Valuations and surface area measures (Q2450170)

Let \(\mathcal{P}_0^n\) denote the set of all convex polytopes \(P \subseteq \mathbb{R}^n\) that contain the origin, \(\mathcal{M}(S^{n-1})\) the space of finite positive Borel measures on the Euclidean unit sphere \(S^{n-1} \subseteq \mathbb{R}^n\), and \(\mathrm{SL}(n)\) the special linear group on \(\mathbb{R}^n\). A valuation on \(\mathcal{P}_0^n\) with values in \(\mathcal{M}(S^{n-1})\) is a map \(\mu:\mathcal{P}_0^n \to \mathcal{M}(S^{n-1})\) such that \(\mu(P \cup Q)+\mu(P \cap Q)=\mu(P) + \mu(Q)\) whenever \(P,Q, P \cup Q, P \cap Q \in \mathcal{P}_0^n\). Moreover, \(\mu\) is called \(\mathrm{SL}(n)\) contravariant of degree \(p \in \mathbb{R}\) if \(\int_{S^{n-1}} fd\mu(\phi P,\cdot)= \int_{S^{n-1}} f \circ (\phi^{-1})^t d\mu(P,\cdot)\) for every \(\phi \in \mathrm{SL}(n)\), \(P \in \mathcal{P}_0^n\), and every continuous \(p\)-homogeneous function \(f: \mathbb{R}^n \setminus \{0\} \to \mathbb{R}\). A first theorem says that a map \(\mu: \mathcal{P}_0^n \to \mathcal{M}(S^{n-1})\), \(n \geq 3\), is an \(\mathrm{SL}(n)\) contravariant valuation of degree \(1\) if and only if there exist constants \(c_1,c_2,c_3,c_4 \in \mathbb{R}\) with \(c_1,c_2,c_1+c_3,c_2+c_4 \geq 0\) such that \(\mu(P,\cdot)=c_1 S(P,\cdot)+ c_2 S(-P,\cdot)+ c_3 S^\ast(P,\cdot)+c_4 S^\ast(-P,\cdot)\) for every \(P \in \mathcal{P}_0^n\). Here \(S(P,\cdot)=\sum_{u \in \mathcal{N}(P)} V_{n-1}(F_u) \delta_u\) is the surface area measure of \(P\) (i.e., \(\mathcal{N}(P)\) consists of all unit facet normals, \(V_{n-1}(F_u)\) is the \((n-1)\)-dimensional volume of the facet corresponding to \(u\), and \(\delta_u\) is the Dirac probability measure supported at \(u\)). \(S^\ast(P,\cdot)\) is defined analogously, involving only facets that do not contain the origin. Similar results are obtained for \(\mathrm{SL}(n)\) (or \(\mathrm{GL}(n)\)) contravariant valuations of degree \(p \neq 1\). They can be expressed by the aid of the \(L_p\) surface area measure.