From John to Gauss--John positions via dual mixed volumes (Q864712): Difference between revisions

For a convex body \(K\subset\mathbb{R}^n\) containing the origin in its interior and any \(i\in\mathbb{R}\), the \(i\)th dual mixed volume is defined as \[ \tilde{W}_i(K)=\frac{1}{n}\int_{\mathbb{S}^{n-1}}\rho_K^{n-i}(u)\,d\sigma(u), \] where \(\rho_K\) is the radial function of \(K\). The authors study the optimization of the dual mixed volumes \(\bigl\{\tilde{W}_i(SK):0\in SK\subset B_n\bigr\}\), where \(B_n\) denotes the Euclidean ball and \(SK\) runs over all positions of \(K\). First the case \(\tilde{W}_i(TK)\) for \(T\in GL(n)\) is considered, and here they give necessary and sufficient conditions for \(I\), the identity, to be the optimal solution. These results extend the one obtained in [\textit{A. A. Giannopoulos, V. D. Milman} and \textit{M. Rudelson}, Lect. Notes Math. 1745, 81--93 (2000; Zbl 0978.52001)], where necessary conditions are given for a \(0\)-symmetric convex body to be in Gauss-John position, i.e., to be the solution of the optimization problem \(\tilde{W}_{n+1}(TK)\). Next the authors study the affine case \(\tilde{W}_i(a+TK)\), \(a\in\mathbb{R}^n\). Since the dual mixed volumes are not affine invariant, these affine optimization problems are different from the linear ones. They show, however, that they are equivalent for \(0\)-symmetric convex bodies and positive indexes. In order to study this problem, the authors prove an interesting extension of the dual Kubota formula that relates the dual mixed volumes \(\tilde{W}_i(L)\) with the parallel sections of a star body \(L\).