On monotone translation-projection covariant Minkowski valuations (Q2664108)

This paper characterizes the orthogonal projection of convex bodies to a fixed linear subspace. Let \({\mathcal K}^n\) be the space of convex bodies (nonempty, compact, convex subsets) in \({\mathbb R}^n\). A map \(\mu: {\mathcal K}^n\to{\mathcal K}^n\) is called a Minkowski valuation if \(\mu(K\cup L)+\mu(K\cap L)=\mu(K)+\mu(L)\) whenever \(K,L,K\cup L\in{\mathcal K}^n\); here \(+\) denotes Minkowski addition. Let \(V\subseteq {\mathbb R}^n\) be a linear subspace and denote by \(p_V\) the orthogonal projection to \(V\). The map \(\mu\) is called translation-projection covariant with respect to \(V\) if \(\mu(K+x)= \mu(K)+p_V(x)\) for all \(K\in {\mathcal K}^n\) and all \(x\in{\mathbb R}^n\). Further, \(\mu\) is called monotone if it is either increasing or decreasing with respect to set inclusion. The following is the main result of this paper. Let \(\mu: {\mathcal K}^n\to{\mathcal K}^n\) be a monotone Minkowski valuation which is projection-translation covariant with respect to some linear subspace \(V\). If \(\mu(K_0)=p_V(K_0)\) for some \(n\)-dimensional, strictly convex body \(K_0\), then \(\mu(K)= p_V(K)\) for all \(K\in {\mathcal K}^n\). The main part of the elementary though lengthy proof concerns the case where \(\dim V=1\). The authors point out that it is unknown whether the assumption `strictly convex' for \(K_0\) can be deleted.