Information Technology Reference
In-Depth Information
3.5. Definition of Neighbourhoods
The neighbourhoods
c
(i,t)
.
H
and
c
(i,t)
.
V
are in general the result of queries executed on
sets of cells, using operators and predefined functions (logical, mathematical,
geographical, etc.). Hence, formally there is no difference between sets
c
(i,t)
.
H
and
c
(i,t)
.
V
and their generating queries. “Cross-references” in queries are contemplated:
references to the state
c
(
i,t
)
.
s
and to parameters
c
(
i,t
)
.
P
of current cell, to layer's
parameters
L
(i,t)
.
P
L
and to global parameters
P
G
, as well as to the target cells' states.
For the way of cell's neighbourhood formulation, the concept of proximity does
not necessarily have a topological meaning: the criteria of Euclidean distance between
cells and predefined patterns (Von Neumann, Moore, Margolus, etc.) are comprised
as special cases of generic queries. Furthermore, the neighbourhood does not remain
constant during the simulation but can evolve since a query can be reiterated with a
defined frequency or can be activated under certain circumstances and configuration
of the scenario.
3.6. Rules of Evolution
In order the cellular automata environment be truly general, a generic transferring
function σ
k
,the transition function φ, as well as eventual functions
f
(
t
) of variation of
parameters regulated by dynamics esternal to the CA, should possibly be described
with an adequate programming language. Such approach will potentially pose no limit
to the generalisation of rules governing the CA.
The σ
k
is implemented in form of a generic analytical function, as one of the
classical aggregation and dis-aggregation available if GIS-based environments (sum,
mean, probabilistic distribution, etc.).
It is important to mention that, in general, the φ is a rule with a modifiable form
during the simulation: the flow-control instructions (for instance IF <
condition
>
THEN
<instruction>
)
permit in fact to subordinate the execution of parts of the rule
to specific conditions of the state of the system. Rules can also be codified in
probabilistic terms, allowing the possibility to design a non-deterministic automaton.
Furthermore, if a parameter is given the role of a counter, the system have the
possibility to establish different frequency of application of transition rules for each
layer and of transferring function between linked layers.
The mechanism of information transferring function between layers (aggregation -
dis-aggregation), a potentially manifold spatial extension of cells belonging to
different layers, as well as the possibility of inclusion of global variable parameters in
transition rules, permit the modelling of “global events” having effects and influences
on local level. This principle constitute a necessary derogation of the principle of
“locality” embodied in classical cellular automata.
3.7. Some Remarks about the Architecture of the System
In order to satisfy all the functional requests described above, it is necessary to design
the general architecture of the software system as modular as possible.
