An adaptive metamodeling-based method for simulation optimization

Summary form only given. In this work, a simulation optimization method is proposed. The method starts with an initial design of experiments with which an incumbent solution is obtained. At each iteration, a metamodel is obtained using the available set of points and it is used to generate a new attractive point where a simulation is performed. The simulated value of the new point is compared against the incumbent for updating purposes. A series of stopping criteria are evaluated and, if none is met, the new point is added to the existing set of points and a new iteration begins. Otherwise, the method stops. Preliminary results on the application of the method to several examples point towards a quick convergence to highly attractive solutions with a low number of simulations. The method is easy to follow and allows to be coded in a convenient manner to be run with low computational resources.