Efficient simulation budget allocation for stochastically constrained simulation optimization with regression in partitioned domains

In this paper, we consider the problem of selecting the best design subject to stochastic constraints. We further improve the selection efficiency by incorporating the information from across the domain into regression equations. The domain of interest is divided into adjacent partitions such that the underlying functions of both the main objective and constraint measures in each partition are approximately quadratic with homogeneous noise. Based on the large deviation theory, we characterize an asymptotically optimal allocation rule by maximizing the rate at which the probability of false selection tends to zero. Numerical experiments demonstrate that the proposed approach can significantly improve the selection efficiency over the existing methods.