Normal sets, polyblocks, and monotonic optimization (Q1573518)

This paper investigates normal sets in order to be able to handle systems of monotonic inequalities and optimization problems involving differences of monotone increasing functions. A normal set \(G\) is a subset of \(\mathbb{R}_+^n\) such that, whenever \(x \in G\), it follows that all \(x' \in \mathbb{R}_+^n\) with \(x' \leq x\) are also in \(G\). The paper reviews after a short introduction the basic properties of normal sets and reverse normal sets. Polyblocks and reverse polyblocks are introduced thereafter. A polyblock is a normal set which is the union of finitely many boxes. Then a section is devoted to systems of monotonic inequalities. Characterizations of solutions are derived and these are used to deal with the maximization or minimization of an increasing function under monotonic constraints. Finally the approach is extended to differences of increasing functions. The applicability of the presented approach is illustrated by examples from generalized multiplicative programming, non-convex quadratic optimization and polynomial programming.