Developing a model for simulation is a difficult task, in which simulation experiments play a critical role. In mod-eling and simulation, domain specific languages are wide-ly used for model description. As more and more efforts have been put in facilitating simulation reproducibility in recent years, an unambiguous and explicit description of simulation experiments increasingly receives attention, with which simulation experiments can be reproduced. This motivates the use of domain specific languages as the means to express experiment specifications.
Domain specific languages can be used to specify differ-ent tasks of simulation experiments, such as experiment configuration, observation, analysis, and evaluation of experimental results. More importantly, they can serve to specify crucial observations from experiments regarding model behavior. Therefore, with a formal description of model behavior, an evaluation based on model checking techniques can also benefit from domain specific languages.
In this paper, we will first discuss how domain specific languages can be used to specify simulation experiments and illustrate it by using the domain specific language SESSL. We aim at dealing with stochastic models. Several problems arise in specifying simulation experiments on stochastic models, such as probability estimation, sto-chastic noises toleration and robustness measurement. We discuss how those problems can be handled by using domain specific languages and show their advantages