Author: Dann Corbit
Date: 21:30:36 12/05/03
Go up one level in this thread
On December 05, 2003 at 22:59:40, Jeremiah Penery wrote:
>After Bob posted results from Opteron, I decided to do some tests also.
>Unmodified Crafty 19.6 compiled with DETECTDRAW on AthlonXP 1600MHz now solves
>297/300 in 5 seconds, missing only #s 2, 141, and 230. #s 2 and 141 are solved
>in about 6-7 seconds, and #230 takes less than 10 seconds.
>
>The draw detection helps solve #100 instantly, but when I analyze this a lot I
>get some strange results that I'm not sure are accurate. Obviously, b6 is the
>easy winning move, and Be3 should also win. However, I've done searches
>starting with Kb3 which also return winning scores, even as I move several moves
>down the line, deep searches still show increasing scores (in the +4 range).
>
>I'd like to know if there is some definitive analysis on this position that
>shows that Kb3 (or other moves!) really can't win. For reference, position is
>given below:
>
>[D]8/k1b5/P4p2/1Pp2p1p/K1P2P1P/8/3B4/8 w - -
From the archives:
Subject : Re: Another solution to WAC 100?
Posted by : Rémi Coulom on May 09, 2001 at 08:15:02
On May 08, 2001 at 00:56:43, Miguel A. Ballicora wrote:
>On May 07, 2001 at 19:29:29, Dann Corbit wrote:
>
>>On May 07, 2001 at 10:42:51, Miguel A. Ballicora wrote:
>>
>>>On May 06, 2001 at 13:46:29, Rémi Coulom wrote:
>>>
>>>>On May 06, 2001 at 08:38:37, Vincent Diepeveen wrote:
>>>>
>>>>>On May 05, 2001 at 08:36:44, Rémi Coulom wrote:
>>>>>
>>>>>>On May 04, 2001 at 17:47:19, Dann Corbit wrote:
>>>>>>
>>>>>>>ce -503; id "WAC.100 after Kb3"; pv Bb8 Bc3 Bxf4 Bxf6 Bd6 Kc3 Bc7 Be7 Be5+ Kd3
>>>>>>>Bd4 Ke2 f4 Kf3 Be3 Bd6 Bd4 Kxf4 Kb6 Be5 Bf2 Bf6 Ka7;
>>>>>>>
>>>>>>>Maybe it isn't so bad. Opinions?
>>>>>>
>>>>>>This line looks OK. I tried with TCB, in four best moves (and it gave me five
>>>>>>!):
>>>>>>
>>>>>> 414178084 ------ Depth = 22
>>>>>> 433380266 +401 1. b6+ Bxb6 2. Kb5 Bc7 3. Be3 Bb8 4. Bxc5+ Ka8 5.
>>>>>> Be3 Bc7 6. Kc6 Ba5 7. Kd6 Kb8 8. Ke6 Ka8 9. Kxf5
>>>>>> Bc3 10. a7 Kb7 11. Kg6 Ka8 12. Kxh5
>>>>>> 459896294 +401 1. Be3 Bd6 2. Ka5 Bc7+ 3. b6+ Bxb6+ 4. Kb5 Bc7 5.
>>>>>> Bxc5+ Ka8 6. Be3 Bd6
>>>>>> 565045738 +290 1. Kb3 Bb8 2. Bc3 Bxf4 3. Bxf6 Bd6 4. Kc3 Kb6 5.
>>>>>> Bd8+ Ka7 6. Kd3 Bf8 7. Ke2 Bd6 8. Kf3 Bh2 9. Be7
>>>>>> Bg1 10. Kf4 Bd4 11. Kxf5 Kb6
>>>>>> 578560247 +290 1. Ka3 Bb8 2. Kb3 Kb6 3. Kc2 Bd6 4. Kb2 Ka7 5.
>>>>>> Bc3
>>>>>> 589681148 +290 1. Bc1 Kb6 2. Kb3 Bd6 3. Ka2 Ka7 4. Bd2 Bc7 5.
>>>>>> Kb3
>>>>>>
>>>>>>So, probably, you have two other alternatives.
