On the lexicographic centre of multiple objective optimization (Q274101)

The authors consider a general Multiple Objective Problem (MOP) and introduce the concept of the MOP's lexicographic centre. It is a refinement of the standard minimax approach to MOPs. This approach requires: (a) minimizing the largest objectives, (b) minimizing the second largest objectives over the set of all optimal solutions in (a), (c) minimizing the third largest objectives over the set of all optimal solutions in (b) and so on. This is a process of recursive optimization. It is shown that the MOP's lexicographic centre exists if the MOP's image set is bounded below and closed. Further, a linear MOP is considered (abbr. MLOP). It is shown that the MLOP's centre is nonempty if and only if there exists an efficient solution of the MLOP. A polynomial-time algorithm to determine whether an efficient solution to an MLOP exists is proposed. The MLOP's lexicographic centre is solved by calculating at most \(p\) dual linear programs, where \(p\) is the number of objective functions. Two numerical examples illustrate the proposed algorithm. Some of the results may be extended to multiple objective convex programming problems.