Click to See Complete Forum and Search --> : GO-Playing Game
chris law
April 22nd, 1999, 01:44 PM
I'm planning to write a GO-Playing program. Anybody who'd like to join forces is welcome...
I'm with the chemistry department of the ETH, programming in my spare time (so the approach will be rather slow).
Any suggestions will be welcome...
chrislaw
Michael Decker
April 22nd, 1999, 02:53 PM
What's a GO-Playing Game?
April 22nd, 1999, 03:11 PM
GO is kinda like chess... more complicated to caclulate moves though, it has some crazy number of possible moves.
April 22nd, 1999, 03:28 PM
Go is not really much like chess at all.
It has a larger board with a grid. Two players place their tiles on the intersections,
alternating one player with white the other black. It's a surround and control kind
of game.
It's not as widely known as chess, but is very much liked by those who play it.
The major reason it is less widely computerized is the much larger board. It
means that brute force methods are very slow indeed. It is necesary to put a
lot of careful thought into strategy, both in selecting the move and in selecting
what moves to consider. Otherwise, the search tree becomes impossibly
huge in only two or three moves. Instead of a maximum 16 chess pieces,
there are hundreds of possible places to put a Go tile. So searching to a
depth of 2 moves for each player is already 100s of millions of positions
in a simple brute force approach. And three deep would be trillions. Clearly
some non-brute force method is required.
chris law
April 23rd, 1999, 01:25 AM
Exactly, this is why I want to have a GO at it!
The general idea is to use a neuronal network/rule based system for general strategies and a rule base/algorithmic (brute force) part to solve small region tactical problems...
I also intend to start with a smaller board (GO-beginners start with 9x9 board, then 13x13). The design goal is to build an expandable base for playing go (in the beginning very weak playing...). I may have to mention that I'm not really a GO player, I used to play chess and I took some glimpses at GO, but the computational problem is more fascinating with GO (the chess programs are too good already, speed is what counts. I don't know any really good Go playing program...)
chrislaw
codeguru.com
Copyright Internet.com Inc., All Rights Reserved.