On sets in \(\mathbb{R}^d\) with DC distance function (Q2011247)

A real-valued function \(f:\mathbb{R}^{d}\mathbb{\rightarrow R}\) is said to be DC if it is the difference of two convex functions. The main result in this paper states that, if \(d=1\) and \(f\) is DC, then the Euclidean distance function to the graph of \(f\) is DC, too. If \(d>1\), the same holds true if \(f\) is semiconcave (that is, if \(f-\alpha ||\cdot ||^{2}\) is concave for some \(\alpha >0,\) with \(||\cdot ||\) denoting the Euclidean norm). A set \(A\subset \mathbb{R}^{d}\) is said to be of positive reach if there exists \(r>0\) such that the metric projection of \(z\) onto \(A\) is a singleton for each \(z\) whose distance to \(A\) is smaller than \(r\). The authors prove that the Euclidean distance function to the closure of the complement of a set of positive reach and the one to its boundary are both DC.