Author: Heiner Marxen
Date: 08:46:50 11/15/98
Go up one level in this thread
On November 14, 1998 at 10:37:55, Hans Havermann wrote: >Valentin Albillo, more than a year ago, posted his "unsolved" >3qk3/8/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - (Test #91: ><http://www.multimania.com/albillo/ajedre9a.htm>). Computationally intensive, >Albillo conjectured "it's a mate in 12". A number of chess engines had a go at >it, but the attempts are old and seem dated. I wonder if this problem has since >been solved. This problem is really hard to solve. According to my problem solving program this is most probably not solvable in 9 moves (or less). [ "most probably" because I accidentally turned castling off ] That computation took 4.5 hours on a fast machine. Trying to solve (or disprove) a mate in ten would take 15x that time. I didn't try, and now I don't have access to that fast/big box anymore. >I thought I should let MacChess (5.0b3) try this on my 300 MHz G3. After 56 >hours, toward the end of 13-ply, MacChess came up with: > > 1. e4 Qd4 > 2. Bb5+ Ke7 > 3. Qg4 Qe5 > 4. Qd7+ Kf8 > 5. d3 Qh5 > 6. Bf4 Qf7 Well, starting from here, I can confirm, that 1.Bh6+ is a mate in 6, that there is no faster way to force mate, nor any other way to mate in 6. I'm working on more, but don't hold your breath. > 7. Bh6+ Kg8 > 8. Bc4 Kh7 > 9. Qxf7+ Kxh6 > >Despite the program's move-extenders (which caused my analysis to go 93 plies >deep on some line!), it couldn't see the follow-up (because, I guess, "g3" is a >non-trivial move and does not engage the move-extender): > >10. g3 Kg5 >11. Qg7+ Kh5 >12. Bf7# > >Anyways, I do not pretend this (necessarily) "solves" the problem. But there >are >people out there with faster computers and better programs than mine. Anyone? Yes, I'm curious, too.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.