You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
FrankWalter 2.3.5 will play the wrong move in this situation: 8/8/8/8/3K4/6r1/4k1p1/6R1 b - - 7 82 when it has access to Win-Draw-Loss tablebase files, but not the Distance-to-Zero files because it relies on these DTZ files too much.
This also causes weird looking moves in the endgame where the winning distance to zero move is giving pieces away.
Solution is to select all moves (taking the 50 move rule into account) that get the maximum result (win or perhaps draw) in a root position and only consider these in the search. Make sure that the WDL information is not terminating the search on ply one.
The text was updated successfully, but these errors were encountered:
As a work-around: please make sure you have the corresponding .rtbz files for any .rtbw syzygy file you have. If you do not have them, it might be better to not provide syzygy tablebases to FrankWalter.
FrankWalter 2.3.5 will play the wrong move in this situation:
8/8/8/8/3K4/6r1/4k1p1/6R1 b - - 7 82
when it has access to Win-Draw-Loss tablebase files, but not the Distance-to-Zero files because it relies on these DTZ files too much.This also causes weird looking moves in the endgame where the winning distance to zero move is giving pieces away.
Solution is to select all moves (taking the 50 move rule into account) that get the maximum result (win or perhaps draw) in a root position and only consider these in the search. Make sure that the WDL information is not terminating the search on ply one.
The text was updated successfully, but these errors were encountered: