Solvability with an bound from an estimate for the formal adjoint
ProvedCauchyRiemannIndex.exists_solution_of_adjoint_estimateLet be continuous loops of symmetric matrices and let be continuous with uniformly as . Let be continuous with for all . Write and
for its formal adjoint. Suppose that for some
Then for every there is with and .
This turns an a priori estimate for the adjoint into surjectivity of with a bounded solution operator into ; need not be asymptotic to anything.
Formalization Note. ; maps on the cylinder are maps of that are -periodic in , and integrals over are over the strip (cylMeasure). Test sections are IsTestSection (smooth, -periodic in , zero for large ); are partialS, partialT, and matrices act through Matrix.toEuclideanLin. is W1p 2 n (triples ), is crApply (fun x => S x + B x) w, and the conclusion bounds the norm of the first component w.1. The bound on is stated for the operator Matrix.toEuclideanLin (B x).
import Definitions.Def_CauchyRiemannIndex_Setting import Mathlib.Analysis.Calculus.ContDiff.Basic import Mathlib.MeasureTheory.Integral.Bochner.Basic
namespace CauchyRiemannIndex
open MeasureTheory Matrix Filter Topology
open scoped ENNReal NNReal ContDiff RealInnerProductSpace
/-- Let `D_S` be asymptotic to loops `S±` of symmetric matrices and let `B` be a continuous
bounded matrix-valued function on the cylinder. If the formal adjoint
`φ ↦ -∂ₛφ + J₀ ∂ₜφ + (S + B)ᵀ φ` of `D_{S+B}` satisfies `‖φ‖₂² ≤ C ‖D_{S+B}^* φ‖₂²` for every
test section `φ`, then for every `f ∈ L²` there is `w ∈ W^{1,2}` with `D_{S+B} w = f` and
`‖w‖₂ ≤ √C ‖f‖₂`. -/
theorem exists_solution_of_adjoint_estimate (n : ℕ) (Splus Sminus : ℝ → Mat n)
(hSplus : IsSymLoop Splus) (hSminus : IsSymLoop Sminus)
(S : ℝ × ℝ → Mat n) (hS : IsAsymptoticTo S Splus Sminus)
(B : ℝ × ℝ → Mat n) (hBc : Continuous B)
(hBb : ∃ M : ℝ, 0 ≤ M ∧ ∀ x (v : Vec n), ‖Matrix.toEuclideanLin (B x) v‖ ≤ M * ‖v‖)
(C : ℝ) (hC : 0 ≤ C)
(hest : ∀ φ : ℝ × ℝ → Vec n, IsTestSection φ →
∫ x, ‖φ x‖ ^ 2 ∂cylMeasure ≤
C * ∫ x, ‖-partialS φ x + Matrix.toEuclideanLin (J₀ n) (partialT φ x) +
Matrix.toEuclideanLin (S x + B x)ᵀ (φ x)‖ ^ 2 ∂cylMeasure)
(f : LpCyl n 2) :
∃ w ∈ W1p 2 n, crApply (fun x => S x + B x) w =ᵐ[cylMeasure] (f : ℝ × ℝ → Vec n) ∧
‖w.1‖ ≤ Real.sqrt C * ‖f‖ := by
sorry
end CauchyRiemannIndex