A class of invariant valuations on \(\operatorname{Lip}(S^{n -1})\) (Q2309105)

The authors establish a classification result for continuous valuations on the space of Lipschitz continuous functions on the unit sphere \(S^{n-1}\), \(\operatorname{Lip}(S^{n-1})\), with some additional invariance properties. This can be seen as a continuation of previous results by \textit{D. A. Klain} [Adv. Math. 125, No. 1, 95--113 (1997; Zbl 0889.52007)], \textit{A. Tsang} [Int. Math. Res. Not. 2010, No. 20, 3993--4023 (2010; Zbl 1211.52013)], \textit{P. Tradacete} and \textit{I. Villanueva} [Adv. Math. 327, 361--391 (2018; Zbl 1400.52015)]. Here, a map \(\mu:\operatorname{Lip}(S^{n-1})\to \mathbb{R}\) is called a valuation whenever \[ \mu(u\vee v) + \mu(u \wedge v) = \mu(u)+\mu(v) \] for every \(u,v\in\operatorname{Lip}(S^{n-1})\), where \(\vee\) and \(\wedge\) denote the pointwise maximum and minimum, respectively. Using radial functions, this is closely related to the study of valuations on star-shaped sets. Furthermore, since the space \(\mathcal{H}(S^{n-1})\) of support functions of convex bodies is contained in \(\operatorname{Lip}(S^{n-1})\), valuations on \(\operatorname{Lip}(S^{n-1})\) can be seen as extensions of valuations on convex bodies. A valuation \(\mu:\operatorname{Lip}(S^{n-1})\to \mathbb{R}\) is said to be rotation invariant if \(\mu(u\circ \varphi)=\mu(u)\) for every \(u\in \operatorname{Lip}(S^{n-1})\) and every \(\varphi\in \operatorname{O}(n)\). Moreover, it is called \(\Lambda\)-invariant if \(\mu(u+l)=\mu(u)\) for every \(u\in \operatorname{Lip}(S^{n-1})\) and every \(l:S^{n-1}\to\mathbb{R}\) which is the restriction of a linear function on \(\mathbb{R}^n\). The last condition corresponds to translation invariance if \(\mu\) is restricted to \(\mathcal{H}(S^{n-1})\). In their main result, the authors show that a map \(\mu:\operatorname{Lip}(S^{n-1})\to \mathbb{R}\) is a continuous, rotation invariant and \(\Lambda\)-invariant valuation if and only if there exist constants \(c_0,c_1,c_2\in\mathbb{R}\) such that \[ \mu(u)=c_0+c_1\int_{S^{n-1}} u(x) \,\mathrm{d}H^{n-1}(x)+c_2 \int_{S^{n-1}} [(n-1)u^2(x)-\|\nabla u(x)\|^2]\,\mathrm{d} H^{n-1}(x) \] for every \(u\in\operatorname{Lip}(S^{n-1})\). Here, \(H^{n-1}\) is the \((n-1)\)-dimensional Hausdorff measure and \(\nabla u(x)\) denotes the spherical gradient of \(u\) at \(x\in S^{n-1}\). Furthermore, continuity is understood with respect to a certain topology which is weaker than the natural topology on \(\operatorname{Lip}(S^{n-1})\). Further examples of valuations which are presented in this paper show that (in contrast to the result above) the space of continuous and rotation invariant valuations is not finite dimensional. In particular, the operators appearing in the main result are extensions of the first three intrinsic volumes, \(V_0\), \(V_1\) and \(V_2\), from \(\mathcal{H}(S^{n-1})\) to \(\operatorname{Lip}(S^{n-1})\). The proof uses a reduction argument which states that every continuous valuation on \(\operatorname{Lip}(S^{n-1})\) is uniquely determined by its values on \(\mathcal{H}(S^{n-1})\) (Proposition 3.1). This allows the authors to apply the classical Hadwiger characterization theorem for convex bodies. Considering this, it is a remarkable fact that the remaining intrinsic volumes do not appear in the main result. In particular, this means that in dimensions three and higher, the volume cannot be extended from convex bodies to \(\operatorname{Lip}(S^{n-1})\). The mathematical reason for this is presented in Proposition 5.1 and its proof. In order to prove the main result, the authors also establish a homogeneous decomposition result for continuous and \(\Lambda\)-invariant valuations on \(\operatorname{Lip}(S^{n-1})\). This resembles the classical McMullen decomposition for continuous and translation invariant valuations on convex bodies and is of interest on its own.