The Hadwiger theorem on convex functions. IV: The Klain approach (Q2679888)

The authors continue their investigation of valuations on spaces of convex functions. A real functional \(Z\) on a space \(X\) of extended real-valued functions is a valuation if \(Z(u)+Z(v)=Z(u\vee v)+Z(u\wedge v)\) for all \(u,v\in X\) for which the pointwise maximum \(u\vee v\) and the pointwise minimum \(u\wedge v\) belong to \(X\). In a former paper [the authors, ``The Hadwiger theorem on convex functions. I'', Preprint, \url{arXiv:2009.03702}], the authors have found a counterpart to Hadwiger's classification of rigid motion invariant continuous valuations on convex bodies for the space \(\mathrm{Conv}_{\mathrm{sc}}({\mathbb R}^n)\) of proper, super-coercive, lower semicontinuous, convex functions on \({\mathbb R}^n\), equipped with the topology induced by epi-convergence. The result was a characterization of certain functional intrinsic volumes (integrals of suitable singular densities with respect to Hessian measures). The main purpose of the present paper is to carry over Klain's treatment of Hadwiger's theorem to function spaces (where, as explained, one must necessarily go via smooth valuations). On the way, further interesting results are obtained. One of them says that a function \(Z:\mathrm{Conv}_{\mathrm{sc}}({\mathbb R}^n)\to{\mathbb R}\) is a simple (i.e., vanishing on functions with lower-dimensional domain), continuous, epi-translation invariant valuation if and only if there exists a continuous function \(\zeta\) with compact support on \({\mathbb R}^n\) such that \(Z(u)= \int_{\mathrm{dom}\,u} \zeta(\nabla u(x))\mathrm{d}x\) for every \(u\in\mathrm{Conv}_{\mathrm{sc}}({\mathbb R}^n)\). Another result concerns a characterization of smooth, epi-translation invariant and rotation invariant valuations on \(\mathrm{Conv}_{\mathrm{sc}}({\mathbb R}^n)\). These results, together with a previously established Cauchy-Kubota formula for convex functions, allow to give a new proof of Hadwiger's theorem for convex functions. The paper concludes with an investigation of the extension of valuations defined on functions with lower-dimensional domains, utilizing a connection to Abel transforms. Fot Part III, see [the authors, Calc. Var. Partial Differ. Equ. 61, No. 5, Paper No. 181, 37 p. (2022; Zbl 07565570)].