Locating lines and hyperplanes. Theory and algorithms (Q1282587)

This book refers to basic problems from continuous location science. The presented material is mainly interesting for path location (e.g., for traffic planning, locating pipelines and related topics). More precisely, the author presents the foundations for the study of median and center hyperplanes in finite-dimensional normed spaces. Given a finite set \(P\) of weighted points in a Minkowski space \(M^n\), \(n\geq 2\), necessary position criteria for median hyperplanes (minimizing the sum of weighted distances with respect to \(P\) and hence referring to a geometric modification of the famous Fermat-Torricelli problem) and center hyperplanes (minimizing the maximum weighted distance with respect to \(P\) and therefore, in the unweighted situation, equivalent to the point set width problem) are derived. These criteria transfer the originally continuous location problems in a certain sense to discrete ones, and so they allow algorithmical approaches to such optimal hyperplanes, although (in general) not every such hyperplane can be found. Thus, parts of the book are directly connected with typical problems from computational geometry (in the spirit of linear fit problems). And since an important building block for all the problems and results is the consideration of vertical distances, direct applications can also be found in robust statistics (linear regression). In addition, the book contains various related results, such as characterizations of smooth norms (with the help of stronger position criteria, for median and center hyperplanes), sharpenings for polyhedral unit balls, extensions to spaces with non-symmetric unit balls, results on hyperplane transversals, and location problems referring to other searched objects (like, e.g., circles).