Unique determination of convex lattice sets (Q527438): Difference between revisions

This paper investigates certain questions from discrete tomography, analogous to some classical results in geometric tomography. Specifically, the well-known theorem of Aleksandrov states that a convex \(0\)-symmetric set in a Euclidean space is uniquely determined by the areas of its projections. A discrete analogue of Aleksandrov's theorem has been proposed in [\textit{R. J. Gardner} et al., Discrete Comput. Geom. 34, No. 3, 391--409 (2005; Zbl 1093.52008)]: given two \(0\)-symmetric convex lattice sets \(K\) and \(L\) in \(\mathbb Z^n\), \(n \geq 2\), such that for every \(u \in \mathbb Z^n\), \(\left| K \mid u^{\perp} \right| = \left| L \mid u^{\perp} \right|\), is it necessarily true that \(K=L\)? Here \(K \mid u^{\perp}\) stands for the projection of the set \(K\) into the hyperplane normal to the vector \(u\), and \(\left| K \mid u^{\perp} \right|\) for its cardinality. In fact, in the same paper this question is answered negatively when~\(n=2\). The paper under review addresses a related question: given two full-dimensional \(0\)-symmetric convex lattice sets \(K\) and \(L\) in \(\mathbb Z^n\) such that for every \(u \in \mathbb Z^n\), \(\left| \partial(K \mid u^{\perp}) \right| = \left| \partial(L \mid u^{\perp}) \right|\), is it necessarily true that \(K=L\)? Here \(\partial (K \mid u^{\perp})\) is the discrete boundary of the projection, i.e. it consists of the points of \(K \mid u^{\perp}\) that lie on the boundary of the convex hull of \(K \mid u^{\perp}\). The main results of this paper answer this question in the affirmative when \(n=3\), as well as when \(n \geq 4\) and convex hulls of \(K\) and \(L\) are zonotopes (finite Minkowski sums of line segments). The proofs employ tools from the geometry of numbers and discrete geometry, including Pick's theorem and Minkowski uniqueness theorem (up to translation) for convex polytopes with given facet areas and normals.