Gated sets in metric spaces (Q910755)

A subspace \({\mathcal A}\) of a metric space \({\mathcal C}\) with metric d is called gated if there is a function \(P_{{\mathcal A}}: {\mathcal C}\to {\mathcal A}\) so that for each \(c\in {\mathcal C}\) and each \(a\in {\mathcal A}\), \(d(c,a)=d(c,P(c))+d(P(c),a)\). If such a function exists it is unique and is called the projection of \({\mathcal C}\) onto \({\mathcal A}\). We also call \({\mathcal P}_{{\mathcal A}}(c)\) the gate of \({\mathcal A}\) with respect to c. The main theorem states that if \({\mathcal A}\) and \({\mathcal B}\) are gated subspaces of a metric space \({\mathcal C}\) with metric d, then \(P_{{\mathcal A}}\) and \(P_{{\mathcal B}}\) induce isometries between \({\mathcal A}\) and \({\mathcal B}\) that are inverse to each other. This paper grew out of an analysis of projection maps in the study of Coxeter complexes and buildings by \textit{J. Tits} [Buildings of spherical type and finite BN-pairs, Lect. Notes Math. 386 (1974; Zbl 0295.20047). The main theorem is used to understand projections in Tits buildings.