Model reduction: some results

Abstract

Simulation is becoming a popular tool, aided by the development of powerful and faster computers, new features like parallel and distributed simulation and easier simulation software that allows to reduce the time to implement a model. In spite of that, the cycle time of a simulation study is still long. The paper presents some remarks on model reduction reasoning and how, through this, the time of a simulation study could be reduced. The discussion is based on the example of Kienbaum and Paul (1994) of the pub problem, using a simplification of their H-ACD representation to describe the original and reduced models. The models are implemented and the results obtained shows their statistical equivalence which suggests the feasibility of this approach.