Covering convex bodies by cylinders and lattice points by flats (Q836226)

The celebrated Tarski plank problem was settled by \textit{T. Bang} [Proc. Am. Math. Soc. 2, 990--993 (1951; Zbl 0044.37802)], who proved that if a convex body in \(d\)-dimensional Euclidean space is covered by a finite collection of planks (strips of space between parallel hyperplanes), then the sum of widths of these planks is at least as large as the minimal width of the body. A related question Bang asks in his paper is whether the sum of the base areas of finitely many cylinders covering a 3-dimensional convex body is at least half of the minimum area of a 2-dimensional projection of the body. The paper under review is concerned with investigating Bang's problem in \(d\)-dimensional Euclidean space. By a cylinder in \(\mathbb R^d\) the authors mean a set \(C \subset \mathbb R^d\) of the form \(C=l+B\), where \(l\) is a line through 0 in \(\mathbb R^d\) and \(B\) is a measurable set in \(E := l^{\perp}\). For a convex body \(\mathbf K \subset \mathbb R^d\), the cross-sectional volume of a cylinder \(C\) with respect to \(\mathbf K\) is defined by \[ \text{crv}_{\mathbf K}(C) = \frac{\text{vol}_{d-1} (C \cap E)}{\text{vol}_{d-1} (P_E \mathbf K)}, \] where \(P_E \mathbf K\) is the orthogonal projection of \(\mathbf K\) on \(E\). With this notation, the authors prove that if a convex body \(\mathbf K \subset \mathbb R^d\) is covered by cylinders \(C_1,\dots,C_N \subset \mathbb R^d\), i.e. \(\mathbf K \subset \bigcup_{i=1}^N C_i\), then \[ \sum_{i=1}^N \text{crv}_{\mathbf K} (C_i) \geq \frac{1}{d}. \] Moreover, if \(\mathbf K\) is an ellipsoid, then \[ \sum_{i=1}^N \text{crv}_{\mathbf K} (C_i) \geq 1. \] The authors also contribute to the study of an analogous discrete problem of covering lattice points by lines, flats, and hyperplanes, proving a few interesting new estimates on the minimal number of such objects required to cover the set of integer points contained in a convex body \(\mathbf K \subset \mathbb R^d\). For instance, they prove that if \(\mathbf K\) is centrally symmetric with respect to 0 and \(H_1,\dots,H_N\) are \((d-1)\)-dimensional hyperplanes in \(\mathbb R^d\) such that \(\mathbf K \cap \mathbb Z^d \subset \bigcup_{i=1}^N H_i\), then \[ N \geq c_0 \frac{\text{w}(\mathbf K,\mathbb Z^d)}{d \ln (d+1)}, \] where \(c_0\) is an absolute positive constant and \(\text{w}(\mathbf K,\mathbb Z^d)\) is the lattice width of \(\mathbf K\), defined as \[ \text{w}(\mathbf K,\mathbb Z^d) = \min \left\{ \max_{x \in \mathbf K} \left< x,y \right> - \min_{x \in \mathbf K} \left< x,y \right> : y \in \mathbb Z^d \setminus \{0\} \right\}, \] where \( \left< \;,\;\right>\) is the usual Euclidean inner product on \(\mathbb R^d\). This bound extends a previous result by \textit{K. Bezdek} and \textit{T. Hausel} [Colloq. Math. Soc. Janos Bolyai. 63, 27--31 (1994; Zbl 0823.52006)].