Definitions for Unconstrained distillation capacities of a pure-loss bosonic broadcast channel
DefinitionQAlgorithms_Papers_Independent_Support50_P48The objects the formalized statements of Unconstrained distillation capacities of a pure-loss bosonic broadcast channel (arXiv:1601.05563) are stated in. Each was checked against the paper by a blind read-back and a moderator.
portWeight: PDF pp.2,6:noneis the environment;some iis receiver i. The nonnegative subtraction agrees with real subtraction by total_le_one.vacuumAmplitude: PDF pp.2,6: the vacuum splitting isometry on the entire number basis. This is the normalized multinomial expansion of the input creation operator; there is no global photon cutoff. All output signal phases are positive.tmsvWeight: PDF p.3, (7): geometric Schmidt probabilities, with finite mean N.ResourceMode: Full mode labels:noneis Alice,some noneis the environment, andsome (some i)is receiver i; PDF pp.3,6, (11),(29).resourceAmplitude: PDF pp.3,6, (7),(11),(29): the actual purified TMSV channel-output amplitude, obtained by applying the vacuum splitting isometry to the input half.purifiedResource: PDF pp.3,6, (11),(29): rank-one purified resource coefficients, before tracing the environment. Normalization is a conclusion of the entropy theorem.subsystem: PDF pp.3,6: actual partial trace over the complementary modes. On all resource states used here the diagonal trace series is convergent.receiverModes: The selected receiver modes in the purified resource, PDF p.6.aliceWith: Alice together with selected receivers; PDF pp.3,6.environmentWith: Environment together with selected receivers; PDF pp.3,6.resourceEntropyBits: PDF pp.3,6: spectral von Neumann entropy of an actual reduced resource, converted from the accepted extended entropy in nats to bits.reverseCutYield: PDF pp.3,6: negative conditional entropy, computed from actual marginals. The entropy theorem supplies finiteness before either toReal is interpreted.extendedCutBound: PDF pp.2,6, (4)-(6),(27): extended endpoint interpretation of the cut bound, inferred from the finite-energy entropy limit. A zero selected signal has bound zero even when the environmental weight is zero.cutOccupation: PDF pp.3,6: reconstruct the full occupation at the T : A complement(T) cut, including the environment that will be traced out.resourceCut: PDF pp.4,6: the actual resource after tracing E, grouped across the selected receiver cut. The right factor includes Alice and all other receivers.relativeEntanglementBits: PDF p.3, (14): infimum of actual extended relative entropy over all normal separable states, using trace-norm closed separability; measured in bits. Arbitrary shields are not bounded and the infimum is extended, not real.residualIntensity: PDF p.6, (30): remaining input intensity before zero-based stage j. The ordering permutes all m+1 ports, including the environment.emittedPhotons: Number of photons already emitted before stage j, PDF p.6, Fig.3.sequentialAmplitude: PDF p.6, Fig.3,(30): the full vacuum network amplitude, with m actual splitters. Each factor is the normalized binomial vacuum splitting amplitude; the final remaining port is fixed by total photon conservation.referencedVacuumOutput: PDF pp.2,6: the full referenced output of a photon-conserving vacuum isometry. Input number indices are fixed by total output photon numbers, so no formal infinite matrix product is used.
import Mathlib
import Definitions.Def_QAlgorithms_Papers_Independent_Dependency50_P30
import Definitions.Def_QAlgorithms_Papers_Independent_Frontier18_P07
/-!
# Pure-loss bosonic broadcast channels (arXiv:1601.05563v3)
The unrestricted operational capacity and merging/simulation interfaces are
explicitly deferred in HARD.md. The three statements here concern actual
number-basis resource states, their entropies/REE, and vacuum network equality.
No capacity is defined to equal a scalar bound. Compilation is pending.
All signal amplitudes use a positive output phase convention. Fixed local phase
rotations recover the minus signs of (1)-(3); these do not change any entropy or
bipartite REE. Occupations run over all natural numbers, without truncation.
