Abstract - Thoeny et al., 1994

Article:

Thoeny, J., Fischlin, A. & Gyalistras, D., 1994.  RASS: Towards Bridging the 
        Gap between Interactive and Off-line Simulations.  In: Halin, J.  (ed.), 
        San Diego, California, The Society for Computer Simulation, 99-103.

Abstract:

Interactive model exploration is an important step in the process of model-building of ill-defined systems such as ecosystems, a task which is well supported by RAMSES (Research Aids for the Modelling and Simulation of Environmental Systems). However, interactive simulation is of minor interest at a later stage of research, when large scale batch oriented simulation experiments are needed.

RASS, the RAMSES Simulation Server, typically located on a high performance computer, translates, compiles, links, and executes in a batch mode interactive RAMSES model definition programs [MDP]. MDPs must be given in source form and formulated according to one or any combination of the following standard formalisms: SQM (Sequential Machine), DESS (Differential Equation System Specification), or DEVS (Discrete EVent System Specification). The simulation results generated and returned by RASS can then be explored interactively by means of the post-analysis component of RAMSES.

RASS was found to allow for automatic translation of interactive programs developed for a graphical user-interface with windows and menus in order to execute them in a batch mode to produce correct and reliable simulation results. Three complex case studies from the field of ecology and engineering showed that performance gains of 1'100-7'200% (overhead included) can be obtained when RASS is run on a SUN S10 server relative to the time needed when running the MDP interactively on an average personal computer. Since all transfers showed to be user-friendly and smooth, we concluded that RASS offers RAMSES simulationists an efficient and attractive alternative for solving interactive MDPs whenever off-line simulations are needed, or when it is a necessity because of the high computing requirements.

Full text (451KB)

back