appearing of computational strategies appropriate to reproduce
complex functions and collective behavior of interacting agents.
The general philosophy inspiring Multi Agent Systems (MAS)
is reminiscent of cellular-automata and adaptive systems [7] whose
basic idea [8] is generalized including the possible influence of
agents in any cell of the grid on the behavior of agents localized
in any other cell.
As a matter of fact, MAS are most often implemented in studies
concerning numerous populations of interacting agents capable of
simple, autonomous actions producing complex, collective behav-
ior,. It is worth noting that single agents lack a full global view of
the system on their own, but show self-organization as well as self-
steering and sophisticated individual behaviors. In other words, the
utility of MAS remains outstanding when one looks for:
l A qualitative account of the global population’s behavior from
the known features of the single agents and of their spatial
location.
l A quantitative refinement of the model parameters describing
the collective behavior with the aim of fitting at best the dynam-
ics observed in experimental data.
Among the several software tools available in this framework
NetLogo™[9] has the advantage of being: (1) frequently updated
by an active users-community, (2) endowed with a relatively
smooth learning curve, and, (3) last but not least, freely available.
NetLogo™is an interpreted programming environment based
upon LOGO™, one of the classical Artificial Intelligence (A.I.)
languages, and is particularly well suited for modeling complex
Fig. 1Left: Development of Multi-Agent Systems (MAS) from other similar disciplines (Modified from [6]).
Right: The space (global domain) in which agents move is a grid of variable size where each single location
(local domain) can be endowed with specific features, like attractive/repulsive ability caused by force-fields of
defined intensity and shape
308 Alfredo Colosimo