Non-convex multi-objective optimization (Q2360302)

The book deals with theoretical properties and solution methods of multi-ubjective optimization problems of the following form: \[ \min_{x \in A} f(x) = (f_1(x),_1, \dots, f_m(x)_m)^T, \] where \(f: A ~\longmapsto~\mathbb R^m\) and \(A = \{ x \in \mathbb R^d ~|~ g_j(x) \leq 0, ~ j = 1, \dots, k \}\). The book has 10 chapters divided into 3 parts. Part I (Chapters 1--4) has an introductory character. The authors introduce definitions of basic concepts together with illustrative examples and explain main ideas of four scalarization approaches to multi-objective problems, namely Chebyshev approach, using achivement scalarization function, an approach based on successive minimization of components \(f_k(x)\) for \(k = 1, \dots, m\) under the constraints \(f_i(x) \leq f_k(x), ~i = 1, \dots, m, i \neq k\), and Pascoletti-Serafini scalarization. Pareto-optimality of the resulting compromise vectors is proved. Existence of polynomial complexity approximations of the NP-complete multi-objective problems is investigated in Chapter 3. A brief review of solution methods for single-objective non-convex optimization problems is presented in Chapter 4. Part II (Chapters 5--8) is devoted to the description of algorithms solving multi-objective optimization problems. The authors extend theoretically substantiated methods of single-objective global optimization problems presented in Chapter 4 of Part I to multi-objective optimization. Optimization algorithms for the case of Lipschitz continuous components of objective function \(f(x)\), statistical models for global optimization as well as branch and probability bounds methods are extended to multi-objective optimization. Further it is shown how visualiation of the Pareto-front estimates can help to solve some of the multi-objective problems. Part III (Chapters 9--10) is devoted to applications of the developed techniques to analysis of business process diagrams. The authors consider also applications to manufacturing and chemical process engineering.