Generalized Confidence Constraints

In robust optimization, finding a solution that solely respects the constraints is not enough. Usually, uncertainty and unknown parameters of model are represented by random variables. such conditions, good to most-likely assignments these Recently, Confidence constraint has been introduced Mercier-Aubin et al. in order enforce this type robustness programming. Unfortunately, it restricted conjunction binary inequalities paper, we generalize any propose an implementation based on Multi-valued Decision Diagrams (MDDs). The defined over vector For given C, threshold, ensures probability for C be satisfied sample variables greater than threshold. We use MDDs represent efficient tool representing combinatorial constraints, thanks their exponential compression power. Here, both decision stored MDD, propagation rules proposed removing values cannot lead solutions. Furthermore, several show can omitted from MDD because lighter filtering algorithms sufficient. This leads gain factor size. experimental results obtained chemical deliveries problem factories – where chemicals consumption uncertain shows efficiency approach.