Home * Evaluation * Game Phases * Endgame * Wrong color Bishop and Rook Pawn

Wrong color Bishop and Rook Pawn is one of typical drawn fortress positions that need to be encoded in the evaluation function, as search would not evaluate them properly. When the stronger side has a pawn on the "a" or "h" file and the bishop can never cover the promotion square because of its "wrong" square color, then the position is drawn when the defending king stands on the promotion square or controls it. The stronger side cannot get more than a stalemate.

In order to detect possibilities of exchanging into such an ending, a code should say something to the effect: if all the features described above are present and the side without a bishop has a couple of pawns more, but nominally still is at a disadvantage, call it a draw.

There is also a position drawn despite having a nominally good bishop: white pawn on a2, black pawn on a3, white king on b1 or c2, and a dark-squared bishop for Black. This works even if all the pieces are moved one file to the right, as there is still no way to outflank white ^[1].

Too far

A position where White to move wins:

8/8/5k2/7P/4B3/5K2/8/8 w - - 0 1

Uri's rule

Uri Blass proposed a rule ^[2] applying Chebyshev distance to the promotion square, considering Tempo, implementation by Gerd Isenberg:

dl = distance(lonesomeKing, promoSquare) + (side2move != loneSome); /* considers tempo */
dw = distance(King, promoSquare);
dp = min (distance(MostAdvancedPawn, promoSquare), 5); /* considers double push */
dp += File(lonesomeKing) == File(MostAdvancedPawn); /* makes dl < dp always true in case of blocked pawn */
if (dl < dw && dl < dp ) draw = true;

Forum Posts

2000 ...

• King, rook pawn and wrong bishop endgames by Dieter Bürssner, CCC, October 18, 2000
• Draw recognition by eval problems by Rafael B. Andrist, CCC, October 17, 2001
• "wrong" bishop + rook pawn - a test position by Jon Dart, CCC, May 31, 2003
• Wrong Colored B and Rook Pawn Positions for Engines (and Humans?) by Dieter Bürssner, CCC, March 04, 2004

2010 ...

• Chess for Android: double blunder between Komodo & Stock by Max May, CCC, May 20, 2011 » Komodo, Stockfish
• Komodo doesn't know the rule "corner promotion+wrong B" by Vincent Lejeune, CCC, October 09, 2013 » Komodo ^[3]
• Discocheck 5.01: Bishop related endgame problems by Mike Scheidl, CCC, November 25, 2013 » DiscoCheck, Color of a Square
• A|H pawn && wrong bishop v K by Colin Jenkins, CCC, September 21, 2015
• Stockfish eval output by Erin Dame, CCC, August 27, 2016 » Stockfish

External Links

• Wrong rook pawn from Wikipedia
• Wrong bishop from Wikipedia
• The wrong bishop by Frederic Friedel, ChessBase News, May 16, 2016
• The wrong bishop – part two by Frederic Friedel, ChessBase News, May 22, 2016 » Cray Blitz, Mikhail Botvinnik , WCCC 1983

References

1. ^ Implementation note: In order for this heuristic to work, program must be able to recognize position as a draw even after the stronger side sacrifices a bishop
2. ^ Re: King, rook pawn and wrong bishop endgames by Uri Blass, CCC, October 18, 2000
3. ^ see game 17.1 in TCEC Season 5 Superfinal, Stockfish vs. Komodo, November 25, 2013, where Komodo had the knowledge, but Stockfish apparently not

What links here?

Up one Level