Author: Roger D Davis
Date: 16:28:54 06/04/04
Go up one level in this thread
On June 03, 2004 at 17:49:22, Steven Edwards wrote: >Symbolic: Status report 2004.06.03 > >I've added a new chess type to the ChessLisp interpreter: the Window type. It's >a pair of score types, alpha and beta, stored in a single Lisp atom. While not >strictly necessary, the Window type makes the Lisp source a bit cleaner and very >slightly faster by avoiding construction and access of a two element list in >many places throughout Symbolic. > >The interface to Symbolic's knowledge sources and been reorganized. Now, all >KSes take exactly two parameters: a common parameter list and a specific >parameter list. The common list has its own family of accessing functions and >includes values for the current node, effort, window, and calling reason string. > The specific parameter list contains values specific to a particular KS and is >often nil. > >I've also done some work on testing construction of large (GByte+) opening books >which in turn tests the book selector code in Symbolic. It is possible that >some of the techniques used in this may also serve for the construction and >access of an extended pattern library. > >Near term future effort will be directed towards Idea formation (by KsSurveyor) >and Plan formation (by KsPlanner). It will be some time until the work is >complete enough for end to end testing. I've always thought that opening books would provide an excellent resource by which to development an engine. Sometimes, many years go by before a particular move is refuted, and then a line that was formerly popular sinks into obscurity. What if a set of several hundred such lines could be gathered together. Then keep tweaking symbolic until the planner tends to find the moves the refute the line. Would be really cool if Symbolic could somehow find moves that the best humans missed for years, until one person finally got it. Roger
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.