Information Technology Reference
In-Depth Information
deals explicitly with parallel execution. The third approach was pro-
posed in 1996 by Ferber and Muller [33] and has also spawned several
follow-up developments. After presenting these three approaches the
agent-object-relationship model by Wagner and two further relevant
approaches are presented briefly.
2.3.1 Klugl (2007)
Motivation
Klugl argues that one of the reasons hindering the wide-spread use
of multi-agent simulation (especially outside universities) is the lack
of a formal basis. This leads to missing standards of specification
and documentation with the consequence of missing reproducibility.
Furthermore, a 'generic formal framework for representing agent-based
simulation would also support differentiation between agent-based
approaches and traditional microscopic ones' [69]. As it is also one
of the key challenges (according to Klugl) to point out advantages of
agent-based simulation in contrast to other paradigms, such a formal
framework would be very useful.
In summary, Klugl points out that 'there is no framework that fully
pins down which elements a Multi-Agent Simulation Model consists
of and what relations might exist between these elements' [69].
Outline of the framework
Klugl distinguishes between structural aspects and processes respect-
ively dynamics. Let
E
=
{e
1
,e
2
,...,e
n
}
be the set of environmental
entities, i.e., the set of objects that exist in an environment. Together
with the set
P
=
{p
1
,p
2
,...,p
m
}
of environmental properties the
complete physical configuration of a multi-agent simulation is given
as
PCON
=
E ∪ P
.Thesetofallpossiblestatesofanentity
e
is de-
noted by
SE
e
, the cartesian product
ES
of all
SE
s gives the set of all
possible states of all entities and properties. The initial configuration
is then given by a function
initializeConfig
:
PCON
→
ES
.
