The rules get more complex. They have to take into account things like reproduction, death due to overpopulation, etc.
I'd argue that while one
could make the rules more complex, cellular automata are mroe to do with
ultra-simplistic (and deterministic) rules creating highly complex behaviour[1].
Departing from the strict field of cellular automata (because I'm in danger of Doing Your Homework For You for the OP if I put down my best ideas in that vicinity) let me compare it with the related concept of 'emergent behaviour', so you can get an idea of what sort of thing you could think about.
An instruction to a simple agent like a a toy bulldozer to "move forward until you're shoving more than a certain amount of blocks, then turn a bit and try again" just clears some seemingly random pathways, but get a bunch of them together and they quickly clear an area of the 'rubble', even without direct communication (of course they bump into each other, and treat each other as unmovable rubble, so back off and turn a bit and try again). This is an idea that might well be employed by advance-guard unmanned rovers to clear future martian landing sites of rubble in preparation for a manned landing...
Or if you want something more contemporary, consider something like the dynamics of a standing crowd at a concert/sports match. Each individual wants to get to somewhere they can see (or, perhaps at certain times, retreat to the food'n'drink stands or the other facilities) and at the same time avoid being squashed by everyone else trying to get there.
And then (topically, for us over here) consider voting tendencies. Each person has a favourite party and a propensity for voting tactically if for some reason they feel their normal voting pattern isn't going to matter too much (maybe going for the least disliked of the two parties in with a chance if their candidate has none, or changing to a no-chance candidate if theirs is going to anyway, but they want to 'send a message' of some kind). And then consider how many people interacting independantly would vote, then how people acting nomimally independantly but aware of everyone else's tendency ("I better vote for <foo>, after all, because the punitive backlash that I'd normally join might be large enough to unseat them").
Now, I'm safe in the knowledge that none of these relate (directly) to cellular automata, but there are similarities there. CAs are often thought toylike, but I've seen (admitedly purely academic) implementations that have been used similar to artificial neural networks to 'process' grid data and come out with an yes/no answer to some question. e.g. "Is the basic shape of the pattern convex?" A bit of a contrived problem being solved there, but it shows both that it's the "field" of the automata which is complex, not the "physics" it obeys, and that there may be purposes. Hunt around. Or (if you promise not to quote me verbatim) edge your piece over into emergent behaviour or game theory so that you can say a little something about the real-world examples I've already mentioned.
[1] The 'instruction set' should be fairly basic. Obviously the more basic they are the less 'programming states' exist. For the most part, I've only dabbled in them as 'playthings', and a randomly set-up CA is as likely to 'fail over' into a boring behaviour as do something interesting[2]
[2] Usually in Conway-esque systems it's set up so that there's a lower limit to 'survival' (below which the lonely cells 'die'), a slightly higher limit where a cell lights up when it has enough neighbours to be 'born', and an upper limit to when they start to die off because of overcrowding. Although there's far more combinations if you consider non-contiguous zones of survival, e.g. (from top of my head) all cases of an even a number of 'lit' neighbours toggles the state, an odd number leaves alone. That could do interesting things. Or not. Also depends on the 'state of play'.
Author
Topic: Cellular Automata (Read 4871 times)