Information Technology Reference
In-Depth Information
Although this distinction is highly desirable, it is not realized in many
available simulation applications [70]. Furthermore, it is important to
stress this distinction, as many publications especially in the field of
multi-agent systems do not emphasize it and use the term
environment
in various contexts [146, 95].
The following subsections deal with the simulated environment
unless explicitly stated otherwise. As 'the simulated environment
is part of the model itself, it should be specified with at least the
same amount of carefulness as the simulated multi agent system' [70].
Nevertheless, 'most researchers neglect to integrate the environment
as a primary abstraction . .. or minimize its responsibilities' [146,
147]. Keeping this in mind, a large amount of issues regarding the
environment itself and the relations of agents and the environment
still needs to be solved.
2.4.1 Definition and properties
The agents of a multi-agent simulation, representing some active entit-
ies of the system under investigation, are usually placed within some
environment and their actions are therefore depending on and limited
to this environment [95]. To put it differently, 'the simulated envir-
onment contains what the agents may perceive and manipulate' [70].
In a more general sense, the environment denotes the common space
in which various types of entities are interacting (either active or
passive). Active entities are commonly represented by the agents,
whereas passive entities can be used to represent arbitrary objects
within the simulated environment which do not exhibit any behavior.
Besides these entities, the environment may carry some global state
variables like overall temperature, also referred to as environmental
properties [70, 52, 69].
The set of environmental entities can usually be partitioned into
disjoint subsets (e.g., the above mentioned active and passive entities).
Often the concept of embedding is used for relating agents to entities.
