Tuesday, December 14, 2010

The Elusive Win in Kramnik-Carlsen

Today at the London Chess Classic, Vladimir Kramnik and Magnus Carlsen played a fantastic game. At move 62, Kramnik exchanged down into what he must have thought would be an easily-winning piece-up endgame. At move 65 the players reached the following position:




Carlsen saw a clever drawing idea, however: he would send his king deep into White's position. If White tried to push his own king forward, Black would attack the White pawns from behind. There thus followed:

65...Kd4 66.Kf3 Kd3 67.g4 Kd2 68.Be6 Kd3

Here Kramnik played 69.Kg3 and Carlsen defended marvelously to achieve the draw.

In this post, I want to examine the alternative 69.g5, which I believe was winning.


69.g5! hxg5 70.Kg4 Ke2 71.g3!


This is the key move. If instead 71.Kxg5 Kf2 72.g4 Kg3, White cannot win. For example, 73.Kf5 g6! 74.Kg5 (74.Kxg6 a2! 75.Bxa2 Kxg4) Kf3!

But after 71.g3!, White is winning:

71...Ke3 (71...Kf2 72.Bd5 Ke3 73.Kxg5 transposes) 72.Kxg5

a) 72...Kf2
73.Kf4! (Not 73.g4, which transposes into the drawing line given above.)

73...g5+ 74.Kg4 Kg2 75.Bb3 (or any other square along the a2-g8 diagonal except d5) Kf2 76.Bd5! Ke3 77.Kxg5 and White wins


b) 72...Ke4 73.Kg4 Ke3 (73...Ke5 74.Bb3 Ke4 75.Ba2 Ke5 (75...Ke3 76.Bb1!) 76.Kf3 and White wins, as the king escapes.)

74. Bb3 Ke4 75.Ba2 Ke3 76.Bb1! (The key move in this line! The bishop needs to leave the a2-g8 diagonal to cut off the e4-square while still covering a2. This tablebase move was mentioned out at chessgames.com by "Kinghunt.")


76...Kd4 77.Kf4
and White wins.

It would of course have been virtually impossible for Kramnik to see all of this over the board, but it is nevertheless a beautiful win.

UPDATE 12/16/2010: Karsten Mueller has posted an explanation of the position after 69.g5! on the Chessbase website, based on an analysis by Hiarcs 13.  It confirms that 69.g5! wins, and adds that 70.g3, rather than 70.Kg4 followed by 71.g3, is apparently the most precise line.  The winning idea is the same, however, and some of the lines transpose.

2 comments:

  1. Incomplete analysis. Taking the pawn after 69.g5 isnt the best reply.

    ReplyDelete