A Consistency-Centric Approach to Set-Based Optimization with Multiple Models of Unranked Fidelity

S-BOMM (Set-Based Optimization with Multiple Models) introduces a consistency-driven multi-fidelity optimization framework that eliminates the assumption of a single ground-truth high-fidelity model. Instead of ranking models by fidelity, it identifies solutions robust across all models using probabilistic consistency metrics. Engineers running simulations with unverified model hierarchies, such as in aerospace or chemical process design, should pay close attention. This directly challenges standard multi-fidelity Bayesian optimization pipelines where high-fidelity labels are treated as oracle ground truth.