Page 2 of 2

Re: Draw position bishop+g3 pawn vs g2 pawn

Posted: Fri Nov 28, 2014 9:18 am
by Edmund
Evert wrote:
Edmund wrote: After all this is still a 7 men endgame.
What assumptions would you make to be able to solve this endgame in just a couple of seconds?
Fix all pawns in place (don't move them, they get taken immediately anyway). Mark as won any position where black can play Kxg2. Recursively mark all precursors as won. The given position ends up not being marked as won, hence it's a draw.

Now, in general you need to be careful about considering pawn slices after pawn trade, and you need to consider whether the position after Kxg2 is actually won, but that can be fairly easily verified statically in the current position.
It seems like it takes a human to make these assumption. hgm however suggested that the engine should know itself.

Ad. fixing the pawns: note that black can force the white pawn push .. to see that the resulting position is also drawn you need to look one ply deeper.

Re: Draw position bishop+g3 pawn vs g2 pawn

Posted: Fri Nov 28, 2014 11:16 am
by Evert
Edmund wrote: It seems like it takes a human to make these assumption. hgm however suggested that the engine should know itself.
No, he suggested that retrograde analysis can show that the position is drawn in an instant.
Ad. fixing the pawns: note that black can force the white pawn push .. to see that the resulting position is also drawn you need to look one ply deeper.
Sure, the pawns are not absolutely fixed, but you can consider pawn slices.