-/
noncomputable section
namespace QAlgorithms.Papers.Independent.Support50.P48
open Filter
open scoped BigOperators ENNReal NNReal Topology
open QAlgorithms.Papers.Independent.Dependency50.P30
(PureLossBroadcastParameters thermalEntropyBits)
open QAlgorithms.Papers.Independent.Dependency50.P33
(State IsDensity IsSeparable)
open QAlgorithms.Papers.Independent.Frontier18.P07
(extendedEntropyNats extendedRelativeEntropyNats)
attribute [local instance] Classical.propDecidable Classical.decEq
/-- PDF pp.2,6: `none` is the environment; `some i` is receiver i.
The nonnegative subtraction agrees with real subtraction by total_le_one. -/
def portWeight {m : ℕ} (p : PureLossBroadcastParameters m) : Option (Fin m) → ℝ≥0
| none => 1 - ∑ i, p.transmissivity i
| some i => p.transmissivity i
/-- PDF pp.2,6: the vacuum splitting isometry on the entire number basis.
This is the normalized multinomial expansion of the input creation operator;
there is no global photon cutoff. All output signal phases are positive. -/
def vacuumAmplitude {J : Type} [Fintype J] (eta : J → ℝ≥0)
(out : J → ℕ) (n : ℕ) : ℝ :=
if (∑ i, out i) = n then
Real.sqrt ((Nat.factorial n : ℝ) / ∏ i, (Nat.factorial (out i) : ℝ)) *
∏ i, Real.sqrt (eta i : ℝ) ^ out i
else 0
/-- PDF p.3, (7): geometric Schmidt probabilities, with finite mean N. -/
def tmsvWeight (N : ℝ≥0) (n : ℕ) : ℝ :=
(N : ℝ) ^ n / ((N : ℝ) + 1) ^ (n + 1)
/-- Full mode labels: `none` is Alice, `some none` is the environment,
and `some (some i)` is receiver i; PDF pp.3,6, (11),(29). -/
abbrev ResourceMode (m : ℕ) := Option (Option (Fin m))
/-- PDF pp.3,6, (7),(11),(29): the actual purified TMSV channel-output
amplitude, obtained by applying the vacuum splitting isometry to the input half. -/
def resourceAmplitude {m : ℕ} (p : PureLossBroadcastParameters m) (N : ℝ≥0)
(out : ResourceMode m → ℕ) : ℝ :=
Real.sqrt (tmsvWeight N (out none)) *
vacuumAmplitude (portWeight p) (fun j => out (some j)) (out none)
/-- PDF pp.3,6, (11),(29): rank-one purified resource coefficients, before
tracing the environment. Normalization is a conclusion of the entropy theorem. -/
def purifiedResource {m : ℕ} (p : PureLossBroadcastParameters m) (N : ℝ≥0) :
Matrix (ResourceMode m → ℕ) (ResourceMode m → ℕ) ℂ :=
fun a b => (resourceAmplitude p N a : ℂ) * (resourceAmplitude p N b : ℂ)
/-- PDF pp.3,6: actual partial trace over the complementary modes. On all
resource states used here the diagonal trace series is convergent. -/
def subsystem {J : Type} (S : Finset J) (X : Matrix (J → ℕ) (J → ℕ) ℂ) :
Matrix (S → ℕ) (S → ℕ) ℂ :=
fun a b => ∑' e : {i : J // i ∉ S} → ℕ,
X (fun i => if h : i ∈ S then a ⟨i, h⟩ else e ⟨i, h⟩)
(fun i => if h : i ∈ S then b ⟨i, h⟩ else e ⟨i, h⟩)
/-- The selected receiver modes in the purified resource, PDF p.6. -/
def receiverModes {m : ℕ} (T : Finset (Fin m)) : Finset (ResourceMode m) :=
T.image (fun i => some (some i))
/-- Alice together with selected receivers; PDF pp.3,6. -/
def aliceWith {m : ℕ} (T : Finset (Fin m)) : Finset (ResourceMode m) :=
insert none (receiverModes T)
/-- Environment together with selected receivers; PDF pp.3,6. -/
def environmentWith {m : ℕ} (T : Finset (Fin m)) : Finset (ResourceMode m) :=
insert (some none) (receiverModes T)
/-- PDF pp.3,6: spectral von Neumann entropy of an actual reduced resource,
converted from the accepted extended entropy in nats to bits. -/
def resourceEntropyBits {m : ℕ} (p : PureLossBroadcastParameters m) (N : ℝ≥0)
(S : Finset (ResourceMode m)) : ℝ≥0∞ :=
extendedEntropyNats (subsystem S (purifiedResource p N)) /
ENNReal.ofReal (Real.log 2)
/-- PDF pp.3,6: negative conditional entropy, computed from actual marginals.
