Author: Robert Hyatt
Date: 11:01:18 08/27/03
Go up one level in this thread
On August 26, 2003 at 18:42:59, Ricardo Gibert wrote: >In the Richard Lang interview at: > >http://www.beepworld.de/members37/computerschach/ >(Just click on the link "ChessGenius Classic 7" in the leftmost column. The >interview also appears in *English*) > >"For example I have never used capture searches and rely instead on a static >swap off routine." > >This seems to indicate that CG does not employ a qsearch. I also understand that >Junior does something similar. I wonder how this is done? I would presume some >type of accuracy tradeoff must be involved, but I wonder what? I'm very curious >about how this is all done and why doesn't everybody do it this way? > >How is all the effort that goes into creating a good eval compatible with such a >handling of non-quiescent positions? It just seems kind of wacky to me. I can tell you what I did in the 1970's. The issue is that the _last_ move in the search path is "iffy". IE if you just do a pure 4 ply search, no extensions, no q-search, the last move in your pv will _always_ be a capture of the most valuable piece that can be taken, whether it is protected or not (IE in a worst case, you end the PV with QxP, where in a real game, the opponent would the reply PxQ.) If you apply a SEE type exchange evaluation to the last piece moved, you stop this nonsense. Of course, that misses overloaded pieces and the like, but it is certainly possible that the errors that produces are not frequent enough to cause a problem. The tree is far more robust than I would have thought possible, until some recent testing has produced some amazing results.
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.