Author: allan johnson
Date: 22:04:06 09/09/02
Go up one level in this thread
On September 09, 2002 at 08:19:04, Gerd Isenberg wrote:
>On September 08, 2002 at 13:30:41, Gerd Isenberg wrote:
>
>>To evaluate this position as "won" for white a few pattern matcher are necessary
>>(bitboards required):
>>
>>1. Endgame with opposite colored bishops ("weaker" side may have an additional
>>knight, "stronger" side has passers).
>>
>>2. "Stronger" King with only one friendly bishop and (passed) pawns stands patt
>>("Stronger" side is noted as black now).
>>
>>3. Black has no capture.
>>
>>4. WB controlles adjacent king square exclusively, there is no other defender
>>rather than the BK (b7).
>>
>>5. All black pawn move targtes to WB-colored squares are controlled by WB on one
>>diagonal but not defended at all (no former duo).
>>
>>6. All other pawn moves to BB-colored squares (here c5 only) are blocked (but
>>not by king) or match condition 5 afterwards (the target-targets are controlled
>>by WB but one the same diagonal and not defended att all).
>>
>>7. No pawn move leaves an escape square for the patt standing BK.
>>
>>8. WK is not (exclusively) involved to trap opposites king and free to walk
>>around.
>>
>>9. All adjacent BK-squares are either occupied by black pawns or controlled by
>>white pieces, so the King is stalemated forever.
>>
>>If all these conditions match, it's time to give a bonus at least results in a
>>draw score:
>>
>> max(-materialbalance(white), 0)
>>
>>... and to disable nullmove for black to move, if not already done in these kind
>>of positions.
>>
>>Now we are looking for a target square for the WK, so that the WB may mate
>>protected by his king. With the help of pattern 3 we find the b8,c8 target set
>>here.
>>
>>If the square set is reachable by WK detouring adjacent to most advanced black
>>passer which stop is controlled by WB, it's a won position.
>>
>>As long as the shortest king way to the target square(s) interrups the bishop
>>ray to black pawn targtes, we force the king toward most advanced black passer
>>(h2). Otherwise we gave a larger bonus the shorter the distance to c8/b8 is.
>>
>>OK, not yet implemented. Worth to try?
>>
>>Gerd
>
>[D]B7/p1p5/kb1p4/p3p3/P1N2p2/6p1/7p/K7 w - - 0 1
>
>oups, Statement 5 was not exact enough:
>
>The potential mate square must also be member of the common diagonal that covers
>all WB-colored black pawn move targets, otherwise black king may escape, if WB
>captures a pushed black pawn.
>
>I hope this should work better:
>
>4. WB controlles adjacent king square exclusively, there is no other defender
>rather than the BK (b7). This square determine a diagonal D (a8h1) for the WB.
>
>5. All black pawn move targtes to WB-colored squares are controlled by WB on the
>same diagonal D, but not defended at all (no former duo).
>
>
>How to implement "If the square set is reachable by WK detouring adjacent to
>most advanced black passer which stop is controlled by WB, it's a won position":
>
>A set of squares is generated, not attacked by black and not occupied by white
>pieces:
>
>0 1 1 1 1 1 1 1
>0 0 0 1 1 1 1 1
>0 0 1 0 1 1 1 1
>0 0 0 1 0 1 1 1
>0 0 0 0 1 0 1 1
>1 1 1 1 0 1 0 1
>1 1 1 1 1 0 1 0
>1 1 1 1 1 1 0 1
>
>Next we exclude all the squares belonging to the diagonal D,
>but not the squares adjacent to most advanced black passer:
>
>0 1 1 1 1 1 1 1
>0 0 0 1 1 1 1 1
>0 0 0 0 1 1 1 1
>0 0 0 0 0 1 1 1
>0 0 0 0 0 0 1 1
>1 1 1 1 0 0 0 1
>1 1 1 1 1 0 1 0
>1 1 1 1 1 1 0 1
>
>Now we look whether there is a '1'-connection between WK and one Target square!
>This may done by a recursive backtracking algorithm. A similar algorithm is
>needed in (blocked) pawn-endings, to look whether a king may break through
>enemies area via a bottleneck square.
>
>To decide whether WK has already passed the bottleneck and may use the
>"shortest" way to his target, we simply split the above bitboard into two sets
>by the Diagonal D (and with pecomputed "above/below" diagonal bitboard[30] for
>each possible diagonal).
>
>Only one testbit-call is necessary to determine whther the king is in the set
>including the Target square.
>
>cheers,
>Gerd
Gerd Could all of what you recommended actually be written into a programme
and work? If it can why don't programmers do so?
al
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.