Simultaneous positive global extension of local invariant contact forms
ProvedContactTopology.simultaneous_global_extension_of_local_invariant_contact_formsLet be the closed unit disk and any open neighborhood of . Suppose
are smooth positive invariant contact forms on . Assume that their normal coefficients have the same cooriented zero set in , and that all coefficients agree on , where .
Then the two sets of coefficients admit simultaneous globally smooth, globally positive invariant extensions to . The extended coefficients agree with the original ones on a single open neighborhood of the whole closed disk, with . The extended normal coefficients have the same cooriented zero set globally, and the two extended forms agree outside the same set .
Here positivity is the actual derivative condition
Only smoothness and positivity on are assumed. The given functions may be arbitrary outside ; the construction does not assume a positive extension or a compression map.
This localization lemma allows invariant-coordinate disk realization results requiring globally positive representatives to be applied to forms given only near a compact disk. The extension preserves the actual coefficients on a neighborhood, so it also preserves their derivatives there. It does not construct contact representatives from arbitrary geometric divided foliations, transfer to curved surface charts, or prove Legendrian classification.
import Mathlib open Set open scoped ContDiff
theorem ContactTopology.simultaneous_global_extension_of_local_invariant_contact_forms
(N : Set (Fin 2 → ℝ)) (hN : IsOpen N) (hDN : ∀ x, (x 0^2+x 1^2) ≤ 1 → x ∈ N)
(u₀ u₁ : (Fin 2 → ℝ) → ℝ) (β₀ β₁ : (Fin 2 → ℝ) → (Fin 2 → ℝ))
(hu₀ : ContDiffOn ℝ ∞ u₀ N) (hu₁ : ContDiffOn ℝ ∞ u₁ N)
(hβ₀ : ContDiffOn ℝ ∞ β₀ N) (hβ₁ : ContDiffOn ℝ ∞ β₁ N)
(hzero : ∀ x ∈ N, u₀ x = 0 ↔ u₁ x = 0)
(hsign : ∀ x ∈ N, u₁ x ≠ 0 → 0 < u₀ x*u₁ x)
(K : Set (Fin 2 → ℝ)) (hKD : K ⊆ {x | (x 0^2+x 1^2) < 1})
(heq : ∀ x ∈ N, x ∉ K → u₀ x = u₁ x ∧ β₁ x = β₀ x) :
let Q : ((Fin 2 → ℝ) → ℝ) → ((Fin 2 → ℝ) → (Fin 2 → ℝ)) → (Fin 2 → ℝ) → ℝ := fun u β x =>
β x 0*fderiv ℝ u x (Pi.single 1 1)-β x 1*fderiv ℝ u x (Pi.single 0 1)+
u x*(fderiv ℝ β x (Pi.single 0 1) 1-fderiv ℝ β x (Pi.single 1 1) 0)
(∀ x ∈ N, 0 < Q u₀ β₀ x) → (∀ x ∈ N, 0 < Q u₁ β₁ x) →
∃ U₀ U₁ : (Fin 2 → ℝ) → ℝ, ∃ B₀ B₁ : (Fin 2 → ℝ) → (Fin 2 → ℝ), ∃ W : Set (Fin 2 → ℝ),
ContDiff ℝ ∞ U₀ ∧ ContDiff ℝ ∞ U₁ ∧ ContDiff ℝ ∞ B₀ ∧ ContDiff ℝ ∞ B₁ ∧
(∀ x, 0 < Q U₀ B₀ x) ∧ (∀ x, 0 < Q U₁ B₁ x) ∧
(∀ x, U₀ x = 0 ↔ U₁ x = 0) ∧ (∀ x, U₁ x ≠ 0 → 0 < U₀ x*U₁ x) ∧
(∀ x, x ∉ K → U₀ x = U₁ x ∧ B₁ x = B₀ x) ∧
IsOpen W ∧ (∀ x, (x 0^2+x 1^2) ≤ 1 → x ∈ W) ∧ W ⊆ N ∧
EqOn U₀ u₀ W ∧ EqOn U₁ u₁ W ∧ EqOn B₀ β₀ W ∧ EqOn B₁ β₁ W := by sorry