Constrained minimum energy designs

Space-filling designs are important in computer experiments, which critical for building a cheap surrogate model that adequately approximates an expensive code. Many design construction techniques the existing literature only applicable rectangular bounded space, but real world applications, input space can often be non-rectangular because of constraints on variables. One solution to generate constrained is first uniformly distributed samples feasible region, and then use them as candidate set construct designs. Sequentially Constrained Monte Carlo (SCMC) state-of-the-art technique generation, it still requires large number constraint evaluations, problematic especially when evaluate. Thus, reduce evaluations improve efficiency, we propose Minimum Energy Design (CoMinED) utilizes recent advances deterministic sampling methods. Extensive simulation results 15 benchmark problems with dimensions ranging from 2 13 provided demonstrating improved performance CoMinED over