Author: Robert Hyatt
Date: 10:15:36 01/30/06
Go up one level in this thread
On January 30, 2006 at 12:22:19, David B Weller wrote: >[quote] >But the main point here is that "the horizon effect" is something the computer >search actively "uses" to hide things that are unpleasant. It is not something >that "just happens" because it can't search deep enough. >[/quote] > >Any threat too! If search doesnt like the place it is in, just threaten to take >a piece which must then evade. > >So between 1. checks [already covered by check extension] and 2. captures[ >covered if using recap ext] _AND_ 3. simple threat moves [which it turns out are >so numerous its hard to extend all them without explosion] This horizon effect >is _all over the tree_! > >This is exactly why I think it is important to find a way of identifying when >the search may be _decieving itself' and 'using' this > >When might this phenomena be more or less likely, with reference to, say, eg >alpha/beta bounds or draft or ply? That is far harder. There is a "null-move threat extensions" idea that works here, but at significant cost. When a move fails high, check the null-move search on that same position to see if it fails low (using a lowered window for safety). If it fails low, something is up here as playing move X fails high, standing pat fails low, so probably move X is holding off some serious threat by the opponent. Extend and search again to see if you can now see what is happening with the increased depth. If you check main.c, you'll see I used this for quite a while in very early versions of Crafty... I think I still have the code around (with comments in it) if you'd like to look at it and test...
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.