Author: Roger D Davis
Date: 21:47:53 06/23/05
Go up one level in this thread
On June 23, 2005 at 16:15:40, Steven Edwards wrote: >Symbolic: Status report 2005.06.23 > >A first version of a tree/node diagnostic HTML output generator has benn >completed. An example output has been posted at: > > http://www.geocities.com/chessnotation/Tree.html > >The above can also be reached from a link on the main page at: > > http://www.geocities.com/chessnotation/ > >Currently, the nodes are output in the order in which they were created, >although this may change. Each output is reported with internal hyperlinks to >various neighbor nodes including all descendent nodes. > >-------- > >I'm now working on the knowledge sequencers for the actual cognitive search. >These are the ones that use a plan to create a portion of the position search >tree and return an analysis. My short term goal is to get enough of this >working to demonstrate the entire plan formation and plan execution process in >an end-to-end fashion. This effort has already resulted in locating a few >undetected bugs where faulty code just happened to work under less then complete >testing. Also, the newly implemented sequencers are getting narration support. > >Related sequencer status: > >CandidateSelector: Somewhat finished, will need more work when progressive >deepening is further developed. > >Selector: Basically finished. > >QuickSelector: Finished, as are the five or so subselector sequencers it calls. > >Surveyor: Only a minimum functionality that enables MateAttack/AvoidMate theme >productions. > >Planner: Basically finished, but much, much more work needed in the actual >planning subfunctions. > >ExpPlan: Basically finished (this does further plan elaboration and minimaxes >the analyses returned by the ExpPlayMove sequencer). > >ExpPlayMove: About halfway there; already does move execute/retract, plan >advance/retreat, and decent narration. Does not yet call ExpPlan as needed for >recursion. > >ExpPlanTerm: Needs much work; it has to decide when the current position needs >either quiescence processing or more planning via a recursive call to the >selector sequencer. > >Quiescence: Not yet implemented; will probably have a very simple capture search >initially. > >-------- > >The total Lisp source including the GA subproject in now 8,124 lines long (ca. >226 KBytes). > >-------- > >More to come. Steven, Sounds like you're getting a lot closer. Will be fascinating to watch Symbolic play it's first game with the planners in place. Can't wait. :) 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.