Application of stochastic approximation methods for stochastic computer models calibration

Computer models are widely used to simulate real processes. Within the computer model, there always exist some parameters which are unobservable in the real process but need to be specified in the model. The procedure to adjust these unknown parameters in order to fit the model to observed data and improve its predictive capability is known as calibration. In this paper, we propose an effective and efficient algorithm based on the stochastic approximation approach for stochastic computer model calibration. We first demonstrate the feasibility of applying stochastic approximation to calibration and apply it to two stochastic simulation models. We compare our proposed SA approach to another direct calibration search method, the genetic algorithm. The results indicate that our proposed SA approach performs equally as well in terms of accuracy and significantly better in terms of computational search time.