The weak gradient of a suitable weak solution pairs as the distributional gradient #
For a suitable weak solution on the unit cylinder, the weak gradient Du satisfies
∫ Du_{kj} ∂ⱼψ = -∫ u_k ∂ⱼ∂ⱼψ for every smooth test function ψ supported in the cylinder.
The proof works on a local box containing the support of ψ, applies the weak derivative on
almost every time slice, and integrates in time. This identifies the viscous term of the weak
momentum equation in the passage to the classical equations used in the proof of
lem:aniso:annulus.
A compact subset of the unit cylinder lies in a local box B(0, r) × (a, b).
The box B(0, r) × (a, b) with r < 1 and -1 < a < b < 0 is a local box of the unit
cylinder.
A spatial partial vanishes where the function vanishes on a neighbourhood.
The spatial partial of a smooth function is its joint derivative in the direction (eⱼ, 0),
hence smooth.
For a suitable weak solution on the unit cylinder, the weak gradient pairs with a test function supported in the cylinder as the distributional gradient.
For a suitable weak solution on the unit cylinder, the weak gradient pairs with a test function supported in the cylinder as the distributional gradient.