Extremal problems for \(L_p\) surface areas and John ellipsoids (Q2320133): Difference between revisions

In the paper under review the authors consider the following two optimization problems for a convex body \(K\subset\mathbb{R}^n\) with \(0\in\operatorname{int}K\): \textbf{P1:} Let \(p>1\). Does there exist a unique \(x_0\in\operatorname{int}K\) such that \[ S_p(K-x_0)=\min_{x\in\operatorname{int}K} S_p(K-x)? \] Here \(S_p(K)\) represents the \(L_p\)-surface area measure of \(K\), namely, \[ S_p(K)=\int_{\mathbb{S}^{n-1}}h_K^{1-p}(u)\,dS(K,u), \] where \(h_K(\cdot)\) stands for the support function of \(K\) and \(S(K,\cdot)\) is its classical surface area measure. \textbf{P2:} Let \(p>0\). Does there exist a unique \(x_0\in\operatorname{int}K\) such that \[ V\bigl(\operatorname{E}_p(K-x_0)\bigr)=\max_{x\in\operatorname{int}K} V\bigl(\operatorname{E}_p(K-x)\bigr)? \] Now \(V\) denotes the volume functional and \(\operatorname{E}_pK\) is the \(L_p\)-John ellipsoid of \(K\), i.e., the unique (\(0\)-symmetric) ellipsoid that solves the constrained optimization problem ``\(\max_EV(E)\) over all \(0\)-symmetric ellipsoids \(E\), subject to \(\overline{V}_p(K,E)\leq 1\)'', where \[ \overline{V}_p(K,E)=\left(\frac{1}{nV(K)}\int_{\mathbb{S}^{n-1}}\left(\frac{h_E(u)}{h_K(u)}\right)^ph_K(u)\,dS(K,u)\right)^{1/p} \] is the normalized \(L_p\)-mixed volume of \(K\) and \(E\). Lutwak, Yang and Zhang defined this ellipsoid in [\textit{E. Lutwak} et al., Proc. Lond. Math. Soc. (3) 90, No. 2, 497--520 (2005; Zbl 1074.52005)] and proved its existence and uniqueness. In this paper the authors completely solve \textbf{P1} if \(K\) is a polytope, and give a positive answer for an arbitrary convex body when \(p\geq n-1\). Regarding the second problem, \textbf{P2}, the solution is also obtained for values of \(p\geq n-1\).