Geoscience Reference
In-Depth Information
- the migration of one household member has a high probability to cause the
same for the other members;
- the migration of a household makes a house available on the real estate market,
etc.
In the CA models, the interactions are made in the form of influence, following a
logic of spatial diffusion of a phenomenon. The MAS models present a very wide
variety of applications:
- in the case of mobile entities, the interaction is reflected by a meeting in the
space, which can result in the transmission of information and can be used to
formalize a diffusion mechanism. Otherwise, it can be reflected by a competition to
occupy a given space and/or exploit resources;
- in the case of fixed entities, the interactions are expressed in the form of
message exchanges;
- the interactions can also be indirect, where the choice of an agent to occupy a
place (a housing or an agricultural parcel) creates a change in context for another
agent (in the same way as seen in the examples mentioned above with regard to
microsimulation).
Another difference between these families of models lies in their objective and
also in the relation to empirical reality. In the case of microsimulation, the
operational dimension of the model is most often favored and the aim is then to test
the effects of different scenarios. The approach is of the “what if?” type: how will
the population be redistributed if the price of oil increases? In the case of climate
change of such magnitude? In the case of the arrival of a significant flow of
refugees? If a new shopping center is created at such a place? In these models, the
transitions of individuals from one state to another are most often governed
according to statistical principles, based on the covariation of demographic and
socio-economic variables. Most MAS are instead developed in a perspective of
supporting reflection and the interest is very specific to the
mechanisms
at the origin
of change. The purpose is to bring forward the emergence of structures or changes to
structures and understand the underlying logics. The practice is more often to
compare the consequences of different hypotheses of the processes in play on a
system's evolution than to study the effect of exogenous events, even if this is
entirely possible from a theoretical point of view
15
, once the model is stabilized and
valid. However, microsimulation and MAS offer different scientific contexts to
explore the effect of exogenous events; in both cases, the model is manipulating the
evolution of an artificial population. In the first case, this population is described in
15 Let us point out that in a number of cases, the differences are more likely a matter of
practices and habits than of constraints imposed by the formalism.
