On the sum of intervals (Q4424445)

This paper deals with the problem of approximating a Euclidean ball in \(\mathbb{R}^n\) with a Minkowski sum of \(N\) segments. Recall that the Minkowski sum of two sets \(A, B \subset \mathbb{R}^n\) is defined as \(A + B = \left \{ a + b ; a \in A, b \in B \right \}\). The author estimates the minimal number \(d(n,N)\) such that it is possible to find a set \(V \subset \mathbb{R}^n\), which is a Minkowski sum of \(N\) segments, and for which \[ D \subset V \subset d(n,N) D \] where \(D\) is the standard Euclidean ball in \(\mathbb{R}^n\). Precise estimates are given for \(d(n,N)\), for all values of \(n\) and \(N\). It is proved that if \(N \geq n\), \[ \frac{1}{C} \min \left \{ \sqrt{n}, \sqrt{\frac{N}{N-n} \log \frac{N}{N-n} } \right \} < d(n,N) < C \min \left \{ \sqrt{n}, \sqrt{\frac{N}{N-n} \log \frac{N}{N-n} } \right \} \] for some universal constant \(C > 0\). The proof of the upper bound is probabilistic, and is almost entirely based on earlier results of the author [Math. USSR, Sb. 48, 173-182 (1984; Zbl 0558.46013)] and \textit{A. Y. Garnaev} and \textit{E. D. Gluskin} [Sov. Math., Dokl. 30, 200-204 (1984); translation from Dokl. Akad. Nauk SSSR 277, 1048-1052 (1984; Zbl 0588.41022)]. The proof of the lower bound is a simple and elegant argument, which uses nothing but some basic notions and tools from the general theory of Banach spaces. For the entire collection see [Zbl 1009.00019].