Author: Mark Rawlings
Date: 11:31:11 08/30/99
Go up one level in this thread
On August 29, 1999 at 20:34:19, Howard Exner wrote:
>On August 29, 1999 at 09:01:44, Mark Rawlings wrote:
>
>>On August 28, 1999 at 19:55:25, Howard Exner wrote:
>>
>>>On August 28, 1999 at 11:07:45, Mark Rawlings wrote:
>>>
>>>>On August 28, 1999 at 09:36:58, Howard Exner wrote:
>>>>
>>>>>On August 27, 1999 at 13:57:54, Mark Rawlings wrote:
>>>>>
>>>>>>I let Crafty with 3+2 tablebases chew on this one overnight after the moves c5
>>>>>>bxc5. After 25 ply, the eval was still +1.07 (Be1 Kc7 Ba5+ Kb8...) This one is
>>>>>>really deep!
>>>>>>
>>>>>>Mark
>>>>>
>>>>>What line of play is Crafty producing?
>>>>
>>>>Here's the Crafty output:
>>>>
>>>>
>>>
>>>> 25 463:44 1.07 2. Be1 Kc7 3. Ba5+ Kb8 4. Ke3 Bd4+
>>>> 5. Ke2 Bf6 6. Kd3 Bb2 7. Kd2 Bf6 8.Ke3 <HT>
>>>
>>>How does Crafty play these lines as black?
>>>I've always wanted to fiqure this position out and now with the
>>>thre of us maybe we can do it.
>>>
>>>A.1. c5 bxc5 2. Be1 Kb8 3. Ba5 Kc8 4. b6 axb6 5. Bxb6 Kb8 6. Kc4 Bd4 7. Bd8
>>>Be3 8. Bxh4 Bxf4 9. Bf6 Bg3 10. h4 f4 11. Kd3 f3 12. Ke3 f2 13. Ke2 Ka7
>>>14. h5 Bf4 15. Bg7 Kxa6 16. Kxf2 Kb5 17. Kf3 Bg5 18. h6
>>>
>>>B.1. c5 bxc5 2. Be1 Kb8 3. Ba5 Kc8 4. b6 axb6 5. Bxb6 Kb8 6. Kc4 Kc8 7. Kb5
>>>Kb8 8. Kc6 Be7 9. Kd7 Bf8 10. Ke6
>>
>>After 3. Ba5 crafty has the following:
>>
>> 19-> 4:06 -1.22 3. ... Bb2 4. Bd8 Bc1 5. Bxh4 Bxf4
>> 6. Be1 Ka8 7. h4 Kb8 8. h5 Bh6 9. Bd2
>> f4 10. Ke4 f3 11. Be3 f2 12. Bxf2 Kc7
>> 13. Be3
>> 20 8:06 -1.18 3. ... Bb2 4. Bd8 Bc1 5. Bxh4 Bxf4
>> 6. Be1 Bh6 7. h4 Kc7 8. Ba5+ Kc8 9.
>> b6 axb6 10. Bxb6 Kb8 11. h5 f4 12.
>> Ke4 Kc8 13. Kf3 c4 14. a7 Kb7
>> 20-> 15:37 -1.18 3. ... Bb2 4. Bd8 Bc1 5. Bxh4 Bxf4
>> 6. Be1 Bh6 7. h4 Kc7 8. Ba5+ Kc8 9.
>> b6 axb6 10. Bxb6 Kb8 11. h5 f4 12.
>> Ke4 Kc8 13. Kf3 c4 14. a7 Kb7
>> 21 28:26 -1.34 3. ... Bb2 4. Bd8 Bc1 5. Bxh4 Bxf4
>> 6. Be7 Be5 7. h4 Kc7 8. h5 Bf4 9. Bf6
>> Kb6 10. Kc4 <HT>
>
>Instead of 6. Be7 input Be1. This looks harder to refute because it can
>keep the black king out of b6 (If Kc7 then Ba5+). Also the h pawn moves
>forward while it appears that black's c and f pawns are halted. For now I
>see npo way out for black.
>
Here's some crafty output after 6. Be1 (didn't have the whole set of 3+2
tablebases for this):
17-> 12:48 -1.69 6. ... Kc7 7. Ba5+ Kb8 8. h4 Bh6 9.
h5 Kc8 10. Bd2 f4 11. Ke4 f3 12. Be3
f2 13. Bxf2 Bg7 14. Be3 Kd7 15. h6
Bc3
18 23:02 -1.75 6. ... Kc7 7. Ba5+ Kb8 8. h4 Bh6 9.
h5 Bg5 10. Bd2 f4 11. Ke4 f3 12. Be3
Bf6 13. Kxf3 Kc7 14. h6 Kb6 15. Bf4
Kc7
18-> 31:04 -1.75 6. ... Kc7 7. Ba5+ Kb8 8. h4 Bh6 9.
h5 Bg5 10. Bd2 f4 11. Ke4 f3 12. Be3
Bf6 13. Kxf3 Kc7 14. h6 Kb6 15. Bf4
Kc7
19 52:19 -1.77 6. ... Kc7 7. Ba5+ Kb8 8. h4 Bh6 9.
h5 Bg5 10. Bd2 f4 11. Ke4 f3 12. Be3
Bf6 13. Kxf3 Kc7 14. h6 Bc3 15. Ke4
Kd7 16. h7 Bh8
I don't think even deep blue would see a forced win from the original position!
Mark
>>
>>No sure thing yet for white. After 10. Ke6 in your line B.1. above, it looks
>>like a win for white:
>>
>> 16-> 43.96 -3.28 8. ... Ka8 9. Kxf5 Bh6 10. Kg4 Bg7
>> 11. Kh5 c4 12. Kg6 Bf8 13. f5 c3 14.
>> Bd4 c2 15. Be3 <HT>
>> 17 1:13 -3.56 8. ... Ka8 9. Kxf5 Bh6 10. Kg4 c4 11.
>> Be3 c3 12. Kh5 Bf8 13. Kg6 Be7 14.
>> f5 Bd8 15. f6 Kb8 16. f7 <HT>
>> 17-> 1:52 -3.56 8. ... Ka8 9. Kxf5 Bh6 10. Kg4 c4 11.
>> Be3 c3 12. Kh5 Bf8 13. Kg6 Be7 14.
>> f5 Bd8 15. f6 Kb8 16. f7 <HT>
>> 18 2:04 -- 8. ... Ka8
>> 18 4:31 -6.45 8. ... Ka8 9. Kxf5 c4 10. Kg6 c3 11.
>> f5 c2 12. Be3 Be7 13. f6 Bd8 14. Kg7
>> Kb8 15. f7 Be7 16. f8=Q+ Bxf8+ 17.
>> Kxf8 <HT>
>>
>>
>>I'll try to find the web site with the solutions to these. Maybe it has the
>>winning line. (If not, it probably isn't a very good problem for a test suite!)
>
>It's definately a difficult problem for computers. The intent of the problem
>could have been the initial position where most machines would just aimlessly
>wander about behind the pawn wall. I human would quickly realize that the only
>chance of a win is the break, c5. Whether that really wins remains but I
>think that we will eventually resolve this one.
>>
>>Mark
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.