Man, this one is a doosey...

It combines:

Cybernetics, AI (Since the problem alone is NP, it becomes an AI problem). Game theory, Path finding and network programming.

If I had three months to research this and get myself a working develment platform with a test server I would take a whack at it but the five-day time period for this problem is just insane.

If you spend a few minutes to read the task description you will see that the problem is not *that* difficult: http://icfpcontest.cse.ogi.edu/task.html

Andreas

On Sat, 31 Aug 2002, Alan Grimes wrote:

...

If you spend a few minutes to read the task description you will see that the problem is not *that* difficult: http://icfpcontest.cse.ogi.edu/task.html

Why do you think they feel the need to specify a 64 mb ram limit? or a 1sec at 1.5ghz...

Don't those numbers seem awfuly large for a simple program?

No.

Those are reasonable *limits*.

Plus, that's runtime consumption, not ANYTHING to do with code size. It really is just a simple game (with a few fun twists) thus some sort of search will be necessary. Given a possible 1000 sqare playing field it's possible to generate fairly deep searches. But the algo for performing the search could be quite straight forward (minimax, for example).

And remember, to qualify, you just need to play the game, not win it. how well you do depends on how well the *other* folks do, too. So the 72 hr (not 5 day) limit is quite reasonable.

See chapter 18 of Peter Norvig's *Paradigms of AI Programming* (the Othello game). That's a pretty good starting point, actually.

(Nicely, he takes you through increasingly sophisticated strategies.)

(No cybernetics either; it's just a board game.)

Cheers, Bijan Parsia.