Author: Steven Edwards
Date: 12:09:12 07/26/04
Go up one level in this thread
On July 23, 2004 at 03:52:04, Daniel Clausen wrote: >Hey Steven > >Could you perhaps shed some light (again) on the overall structure of your baby? >Especially the difference between the toolkit and Symbolic and how they interact >together. The toolkit is written entirely in portable C++. It consists of a big bunch of classes that are organized into chess and non-chess funcitions. It also includes a ChessLisp interpreter that uses the rest of the toolkit to run Symbolic' Lisp code. >When I hear "toolkit" in this context, I'm thinking about some kind of framework >and/or support libraries. In your last (or secondlast) post you mention though, >that this toolkit beat Crafty along the way. I was a bit speechless after that >post! :) [my support classes convert moves to SAN-notation, they don't beat >Crafty! :D] The toolkit has a number of classes that inherit from its CTSearch class, and one of these is CTPVSearch which is a mulithreaded bag-o-tricks A/B search that is used for testing and also for tempororily filling gaps in Symbolic's Lisp encoded chess knowledge. It's the CTCogSearch class that actually creates and runs a copy of the ChessLisp interpreter to pick a move. >Most people are happy when their creation is near Craftys level, and you have a >"toolkit" which beats it? What will Symbolic do one day!? Shred its opponent to >pieces in 15 moves? :) Well, the toolkit does not yet achieve an overall even result against Crafty, although this may not be too far off. But that's unimportant; the goal here is to get the Lisp part of Symbolic to play decent chess (1800 Elo or USCF Class A) entirely by explicit reasoning using pattern recognition and planning. The road is long.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.