Skew integration by parts for the first coefficient derivative in the drift error.
theorem
AVenhance.Infra.Section4.iterate_matrix_gradient_smooth
{A : Homogenization.Vec 2 → Matrix (Fin 2) (Fin 2) ℝ}
{v : Homogenization.Vec 2 → ℝ}
(hA : ContDiff ℝ (↑⊤) A)
(hv : ContDiff ℝ (↑⊤) v)
:
The actual matrix-gradient flux is smooth.
theorem
AVenhance.Infra.Section4.iterate_matrix_gradient_periodic
{A : Homogenization.Vec 2 → Matrix (Fin 2) (Fin 2) ℝ}
{v : Homogenization.Vec 2 → ℝ}
(hv : ContDiff ℝ (↑⊤) v)
(hA : IsZ2Periodic A)
(hvp : IsZ2Periodic v)
:
IsZ2Periodic fun (x : Homogenization.Vec 2) => (A x).mulVec (spaceGrad v x)
Periodicity of the actual coefficient and scalar supplies flux periodicity.
theorem
AVenhance.Infra.Section4.iterate_stream_pairing_transfer
{ψ u v : Homogenization.Vec 2 → ℝ}
(hψ : ContDiff ℝ (↑⊤) ψ)
(hu : ContDiff ℝ (↑⊤) u)
(hv : ContDiff ℝ (↑⊤) v)
(hψp : IsZ2Periodic ψ)
(hup : IsZ2Periodic u)
(hvp : IsZ2Periodic v)
:
The first stream-coefficient derivative can be transferred to the coefficient once more, avoiding Young's inequality on the large first jet.
theorem
AVenhance.Infra.Section4.iterate_scalar_transport_abs_bound
{H a : ℝ}
(hH : 0 < H)
(b v : Homogenization.Vec 2)
(hb : ∀ (j : Fin 2), |b j| ≤ H)
:
Pointwise bilinear bound after the skew transfer. The scalar norm is undifferentiated; both factors can then be controlled by lower-order energy.
theorem
AVenhance.Infra.Section4.iterate_continuous_cell_integrable
{f : Homogenization.Vec 2 → ℝ}
(hf : Continuous f)
:
Continuous cell integrands are integrable; this discharges the spatial integrability in the forcing estimates.
theorem
AVenhance.Infra.Section4.iterateSpatialWord_sq_le_tail_gradient
(f : Homogenization.Vec 2 → ℝ)
(i : Fin 2)
(w : List (Fin 2))
(x : Homogenization.Vec 2)
:
iterateSpatialWord (i :: w) f x ^ 2 ≤ Homogenization.vecNormSq (spaceGrad (iterateSpatialWord w f) x)
Any nonempty ordered scalar word is a component of a lower-order gradient.