Documentation

LeanPool.EllipticPDE.Extension.EvenReflection

Extension across a flat boundary by reflection #

A Sobolev class on the half space extends across the interface by reflecting it: the value at a point below the interface is the value at its mirror image. The gradient extends the same way, with a sign in the normal direction, and the extended pair is a weak gradient on the whole space.

The identity is tested against an arbitrary test function of the whole space. Splitting the integral at the interface and reflecting the lower half turns it into an integral over the half space, tested against φ + s (φ ∘ R) with s the sign of the direction. In the normal direction that combination is odd, so it vanishes on the interface, which is exactly the hypothesis under which the boundary term of EllipticPdes.Extension.integral_partialD_of_eq disappears.

Main declarations #

References #

L. C. Evans, Partial Differential Equations (2nd ed.), §5.4 Theorem 1.

The open half space below the interface.

Equations
Instances For

    Up to the null interface, the whole space is the union of the two open half spaces.

    Splitting of an integral over the whole space at the interface.

    The reflected extension #

    noncomputable def EllipticPdes.Extension.evenExt {d : ℕ} (j : Fin d) (u : EuclideanSpace ℝ (Fin d) → ℝ) :

    Reflected extension of a function: below the interface it takes the value at the mirror image.

    Equations
    Instances For
      noncomputable def EllipticPdes.Extension.evenExtGrad {d : ℕ} (j : Fin d) (g : Fin d → EuclideanSpace ℝ (Fin d) → ℝ) (k : Fin d) :

      Reflected extension of a gradient, with a sign in the normal direction.

      Equations
      Instances For

        Integral over the lower half space as the reflected integral over the upper one.

        theorem EllipticPdes.Extension.partialD_add {d : ℕ} {f h : EuclideanSpace ℝ (Fin d) → ℝ} (k : Fin d) {x : EuclideanSpace ℝ (Fin d)} (hf : DifferentiableAt ℝ f x) (hh : DifferentiableAt ℝ h x) :
        Sobolev.partialD k (fun (y : EuclideanSpace ℝ (Fin d)) => f y + h y) x = Sobolev.partialD k f x + Sobolev.partialD k h x

        The partial derivative is additive.

        theorem EllipticPdes.Extension.partialD_const_mul {d : ℕ} {f : EuclideanSpace ℝ (Fin d) → ℝ} (k : Fin d) (c : ℝ) {x : EuclideanSpace ℝ (Fin d)} (hf : DifferentiableAt ℝ f x) :
        Sobolev.partialD k (fun (y : EuclideanSpace ℝ (Fin d)) => c * f y) x = c * Sobolev.partialD k f x

        The partial derivative commutes with a constant multiple.

        theorem EllipticPdes.Extension.reflectLI_eq_self_of_interface {d : ℕ} {j : Fin d} {x : EuclideanSpace ℝ (Fin d)} (hx : x.ofLp j = 0) :
        (reflectLI j) x = x

        The reflection fixes the interface.

        Weak gradient of the reflected extension on the whole space. Splitting the integral at the interface and reflecting the lower half tests the class against φ + s (φ ∘ R), which in the normal direction is odd and so vanishes on the interface. That is the hypothesis under which no boundary term survives.

        The bound #

        theorem EllipticPdes.Extension.evenExt_ae_eq {d : ℕ} (j : Fin d) (u : EuclideanSpace ℝ (Fin d) → ℝ) :
        evenExt j u =ᵐ[MeasureTheory.volume] fun (x : EuclideanSpace ℝ (Fin d)) => (halfSpace j).indicator u x + (halfSpaceNeg j).indicator (fun (y : EuclideanSpace ℝ (Fin d)) => u ((reflectLI j) y)) x

        The extension is, almost everywhere, the sum of the class and its reflection, each on its own side of the interface.

        Reflection of the lower half space onto the upper one, preserving measure.

        Bound for the extension in every Lᵖ seminorm. The reflection preserves measure, so each side contributes the seminorm on the half space.

        Integrability of the extension #

        theorem EllipticPdes.Extension.evenExtGrad_ae_eq {d : ℕ} (j : Fin d) (g : Fin d → EuclideanSpace ℝ (Fin d) → ℝ) (k : Fin d) :
        evenExtGrad j g k =ᵐ[MeasureTheory.volume] fun (x : EuclideanSpace ℝ (Fin d)) => (halfSpace j).indicator (g k) x + (halfSpaceNeg j).indicator (fun (y : EuclideanSpace ℝ (Fin d)) => reflectSign j k * g k ((reflectLI j) y)) x

        The extended gradient is, almost everywhere, the sum of the component and its signed reflection, each on its own side of the interface.

        Measurability of the reflected extension, from the description of it as a sum of two indicators.

        Integrability of the reflected extension. Each side of the interface contributes the integral over the half space, the reflection preserving measure.

        Integrability of the extended gradient, componentwise.

        The gradient's bound #

        The sign a reflection attaches to a direction has absolute value 1.

        Bound on the extended gradient in every Lᵖ seminorm. The reflection preserves measure and the sign has absolute value 1, so each side contributes the seminorm on the half space.