# Sharp pawn endgame - can it be calculated to the end? With Computer engine?

White to move, here is a brief summary of my analysis where I attempted to determine exact evaluation. Number of pawns is equal but White's 2 pawns on the queenside are held by one Black's pawn, so in a sense Black has an extra pawn on the rest of the board.
`1. a4` with pawn breaks on both sides of the board by both White and Black (!) leads to a difficult queen endgame for White.
`1. Kc3` (played by me in the game) is a blunder that should lose after 1...f4! since then Black puts pawns on e4 and f4 and goes to the kingside with his king to pickup White's pawns.
`1. Kd3` is possibly best as now `1...f4` is met with `2.Ke4`. But Black instead plays `1...Kf7` and manoeuvres, waiting for zugzwang. Can Black win? Can a computer engine be used in this day and age to determine exactly if this position is a draw or win for Black?

This is a pawn endgame - which are supposed to be evaluated precisely... I feel this is likely a draw, but can't prove exactly. Is there Mr. Grigoriev in the audience?

``````8/7p/4k3/1p2ppP1/1P5P/8/P1K5/8 w - - 0 38
``````
-
This endgame can be calculated to the end with an engine if tablebases are provided as well. White can force a draw in this position. If you are interested I could post an answer. Best regards. – AlwaysLearningNewStuff Mar 1 '14 at 20:15

It's clear that Black has the following advantages:

1) His king is better centralized.

2) He has two connected passed pawns.

3) White has no passed pawns because one Black pawn holds up two White ones on both the queen and king sides.

White's main advantage is that he has two rook pawns, which could lead to "outside" passed pawns. Even so, it might take the sacrifice of a pawn for White to get a passed pawn.

So I would give Black the advantage overall, and say that he should win with proper play. White should draw with great difficulty if at all.

I'm basing this judgment on "pattern recognition," because I am a only a 1500 player.

-
White can potentially create two passed pawns - if he plays h5 and g6, or if he plays a4. In most lines - only one pawn needs to queen. – Joe Nov 3 '12 at 23:03

This is actually a drawn position with correct play. As expected, the two sides with pawns are actually more important in the end-game than the center pawns (because the center pawns can be effectively barricaded.) Here is a full analysis done GNU chess (your move kd3 is correct, however, the difficult moves come later on in the game when there are two queens on the board.)

GNU Chess comes up with this:

``````[fen "8/7p/4k3/1p2ppP1/1P5P/8/P1K5/8 w - - 0 38"]

1. Kd3 f4 2. h5 f3 3. Ke3 e4 4. g6 hxg6 5. h6 Kf7 6. a4 bxa4 7. b5 a3 8. b6 a2 9. b7 a1=Q 10. b8=Q Qe1+ 11. Kf4 Qh4+ 12. Ke3 g5 13. h7 Qxh7 14. Qb7+ Kg8 15. Qc8+ Kf7 16. Qd7+ Kg6 17. Qe6+ Kh5 18. Qh3+ Kg6 19. Qe6+ Kh5 20. Qh3+ Kg6 21. Qe6+ 1/2-1/2
``````
-
awesome, is there a way to get this as pgn? Another question is - did you just plug it into this engine and that's what it came up with? How much time per move was it given? The big dilemma I have is how much to trust engines and which ones ... – Joe Nov 11 '12 at 16:22
@Joe: This is GNU chess, and although it has never been given an official rating, according to the FAQ gnu.org/software/chess/chess_faq.html#B.2 it was rated ~2200 over 10 years ago. On the advent of modern PCs, it has a much higher rating. I've seen it playing on FICs games with a rating over 2800. The time thinking per move is actually not a parameter in GNU chess, but the depth of search was set to the default (which is 20 I think.) It has horizon expansion I think so that is a soft limit. – ldog Nov 13 '12 at 7:07
@ldog: I agree that this position looks like a draw after 1 Kd3 e4+. But the line given by GNU Chess looks suicidal for White! Instead of 2...f3, why doesn't 2...Kf5 win quickly? If 3 g6 hxg6 4 h6 Kf6 wins. – RoadWarrior Sep 16 '13 at 23:35