Closed viktpal closed 1 year ago
Thanks for this! This is remarkably the first counterexample to the soundness of the tool.
I didn't have pawn attacks into account when identifying visitors to the king's region. Here is a fix: https://github.com/miguel-ambrona/D3-Chess/pull/42
I'll merge it in about 24 h in case you want to have a look before that.
It seems to be working great now. Similar example:
8/1p3b2/pPp1p1p1/BkP1P1P1/1P6/BPB5/PKP5/8 b - - white
Great, thanks!
The semi-static algorithm is unsound because it doesn't take into account possible pawn checkmate from outside the king's region. Example:
1b3kBR/4pP1P/1p1pP2P/1P1P4/8/K5p1/6P1/1B6 b - - white
The output of the algorithm is “unwinnable”, which is not true.