Timman-Berg: A Study-like Conclusion
In the game Jan Timman-Emanuel Berg, 14th Sigeman & Co. 2006 (played this past Friday), the players reached this position after Black's 65th move:
White to move and win. (Remember, R + N vs. R is a draw except in special circumstances, none of which will apply here.) When you think you've got it worked out, click here for the whole game and the solution (and don't forget to check out the second, bonus game as well).
White to move and win. (Remember, R + N vs. R is a draw except in special circumstances, none of which will apply here.) When you think you've got it worked out, click here for the whole game and the solution (and don't forget to check out the second, bonus game as well).
Posted by Dennis Monokroussos on
Sunday April 30, 2006 at 3:24am
Here it is: 1.Nf5 Rg8 2.g7 b3 3.Rf8 b2 4.Rxg8 b1(Q) 5.Ng3+ Ke3 (otherwise White removes his R from the queening square with check and then promotes) 6.Rf8, and it looks to me like Black can give no more than a single check either on the 2nd rank or on g7, before White queens his pawn and wins with his extra material.
I tried this with other engines too: Hiarcs 10, Fruit and Rybka, and they all did the same thing. That led me to think that it had something to do with the way tablebases are accessed, but then I switched on my favorite engine brand, Shredder, and it not only saw that 1.Nf5 won before I let it see 1...Rg8, it even saw 1.Nf5 as a winner (along with 1.Rf7) from the diagram position above. (I sure hope Shredder 10 comes out soon!)
At any rate, White wins, but just by a hair, after 1.Nf5 Rg8 2.g7 b3 3.Rf8 Rxg7 4.Nxg7 Kd3 5.Ne6! b2 6.Nc5+!, as the knight will be able to give itself up for the passed pawn (6...Kc2 7.Rf2+). So you're right, Jon, and thanks for accidentally leading me to discover something odd about chess engines!
I don't remember if I ever optimized Fritz's hash tables for my machine; I think I did, using the instructions given in Krongard's How to Use Computers to Improve Your Chess.
Jon: I remember seeing that book in the stores once upon a time, and thought that I knew too much of what was in the book to justify its purchase. If you can offer a concise summary of hash table optimization, however, that might prove useful to the readers of this blog - myself included!