Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / The operational definition of the endgame for programmers?
- - By rocket (****) [se] Date 2018-12-02 16:55
Elite level players differ on when the endgame phase enters a given game. Some say queens off the board, or only queens, others when queen plus one minor piece for each side is exchanged, and a third group submits that the engame phase enters when the king is supposed to be activated in a given game.

How is the endgame defined by you computer programmers in your engines? Queens off?
Parent - By rocket (****) [se] Date 2018-12-02 16:59 Edited 2018-12-02 17:02
How about my definition? "The endgame starts when the kings, minor pieces and pawns engage without any positional conciderations and kingside attacks"

If an engine performs badly in a game with queens off by simply weakening pawn structures, it's clearly  subpar middlegame evaluations. I think when positional conciderations are off and only long term strategy/calculation is in effect, and the king takes an active role, then the endgame is said to have begun.
Parent - - By Lazy Frank (****) [gb] Date 2018-12-02 18:30
Queens off can be middle game.
I think (strongly subjective) endgame phase starts when king role in the game increases (attack/defend some point, force opponent to exact play).
Parent - - By rocket (****) [se] Date 2018-12-02 19:49
That definition is inadequate. See Nigel Shorts famous king walk mate .

http://www.chessgames.com/perl/chessgame?gid=1124533
Parent - - By Lazy Frank (****) [gb] Date 2018-12-02 20:35
Nigel game ...
King walk exactly is possible because all pieces is locked (paired) to each other and king create mate threat.
Parent - - By rocket (****) [se] Date 2018-12-02 21:36
So your definition fails. That was not an endgame.
Parent - - By Lazy Frank (****) [gb] Date 2018-12-02 22:00
Ok ... there is always exceptions (to any definition). Good luck!
Parent - - By rocket (****) [se] Date 2018-12-03 12:58
Not mine. :lol:
Parent - By rocket (****) [se] Date 2018-12-04 15:30
Silman is not a programmer, and players differ when the endgame is said to start.
Parent - By Vegan (****) [ca] Date 2018-12-08 15:22
stages in a game are relative

EGTB kick in with 6 pieces
Parent - By h.g.muller (****) [nl] Date 2018-12-09 12:48
Almost no contemporary chess engine makes any attempt to define when the end-game starts. They just gradually taper the evaluation from opening-like to end-game like, as a function of the total material on the board. Where stronger pieces get more weight in determining the material. E.g. Q=6, R=3, N=B=1 (or Q=4, R=2, N=B=1). When total material is 32 (or 24) they use 100% opening evaluation, when it is 0 they use 100% end-game evaluation. When it is 16 (or 12) they use the 50-50 average, etc.
Parent - By sovi (**) [cz] Date 2018-12-09 16:39 Edited 2018-12-09 16:52
toga-30-ja-release\src\search.cpp  (dead code)
// analyze game for evaluation
  
   /*if (SearchCurrent[ThreadId]->board->piece_size[White] < 3
     && SearchCurrent[ThreadId]->board->piece_size[Black] < 3
     && (SearchCurrent[ThreadId]->board->pawn_size[White]+SearchCurrent[ThreadId]->board->pawn_size[Black]) > 0){
     trans_endgame = true;
   }
   else{
     trans_endgame = false;
   } */
Up Topic The Rybka Lounge / Computer Chess / The operational definition of the endgame for programmers?

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill