Constructing a polytope to approximate a convex body (Q1900094)

The paper is a sequel to the authors' paper in Stud. Math. 111, No. 1, 81-95 (1994; Zbl 0808.52001)] where the authors developed a method to construct a convex polytope \(P\) with at most \(n\) vertices lying on the boundary of a convex body \(K\) in \(\mathbb{R}^d\), which satisfies \(|K \backslash P |/ |K |\leq f (d)/n^{2/(d - 1)} (|A |\) denoting the \(k\)-dimensional volume of the measurable set \(A \subset \mathbb{R}^d\), where \(k\) is the dimension of the minimal flat containing \(A\).) In this paper they establish \(f(d) \leq c \cdot d\) \((c\) is an absolute constant) for any convex body \(K\) in \(\mathbb{R}^d\), and this is the smallest known general upper bound for \(f(d)\). The result is a consequence of the Brunn-Minkowski theorem and the following proposition established in the paper: Proposition: Let \(f\) be a nonnegative, concave function on the interval \([a,b]\). For every \(p \geq 1\) and positive integer \(n \geq n_0 (p)\) there exists a piecewise linear function \(g\) on \([a,b]\), having at most \(2n\) nodes, all of them on the graph of \(f\), which satisfies \[ \int^b_a \bigl( f(x)^p - g(x)^p \bigr) dx \leq {c \over n^2} \int^b_a f(x)^p dx. \]