Vector Norm Aggregation #
Part of the Caffarelli–Kohn–Nirenberg partial regularity proof.
Aggregation of componentwise vector inequalities to the Euclidean norm #
The Poincare and H¹ interpolation inequalities of CKN.Setting.VectorInequalities are
stated componentwise: each coordinate u_i of a velocity field u : Vec3 → Vec3 is
controlled by the scalar theorems. The estimates of paper/ckn.tex instead use the
Euclidean norm |u| and the Frobenius norm of the spatial gradient,
|∇u|² = spatialGradientSq. This file supplies the finite-dimensional norm comparisons
that turn the componentwise statements into their Euclidean counterparts.
The two elementary inequalities on ℝ³ are
|u|^p ≤ 3^{max(0,p/2-1)} Σ_i |u_i|^p(the sharp comparison of the Euclidean norm with the componentℓᵖmass), andΣ_i |u_i|^p ≤ 3 |u|^p.
Both are proved pointwise and then integrated, first on the ℝ≥0∞ side (the form in which
the scale quantities γ, α, β store their local Lᵖ masses) and then for the
spatial balls used by the Poincare and Sobolev statements.
Pointwise comparison on ℝ³ #
The Euclidean norm raised to p ≥ 1 is controlled by the component ℓᵖ mass with
the sharp finite-dimensional constant: |u|^p ≤ 3^{max(0,p/2-1)} Σ_i |u_i|^p.
The ℝ≥0∞ integrable form used by the scale quantities #
Pointwise comparison on the ℝ≥0∞ side: the p-th power of the Euclidean density
is dominated by the component ℓᵖ mass with the sharp finite-dimensional constant.
Integrated form of the sharp comparison: the Lᵖ mass of the Euclidean density is
controlled by the sum of the component Lᵖ masses.