Single-valuedness of the closed reflected graph
ProvedEthierKurtz.oblique_closure_singlevaluedmarkov-processesprobability
Single-valuedness of the closed reflected diffusion graph.
Let be a bounded connected open set with boundary, and let be uniformly elliptic with H"older coefficients. Let be a reflection field uniformly oblique to the outward unit normal on the boundary, and let be the operator graph defined by the interior equation (1.15) together with the oblique boundary condition of (1.19)--(1.20). With the closure of in , we have
That is, closing the graph never pairs one function with two distinct images; is the graph of a (single-valued) operator. This is the dissipativity / positive-maximum-principle half of Theorem 1.5.
Formalization Note Lean states the conclusion with and quantifies over bounded continuous functions on .
Preamble
import Definitions.Def_EthierKurtz_BoundaryCTwiceHolder import Definitions.Def_EthierKurtz_BoundaryCOnceHolder import Definitions.Def_EthierKurtz_IsOutwardUnitNormal import Definitions.Def_EthierKurtz_obliqueDiffusionGraph import Definitions.Def_EthierKurtz_IsStronglyContinuousContractionSemigroup open Filter open scoped Topology BoundedContinuousFunction namespace EthierKurtz
Formal statement
/-- Single-valuedness of the closed obliquely reflected diffusion graph:
under the hypotheses of Theorem 1.5, no continuous function on the closed
region is paired with two distinct images in the closure of the graph. -/
theorem oblique_closure_singlevalued (n : ℕ) (hd : 2 ≤ n + 1)
(Ω : Set (EuclideanSpace ℝ (Fin (n + 1))))
(hbounded : Bornology.IsBounded Ω) (hconnected : IsConnected Ω)
(hopen : IsOpen Ω) (μ : ℝ) (hμ : 0 < μ ∧ μ ≤ 1)
(hboundary : BoundaryCTwiceHolder Ω μ)
(a : EuclideanSpace ℝ (Fin (n + 1)) → Matrix (Fin (n + 1)) (Fin (n + 1)) ℝ)
(b c normal : EuclideanSpace ℝ (Fin (n + 1)) → EuclideanSpace ℝ (Fin (n + 1)))
(ha : ∀ x ∈ Ω, (a x).PosSemidef)
(haHolder : ∀ i j, ComponentHolder Ω μ (fun x => a x i j))
(hbHolder : ∀ i, ComponentHolder Ω μ (fun x => b x i))
(helliptic : ∃ ε : ℝ, 0 < ε ∧ ∀ x ∈ Ω,
∀ θ : EuclideanSpace ℝ (Fin (n + 1)), ‖θ‖ = 1 →
ε ≤ ∑ i, ∑ j, θ i * a x i j * θ j)
(hc : ∀ i, BoundaryCOnceHolder Ω μ (fun x => c x i))
(hnormal : ∀ x ∈ frontier Ω, IsOutwardUnitNormal Ω x (normal x))
(hoblique : ∃ ε : ℝ, 0 < ε ∧ ∀ x ∈ frontier Ω,
ε ≤ ∑ i, c x i * normal x i) :
let A := closure (obliqueDiffusionGraph Ω μ a b c)
∀ f g₁ g₂, (f, g₁) ∈ A → (f, g₂) ∈ A → g₁ = g₂ := by sorry
Source
Stewart N. Ethier and Thomas G. Kurtz, Markov Processes: Characterization and Convergence, Wiley, 1986. Chapter 8, Section 1, Theorem 1.5, printed p. 369 (PDF p. 378); graph (1.20) and boundary equations (1.19)--(1.20).