A probabilistic solution-generator for simulation

In a discrete-event simulation study, we are usually faced with dealing with an extremely large design space in a random environment. Our research goal in the paper is to construct a probabilistic solution generator (PSG) to generate a small candidate design set (CD). The CD should contain the solutions meeting at least a certain performance level with a high probability. The PSG is constructed using a rough-cut evaluation method rather than the simulation model in order to sift through the large design space quickly. A rough-cut evaluation of designs may be achieved through an approximate method, a heuristic approach, or a short simulation run possibly with a high error. Subsequently, the designs in the CD are to be analyzed thoroughly in the simulation model to select the best with a high degree of certainty with one of the well-known statistical comparison and selection techniques.