Information Technology Reference
In-Depth Information
defines which information an agent
a
may perceive from its relevant
individual environment
RU
a
and maps this information to the physical
state of the agent. The behavior of an agent is then defined by the
function
act
a
:
S
a
→
ACT
a
(2.7)
where
ACT
a
denotes the set of possible actions of an agent
a
.Asan
agent can only control its effectors, but not the effects of its action
in the environment, Klugl separates (in accordance with [33], see
also sec. 2.3.3) the agents actions from the environmental outcome.
Therefore, the function
execute
:
ACT
a
1
×···×
ACT
a
n
→
ES
(2.8)
of the environment actually computes the effects of all attempted
actions.
The global state set
GS
=
ES×MS
(with
MS
=
SA
1
×···×SA
i
)
is defined as the set of all possible physical states and mental states.
A simulation starts with a given state
gs
(0)
∈ GS
(usually determined
by the
initConfig
function) and consists of the iterated application
of the update functions defined by the model onto this state. The
result is a sequence (
gs
(
t
)
t∈T
) of states indexed and ordered according
to the simulation time, the so-called simulation trajectory.
Strengths and weaknesses
The formal framework proposed by Klugl is to be understood as a first
attempt, although unfortunately no further work on this framework
was published and development seems to have come to a stop. Besides,
there are several strengths and weaknesses. The framework aims to
provide a formally grounded terminology about agent-based models
and shows the model structure and dynamics in a quite concise way,
making consequent use of a rigorous mathematical notation. The
major drawbacks are:
1. Insucient integration of simulation time