Boundary maximum-principle step for resolvent positivity
DisprovedEthierKurtz.resolvent_positive_boundary_stepThis is the boundary step of the positive maximum principle for the resolvent problem with oblique boundary condition.
Let have twice-H"older boundary, and let be a frontier point where a function , twice continuously differentiable inside and continuous up to the closure, attains its global minimum . Suppose the oblique boundary data are satisfied at : the boundary gradient annihilates the obliquefield , and is uniformly transverse to the outward unit normal. Suppose further that the elliptic identity and the resolvent inequality hold as stated. Then this situation is impossible.
This is the Hopf-boundary-point step of the argument: at a boundary minimum, either the Hopf lemma forces a strictly positive outward normal derivative, contradicting the homogeneous oblique condition, or the degenerate case is ruled out through the operator identity. It isolates the entire boundary analysis of the resolvent-positivity proof in one reusable statement.
Formalization Note Boundary regularity and the outward normal are the mission predicates BoundaryCTwiceHolder and IsOutwardUnitNormal; the boundary gradient is the value at of the continuous extension supplied by EthierKurtz.oblique_boundary_differentiability_step.
import Mathlib import Definitions.Def_EthierKurtz_BoundaryCTwiceHolder import Definitions.Def_EthierKurtz_IsOutwardUnitNormal open Filter open scoped Topology BoundedContinuousFunction
namespace EthierKurtz
theorem resolvent_positive_boundary_step {n : ℕ}
{Ω : Set (EuclideanSpace ℝ (Fin (n + 1)))} {μ : ℝ}
{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))}
{u g : EuclideanSpace ℝ (Fin (n + 1)) → ℝ} {x₀ : EuclideanSpace ℝ (Fin (n + 1))}
{J0 : EuclideanSpace ℝ (Fin (n + 1)) →L[ℝ] ℝ} {lam : ℝ}
(hx : x₀ ∈ frontier Ω)
(hboundary : BoundaryCTwiceHolder Ω μ) (hμ : 0 < μ ∧ μ ≤ 1)
(hC : ContDiffOn ℝ 2 u Ω)
(hcont : ContinuousOn u (closure Ω))
(hcontG : ContinuousOn g (closure Ω))
(hD : HasFDerivAt u J0 x₀)
(hJ0b : J0 (c x₀) = 0)
(hnor : IsOutwardUnitNormal Ω x₀ (normal x₀))
(hob : ∃ ε : ℝ, 0 < ε ∧ ε ≤ ∑ i, c x₀ i * normal x₀ i)
(hIdΩ : ∀ x ∈ Ω, g x = (1 / 2 : ℝ) * (∑ i : Fin (n + 1), ∑ j : Fin (n + 1),
a x i j * fderiv ℝ (fun y => fderiv ℝ u y (EuclideanSpace.single j 1)) x
(EuclideanSpace.single i 1)) + fderiv ℝ u x (b x))
(ha : ∀ x ∈ Ω, (a x).PosSemidef)
(hell : ∃ 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)
(hminglob : ∀ y ∈ closure Ω, u x₀ ≤ u y)
(hu0 : u x₀ < 0)
(hlam : 0 < lam)
(hop : 0 ≤ lam * u x₀ - g x₀) :
False := by sorry
end EthierKurtz