The entropy theorem supplies finiteness before either toReal is interpreted. -/
def reverseCutYield {m : ℕ} (p : PureLossBroadcastParameters m) (N : ℝ≥0)
(T : Finset (Fin m)) : ℝ :=
(resourceEntropyBits p N (aliceWith (Finset.univ \ T))).toReal -
(resourceEntropyBits p N (aliceWith Finset.univ)).toReal
/-- PDF pp.2,6, (4)-(6),(27): extended endpoint interpretation of the cut
bound, inferred from the finite-energy entropy limit. A zero selected signal
has bound zero even when the environmental weight is zero. -/
def extendedCutBound {m : ℕ} (p : PureLossBroadcastParameters m)
(T : Finset (Fin m)) : ℝ≥0∞ :=
let selected : ℝ≥0 := ∑ i ∈ T, p.transmissivity i
let loss := portWeight p none
if selected = 0 then 0 else if loss = 0 then ⊤ else
ENNReal.ofReal (Real.logb 2 (((loss : ℝ) + (selected : ℝ)) / (loss : ℝ)))
/-- PDF pp.3,6: reconstruct the full occupation at the T : A complement(T)
cut, including the environment that will be traced out. -/
def cutOccupation {m : ℕ} (T : Finset (Fin m)) (a : T → ℕ)
(b : ℕ × ({i : Fin m // i ∉ T} → ℕ)) (e : ℕ) : ResourceMode m → ℕ
| none => b.1
| some none => e
| some (some i) => if h : i ∈ T then a ⟨i, h⟩ else b.2 ⟨i, h⟩
/-- PDF pp.4,6: the actual resource after tracing E, grouped across the
selected receiver cut. The right factor includes Alice and all other receivers. -/
def resourceCut {m : ℕ} (p : PureLossBroadcastParameters m) (N : ℝ≥0)
(T : Finset (Fin m)) :
Matrix ((T → ℕ) × (ℕ × ({i : Fin m // i ∉ T} → ℕ)))
((T → ℕ) × (ℕ × ({i : Fin m // i ∉ T} → ℕ))) ℂ :=
fun a b => ∑' e : ℕ, purifiedResource p N
(cutOccupation T a.1 a.2 e) (cutOccupation T b.1 b.2 e)
/-- PDF p.3, (14): infimum of actual extended relative entropy over all
normal separable states, using trace-norm closed separability; measured in bits.
Arbitrary shields are not bounded and the infimum is extended, not real. -/
def relativeEntanglementBits {I J : Type} (X : Matrix (I × J) (I × J) ℂ) : ℝ≥0∞ :=
(⨅ sigma : {rho : State (I × J) // IsSeparable rho.coefficients},
extendedRelativeEntropyNats X sigma.val.coefficients) / ENNReal.ofReal (Real.log 2)
/-- PDF p.6, (30): remaining input intensity before zero-based stage j.
