Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Test Position:Interesting Puzzle

Author: Dieter Buerssner

Date: 09:04:06 09/08/05

Go up one level in this thread


On September 08, 2005 at 11:46:19, Uri Blass wrote:

>On September 07, 2005 at 17:44:28, F. Huber wrote:
>
>>On September 07, 2005 at 16:47:47, Joseph Tadeusz wrote:
>>
>>>On September 07, 2005 at 05:14:49, Madhavan wrote:
>>>
>>>>[D] 8/p5p1/1pP3p1/p5p1/k3p3/4p3/K7/8 w - - 0 1
>>>>
>>>>White to play and win
>>>>
>>>>Post the winning moves
>>>>
>>>>Spike is getting trapped,the engine couldn't find the solution immediately.the
>>>>evaluation got from the engine shows +0.00
>>>>
>>>>The position would look losing in a real life tournament.You have to look hard
>>>>and find the winning line.I wonder if this kind of puzzle would be even
>>>>difficult for a Grandmasters?
>>>
>>>[Event "Edited position"]
>>>[Site "linux"]
>>>[Date "2005.09.07"]
>>>[Round "-"]
>>>[White "-"]
>>>[Black "-"]
>>>[Result "1-0"]
>>>[FEN "8/p5p1/1pP3p1/p5p1/k3p3/4p3/K7/8 w - - 0 1"]
>>>[SetUp "1"]
>>>
>>>1. c7 e2 2. c8=Q e1=Q 3. Qc4+ Qb4 4. Qc6+ Qb5 5. Qxe4+ Qb4 6. Qd3 g4 7.
>>>Qd7+ Qb5 8. Qxg4+ Qb4 9. Qd7+ Qb5 10. Qd4+ Qb4 11. Qd3 g5 12. Qd7+ Qb5 13.
>>>Qd4+ Qb4 14. Qd3 g6 15. Qd7+ Qb5 16. Qd4+ Qb4 17. Qd3 a6 18. Qd7+ Qb5 19.
>>>Qd4+ Qb4 20. Qd3 g4 21. Qd7+ Qb5 22. Qxg4+ Qb4 23. Qd7+ Qb5 24. Qd4+ Qb4
>>>25. Qd3 g5 26. Qd7+ Qb5 27. Qd4+ Qb4 28. Qd3 g4 29. Qd7+ Qb5 30. Qxg4+ Qb4
>>>31. Qd7+ Qb5 32. Qd4+ Qb4 33. Qd3 b5 34. Qc2+ Qb3+ 35. Qxb3#
>>>{White mates} 1-0
>>
>>Your solution is not ok - after move 32... it´s only a #10:
>>(checked with ChestUCI Ver.4.2)

I think Joseph did not mean to show the shortest mate, but to show the
"principle" solution: always force black to advance a pawn with Qd3, until he
has no "good" pawn move anymore. At various points, black can sacrifice his Q,
and white will have an easy win.

>>This is the correct continuation:
>>33.Qd3 Qb2+ 34.Kxb2 Kb4 35.Qc3+ Kb5 36.Qc7 Kb4 37.Qc6 a4 38.Kc2 a3
>>39.Qc3+ Ka4 40.Qc4+ Ka5 41.Kb3 b5 42.Qc7#  {Mate} 1-0
>>
>>Regards,
>>Franz.
>
>
>It is possible that 17...g4 it leads to longer mate but I am not sure
>
>18.Qd7+ Qb5 19.Qxg4+ Qb4 20.Qd7+ Qb5 21.Qd4+ Qb4 22.Qd3 g5 23.Qd7+ Qb5
>24.Qd4+ Qb4 and the question is if 25.Qd3 is forcing mate that is short enough.

White has at least a mate of the same length. After 17...g4 I get:


8/p7/1p4p1/p7/kq4p1/3Q4/K7/8 w - -

15/41  0:00   +M25     18.Dd7+ Db5 19.Dxg4+ Db4 20.Dd7+ Db5
                       21.Dd4+ Db4 22.Dd3 g5 23.Dd7+ Db5
                       24.Dd4+ Db4 25.Dd3 g4 26.Dd7+ Db5
                       27.Dd2 (930.416) 1238 TB:78



After 2...e1Q I get:

2Q5/p5p1/1p4p1/p5p1/k3p3/8/K7/4q3 w - -

Engine: Yace via ssh (700 MB)
von Dieter Buerssner

10/32  0:00   +M40     3.Dc4+ Db4 4.Dc6+ Db5 5.Dxe4+ Db4
                       6.Dd3 g4 7.Dd7+ Db5 8.Dxg4+ Db4
                       9.Dd7+ Db5 10.Dd4+ Db4 11.Dd3 a6
                       12.Dd7+ Db5 (97.621) 1394

I was too lazy, to get a mate score in the front. Now black would just not
convert - most probably white will mate much faster, then.

This was the tree I analysed from back to front to get the indicated result.


[Event "Edited position"]
[Site "linux"]
[Date "2005.09.07"]
[Round "-"]
[White "-"]
[Black "-"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/p5p1/1pP3p1/p5p1/k3p3/4p3/K7/8 w - -"]

1. c7 e2 2. c8=Q e1=Q 3. Qc4+ Qb4 4. Qc6+ Qb5 5. Qxe4+ Qb4
6. Qd3 g4 (6... a6 7. Qd7+ Qb5 8. Qd4+ Qb4 9. Qd3 Qc3
(9... Qd2+ 10. Qxd2) (9... Qa3+ 10. Qxa3+) 10. Qxc3 {M9})
(6... Qb2+ 7. Kxb2 Kb4 8. Qd4+ Kb5) 7. Qd7+ Qb5 8. Qxg4+
Qb4 9. Qd7+ Qb5 10. Qd4+ Qb4 11. Qd3 g5 12. Qd7+ Qb5
13. Qd4+ Qb4 14. Qd3 g6 (14... g4 15. Qd7+ Qb5 16. Qxg4+
Qb4 17. Qd7+ Qb5 18. Qd4+) (14... a6 15. Qd7+ Qb5 16. Qd4+)
15. Qd7+ Qb5 16. Qd4+ Qb4 17. Qd3 a6 (17... g4 18. Qd7+ Qb5
19. Qxg4+ Qb4 20. Qd7+ Qb5 21. Qd4+ Qb4 22. Qd3 g5 23. Qd7+
Qb5 24. Qd4+ Qb4 25. Qd3 {M18}) 18. Qd7+ Qb5 19. Qd4+ Qb4
20. Qd3 g4 21. Qd7+ Qb5 22. Qxg4+ Qb4 23. Qd7+ Qb5 24. Qd4+
Qb4 25. Qd3 g5 26. Qd7+ Qb5 27. Qd4+ Qb4 28. Qd3 g4
29. Qd7+ Qb5 30. Qxg4+ Qb4 31. Qd7+ Qb5 32. Qd4+ Qb4
33. Qd3 Qb2+ 34. Kxb2 Kb4 35. Qc3+ Kb5 36. Qc7 Kb4 37. Kc2
a4 38. Kd3 Ka5 39. Qc3+ Kb5 40. Qc4+ Ka5 41. Kc3 b5
42. Qc7# 1-0

It will be very difficult for an engine, to find this from the front. I think
impossible for any "normal" playing engine.

Regards,
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.