No label defined (Q3972041)

[For the entire collection see Zbl 0699.00020.] Let \(X\) be a subset of \(\mathbb{R}^ n\) and suppose \(f:X \to \mathbb{R}\) is ``weakly piecewise linear'' in the following sense: there is a finite collection of closed piecewise linear sets \(\{P_ i\}\) such that \(X \subseteq \bigcup P_ i\) and for each \(i\), the restriction of \(f\) to \(X \cap P_ i\) is (affine) linear. Is \(f\) ``strictly piecewise linear'', i.e., is \(f\) the restriction to \(X\) of a piecewise linear function defined on \(\mathbb{R}^ n\)? In general the answer is no, though it is not hard to think of conditions on \(X\) which are sufficient for a yes answer, e.g., \(X\) itself is a closed p.l. set. We show in this paper that there is a geometric condition on the embedding \(X \subseteq \mathbb{R}^ n\) which is necessary and sufficient for every weakly p.l. \(f:X \to \mathbb{R}\) to be strictly p.l.. It is a kind of local connectedness near the points of the remainder of a certain non-Hausdorff completion of \(X\). The proof involves facts about vector-lattice ideals and, implicitly, the duality between finitely presented vector-lattices and piecewise linear sets and maps due to Beynon (1975) and anticipated by Baker (1968). While the Baker-Beynon theory has been used in the past to derive results about vector-lattices from well known geometric facts, the present paper is the first instance known to the author in which it is used to derive a theorem in elementary geometry from algebraic facts about vector- lattices.