|
|
|
|
|
| (..Declaration Part) | Contents | Index | (..Table Part..) |
Finally we get to the main purpose of a SDL program: the definition of one
(or more than one) local rules
for the CA in form of a rule table.
The local rule
controls the whole course of the simulation. At each discrete time step all
sites of the retina are updated
simultaneously depending on the entries of the
table. (Usually they themselves refer to the states of certain cells in the
neighbourhood.)
(The update is indeed done at the same time in case SCARLET is
running on a computer which has got as many processors as sites. Otherwise
the procedure is replaced with an equal serial one.)
Now one of SCARLET'S advantages compared with an own implementation
becomes visible: because SCARLET uses internal procedures to control
the simulation (e.g., to apply the table to each cell, to create the needed
data types for the retina itself as well as for the results, ...) our original
task is reduced to constructing the rule table(s)
in a rather abstract form.
Nevertheless we have to talk a bit about how such a table will be
interpreted
by SCARLET, at least as far as it affects our proceding.
But before we can do this, it is necessary to introduce a special syntax for assignments to cells of the retina, which is only used in SDL.
|
|
|
|
|
| (..Declaration Part) | Contents | Index | (..Table Part..) |