Information Technology Reference
In-Depth Information
the simulation engines with the GRAMS reference model is verified
by a test suite built around the widely known Tileworld-model.
The example implementation was further used for empirical meas-
urements. Multiple benchmark suites were defined and subsequently
executed on different simulation engines. The limiting influence of
constraint evaluation on overall runtime was shown in benchmark
setups with up to 10000 agents. Besides this, the practical evalu-
ation confirmed the tacit assumption that the choice of an optimal
partitioning strategy is not domain independent, but has to take into
account the specific domain of the model at hand.
