A Model-Driven Approach for Conducting Simulation Experiments

With the increasing complexity of simulation studies, and thus experiments, there is a high demand for better support them to be conducted. Recently, model-driven approaches have been explored facilitating specification, execution, reproducibility experiments. However, more general approach that suited variety modeling areas, experiment types, tools, which also allows further automation, still missing. Therefore, we present novel engineering (MDE) framework studies extends state-of-the-art conducting experiments in following ways: (a) Providing structured representation various ingredients form meta models collecting repository improves knowledge sharing across application domains approaches. (b) Specifying quasi-standardized (e.g., via GUI) and, subsequently, performing automatic generation specifications language choice increases both productivity quality complex (c) Automatic code transformation between specification languages enables reusability (d) Integrating using command-line interface automation subprocesses within study. We demonstrate advantages practicality our real from three different fields (stochastic discrete-event cell signaling pathway, virtual prototyping neurostimulator, finite element analysis electric fields) types (global sensitivity analysis, time course convergence testing). The proposed can starting point therefore, assist systematic effective manner. For example, based on this MDE framework, automatically selecting parametrizing experimentation methods, or planning follow-up activities depending context study, could developed.