Geoscience Reference
In-Depth Information
(a)
(b)
FIGURE 9.6
The use of relationships for defining the neighbourhood of buildings based on Voronoi parti-
tioning showing (a) the spatial view and (b) the relationship table.
1. How does an agent account for empty cells within the neighbourhood and for itself?
2. How does an agent located within an unfriendly neighbourhood behave in cases where all
vacant places are even worse?
3. Does an agent know about the state of all or only sufficiently close cells?
4. Does the choice of cell to relocate to depend on the distance between cells?
5. Does an agent move into a cell if the fraction of friends within the neighbourhood of the
vacant cell is the same as or higher than the current neighbourhood but still lower than F?
6. Are random (independent of the state of the neighbourhood) migrations possible?
7. How does an agent choose one of several vacant cells, if the number of friends is sufficiently
high for each of them? Is an agent a
satisficer
or
maximiser
(Simon, 1969)?
8. Is information about an agent's move immediately available to the other agents?
9. Does an agent react to vacant cells only? Do agents swap their locations?
These and many other factors such as the form of the city or the ratio of the numbers in each group
may influence Schelling model dynamics. The study of the model (Benenson and Hatna, 2011;
Hatna and Benenson, 2012) shows that externally minor variations in conceptual and technical
aspects of the behavioural rules can essentially influence Schelling model dynamics. Surprisingly,
the majority of variations of the behavioural rules of Schelling have not yet been investigated.
9.3.3 t
eMPoral
and
S
Patial
r
eSolution
and
the
o
rder
of
u
Pdating
An abstract view of residential behaviour makes the choice of spatial and temporal resolution for the
Schelling model relatively simple. Spatially, a cell symbolises a one-family building; temporally, it
is convenient to consider that the residential decision is done by the householder, most frequently,
once every few months. With regard to the neighbourhood size, it makes sense to consider the
reactions of the agent to the neighbours within 3 × 3, 5 × 5 or 7 × 7 neighbourhoods, but the modeller
should stop at some reasonable size of the neighbourhood: Schelling's idea is that a householder
reacts to the fraction of friends among immediate neighbours. Note that a major parameter in the
Schelling model, that is, the desirable fraction of friends F, does not depend on the time step of
the model. The temporal resolution of the model is thus only important for establishing the rate of
occasional migration if the latter is included in the model. Assuming that urban residents migrate,
