The forced heat response of a double-divergence-free tensor is divergence free #
Let G ∈ L² be a tensor supported in the slab Q_τ = ℝ³ × (0, τ) with
∑_ij ∂_i ∂_j G_ij(·, s) = 0 in the sense of distributions for almost every
time s. Then the forced heat response Z_i = ∑_j ∂_j W₊ ⋆ G_ij is weakly
divergence free on almost every time slice of the slab. This is the
divergence-free clause of the local solution in prop:pv-local-solution,
where G = -(F + p I) with p the Riesz pressure of F.
The proof is by duality: at almost every time the kernel integrals converge
absolutely against the weight forcedHeatWeight
(kernel_weighted_lintegral_le), and forcedHeat_divergence_pairing_eq_zero
evaluates the pairing with ∂_i ψ as zero. A general tensor is first replaced
by a measurable modification, which does not change the response.
For a measurable square-integrable tensor supported in the slab, the weighted kernel integrals defining its forced heat response are finite on almost every time slice of the slab.
Almost every time slice of a square-integrable tensor supported in the slab is square integrable.
The divergence-free property of the forced heat response for a measurable
double-divergence-free tensor in L² supported in the slab.
The forced heat response of a double-divergence-free tensor is divergence
free: for G ∈ L² supported in Q_τ with ∑_ij ∂_i ∂_j G_ij(·, s) = 0 in the
sense of distributions for almost every s, almost every time slice of
Z = forcedHeat G in (0, τ) satisfies ∫ Z · ∇ψ = 0 for every test function
ψ. This is the divergence-free clause of prop:pv-local-solution.