Proposition 4(i) — Lipschitz dependence of the controlled diffusion on its initial point
ProvedManyServerQED.Scheduling.proposition_4_idiffusiongronwallp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1stochastic-differential-equations
Let be diffusion data. There is a constant , depending only on these data, with the following property. For every admissible system , all , and controlled processes (for and ) and (for and the same ),
Here is the norm.
This estimate is the input to the continuity of the value function (Proposition 5(ii)).
Formalization Note The paper writes and for vectors; we read as the paper's norm . The constant is quantified before the system, the initial points and the time.
Preamble
import Mathlib import Definitions.Def_ManyServerQED_Scheduling_Diffusion open MeasureTheory ProbabilityTheory Filter Topology open scoped ENNReal NNReal
Formal statement
namespace ManyServerQED.Scheduling
/-- Proposition 4(i) (p. 23): there is a constant `c`, depending only on the diffusion data, such that
for every admissible system `π` and controlled processes `X`, `X̄` associated with `(x, π)` and
`(x̄, π)`, `P`-a.s. `‖X_t − X̄_t‖ ≤ ‖x − x̄‖(1 + e^{ct})` for all `t ≥ 0` (ℓ¹ norm). -/
theorem proposition_4_i {k : ℕ} [NeZero k] (D : DiffusionData k) :
∃ c : ℝ, ∀ (π : AdmissibleSystem k) (x xbar : Fin k → ℝ) (X Xbar : ℝ≥0 → π.Ω → Fin k → ℝ),
IsControlledProcess D π x X → IsControlledProcess D π xbar Xbar →
∀ᵐ ω ∂π.P, ∀ t : ℝ≥0,
l1norm (X t ω - Xbar t ω) ≤ l1norm (x - xbar) * (1 + Real.exp (c * t)) := by sorry
end ManyServerQED.Scheduling
Source
Atar, Mandelbaum & Reiman, Scheduling a Multi Class Queue with Many Exponential Servers: Asymptotic Optimality in Heavy Traffic, arXiv:math/0407058v1 (reprint of Ann. Appl. Probab. 14(3), 2004), p. 23, Proposition 4(i)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.