The ordering permutes all m+1 ports, including the environment. -/
def residualIntensity {m : ℕ} (eta : Fin (m + 1) → ℝ≥0)
(order : Equiv.Perm (Fin (m + 1))) (j : ℕ) : ℝ :=
1 - ∑ k ∈ Finset.univ.filter (fun k : Fin (m + 1) => k.val < j), (eta (order k) : ℝ)
/-- Number of photons already emitted before stage j, PDF p.6, Fig.3. -/
def emittedPhotons {m : ℕ} (order : Equiv.Perm (Fin (m + 1)))
(out : Fin (m + 1) → ℕ) (j : ℕ) : ℕ :=
∑ k ∈ Finset.univ.filter (fun k : Fin (m + 1) => k.val < j), out (order k)
/-- PDF p.6, Fig.3,(30): the full vacuum network amplitude, with m actual
splitters. Each factor is the normalized binomial vacuum splitting amplitude;
the final remaining port is fixed by total photon conservation. -/
def sequentialAmplitude {m : ℕ} (t : Fin m → ℝ)
(order : Equiv.Perm (Fin (m + 1))) (out : Fin (m + 1) → ℕ) (n : ℕ) : ℝ :=
if (∑ i, out i) = n then
∏ j : Fin m,
Real.sqrt (Nat.choose (n - emittedPhotons order out j.val)
(out (order j.castSucc)) : ℝ) *
Real.sqrt (1 - t j) ^ out (order j.castSucc) *
Real.sqrt (t j) ^ (n - emittedPhotons order out j.val - out (order j.castSucc))
else 0
/-- PDF pp.2,6: the full referenced output of a photon-conserving vacuum
isometry. Input number indices are fixed by total output photon numbers, so
no formal infinite matrix product is used. -/
def referencedVacuumOutput {J R : Type} [Fintype J]
(amplitude : (J → ℕ) → ℕ → ℝ) (X : Matrix (R × ℕ) (R × ℕ) ℂ) :
Matrix (R × (J → ℕ)) (R × (J → ℕ)) ℂ :=
fun a b => (amplitude a.2 (∑ i, a.2 i) : ℂ) *
X (a.1, ∑ i, a.2 i) (b.1, ∑ i, b.2 i) *
(amplitude b.2 (∑ i, b.2 i) : ℂ)
/-- The vacuum sector has amplitude one, even if some port weights vanish. -/
example {J : Type} [Fintype J] (eta : J → ℝ≥0) :
vacuumAmplitude eta (fun _ => 0) 0 = 1 := by
simp [vacuumAmplitude]
/-- The TMSV zero-photon probability is 1/(N+1), including N=0. -/
example (N : ℝ≥0) : tmsvWeight N 0 = 1 / ((N : ℝ) + 1) := by
simp [tmsvWeight]
/-- The empty cut never becomes infinite at the lossless endpoint. -/
example {m : ℕ} (p : PureLossBroadcastParameters m) : extendedCutBound p ∅ = 0 := by
simp [extendedCutBound]
/-- The environment coordinate is retained explicitly in the cut reconstruction. -/
example {m : ℕ} (T : Finset (Fin m)) (a : T → ℕ)
(b : ℕ × ({i : Fin m // i ∉ T} → ℕ)) (e : ℕ) :
cutOccupation T a b e (some none) = e := by rfl
/-- With no photons, every physical splitter acts trivially on the vacuum. -/
example {m : ℕ} (t : Fin m → ℝ) (order : Equiv.Perm (Fin (m + 1))) :
sequentialAmplitude t order (fun _ => 0) 0 = 1 := by
simp [sequentialAmplitude, emittedPhotons]
end QAlgorithms.Papers.Independent.Support50.P48
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
QAlgorithms.Papers.Independent.Support50.P48.ResourceMode
For each natural number , this abbreviates the type of labels consisting of one outer absent value , one outer present value containing an inner absent value , and one outer present value containing an inner present index for each . These are distinct labels, so the type has elements. In particular, when it still has the two labels and .
QAlgorithms.Papers.Independent.Support50.P48.aliceWith
For every and every finite subset , this defines the finite set
in the label type whose elements are (outer absent), (outer present with inner absent), and (outer present with inner present index ). The set contains and exactly the doubly present labels indexed by ; it does not contain . For , including when , it is the singleton .
QAlgorithms.Papers.Independent.Support50.P48.cutOccupation
For every natural number , let the mode labels consist of a distinguished label , a distinguished label , and receiver labels for ; these correspond respectively to none, some none, and some (some i). For any subset of the receiver indices, any function , any pair with and , and any , the definition produces the occupation function
There is no restriction on these occupations or their sum. Empty , empty complement, and are all allowed; a function on an empty index set is the unique empty function.
QAlgorithms.Papers.Independent.Support50.P48.emittedPhotons
For every natural number , every permutation of , every occupation function , and every natural number , the definition gives
No total occupation is prescribed. This includes ; , and for every , .
