Documentation

LeanPool.EllipticPDE.Regularity.Local.Classical

Interior regularity with coefficients on the domain #

Evans, Partial Differential Equations (2nd ed.), §6.3.1 poses its interior regularity theorems on a bounded open U ⊂ ℝⁿ for coefficients given on U alone, with no bound on a^{ij} over U, and for u ∈ H¹(U) with no boundary condition. The local chain of this development (higher_interior_regularity_W12) runs on a global FullEllipticOp. This file reads the local chain back on plain functions on U.

For an open V with closure V compact in U, exists_localOp_C1 or exists_localOp_Ck gives an open W between closure V and U and a global operator agreeing with the given coefficients on W. The solution restricted to W is a local weak solution of that operator in W12 W (isLocalWeakSolution_of_localWeakSol), and the local chain on W at the compact closure V gives the family on V (exists_family_of_localWeakSol). The constant is fixed by V, U and the coefficients before the solution and the datum are, as in Evans.

Main declarations #

Extending the class of g on U by zero and restricting to V ⊆ U gives the class of g on V.

A function in L^∞(U) is essentially bounded on U.

theorem EllipticPdes.Regularity.exists_family_of_localWeakSol {n : ℕ} {U W K V : Set (EuclideanSpace ℝ (Fin (n + 1)))} (hU : IsOpen U) (hWo : IsOpen W) (hWU : W ⊆ U) (hK : IsCompact K) (hKW : K ⊆ W) (hVm : MeasurableSet V) (hVK : V ⊆ K) (Op : Sobolev.FullEllipticOp (n + 1)) {k : ℕ} (hreg : LocalRegularityAt Op ⋯ k) {a : EuclideanSpace ℝ (Fin (n + 1)) → Fin (n + 1) → Fin (n + 1) → ℝ} {b : EuclideanSpace ℝ (Fin (n + 1)) → Fin (n + 1) → ℝ} {c : EuclideanSpace ℝ (Fin (n + 1)) → ℝ} (ha : ∀ x ∈ W, Op.a x = a x) (hb : ∀ (i : Fin (n + 1)), ∀ᵐ (x : EuclideanSpace ℝ (Fin (n + 1))) ∂MeasureTheory.volume.restrict W, Op.b x i = b x i) (hc : ∀ᵐ (x : EuclideanSpace ℝ (Fin (n + 1))) ∂MeasureTheory.volume.restrict W, Op.c x = c x) :
∃ (C : ℝ), 0 ≤ C ∧ ∀ (f u : EuclideanSpace ℝ (Fin (n + 1)) → ℝ) (G : Fin (n + 1) → EuclideanSpace ℝ (Fin (n + 1)) → ℝ) (hf : MeasureTheory.MemLp f 2 (MeasureTheory.volume.restrict U)) (hu : MeasureTheory.MemLp u 2 (MeasureTheory.volume.restrict U)), (∀ (i : Fin (n + 1)), MeasureTheory.MemLp (G i) 2 (MeasureTheory.volume.restrict U)) → ∀ (Hf : HasIteratedWeakDerivOn U k (MeasureTheory.MemLp.toLp f hf)) (M : ℝ), IteratedL2Bound Hf M → Embedding.HasWeakGradOn U u G → LocalWeakSol U a b c f u G → ∃ (H : HasIteratedWeakDerivOn V (k + 2) (MeasureTheory.MemLp.toLp u ⋯)), IteratedL2Bound H (C * (M + ‖MeasureTheory.MemLp.toLp u hu‖))

Transfer of the local chain to plain functions on U. Let Op agree with a, b, c on an open W ⊆ U (everywhere for a, almost everywhere for b and c) and satisfy the order-k conclusion of the local chain on W. For a compact K ⊆ W and a measurable V ⊆ K there is a constant such that every u with weak gradient G, both in L²(U), solving the local weak formulation on U with a datum f whose family of k weak derivatives on U is bounded by M, has weak derivatives up to order k + 2 on V, bounded by C (M + ‖u‖_{L²(U)}).

In dimension zero every class has a family of every order, the family being constant, since there is no direction to differentiate in.

Equations
Instances For
    noncomputable def EllipticPdes.Regularity.C1OpOn.ofMemLp {d : ℕ} {U : Set (EuclideanSpace ℝ (Fin d))} {a : EuclideanSpace ℝ (Fin d) → Fin d → Fin d → ℝ} {b : EuclideanSpace ℝ (Fin d) → Fin d → ℝ} {c : EuclideanSpace ℝ (Fin d) → ℝ} (ha : ∀ (i j : Fin d), ContDiffOn ℝ 1 (fun (x : EuclideanSpace ℝ (Fin d)) => a x i j) U) (hb : ∀ (i : Fin d), MeasureTheory.MemLp (fun (x : EuclideanSpace ℝ (Fin d)) => b x i) ⊤ (MeasureTheory.volume.restrict U)) (hc : MeasureTheory.MemLp c ⊤ (MeasureTheory.volume.restrict U)) {θ : ℝ} (hθ : 0 < θ) (hell : ∀ᵐ (x : EuclideanSpace ℝ (Fin d)) ∂MeasureTheory.volume.restrict U, ∀ (ξ : Fin d → ℝ), θ * ∑ i : Fin d, ξ i ^ 2 ≤ ∑ i : Fin d, ∑ j : Fin d, a x i j * ξ i * ξ j) :
    C1OpOn d U

    The coefficients of Theorem 1 as a C1OpOn, with one essential bound for all transport components.

