Whole-space heat gain of every order #
The whole-space form of lem:local-heat-gain, for every order m ≥ 1, as
used in prop:lps-smoothing. Let a < a + σ < a + β. If
z ∈ L²((a, a + β); H^m(ℝ³)) solves ∂ₜ z - Δ z = G in distributions on
ℝ³ × (a, a + β) with G ∈ L²((a, a + β); H^{m-1}(ℝ³)), then
z ∈ C([a + σ, a + β]; H^m(ℝ³)) ∩ L²((a + σ, a + β); H^{m+1}(ℝ³)) with
‖z‖_{C H^m} + ‖z‖_{L² H^{m+1}} ≤ C (‖z‖_{L² H^m} + ‖G‖_{L² H^{m-1}}),
where C depends only on m, σ and β.
The proof multiplies by a time cutoff c(t) vanishing for t ≤ a + σ/2 and
equal to one for t ≥ a + σ. Each spatial derivative ∂^γ (c z), |γ| ≤ m,
solves the heat equation on ℝ³ × (a, a + β) with a square-integrable source
in source or divergence form built from c ∂^γ G + c' ∂^γ z, and has zero
initial trace, so the energy estimate of lem:localized-vorticity-energy
applies.
lem:local-heat-gain on the whole space and for every order m ≥ 1
(prop:lps-smoothing). The family Dz' of the gained regularity agrees with
the given family Dz through order m.