Author: Oliver Roese
Date: 07:49:11 10/18/00
Go up one level in this thread
On October 18, 2000 at 10:10:51, Dieter Buerssner wrote: >I am trying to find a relatively fool proof way, to detect, if >an endgame with a rook pawn (or a doubled rook pawn) and the wrong >bishop against a lonely king, is a draw. Ideally I would like an >algorithm, that only needs the loosing king square (perhaps the >winning king square), the pawn square and the side to move, and it >should detect, if this is a draw. > >I tried so far, to only use the pawn square and the losing king >square. When the loosing king is closer to the corner, than the >pawn, it is a draw. When the distance is the same, the side to move >will be better. But this does not allways work. I think, it would be better, to >be certain of the draw, and in compilcated cases, that can not be decided, give >a small positive score to the bishop side. Have you any ideas about a simple >algorithm? Would it be different, when the rook pawn >is doubled? And when the losing side has a pawn, that say is not too advanced? > >One nice example from Tarrasch "Das Schachspiel": > >[D] 8/4k3/8/7P/4B3/5K2/8/8 w - - 0 1 > >White wins with 1.h6 Kf7 2.Bh7 > Yeah, this is a win after 2..Kf6 Kf4 and white breaks through to g7. >Which programs, that have knowledge about this sort of endgame will >show a winning score without search and TBs? I tried this with >Crafty, and it shows a draw score up to depth 6. This is of course >problem in this game, because it only takes a fraction of a second >to reach depth 7, but it might be dangerous, when the position >is reached in the search. > >Regards, >Dieter Once a while i had a look at craftys sourcecode, and if i recall it correctly there is dedicated code to handle this special endgame. It might be possible, that you have found an evaluationbug. I am currently not at home and have no access to the source code to check this out. Attract dr hyatts attention to see thats going on here. Oliver
This page took 0.01 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.