    Equations
    • One or more equations did not get rendered due to their size.
    Instances For
      theorem EllipticPdes.Regularity.interior_H2_regularity_evans {d : ℕ} {U : Set (EuclideanSpace ℝ (Fin d))} (hU : IsOpen U) {a : EuclideanSpace ℝ (Fin d) → Fin d → Fin d → ℝ} {b : EuclideanSpace ℝ (Fin d) → Fin d → ℝ} {c : EuclideanSpace ℝ (Fin d) → ℝ} (ha : ∀ (i j : Fin d), ContDiffOn ℝ 1 (fun (x : EuclideanSpace ℝ (Fin d)) => a x i j) U) (hb : ∀ (i : Fin d), MeasureTheory.MemLp (fun (x : EuclideanSpace ℝ (Fin d)) => b x i) ⊤ (MeasureTheory.volume.restrict U)) (hc : MeasureTheory.MemLp c ⊤ (MeasureTheory.volume.restrict U)) {θ : ℝ} (hθ : 0 < θ) (hell : ∀ᵐ (x : EuclideanSpace ℝ (Fin d)) ∂MeasureTheory.volume.restrict U, ∀ (ξ : Fin d → ℝ), θ * ∑ i : Fin d, ξ i ^ 2 ≤ ∑ i : Fin d, ∑ j : Fin d, a x i j * ξ i * ξ j) {V : Set (EuclideanSpace ℝ (Fin d))} (hVo : IsOpen V) (hVc : IsCompact (closure V)) (hVU : closure V ⊆ U) :

      Interior H²-regularity (Evans, Partial Differential Equations (2nd ed.), §6.3.1, Theorem 1, p. 327), with the unnecessary boundedness and symmetry assumptions omitted. U ⊆ ℝᵈ is open, a^{ij} ∈ C¹(U), b^i, c ∈ L^∞(U), the operator is uniformly elliptic with a constant θ > 0 for almost every x ∈ U (§6.1.1, (4)). For each open V ⊂⊂ U there is a constant C, depending on V, U and the coefficients alone, such that every u ∈ H¹(U), given as u and its weak gradient G in L²(U), that is a weak solution of L u = f in U for f ∈ L²(U) has weak derivatives up to order two in L²(V), which is u ∈ H²(V), with

      ‖u‖_{H²(V)} ≤ C (‖f‖_{L²(U)} + ‖u‖_{L²(U)}).

      Since every V ⊂⊂ U is covered, this is also u ∈ H²_loc(U). The norm is iteratedNorm, the sum over lists of directions. The weak formulation is tested against C_c^∞(U), as in Remark (ii) after the theorem; a solution tested against H₀¹(U) is one. The boundedness of U and symmetry of a^{ij} assumed by Evans are unnecessary here.

      theorem EllipticPdes.Regularity.higher_interior_regularity_evans {d : ℕ} {U : Set (EuclideanSpace ℝ (Fin d))} (hU : IsOpen U) (m : ℕ) {a : EuclideanSpace ℝ (Fin d) → Fin d → Fin d → ℝ} {b : EuclideanSpace ℝ (Fin d) → Fin d → ℝ} {c : EuclideanSpace ℝ (Fin d) → ℝ} (ha : ∀ (i j : Fin d), ContDiffOn ℝ (↑(m + 1)) (fun (x : EuclideanSpace ℝ (Fin d)) => a x i j) U) (hb : ∀ (i : Fin d), ContDiffOn ℝ (↑(m + 1)) (fun (x : EuclideanSpace ℝ (Fin d)) => b x i) U) (hc : ContDiffOn ℝ (↑(m + 1)) c U) {θ : ℝ} (hθ : 0 < θ) (hell : ∀ᵐ (x : EuclideanSpace ℝ (Fin d)) ∂MeasureTheory.volume.restrict U, ∀ (ξ : Fin d → ℝ), θ * ∑ i : Fin d, ξ i ^ 2 ≤ ∑ i : Fin d, ∑ j : Fin d, a x i j * ξ i * ξ j) {V : Set (EuclideanSpace ℝ (Fin d))} (hVo : IsOpen V) (hVc : IsCompact (closure V)) (hVU : closure V ⊆ U) :

      Higher interior regularity (Evans, Partial Differential Equations (2nd ed.), §6.3.1, Theorem 2, p. 332), with the unnecessary boundedness and symmetry assumptions omitted. Let m be a nonnegative integer, U ⊆ ℝᵈ open, a^{ij}, b^i, c ∈ C^{m+1}(U), the operator uniformly elliptic with a constant θ > 0 for almost every x ∈ U (§6.1.1, (4)). For each open V ⊂⊂ U there is a constant C, depending on m, U, V and the coefficients alone, such that every u ∈ H¹(U) that is a weak solution of L u = f in U for f ∈ H^m(U) has weak derivatives up to order m + 2 in L²(V), which is u ∈ H^{m+2}(V), with

      ‖u‖_{H^{m+2}(V)} ≤ C (‖f‖_{H^m(U)} + ‖u‖_{L²(U)}).

      Since every V ⊂⊂ U is covered, this is also u ∈ H^{m+2}_loc(U). f ∈ H^m(U) is f ∈ L²(U) with a family Hf of weak derivatives up to order m in L²(U), and both norms are iteratedNorm. The weak formulation is tested against C_c^∞(U). The boundedness of U and symmetry of a^{ij} assumed by Evans are unnecessary here.