Author: Omid David
Date: 14:52:10 10/04/02
Go up one level in this thread
On October 04, 2002 at 16:42:52, Jeremiah Penery wrote: >On October 04, 2002 at 06:26:13, Omid David wrote: > >>That's right, if a general "goal-oriented" blockade exists, my algorithms with >>detect it (without any risk of mistakes). However its being an algorithm rather >>than a simple static evaluation, results in its being too expensive to conduct >>at all nodes. However, I believe a wise distribution of it along the search >>tree, will make it profitable. > >I'm curious to see what your program says about this position: > >[D]2k5/2p5/1q1p4/pPpPp1pp/N1P1Pp2/P4PbP/KQ4P1/8 w - - 0 1 > >My stuff completely fails here now, wanting to play Nxb6 which supposedly leads >to a draw. Mine also doesn't detect the draw and naturally chooses Nxb6. I think the reason is that it doesn't reach a clear-cut conclusion about the position arising after: 1.Nxb6 cxb6 2.Qc3 followed by 3.Kb3, 4.Ka4, and 5.Qxa5 However after playing Nxb6, it won't step into the above mentioned variation, since on every node the blockade-detection algorithm will say "I don't know", and so the decision will be made by the regular search algorithms.
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.