Affine vs. Euclidean isoperimetric inequalities (Q2326670)

The classical Euclidean isoperimetric inequality says that among all measurable sets \(S\) of \(\mathbb{R}^n\) with given finite \((n - 1)\)-dimensional measure of the boundary \(\partial S\), the volume of \(S\) is maximized when \(S\) is a ball. This well-known inequality together with the Petty projection inequality -- the only affine invariant one -- are special cases in one of the main results in the paper, Theorem 1. It states that for an even, zonal measure \(\mu\) on \(\mathbb{S}^{n-1}\), among sets of finite perimeter \(L\subset \mathbb{R}^n\), with convex body \(\Phi^{\mu}L\), the volume product \(|(\Phi^{\mu}L)^\mathrm{o}||L|^{n-1}\) is maximized by Euclidean balls \(L\), for \(\mu\) not discrete, and for \(\mu\) discrete, the maximizer \(L\) is an ellipsoid. The classical Sobolev embedding theorem \(W^{1,p}(\mathbb{R}^n)\subset L^{p^{*}}(\mathbb{R}^n)\) with \(p^{*}=np/(n-p)\), \(1\leq p<n\), for \(p = 1\) is essentially the same as the isoperimetric inequality for functions of bounded variation. Among other results, the authors give a family of sharp Sobolev-type inequalities parametrized by even, zonal measures on \(\mathbb{S}^{n-1}\), in which this classical and the affine \(L_p\) Sobolev inequality appear as members.