QAlgorithms.Papers.Independent.Support50.P48.environmentWith
For every and every finite subset , this defines the finite set
in the label type whose elements are (outer absent), (outer present with inner absent), and (outer present with inner present index ). The set contains and exactly the doubly present labels indexed by ; it does not contain . For , including when , it is the singleton .
QAlgorithms.Papers.Independent.Support50.P48.extendedCutBound
For every natural number , including , let specify nonnegative real numbers for with , and let be any subset of these indices, including the empty subset. Put and . The definition gives the following value in :
The subtraction defining is originally nonnegative-real subtraction, which agrees here with ordinary subtraction because . The first branch takes precedence when both and vanish. In particular, , and necessarily , give value zero.
QAlgorithms.Papers.Independent.Support50.P48.portWeight
For every natural number and every family of nonnegative real numbers satisfying , supplied as the transmissivities of , this defines a nonnegative real weight on the labels consisting of an additional label and the indices : the weight of is , and the weight of is . Subtraction takes place in the nonnegative reals, where it is truncated at zero; the assumed bound on the sum makes this the ordinary nonnegative difference. The case is included: there is only the label , whose weight is .
QAlgorithms.Papers.Independent.Support50.P48.purifiedResource
For every , every parameter family with nonnegative transmissivities satisfying , and every nonnegative real , this defines a complex matrix indexed in both directions by all occupation functions on the distinct labels for , corresponding respectively to the outer absent, outer present and inner absent, and doubly present labels. Put , and for each occupation function define
The matrix entry at occupation functions is , with both real factors embedded into . The definition multiplies these factors directly. Empty products are , empty sums are , and , so and vanishing transmissivities are included. At , its only nonzero entry is at the pair of everywhere-zero occupation functions.
QAlgorithms.Papers.Independent.Support50.P48.receiverModes
For every and every finite subset , this defines the finite subset
of the label type consisting of (the outer absent label), (outer present with inner absent), and (outer present with inner present index ). Thus it contains exactly the doubly present labels whose indices belong to . It is empty when is empty, including necessarily when .
QAlgorithms.Papers.Independent.Support50.P48.referencedVacuumOutput
For every finite type , every type , every arbitrary real-valued function , and every complex matrix indexed by , define for an occupation function . The definition produces the complex matrix indexed by with entries
where both real amplitude factors are embedded in . There are no normalization, positivity, or density assumptions on or . Empty and empty are allowed; if is empty, its unique occupation function has total occupation zero.
QAlgorithms.Papers.Independent.Support50.P48.relativeEntanglementBits
For any types , with no countability or nonemptiness assumptions, and any complex matrix indexed by , the definition assigns a value in . To describe it, call a matrix on an index set a density matrix precisely when it admits sequences with and , every finite quadratic form is real and nonnegative, and the real diagonal entries are summable with . For any matrix on , put
where the infimum of an empty family in is . A density matrix on is eligible for the infimum below precisely when, for every , there is a finite real convex combination of matrices with entries , where and are density matrices on and , such that . For finite , write for the corresponding principal submatrices and define
Here matrix logarithms denote the continuous functional calculus of the real logarithm used in the definition, including its total-function conventions; the scalar real logarithm has . The requested value is
No density, positivity, or summability assumption is imposed on . The supremum includes the empty principal submatrix, and if there are no eligible density matrices, the result is ; in particular, this occurs when is empty.
QAlgorithms.Papers.Independent.Support50.P48.residualIntensity
For every natural number , every function , every permutation of , and every natural number , the definition gives the real number
There is no assumption that the weights sum to one or at most one, so may be negative. The definition includes and all : , and for it equals .
QAlgorithms.Papers.Independent.Support50.P48.resourceAmplitude
For every , every parameter family with nonnegative transmissivities for and , every nonnegative real , and every occupation function on the labels —respectively the outer absent label, the outer present and inner absent label, and the doubly present label indexed by —put and . This defines the real number
All square roots are real square roots, and powers include . The case uses empty products and sums, leaving the condition . When , the value is on the occupation function that is zero everywhere and on every other occupation function.
