Observation of large-scale multi-agent based simulations
arXiv:1311.0758 · doi:10.1007/978-3-642-28400-7_8
Abstract
The computational cost of large-scale multi-agent based simulations (MABS) can be extremely important, especially if simulations have to be monitored for validation purposes. In this paper, two methods, based on self-observation and statistical survey theory, are introduced in order to optimize the computation of observations in MABS. An empirical comparison of the computational cost of these methods is performed on a toy problem.