Geoscience Reference
In-Depth Information
used to model the interactions between the developers, the policies, the jobs and the
households. The model has been applied to numerous American cities (Seattle,
Detroit and San Francisco, for example) and in the world (Rome, Beijing and Seoul,
among others). From an epistemological point of view, this model is the opposite of
the Schelling's model mentioned in the introduction of this chapter. It contains a
multitude of processes and once well calibrated, it is operational, allowing relatively
safe estimates in the short term. Nonetheless, it is not appropriate to explore how
interactions operate between different urban processes and identify changes in the
founding mechanisms of urban dynamics.
4.3. Computing models: simulation and emergence
As it has been pointed out in section 4.1, the AC and MAS models favor an
approach in terms of “generating mechanisms” and place the emphasis on the
organization, which emerges at a higher observation level, the driving force being
the interactions operating at a lower level. A simulation model can play the role of a
“laboratory” where the possible consequences of hypotheses concerned with the
formation of the interactions between the considered entities (for example, forming
an artificial society, an artificial system of population or an artificial school space)
are tested. As such, the idea is similar to the microsimulation models presented
above. The difference is probably more attributable to the practices than to the
methodologies [BOM 04]. The microsimulation models aim to rebuild, in the finest
possible manner, an artificial population in order to test the effects of different types
of changes in political or planning areas. More often interested in stylized facts,
many MAS applications are relatively parsimonious in terms of modeled
mechanisms (such is the case of Schelling's model mentioned in section 4.1). Others
incorporate several dimensions but the behaviors of the agents tend to be governed
by cognitive principles rather than by statistics. The initial situation of a
microsimulation more often corresponds to a representative population of the target
population, so that the reasoning in terms of “what if?” allows to produce relatively
“safe” measurements of the consequences of a change of a political nature. For its
part, a MAS model more often allows to identify the functioning of the process
behind a emergence, which often makes it start from an initial neutral situation, an
empty space or a space where the elementary objects are randomly distributed.
Further on, we present three families of computer simulation models:
- random walk models in order to simulate diffusion phenomena;
- CA models with applications to the evolution of land use;
- agent-based models, while distinguishing the applications where the agents
represent simple entities (individuals) or composite entities (settlements or cities).
Search WWH ::




Custom Search