9.4. Global solver
\({\rm I\!R}^p\) spaces have no ordering relations when \(p\) is greater than 1. As a consequence, finding the minimum of a function the output of which is defined in such a space has no meaning. Multicriteria optimisation solves this problem by finding the inputs corresponding to the Pareto frontier in the costs space.
The evolutionary algorithm library Vizir, allows to perform multicriteria optimisation with a global approach, and is available in Uranie. It is first introduced in a broad picture (even though more details can be found in [Bla17]) before considering the different solvers available and their possible configuration. We have also illustrate its use in a simple example that can be found in Macro “reoptimizeHollowBarCodevizir.C”.