : IsPositive (0 : H →L[ℂ] H) := by
constructor
· simp
· intro x
simp
example : (zeroPositive (H := H)).val.val = 0 := rfl
example (x : H) : (inner ℂ x ((zeroPositive (H := H)).val.val x)).re = 0 := by simp [zeroPositive]
example (L : H →L[ℂ] H) (h : UniformlyPositive L) : IsPositive L := h.1
example (L : H →L[ℂ] H) (h : StrictlyPositiveQuadraticForm L) : IsPositive L := h.1
end QAlgorithms.Papers.Independent.Intakethird100.P88
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
QAlgorithms.Papers.Independent.Intakethird100.P88.BoundedSelfAdjoint
For any type equipped with a normed additive commutative group structure, a complex inner product space structure, and completeness, this denotes the type of continuous complex-linear operators satisfying , where is the Hilbert-space adjoint. There is no restriction on the dimension of , and the zero Hilbert space is included.
QAlgorithms.Papers.Independent.Intakethird100.P88.FinitePOVM
For a complex Hilbert space , any natural number , and a family of continuous complex-linear operators , this property means that every is self-adjoint and satisfies for all , and that
The value is included: then positivity is vacuous and the sum condition requires the zero operator to equal the identity, which holds precisely when is the zero space.
QAlgorithms.Papers.Independent.Intakethird100.P88.StrictlyPositiveQuadraticForm
For any complex Hilbert space and continuous complex-linear operator , this property means that , that for every , and that
for every . No uniform positive lower bound in terms of is required. The zero Hilbert space is included, where the condition concerning nonzero vectors is vacuous and its unique operator satisfies the property.
QAlgorithms.Papers.Independent.Intakethird100.P88.UniformlyPositive
For a complex Hilbert space and a bounded complex-linear operator , this predicate means that , that for every , and that there exists a real number such that
The same must work for all vectors, including . No assumption that is nonzero is imposed; on the zero space these conditions hold.
QAlgorithms.Papers.Independent.Intakethird100.P88.exponentialProgram
Let be a complex Hilbert space, any index type, and a Hilbert basis, with . An operator is trace class here when there exist sequences with summable and for every ; positivity means and for all . Given a positive trace-class operator with , any positive trace-class operator , and a Boolean parameter , this definition is
Each bracketed expression is calculated in , then embedded into the extended reals before taking the supremum. There is no trace-one or nonzero assumption on . In particular, is allowed, and the logarithmic branch uses the total real logarithm, for which .
QAlgorithms.Papers.Independent.Intakethird100.P88.logarithmicExpectation
Let be a complex Hilbert space, any index type, and a Hilbert basis. Write . Let be a bounded complex-linear operator that is self-adjoint, satisfies for every , has , and admits sequences , indexed by , such that is summable and for every . For any bounded complex-linear operator , this definition is the extended real number
where each real value is embedded into the extended reals before taking the infimum. The index includes , whose cutoff is . No positivity or self-adjointness hypothesis is imposed on ; the expression uses the total continuous-functional-calculus operation on every such input.
QAlgorithms.Papers.Independent.Intakethird100.P88.measuredKLSummand
For arbitrary real numbers , this defines an extended-real value by
The finite expression uses the total real logarithm: at a negative argument it equals the logarithm of the absolute value. No nonnegativity assumptions are imposed on or . In particular, the first branch takes precedence, so .
QAlgorithms.Papers.Independent.Intakethird100.P88.normPositiveTopology
Let be a complex Hilbert space. Call a continuous complex-linear operator trace class when it has a representation for every , with , and call it positive when and for every . Define
where the infimum is in , , and conversion to sends to . This defines the topology on positive trace-class operators induced by their inclusion into the trace-class operators, whose topology is itself induced by into a product of copies of . Equivalently, it is the coarsest topology making continuous for every trace-class operator . Positive operators of arbitrary trace, including zero, are included.
QAlgorithms.Papers.Independent.Intakethird100.P88.positiveProgram
Let be a complex Hilbert space, any index type, and a Hilbert basis, and put . Trace class means admitting sequences such that is summable and for every ; positivity means self-adjointness and nonnegative real quadratic form. Given a positive trace-class operator with , any positive trace-class operator , and Boolean parameters , define to consist of bounded complex-linear positive operators satisfying, if , the existence of a real with for every , and, if , the condition for every nonzero . The definition returns
The inner real trace values and the real correction term are embedded into the extended reals, where the infimum, addition, and outer supremum are taken. The inner index includes . No trace-one or nonzero condition is imposed on , and in the logarithmic correction.
QAlgorithms.Papers.Independent.Intakethird100.P88.traceClassMeasuredEntropyBits
Let be a complex Hilbert space, any type, and a Hilbert basis. A trace-class operator here admits a representation for every , with ; positivity means self-adjointness and nonnegative real quadratic form. Let be a positive trace-class operator with , where , and let be any positive trace-class operator, with no trace normalization required. This defines the extended-real quantity
where if , if and , and otherwise , using the total real logarithm. Thus a zero first argument contributes zero even when the second argument is also zero; is allowed. The supremum includes , although an empty operator family cannot sum to the identity on a nonzero , and no countability assumption on is imposed.
QAlgorithms.Papers.Independent.Intakethird100.P88.twoPower
For a complex Hilbert space and a bounded complex-linear self-adjoint operator , this definition returns the bounded operator obtained by applying continuous functional calculus to with the real function :
Here the input includes the condition . There is no positivity assumption on , and no nonzero or finite-dimensional assumption on .
QAlgorithms.Papers.Independent.Intakethird100.P88.weakOperatorPositiveTopology
For any complex Hilbert space , consider the continuous complex-linear operators that satisfy , for every , and admit sequences with and for every . This defines on these positive trace-class operators the coarsest topology making
continuous for each , by mapping into the product space of all these values. There is no trace normalization, and both the zero operator and the zero Hilbert space are allowed.
QAlgorithms.Papers.Independent.Intakethird100.P88.weakOperatorStateTopology
Let be a complex Hilbert space and any Hilbert basis. A state means a continuous complex-linear operator that has a representation for every , with , and satisfies , for every , and . This defines on the type of such states the topology induced by
where has the product topology; thus it is the coarsest topology making every displayed quadratic-form evaluation continuous. No countability or nonzero-dimension assumption is imposed; in zero dimension the state type is empty.
QAlgorithms.Papers.Independent.Intakethird100.P88.weakPositiveTopology
Let be a complex Hilbert space, any index type, and a Hilbert basis of . An operator is called trace class here if there exist sequences , indexed by , such that and for every ; it is positive if and for every . On the type of positive trace-class operators, this defines the coarsest topology making every real-valued function
continuous, as ranges over all continuous complex-linear self-adjoint operators on . The target function space carries the product topology. No trace normalization is imposed on , so the zero operator is included, and no countability assumption on is made.
QAlgorithms.Papers.Independent.Intakethird100.P88.weakStateTopology
Let be a complex Hilbert space and a Hilbert basis indexed by any type . A state here is a continuous complex-linear operator admitting sequences with and for every , satisfying , for every , and . This defines on these states the coarsest topology making
continuous for every continuous complex-linear self-adjoint operator , using the product topology on the collection of these real values. There is no countability assumption on ; if is zero, there are no states satisfying the trace-one condition.
QAlgorithms.Papers.Independent.Intakethird100.P88.zeroPositive
For any complex Hilbert space , this definition constructs the zero bounded complex-linear operator , packaged together with its trace-class and positivity properties. Its trace-class representation uses the two sequences for every , so that and for every . Its positivity properties are and for every . No trace-one condition is part of this construction.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.