Saddle points and scalarizing sets in multiple objective linear programming (Q406635)

The article of Daniel Gourion and Dinh The Luc is a valuable contribution to the areas of both linear programming (LP) and multiobjective optimization. This research takes place in a wide set-theoretical and linear-algebraic setting, and elegantly generalizes approaches and results from LP and so-called vector optimization. The paper certainly stimulates experts to important future research and also application. The major goal of this article is to investigate saddle points of the vector Lagrangian function that comes from a problem of multiple objective linear programming. The authors introduce three different concepts of saddle points and they establish corresponding characterizations via the solution of systems of equalities and inequalities accordingly. They deduce dual problems and demonstrate a relationship that exists between saddle points and dual solutions, enabling them to receive an explicit expression of the scalarizing set of a considered saddle point, expressed by the help of normal vectors to the problems value set. Eventually, the authors introduce a procedure to compute saddle points associated to nondegenerate vertices and the related scalarizing set. The paper is mathematically deep, well structured, carefully demonstrated and well written. The five sections of this article are as follows: 1. Introduction, 2. Vector Lagrangian function, 3. Saddle points and dual solutions, 4. Scalarizing sets, and 5. Computing saddle points and scalarizing sets. In the future, further extensions, strong results and numerical algorithms can be expected, initialized in the scientific community by this present research article. This progress could initiate and support important achievements in science, engineering, economics, decision making, finance and OR, and, eventually, for improvements of living conditions of the people.