Author: Christophe Theron
Date: 15:58:36 09/27/01
Go up one level in this thread
On September 27, 2001 at 18:01:32, Roy Eassa wrote:
>On September 27, 2001 at 17:47:47, Robert Hyatt wrote:
>
>>On September 27, 2001 at 17:33:58, Roy Eassa wrote:
>>
>>>On September 27, 2001 at 17:07:23, Andy Serpa wrote:
>>>
>>>>I am intrigued by this position from an old book:
>>>>
>>>>[D]rnb2r1k/pp1p2pp/2p1P3/2q1p3/2B3Q1/2N4P/PPP3P1/R1B4K w - - 0 1
>>>>
>>>>
>>>>In the book, White saves himself with the following manoeuvre:
>>>>
>>>>1. b4 Qxb4 2. Rb1 Qxc3 3. Bh6 gxh6 4. e7 Re8 5. Rf1
>>>>and white mates in a few moves.
>>>>
>>>>
>>>>I'm curious what a fast computer comes up with in this position...
>>>
>>>
>>>Starting two moves in (I think it's too hard to start at the beginning), after
>>>1. b4 Qxb4 2. Rb1 Qxc3:
>>>
>>>[D]rnb2r1k/pp1p2pp/2p1P3/4p3/2B3Q1/2q4P/P1P3P1/1RB4K w - - 0 1
>>>
>>>Here's what I get with Gambit Tiger 2.0 Aggressive on an Athlon 1.2 (in 1
>>>minute, it sees it has a perpetual. In 1.5 minutes, it sees it has an
>>>advantage. In 4 minutes, it sees it has a win.):
>>>
>>>00:00:00.8 -4.14 8 161550 Bd3 Qe1+ Kh2 Qf2 e7 Re8 Bxh7 Rxe7 Bg6
>>>00:00:00.1 -3.24 8 312659 Bd3
>>>00:00:01.7 -3.24 9 391208 Bd3 Qe1+
>>>00:00:01.4 -3.68 9 670505 Bd3 Qc5 Qh4 e4 Qxe4 Qh5 e7 Re8 Ba3 d5
>>>00:00:02.2 -3.02 9 885911 e7 Re8 Bh6 gxh6 Rf1 Qa3 Rf7 Qc1+ Kh2 Qf4+ Rxf4 exf4
>>>00:00:05.0 -2.12 10 2037930 e7
>>>00:00:06.5 -2.12 11 2496977 e7
>>>00:00:11.9 -1.60 11 4422079 e7 Re8 Bd3 d5 Qf3 Nd7 Qf7 Rxe7 Qxe7 Qc5 Qe6 Qf8 Bb2
>>>00:00:14.9 -1.60 12 5492547 e7 Re8
>>>00:00:48.3 -1.36 12 17494341 Bd3 Qe1+ Kh2 Rf1 Bxf1 Qxf1 Ba3 Qf6 e7 Qf7 Rf1 Qg8
>>>Qf5
>>>00:00:55.1 -0.70 12 20234211 Bh6
>>>00:00:57.8 -0.31 12 21078455 Bh6 gxh6 e7 Re8 Rf1 Qa3 Rf7 Qc1+ Kh2 Qg5 Rf8+ Kg7
>>>Rf7+ Kg8 Rf8+ Kg7
>>>00:01:33.7 0.59 13 35309917 Bh6
>>>00:04:00.9 3.84 13 94628156 Bh6 Rg8 e7 Qxc4 Qxc4 d5 Bxg7+ Kxg7 Qh4 Nd7 Qg5+ Kf7
>>>Rf1+ Ke6 Qxg8+ Kxe7 Rf7+ Kd6
>>
>>
>>I think that is too late to start searching. Qc3 probably loses. Question is,
>>is there an alternative that doesn't, such as Qa3 instead. I haven't studied it
>>much, but the queen at c3 is locked out of the game which is how white wins.
>
>
>I chose that point not because I thought Black could save himself (it's
>definitely too late), but because it serves as a pretty good computer chess
>puzzle: "How long does it take your program to see that 3. e7 and/or 3. Bh6 is
>winning (advantage > 1.5 pawns, say)?"
>
>A different puzzle would be, "How long does it take for your program to avoid
>2...Qxc3?". That puzzle is harder, from what I can see, but also a valid one to
>try.
Chess Tiger 14.6 on K6-2 450MHz, 6Mb hash tables, would reject Qxc3 and play Qe7
in 8min 39s.
Christophe
>Finally, I think that the original puzzle, "How long does it take for your
>program to see that 1. b4 is best?" (or wins, or whatever) would probably take
>an awful long time to solve.
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.