ARGONAUT: AlgoRithms for Global Optimization of coNstrAined grey-box compUTational problems

The algorithmic framework ARGONAUT is presented for the global optimization of general constrained grey‐box problems. ARGONAUT incorporates variable selection, bounds tightening and constrained sampling techniques, in order to develop accurate surrogate representations of unknown equations, which are globally optimized. ARGONAUT is tested on a large set of test problems for constrained global optimization with a large number of input variables and constraints. The performance of the presented framework is compared to that of existing techniques for constrained derivative‐free optimization.