Proof of Proposition 3.1, Existence — the successive approximations (A.3) converge uniformly on bounded intervals to a càdlàg solution of (3.1)
ProvedReedGGN.Regulator.existencep2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1regulator-mapsuccessive-approximations
Let , , , and be as in the uniqueness milestone: for all and . Let be càdlàg and let be the successive approximations (A.3):
Then there is a càdlàg path such that as for every , and solves (3.1).
This is the existence half of Proposition 3.1 in the non-degenerate case, by the method of successive approximations.
Formalization Note The paper writes (A.3) "for "; the recursion starts at . The paper's limit is taken "under the supremum metric ", i.e. uniformly on every bounded interval , which is what is stated.
Preamble
import Mathlib import Definitions.Def_ReedGGN_Regulator_PathSpace import Definitions.Def_ReedGGN_Regulator_Equation
Formal statement
namespace ReedGGN.Regulator
open MeasureTheory Filter Topology
/-- Proof of Proposition 3.1, Existence (pp. 32–34): under the window condition of the
non-degenerate case, for every càdlàg input `x` the successive approximations (A.3),
`u_0 = 0`, `u_{n+1} = x + Ψ^a_B(u_n)`, converge uniformly on every `[0, T]` to a càdlàg
path `u⋆`, and `u⋆` solves (3.1). -/
theorem existence (μ : Measure ℝ) [IsProbabilityMeasure μ] (a δ ε : ℝ)
(hδ : 0 < δ) (hε₀ : 0 < ε) (hε₁ : ε < 1)
(hB₀ : μ (Set.Icc 0 δ) < ENNReal.ofReal ε)
(hB : ∀ y, 0 ≤ y → μ (Set.Ioc y (y + δ)) < ENNReal.ofReal ε)
(x : ℝ → ℝ) (hx : IsCadlag x) :
∃ uStar : ℝ → ℝ, IsCadlag uStar ∧
(∀ T, 0 ≤ T → Tendsto (fun n => supNorm T (picard μ a x n - uStar)) atTop (𝓝 0)) ∧
SolvesRegulator μ a x uStar := by sorry
end ReedGGN.Regulator
Source
Reed, The G/GI/N Queue in the Halfin–Whitt Regime, arXiv:0912.2837v1, pp. 32–34, proof of Proposition 3.1 (Existence), Eq. (A.3)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.