Author: Heiner Marxen
Date: 14:55:59 05/01/01
Go up one level in this thread
On May 01, 2001 at 12:34:19, Heiner Marxen wrote:
>On May 01, 2001 at 07:55:26, leonid wrote:
>
>>On May 01, 2001 at 01:02:33, Paul Byrne wrote:
>>
>>>On April 29, 2001 at 07:32:15, leonid wrote:
>>>
>>>>Hello!
>>>>
>>>>I know that I should never indicate here any position like this, but it is just
>>>>too seductive! Only really dedicated man or highly performant mate solver can
>>>>crunch it.
>>>>
>>>>[D]Qrbqkbr1/Qn3qPP/NQqQq1nB/1qNqQBq1/1Qq1q2Q/8/3RR3/4K3 w - -
>>>>
>>>>Please indicate your result.
>>>>
>>>>Thanks,
>>>>Leonid.
>>>
>>>That was a tough one! My PN search went through 512MB of memory without solving
>>>it. PN2 got Qbxd8+ winning in 1316.46 seconds (100043829 nodes) -- the PN2 code
>>
>>Hi, Paul!
>>
>>What is PN2?
>
>I strongly suspect it is a modified version of "proof number search".
>I seem to remember that Aske Plaat wrote about it. You should find
>something with your favorite search engine (e.g. Google).
>
>>Your move is what mine found.
>>
>>Your solution was quick compared with mine. I actually had the impression that
>>this position contain the mate and asked to search this position. Very deep
>>selective search, 13 moves deep. Left my solver thinking in the night. Celeron
>>600Mhz. No hash. It took almost 4 hours (3 h 56 min) to solve in 13 moves. But
>>at what depth your found? Maybe your said that mate existe in 12 or even 11.
>>
>>Cheers,
>>Leonid.
>
>Hi Leonid,
>
>According to Chest it is a mate in 13, i.e. no solution in 12 or less moves.
>On a K7/600 with 350 MB hash Chest spends nearly 37 hours to find that
>there are two key moves for the mate in 13: Qdxd8+ and Qbxd8+.
>I have not yet a PV (the solution tree is not contained in the hash, any more,
>and is currently recomputed). I will follow-up when it is there.
Here are two PV lines from Chest:
Qbxd8+ Qxd8 Qhxd8+ Nxd8 Qxd8+ Qxd8 gxf8Q+ Rxf8 Qxf7+ Rxf7 h8=Q+ Nf8
Qxf8+ Rxf8 Rxd8+ Ke7 Bg5+ Rf6 Bxf6+ Kf7 Rf8+ Kxf8 Nxe6+ Ke8
Qe7#
Qdxd8+ Qgxd8 Qhxd8+ Nxd8 Qxd8+ Qxd8 gxf8Q+ Rxf8 Qxf7+ Rxf7 h8=Q+ Nf8
Qxf8+ Rxf8 Rxd8+ Ke7 Bg5+ Rf6 Bxf6+ Kf7 Rf8+ Kxf8 Nxe6+ Ke8
Qe7#
They are basically equal. Other black moves lead to shorter solutions, except:
8... Kxd8 Nxe6+ Bxe6 Bg5+ Kc8 Qexb8+ Qxb8 Qaxb8+ Kd7 Qe7#
Kd7 Nxb8+ Kc8 Nxc6+ Kd7 Qed6#
Qb8 Qbb7#
Qxb8 Qe7+ Kc8 Qexb8#
Rxb8 Qe7+ Kc8 Qd8#
Rf6 Qxf6+ Kd7 Qfe7+ Kc8 Qd8#
Total computation time is 43.2 hours.
Heiner
>Here is the timing statistics:
>
>depth time[s] speed nodes in nodes out
># 1 0.00 0.94 1- 0
># 2 0.01 1.00 1- 0
># 3 0.03 0.97 88- 0
># 4 0.20 1.12 612- 0
># 5 1.06 1.31 3484- 0
># 6 4.64 1.66 16340- 0
># 7 17.77 3.27 67644- 0
># 8 79.58 5.29 335490- 0
># 9 385.99 7.30 1776286- 30
># 10 1492.82 9.43 7245314- 342027
># 11 5287.27 11.55 28318639- 19570743
># 12 31346.51 10.79 174568973- 165821072
># 13 132724.97 10.18 765545533- 756797632
>
>Cheers, Heiner
>
>
>>>is set up for suicide chess though and could probably be sped up a good bit for
>>>regular chess.
>>>
>>>-paul
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.