A Model Driven Engineering Framework for Simulation Experiment Management

Abstract

Simulation experiments are a convenient and useful means to gain insight into the operation of scientific models. They are conducted to address specific goals and evaluate specific questions about the model. These simulation models are complex, with many possible factors and outcomes. Also, a model that represents certain key characteristics or behaviors of the system can be analyzed to show the eventual real effects of alternative conditions and courses of action. The strength of simulation is that it enables precisely this “what if” hypotheses analysis, under certain assumptions. Efficient experiment designs are necessary for understanding the impact of these factors and their interactions on the model outcomes that establish the dependencies among goals, hypotheses and experiments with the factors of the model. In our study, we propose a model discovery process by devising questions about the model, designing experiments to validate these hypotheses, executing them, drawing inferences and refining it in an iterative manner to support temporal evidences about the model that have a degree of acceptability of its own. Using the cognitive theory of coherence, we establish links between hypotheses and temporal evidences. We use the principles of model driven engineering and domain specific languages to streamline the discovery process through scientific experimentation.