Author: Robert Hyatt
Date: 21:38:14 02/03/00
Go up one level in this thread
On February 03, 2000 at 19:44:51, Amir Ban wrote:
>On February 03, 2000 at 19:25:16, Frederic Friedel wrote:
>
>>On February 03, 2000 at 18:42:50, Dann Corbit wrote:
>>
>>>So, I think the problem is not to find the key moves but to prove the
>>>checkmates.
>>
>>The problem is to find the _solution_, not the key move. Kasparov had to do
>>this, and John Nunn gave me critical lines he had to find. I was not allowed to
>>prompt for the best defence. I only helped him in position 6 by giving him the
>>best defence, which he had missed. After 1.Kh2 f6 there is a critical move which
>>I challenge any program to find.
>>
>
>Here's your challenge:
>
>[D]8/7p/5p2/6k1/8/6P1/5PPK/8 w - - ; bm Kh1;
>
>It's not practical to ask programs to provide solutions in the form of complete
>variations and subvariations. It's more practical to add to the 6 original
>positions positions from critical variations and ask the program to make the
>correct choice. This will enable to present this contest as a regular search for
>key move suite.
>
>Amir
>
Here is a more important question. To prove that Kh1 is better than Kg1. I
get a fail high on Kg1 at depth=28, after about 1 minute. Score jumps to +2.5
at that point... Here is the PV/etc at the end of depth=28:
28-> 1:19 2.34 1. Kg1 h5 2. Kf1 Kf5 3. f3 Ke6 4. g4
hxg4 5. fxg4 Ke5 6. g3 f5 7. gxf5 Kxf5
8. Kf2 Kg5 9. Kf3 Kf5 10. g4+ Kg5 11.
Kg3 Kg6 12. Kh4 Kh6 13. g5+ Kg7 14.
Kh5 Kh7 15. g6+ Kg8 16. Kh6 Kh8 17.
Kg5
I searched Kh1 (made it only consider Kh1 at the root) and the score was
roughly the same, to the same depth. By -- oops... at depth=29, crafty
fails high on Kh1 after 3 minutes. So maybe the above can be forgotten. :)
I started the search from scratch when I saw the fail high. It takes just
over 3 minutes and 29 plies to finally give up on Kg1 and switch to Kh1 (this
all on my quad xeon 400). I couldn't get a score back as the time limit was
set to 240 seconds, and someone started a game on ICC and I had to stop. I
will try to find the 'score' for this thing later. But Kh1 _can_ be found.
I should have waited until the terms of the 'bet' were announced. :)
at depth=29 score went to +3.2, at depth=30, it failed high again before I
stopped it.
29 3:14 ++ 1. Kh1!!
>
>>Maybe I'll post the article (it's fairly long)?!
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.