European Conference on Operational Research

p. 271-272

Presentation's date: 2016-07-05

Abstract:

When a deterministic global optimization algorithm, such as the Ex- tended Cutting Angle Method (ECAM) is used for solving Lipschitz programs, the algorithm always converges to the same solution for the global optimum irrespective of the starting point. This is a very usual fact of the deterministic global optimization algorithms. While this could be good enough in many situations, it is not convenient in many other cases in which there may be different solutions for the global optimum and the realistic solution depends on the physical (or engi- neering) interpretation of these solutions. To overcome this limitation, this article provides the Mesas-Ferrer-Sainz method (MFS) for sub- dividing the domain of definition of a given optimization problem in such a way that it is possible to find all its global optimal solutions, especially when the set of global optimal solutions is finite. The main characteristic of MFS is its generality so it can be applied to any de- terministic global optimization algorithm for solving continuous opti- mization problems. As an example of application, the MFS method will be used for the resolution of an electrical engineering problem: The Load Flow in an Electrical Network.]]>