Introduction - Modeling and Simulation of Complex Systems - page 6

1.3 Approach of this thesis and outline
The following approach is taken in this thesis to address the outlined
goals:
1. Providing a solid foundation of agent-based modeling and simula-
tion:
The solid foundation is provided in this thesis by definition of the
General Reference Model for Agent-based Modeling and Simulation
(GRAMS) . In a domain-independent way, the GRAMS reference
model defines on a conceptual level the basic building blocks of
agent-based models and their relations. Based upon this, con-
straints for the simulation of such a model are defined. Such a
clear and common understanding is the prerequisite for effective
model development (and subsequently for mastering the shift to
micro-modeling and enabling effective collaborative development).
Following well-established patterns, the GRAMS reference model
distinguishes between a simulation model and a simulation engine
executing a model (within the defined constraints). Although
seemingly obvious, many agent-based simulation studies often
intermingle these aspects. The clear distinction makes it possible
to develop and use different simulation engines executing the same
model and producing identical results.
2. Enabling small-scale development and smooth transition to large-
scale models:
Based on the GRAMS reference model, this thesis discusses various
possibilities of executing the simulation of agent-based models in
a parallel and distributed fashion. The objective is to develop
a model just once, and to enable its execution by different sim-
ulation engines producing identical results. Therefore, different
types of simulation engines are described, including single-threaded
simulation engines and a multi-threaded simulation engine. The
multi-threaded simulation engine basically serves two purposes:
Firstly, to evaluate different strategies for parallelizing execution of
an agent-based simulation and measuring achieveable performance