home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.theory.cell-automata
- Path: sparky!uunet!haven.umd.edu!darwin.sura.net!gatech!bloom-beacon!INTERNET!dont-send-mail-to-path-lines
- From: rudy@autodesk.COM (Rudy Rucker)
- Subject: Go Bugs
- Message-ID: <9208161929.AA06694@air.acad>
- Sender: daemon@athena.mit.edu (Mr Background)
- Organization: The Internet
- Distribution: inet
- Date: Sun, 16 Aug 1992 19:29:44 GMT
- Lines: 37
-
-
- Antti Karttunen sends
-
- I recently downloaded Core Wars like game from Japanese ftp-site
- called 'Jintori', which seems to be somewhat influenced by Go. It has
- two bugs competing in two-dimensional space (instead of the traditional
- one-dimensional Core Wars space), and the winner bug is the one which have
- conquered more area in the end (by surrounding it). I wonder if this could
- be made still more Go-like so that bugs' existence could be determined by
- their liberties, count of eyes, and so on. In the end we maybe could
- have the Go-playing program where the stones 'themselves' decide how to
- play. Or even 'artificial life' system (See for example the July 1992
- Scientific American article about the Genetic Algorithms) where the better
- and better Go-playing algorithms would evolve.
-
- What an interesting idea! A couple of years ago I started simiulated
- 2D Turing machine programming. The first ones I did were "Langton
- vants" which I ran in a CA. Then I started doing free-ranging
- linked list vants with many states which were available as a
- shareware program called ant21. These guys have lookup tables which
- are bred with genetic algorithms. The fitness functions I've used
- at present have been based on splitting the bugs up into colonies,
- and having them leave evanescent trails; the bugs get points by
- running across a prey colony's trails and lose points be encountering
- pred colony trails. But Karttunen's rap suggests I might also keep
- track of whenever a given bug's trail encloses a region and add this
- region's pixel area to the bug's score. What it is to properly "enclose"
- a region is defined by the rules of Go. The difference between Go Bugs
- and a proper Go player is that each Go Bug plays its next stone within
- some (preset but preadjustable) radius. Real Go Players often keep
- things happening at quite a few board location, they maintain multiple
- Go Bugs. Well, th en a Go Player could be a colony of Go Bugs.
-
- A 2D Turing machine is like a CA cell that has come untethered. It
- reads its neighbor cells and changes its state. But it also can
- move, though when it moves, it paints the "aether" under it so that
- its computation-record remains.
-