QAlgorithms.Papers.Independent.Support50.P48.resourceCut
For every natural number , let for satisfy , let , and let . Set , and use mode labels . For an occupation function on these labels, define the real number
For , , and , let assign to , to , to in , and to the other receiver labels. The definition produces the complex matrix indexed by such pairs whose entries are
with the real factors embedded in . This is a definition for all the stated inputs, without a summability hypothesis; the formal infinite sum is zero if its summands are not summable. It includes , , and empty or complement, using empty products equal to one and zeroth powers, including , equal to one.
QAlgorithms.Papers.Independent.Support50.P48.resourceEntropyBits
For every , every parameter family with nonnegative transmissivities satisfying , every nonnegative real , and every finite subset of the labels , where these labels are respectively outer absent, outer present with inner absent, and doubly present with index , this defines an extended nonnegative real number as follows. Put and, for each , set
For occupations , define the complex matrix
where joins the occupations on the two disjoint sets, the real products are embedded into , and the infinite sum is totalized, with value for a nonsummable family. For each finite set of occupations , let . The defined value is
where is regarded as an extended nonnegative real, is the continuous functional calculus expression for the real natural logarithm, and the supremum and division are in the extended nonnegative reals. The real logarithm is totalized, in particular . The supremum includes the empty , and its value may be . The definition includes , , , and zero transmissivities, using empty products equal to , empty sums equal to , and .
QAlgorithms.Papers.Independent.Support50.P48.reverseCutYield
For every , every parameter family with nonnegative transmissivities satisfying , every nonnegative real , and every finite subset , this defines a real number using the distinct labels , corresponding respectively to the outer absent, outer present and inner absent, and doubly present labels. Put , and for put
For each subset , define for occupations , where combines occupations on disjoint sets and the sum is the totalized complex infinite sum, equal to if its family is not summable. For a finite set of occupations on , write , and define
Here is embedded into the extended nonnegative reals, denotes continuous functional calculus for the real natural logarithm, and the supremum and division defining take place in the extended nonnegative reals. The real number defined by the declaration is
where sends a finite extended nonnegative real to its real value and sends to . Thus the subtraction uses these real conversions even when an entropy is infinite. All finite subsets , including the empty set, enter the suprema; empty products and sums are and , , and . The cases , , and zero transmissivities are included. For , including necessarily when , the two converted quantities coincide and the defined value is .
QAlgorithms.Papers.Independent.Support50.P48.sequentialAmplitude
For every natural number , every real-valued function on , every permutation of , every occupation function , and every , put for . The definition gives the real amplitude
where is subtraction in , and a binomial coefficient is zero when its lower argument exceeds its upper argument. No assumption is imposed: the real square root is zero on negative inputs. Zeroth powers are one, including . When , the product is empty, so the amplitude is one if and zero otherwise.
QAlgorithms.Papers.Independent.Support50.P48.subsystem
For any type , any finite subset , and any complex matrix indexed by all functions , this defines a complex matrix indexed by functions . For and , let be the function on equal to on and to outside , and define similarly. The entry at is
where this is the totalized infinite sum over all such functions . There are no finiteness or countability assumptions on , and no summability, positivity, or normalization assumptions on ; for a nonsummable family of complex entries, the totalized sum is . If is empty, the resulting matrix has one entry, namely the totalized sum of all diagonal entries of . If contains every element of , the complement has a unique occupation function, and the formula recovers under the natural identification of indices.
QAlgorithms.Papers.Independent.Support50.P48.tmsvWeight
For every nonnegative real number and natural number , this defines the real number
The denominator is strictly positive. The value at is ; when , the value is at and at every positive , using .
QAlgorithms.Papers.Independent.Support50.P48.vacuumAmplitude
For any type equipped with a finite enumeration, any nonnegative real weights , any occupation function , and any , this defines the real number
There is no assumption that the weights sum to one or are individually at most one. Empty sums and products are respectively and , so when is empty the value is for and otherwise. Powers use the convention ; in particular, when every occupation and are zero, the value is even if some weights vanish. The factorial denominator is always positive.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.