|
|
|
Towards an Agent-Based Modeling Platform with Formal Semantics Based on the SARL Programming Language.
Abstract
Agent-based modelling and simulation (ABMS) is a powerful technique for modelling
complex systems that consist of multiple interacting entities. In ABMS, each entity
is represented as an autonomous agent that observes and acts on an environment.
The field of ABMS, however, is very divided and, although most research groups
have similar interpretations of the concepts in ABMS, a universal agreement on a
definition of an agent-based model remains absent. This problem manifests itself in
the available agent-based modelling and simulation tools, since no tool is available
which has formal or even precise semantics. The lack of uniformity results in a
field where models are created on an individual basis and reuse-ability is scarce.
In this study, we tackle this problem by investigating how agent-based modelling
is currently realized and propose a formalism with precise semantics that captures
the essential characteristics of agent-based modelling. Additionally, we present a
tool which implements the proposed formalism. This new tool, SARLforSIM, is
evaluated against a selection of current agent-based modelling and simulation tools.
We performed a comparative study by implementing the same population dynamics
model in each of the tools and investigated how they realize the features of an agent-based model. Finally, we also investigated the reproducibility of the model in the
different tools
Download Link
The thesis report can be downloaded here.
Mindmap
|
