Information Technology Reference
In-Depth Information
detailed analysis of the influence-reaction approach will be given in
Section 2.3.3, and an exhaustive discussion as well as more details on
the role of the environment are given in Section 2.4. Summing up, the
framework proposed by Klugl provides formal descriptions of many
ideas which are central to agent-based modeling. Unfortunately, the
lack of a well-integrated simulation time and the time-stepped exe-
cution are great disadvantages (especially as no further development
takes place).
2.3.2 Scheutz, Schermerhorn (2006)
Motivation
Scheutz and Schermerhorn state that 'it is very dicult ... to utilize
the potential parallelism present in many agent-based models for
running parts of the models in parallel' [114]. They believe, that
'an algorithm that can automatically and dynamically distribute a
given agent-based model over a dynamically changing set of CPUs
. . . would be of great utility' [114]. They also point out very clearly,
that a prerequisite for the definition of any mechanism for automatic
parallelization or load distribution is a 'formal description of the
notion of
agent-based model
'[114], as otherwise all approaches are
only applicable and valid for specific model instances.
Therefore, their motivation for defining a formal model of a multi-
agent simulation is derived from the intention to design algorithms for
automatic parallelization of multi-agent simulations defined within
their proposed framework.
Outline of the framework
In short, the framework defined by Scheutz and Schermerhorn char-
acterizes 'a simulation of an agent-based model as a computational
process that starts in some initial condition and then updates the
