The Laplacian of a convolution #
On a finite-dimensional real inner product space, convolution with a locally integrable function
commutes with the Laplacian on C² functions with compact support:
Δ (f ⋆ g) = f ⋆ Δ g.
Each second directional derivative passes onto g by the rule ∂ᵥ (f ⋆ g) = f ⋆ ∂ᵥ g
(HasCompactSupport.fderiv_convolution_right_apply), and the Laplacian is their sum along an
orthonormal basis (TauCeti.laplacian_eq_sum_fderiv_fderiv_apply).
This is how a convolution against a fundamental solution, such as the Newtonian potential, is
shown to solve Poisson's equation: the Laplacian moves onto the smooth compactly supported
factor, where the distributional identity for the kernel applies
(TauCeti.laplacian_convolution_eq_neg_of_integral_laplacian_mul).
Main declarations #
HasCompactSupport.laplacian_convolution_right:Δ (f ⋆[L, μ] g) = f ⋆[L, μ] Δ g.TauCeti.laplacian_convolution_eq_neg_of_integral_laplacian_mul: if-ΔK = δin the sense of distributions, then-Δ (K ⋆ f) = ffor everyC²functionfwith compact support.
The Laplacian of a convolution. If f is locally integrable and g is C² with compact
support, then Δ (f ⋆ g) = f ⋆ Δ g.
Convolution with a fundamental solution solves Poisson's equation. Let K be locally
integrable with -ΔK = δ in the sense of distributions, tested against real C² functions with
compact support: ∫ Δφ(y) K(x - y) dy = -φ x for every such φ and every x. Then for every
f of class C² with compact support, Δ (K ⋆ f) = -f.