Author: Anthony Cozzie
Date: 17:15:28 10/04/03
Go up one level in this thread
On October 04, 2003 at 18:30:05, Dan Andersson wrote: > I'm currently trying out Lambda Search and Generalized Threats Search. Before >that it was PN,PN^2 and PSS. Combining, mixing and matching ideas. Too many >ideas, too little time... >PSS: http://www.t-t.dk/go/cg2000/index.html >Lambda Search: http://www.t-t.dk/go/cg2000/index.html >GTS: http://www.ai.univ-paris8.fr/~cazenave/ > >MvH Dan Andersson I took a look at some of those references. I may be making some mistakes as I did not completely analyze the papers. Lamba search seems to be uninteresting for chess (except for mate solvers) as it assumes 0/1 (white can mate/white can't mate). PSS (did you notice that that Muller guy spent **12** years as a Ph.D or postdoc) seems to have very limited applications to chess. Go subdivides spatially quite nicely, chess does not: a Bishop on B1 has an effect on the King at G8, etc. Plus I admit I got a little lost as every other word was "combinatorics". It seems to me that Alpha-beta (and variants) work fine for chess search. Most of the work left more engineering than science: figure out good evaluation and extension/pruning methods. Go, being a much more complex game, is still in the "science" stage. Alpha-beta is prohibitively expensive there. anthony
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.