Iterated strong-loss and large-modulation asymptotic
OpenQAlgorithms.Papers.Independent.Intakenext100.P72.strongLossLargeModulationFor and , let and denote the coherent and EPR signed expressions at pure-loss noise , defined in (54) and (56). Their right-hand strong-loss coefficients agree:
The normalized quotient is tested only at positive . This pins the source's approximation by first taking the strong-loss coefficient at finite modulation and then taking large modulation. It does not replace the finite- coefficient by a constant or assert capacity optimality.
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P72
noncomputable section
open Filter
open scoped Topology
namespace QAlgorithms.Papers.Independent.Intakenext100.P72
/-- Grosshans--Grangier, PDF p.5, IV.B, (59).
The source's approximation is the iterated strong-loss/large-modulation limit.
The G→0 limit is one-sided and excludes G=0 from the normalized quotient.
The finite-V coefficient retains the factor (1-1/V). -/
theorem strongLossLargeModulation :
(∀ V : ℝ, 1 ≤ V →
Tendsto (fun G : ℝ => eprExpression G ((1 - G) / G) V / G)
(nhdsWithin 0 (Set.Ioi 0)) (nhds (strongLossCoefficient V)) ∧
Tendsto (fun G : ℝ => coherentExpression G ((1 - G) / G) V / G)
(nhdsWithin 0 (Set.Ioi 0)) (nhds (strongLossCoefficient V))) ∧
Tendsto strongLossCoefficient atTop (nhds (1 / (2 * Real.log 2))) := by sorry
end QAlgorithms.Papers.Independent.Intakenext100.P72
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Define for every real , using total real division, so that . For every real , both limits below hold as tends to zero through strictly positive real values:
In addition,
The two limits at exclude itself. The case is included, and then both limiting values are zero.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.