The Fermat-Torricelli problem. I: A discrete gradient-method approach (Q368737): Difference between revisions

The article of Yaakov S. Kupitz, Horst Martini and Margarita Spirova is a valuable contribution to the field of discrete mathematics, geometry and combinatorics, but also to discrete-continuous calculus and optimization, particularly, as it employs the use of a special and, in fact, discrete area of derivative-free calculus and optimization called discrete-gradient method. The paper is well structured, well written and explained, and well demonstrated and illustrated by proofs, applications and figures. In order to give some potential areas of special application which could be considered for the authors' results and methodology, there are: location problems, transportation and allocation problems, image processing, tomography, humanitarian logistics, special problems in fluid dynamics, portfolio optimization and indication of financial bubbles, actuarial sciences, risk management and the design of information flows and their technology. The authors provide a discrete geometric (differential-free) proof of the theorem underlying the solution of the well-known Fermat-Torricelli problem, referring to the unique point having minimal distance sum to a given finite set of non-collinear points in \(d\)-dimensional space. Furthermore, they extend this problem to the case that one of the given points is replaced by an affine flat, and they also provide a partial result in the case where all given points are generalized to affine flats (of various dimensions), with illustrative applications of these theorems. It may be expected that this new powerful tool will lead to strong improvements in linear integer- or mixed-integer programming and, based on that understanding of the linear case, also to nonlinear integer- and mixed-integer programming later on. The five sections of this article are as follows: 1. Introduction, 2. The basic characterization, 3. A generalization, 4. Applications and further generalizations, and 5. Concluding remarks. Indeed, in the future, further strong results and algorithms could be expected, initialized by this paper. Those advances might foster and further initiate emerging contributions to science and engineering, e.g., in data mining, image processing and inverse problems, to economics and finance, to OR, to medicine and healthcare, and, eventually, to the living conditions on earth.