>>>>>>
>>>>>>Rémi
>>>>>
>>>>>Noop. The line shown are wrong. You need to play b6 one day to win this
>>>>>position i don't see it in the line!
>>>>>
>>>>>Best regards,
>>>>>Vincent
>>>>
>>>>What better defense do you propose for Black in Dann's line? White goes and
>>>
>>>Standing still would be the plan. Bxf4 is a bad move. How does white win if
>>>Black does not take the pawn in Bxf4 allowing the white king to enter
>>>into the king side? The only way I see white can win involve always to come back
>>>with the king to a4. f6 pawn is worthless, white can take it any time.
>>>I think that Kb3 is not a solution.
>>>
>>>I posted something before:
>>>http://www.icdchess.com/forums/1/message.shtml?168045
>>
>>I am not fully convinced either way (of course, I'm a real chess blockhead
>>sometimes). I see one way to get a better view is to go to that position, and
>>spawn all possible moves, then analyze each of them at long time control (in
>>othter words -- what really does happen if we don't take it -- does it stay
>>sealed?).
>
>As an example, as black, place the Bishop in b6 and play Ka8-a7-a8-a7 for ever.
>There is only way to penetrate into the black position = through a5.
>However, white should play Ba5 (with the wK in a4 of course) at the right
>moment, which is when the bK is in a8. Otherwise, with the bK in a7
>black can play Bc7 and after Ba5 Bd6. White can't then play b6 because
>a6 is hanging. That's how 1.Be3 wins.
>
>1.Be3 Bb6
>2.Bf2 (needed waste of tempo) Ka8
>3.Be1 and next Ba5 wins.
>
>There is an easy way to confirm whether the program you mention that saw a win
>with Kb3 is true. Set the position, let the program think until it plays
>Kb3. You, as black, play Ka8-a7-a8-a7 and Bb6 when necessary. See if it can beat
>you without placing back the king in a4 and using the plan I told you.
>If it really "saw" a win already, it should beat you easily.
>
>My opinion is that the program saw that Kb3 wins the pawn in f6.
>
>Regards,
>Miguel
Yes, you are right. I am convinced now. I will let my program search in n-best
move after 1. Kb3 Kb8 2. Bc3 Ka7 3. Bxf6 overnight to make sure. But this is the
kind of position the good old human understanding handles better. My problem is
I am such a poor chess player I have to trust my stupid chess program most of
the time.
Greetings,
Remi
======================================================================
Also:
======================================================================
Subject : Re: WAC.100 --> I know this has been discussed recently, but I don't
have it
Posted by : Dieter Buerssner on October 07, 2001 at 09:39:09
On October 04, 2001 at 23:33:42, Dann Corbit wrote:
>I have the CCC archives, but there is a big hole in them, since the ftp site has
>not been working for some time. Anyway, it sure looks to me like Ke3 is also a
>solution. Can someone refute this move sequence, or show me a perpetual check
>or some other escape? It looks to me like black is simply toast, and rather
>quickly at that!
>Position after Kb3:
>[D] 8/k1b5/P4p2/1Pp2p1p/2P2P1P/1K6/3B4/8 b - -
>
>[Event "?"]
>[Site "?"]
>[Date "????.??.??"]
>[Round "-"]
>[White "?"]
>[Black "?"]
>[Result "*"]
>[FEN "8/k1b5/P4p2/1Pp2p1p/2P2P1P/1K6/3B4/8 b - - 0 1"]
>[SetUp "1"]
>
>{--------------
>. . . . . . . .
>k . b . . . . .
>P . . . . p . .
>. P p . . p . p
>. . P . . P . P
>. K . . . . . .
>. . . B . . . .
>. . . . . . . .
>black to play
>--------------}
>1... Bd6 2. Ba5 Bxf4
1...Bd6 is wrong. I think, 1...Bb8 draws (I hope I remember this correctly from
an analysis I have done some time ago). At least no engine was able to win, and
also Miguel has given convincing arguments.
It will be better to start from the initial position of WAC 100, instead of the
setup positions from FEN strings, to exclude the posssibility of repetition of
the original position.
Otherwise, one could argue, that the position is won after Kb3 (which would be a
wrong argument for a chess engine in my opinion).
Dieter
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.