Minkowski type theorems for convex sets in cones (Q2043690)

Let \({\mathcal{S}}^{n-1}\) be the unit sphere of the Euclidean space \(\mathbb{R}^n\), \(n\ge 2\) (with scalar product \(\langle\cdot, \cdot\rangle\), norm \(\Vert \cdot\Vert \), and origin 0). Let \(K\subset {\mathbb{R}}^n\) be a convex body, that is a nonempty compact convex set. For a Borel set \(\omega\subset {\mathcal{S}}^{n-1}\), let \(\tau(K, \omega)\) be the set of all boundary points of \(K\) at which there exists an outer normal vector falling in \(\omega\). Then \(S_{n-1}(K, \omega) := {\mathcal{H}}^{n-1}(\tau(K, \omega))\) defines the surface area measure \(S_{n-1}(K, \cdot)\) of the convex body \(K\) (here \({\mathcal{H}}^{n-1}\) denotes the \((n-1)\)-dimensional Hausdorff measure). The Minkowski theorem states that a Borel measure \(\varphi\) on \({\mathcal{S}}^{n-1}\) is the surface area measure of some convex body \(K\) if and only if \(\varphi\) is finite, not concentrated on a great subsphere, and satisfies \(\int_{{\mathcal{S}}^{n-1}}u\, \varphi(du) = 0\). Moreover, \(K\) is uniquely determined up to a translation, see [the author, Convex bodies: the Brunn-Minkowski theory. Cambridge: Cambridge University Press (2014; Zbl 1287.52001)]. In the present paper, the author considers the same problem for the case when \(K\) is an unbounded convex body, which is contained in a given convex cone and such that its behavior at infinity is determined by the cone \(C\subset {\mathbb{R}}^n\) with the apex at the origin and with interior points. All the above definitions remain the same, only \(S_{n-1}(K, S_K)\) is no longer a finite measure (\(S_K\) is defined as the set of all unit vectors that are outer normal vectors of \(K\) at boundary points). The author considers the Minkowski problem with boundary condition, where a Borel measure is given on a spherically convex open set \(\Omega_{C}\subset {\mathcal{S}}^{n-1}\). \(K\) is called \(C\)-asymptotic if for \(x\in bd\,C\) the distance of \(x\) from \(K\) tends to zero as \(\Vert x\Vert \to \infty\). \(K\) is called \(C\)-close if \(C\setminus K\) has finite Lebesgue measure. Finally, the set \(K\) is called \(C\)-full if \(C\setminus K\) is bounded. \textit{A. Khovanskiĭ} and \textit{V. Timorin} [Discrete Comput. Geom. 52, No. 4, 806--823 (2014; Zbl 1397.52003)] introduced coconvex sets as the differences \(C\setminus K\), where \(K\) is a \(C\)-full set. In particular, they introduced mixed volumes and derived Alexandrov-Fenchel inequalities for coconvex sets. Coconvex sets of finite Lebesque measure were treated in [the author, Adv. Math. 332, 199--234 (2018; Zbl 1396.52007)]. Here, Minkowski-type theorems for convex sets in cones were first obtained. Note that the natural domain of definition for the surface area measure of a \(C\)-asymptotic convex set \(K\) is the open subset \(\Omega_C := {\mathcal{S}}^{n-1}\bigcap int\, C^0\) of the unit sphere, where \(C^0 := \{x\in {\mathbb{R}}^n: \langle x, y\rangle \le 0\,\, \forall y\in C\}\) is the polar cone of \(C\). Thus the surface area measure of \(K\) is defined solely for Borel sets \(\omega\) from \(\Omega_C\). The following two theorems were proved: Theorem A. Let \(\varphi\) be a finite Borel measure on \(\Omega_C\) with compact support in \(\Omega_C\). Then there exists a \(C\)-full set \(K\) such that \(S_{n-1}(K, \cdot) = \varphi\). Theorem B (uniqueness) Let \(K\), \(L\) be \(C\)-close sets such that \(S_{n-1}(K, \cdot) = S_{n-1}(L, \cdot)\). Then \(K=L\). The main results of the paper are the following two theorems. Theorem 1. Let \(\varphi\) be a finite Borel measure on \(\Omega_C\). Then there exists a \(C\)-close set \(K\) such that \(S_{n-1}(K, \cdot) = \varphi\). Although the measure \(\varphi\) in Theorem 1 is finite, and for body \(K\) \(S_{n-1}(K, \cdot) = \varphi\), the set \(K\) will not be a \(C\)-full. If \(n\ge 3\), one can construct \(C\)-close sets with finite surface area measure for which \(C\setminus K\) is unbounded. An example is given in Section 4. Theorem 2. Let \(\omega\) be a compact subset of \(\Omega_C\), and let \(K\), \(L\) be \(C\)-full sets whose surface area measures are concentrated on \(\omega\). There is a constant \(c\), depending only on \(C\), \(\omega\) and the upper bound for \(S_{n-1}(K,\cdot)\), \(S_{n-1}(L,\cdot)\), such that \(d_H(K, L) \le c\, \delta_{L,P}(S_{n-1}(K, \cdot),S_{n-1}(L, \cdot))^{1/n}\), where \(d_H(K, L)\) is the Hausdorff distance of \(C\)-full sets \(K\) and \(L\), and \(\delta_{LP}(\cdot, \cdot)\) is the Lévy-Prokhorov distance for two finite Borel measures. The Lévy-Prokhorov distance of surface area measures of convex bodies plays an important role in a characterization of Blaschke addition, cf. [\textit{R. J. Gardner} et al., Adv. Math. 254, 396--418 (2014; Zbl 1291.52011)]. It appears also in a stability theorem for convex bodies, see [\textit{D. Hug} and the author, in: IV international conference on ``stochastic geometry, convex bodies, empirical measures and applications to engineering science'', Tropea, Italy, September 24--29, 2001. Vol. II. Palermo: Circolo Matematico di Palermo. 21--51 (2002; Zbl 1113.52013)]. The paper is organized as follows. Section 2 contains some preliminaries. Theorem 1 is proved in Section 3 and Theorem 2 in Section 5. Section 4 contains examples and a necessary condition.