Agent-based modeling and simulation - Modeling and Simulation of Complex Systems - page 42

teams. Furthermore, the unclear basis complicates the development of
alternative simulation engines which adhere to a common set of spe-
cifications (e.g., a sequential simulation engine or a massively parallel
simulation engine, capable of executing exactly the same model).
Insu cient definition of simulation time and time-advancement
mechanism Many approaches miss a consistent and well-defined
description of simulation time or a definition of the time-advancement
mechanism. Closely related to this is the vague definition of syn-
chronization aspects in the approaches of Klugl and Scheutz and
Schermerhorn. As 'synchronization is often the technically most dif-
ficult aspect in implementing an agent-based simulation' [69, p. 22],
the missing accuracy and clarity in this point is a major drawback of
these two approaches. Furthermore, both approaches do not specify
whether the simulation steps are all of the same length (in matters
of time), and both approaches do not make any statement on the
granularity or atomicity of the actions of the agents. This means,
it is unclear whether each action has to be executed within a single
simulation step or within multiple simulation steps. Therefore, it is
unclear how much time actions consume or if the agents behavior is
based on some kind of atomic actions .
Any answer to this question will also have to deal with the problem
of conflicting actions. The approaches of De Vries et al. and Ke/Hu
both deal with the matter of actions with duration and have further in
common that they distinguish between a macro-time and a micro-time
scale. Actions are commonly defined on the macro-time level while
the concurrent execution usually takes place on the micro-time level.
Therefore these two approaches provide a useful starting point for
integrating this missing aspect.
The whole area of interference of actions is treated only very sparse
in the presented approaches, except for the approaches of De Vries et
al. and Ke/Hu that may thus provide useful starting points.