System models are often studied through the use of computer simulation. This approach can be very attractive as it provides a simple dynamic view of the system under study, which can be used for debugging, and performance data can be gathered by repeated experiment. However, simulation has the limitations that all experiments do, the cost of each data point, potential to avoid particular cases and difficulty in establishing precisely what has been measured. A dual approach of using a simulation style to present and debug programs followed by a formal mathematical analysis would have the benefits of both approaches. We demonstrate how a large fragment of a process oriented simulation language DEMOS can be translated into Petri Nets. This translation has benefits in that it permits the simular access to the analysis tools for Petri Nets (we demonstrate automated translation both to the Sharpe and SPNP formats) and provides a succinct formal account of DEMOS. The current compiler is abstracted from the underlying Petri Net representation syntax and consequently can be rapidly retargeted at an arbitrary Petri Net representation. Given the ratio between the Petri Net representation and the DEMOS shown herein it may well be the case that the later is a significantly better approach to introducing Petri Net analysis. 25 Pages
PDF
download
878987797
028-85220240
