Author: blass uri
Date: 12:49:42 05/18/00
Go up one level in this thread
On May 18, 2000 at 14:30:33, Ernst A. Heinz wrote: >Hi Blass, > >>I agree that this is not a bug because the decision to let Bxa7 was a design >>decision but it is not clear that this decision was correct. > >As I said, I very much prefer the implementation of _general >principles_ rather than the addition of ever more code for >the handling of special cases. IMO, this is the correct >design decision. > >But it does not make much sense to discuss the "trapped >Bishop" phenomenon in more detail, unless we have some hard >facts to support either point of view. > >>This is a position from the game Junior5.4-Crafty16.6 >> >>[D]r2nkb1r/pp1n2pp/8/1P1pp3/8/2P1B3/1P3PPP/RN2KB1R w KQkq - 0 1 >> >>Junior played the move 17.Rxa7 and was in trouble after Rxa7 18.Bxa7 b6 >>Junior saved a draw because of the fact that crafty did not use tablebases so >>Junior could escape to a KR vs KRN endgame. >> >>I do not think that Rxa7 is a good move >>The game was played with p200 under Junior5 interface(engine-engine game). >> >>Can your program avoid Rxa7? > >Yes, "DarkThought WCCC'99" locks onto Be2 in iteration #9 after >less than a second. > >=Ernst= I see that dark thought knows some things that most fast searchers do not know. In the example of avoiding Rxa7 it knows that the fact that the bishop at a7 cannot move is more important than a pawn. I guess that in the previous example the reason that it liked Bxa7 is the king safety propblem of black. I know that your program liked Nxe6 in the famous position from Rebel's homepage and that it also could find Nxe6 in another position from an old game of chess programs from 1974(In this case it was clear that Nxe6 was the right move but other programs could not find it) I do not know if you do something right about king safety or if your king safety is too big and it can help to solve some test positions but can cause also some wrong sacrifices in games Here is a test position from the game Yudasin-Junior5 when black almost went wrong by playing Bxh2+ [D]r1b1k2r/1pqp1ppp/p1Nbp3/8/4P3/2N5/PPP2PPP/R1BQR1K1 b kq - 0 1 Can your program avoid Bxh2+? Uri
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.