Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A Tough Endgame For Computers! Interesting....

Author: Martin Giepmans

Date: 11:31:20 04/08/02

Go up one level in this thread


On April 08, 2002 at 10:49:00, Terry McCracken wrote:

>On April 08, 2002 at 09:45:43, Martin Giepmans wrote:
>
>>On April 07, 2002 at 15:48:22, Terry McCracken wrote:
>>
>>>Here's an Endgame that should make _all_ chess engines really work!
>>>
>>>Fritz 7 sees the draw but doesn't understand the loss after 1.Qe8+..Qf8??
>>>
>>>Played by Mieses-NN 1903. Mieses won after Qf8??, I'll give the main line.
>>>
>>>
>>>1.Qe8+ Qf8 2.h5 Nf7 3.Kc2 Nd8 4.Kc3 Nf7 5.a4 Nd8 6.Kc4 Nf7 7.Kd5 Nd8 8.b4 Nf7
>>>9.Kc6 Nd8+ 10.Kc7 Line
>>>
>>>After Qe8+..Kh7! is correct. After 2.Qxd8...Qd2 will draw with perpetual check.
>>>
>>>[D]3n2k1/p2P2p1/1p6/7Q/5q1P/8/PP6/1K6 w - -
>>>
>>>Terry
>>
>>I get this:
>>
>>SpiderChess (amd 1200)
>>05  00:00:00  + 0.77  H5-E8 F4-F8 B1-C2 G7-G6 E8xF8 G8xF8
>>06  00:00:00  + 0.33  H5-E8 F4-F8 B1-C2 G7-G6 E8xG6 G8-H8 C2-D3
>>07  00:00:00  + 0.10  H5-E8 G8-H7 E8xD8 F4-F5 B1-C1 F5-C5 C1-D2 C5-D5 D2-C3
>>                      D5xA2
>>08  00:00:00  + 0.10  H5-E8 G8-H7 E8xD8 F4-F5 B1-C1 F5-C5 C1-D2 C5-D5 D2-C3
>>                      D5-E5 C3-B4 E5xB2 B4-C4
>>09  00:00:02  + 0.10  H5-E8 G8-H7 E8xD8 F4-F5 B1-C1 F5-C5 C1-D2 C5-D5 D2-C3
>>                      D5-E5 C3-D3 E5-F5 D3-C4
>>10  00:00:11  + 0.10  H5-E8 G8-H7 E8xD8 F4-F5 B1-C1 F5-C5 C1-D2 C5-D5 D2-C3
>>                      D5-E5 C3-D3 E5-F5 D3-C4 F5-F7 C4-B5
>>
>>(score +0.10 means that the engine thinks that it is probably a perpetual)
>>
>>After Qe8 Qf8 the score rises immediately:
>>05  00:00:00  + 0.75  B1-C2 D8-F7 C2-D3 G7-G5 E8xF8 G8xF8
>>06  00:00:00  + 0.79  B1-C2 D8-B7 C2-D3 G7-G5 E8xF8 G8xF8
>>07  00:00:00  + 0.61  B1-C2 G7-G5 E8-G6 G8-H8 G6-H5 H8-G7 H5xG5 G7-H7 C2-D3
>>08  00:00:00  + 0.65  B1-C2 G7-G5 H4xG5 G8-G7 E8xF8 G7xF8 G5-G6 F8-G7 C2-D3
>>                      G7xG6
>>09  00:00:01  + 0.45  B1-C2 G7-G6 E8xG6 G8-H8 G6-E8 H8-G7 E8-E5 G7-G8 E5-G5
>>                      G8-H7 C2-C3 F8-F3 C3-B4 D8-C6 B4-B5
>>10  00:00:05  + 0.55  B1-C2 G7-G6 E8xG6 G8-H8 G6-E8 H8-G7 E8xF8 G7xF8 H4-H5
>>                      F8-E7 H5-H6 D8-F7 D7-D8D F7xD8 H6-H7
>>11  00:00:11  + 0.70  B1-C2 D8-F7 H4-H5 B6-B5 C2-D3 B5-B4 D3-D4 B4-B3 A2xB3
>>                      G7-G5
>>12  00:00:48  + 0.86  B1-C2 B6-B5 H4-H5 A7-A6 C2-D3 A6-A5 D3-E4 A5-A4 B2-B3
>>                      A4xB3 A2xB3 G7-G5
>>13  00:01:44  + 0.88  B1-C2 A7-A5 H4-H5 A5-A4 C2-D3 B6-B5 D3-E4 D8-F7 E4-D4
>>                      A4-A3 B2xA3 G7-G5
>>
>>Apparently this is mainly eval, allthough the last pv begins to look right ;)
>>
>>Cheers,
>>Martin
>
>
>I'm surprised SpiderChess considers 1..Kh7! It is the _only_ move that survives,
>and it does indeed looks like it wants to draw with perpetual!
>
>It may be the only engine that does!
>
>1..Qd2 also ensures perpetual and draws faster I believe.
>
>When you feed it 1..Qf8?, it still has problems seeing the loss, but does see
>White is better, which is more than I can say for the top commercial programmes.
>
>But it doesn't play 2.h5! straight away and I think that White can't win if it
>isn't played immeadiatly after 1..Qf8?
>
>1.Qe8+..Qf8? 2.Kc2?..g6! and it appears to be a draw with Qxg6+..kh7= with
>another perpetual. However I haven't dug real deep here.
>
>I think you miss the win with White and let Black go. 1.Qe8+..Qf8? 2.h5! ++/-
>and White will win.
>
>Still your eval is very good for finding perpetuals:o)
>
>Let SpiderChess work much longer and see if it changes 2.Kc2? to 2.h5! sealing
>Black's fate.
>
>Terry

The default ("positional") version of SpiderChess still wants to play Kc2 after
10 minutes.
With the tactical version I get:

(position after Qe8 Qf8):
.........
10  00:00:19  + 0.55  B1-C2 G7-G6 E8xG6 G8-H8 G6-E8 H8-G7 E8xF8 G7xF8 H4-H5
                      F8-E7 H5-H6 D8-F7 D7-D8D F7xD8 H6-H7
    00:00:25  + 0.70  H4-H5 D8-F7 B1-C2 B6-B5 C2-D3 B5-B4 D3-D4 B4-B3 A2xB3
                      G7-G5 E8xF8 G8xF8
11  00:00:39  + 1.21  H4-H5 D8-F7 B1-C2 B6-B5 C2-D3 B5-B4 D3-D4 B4-B3 A2xB3
                      A7-A5 E8xF8 G8xF8
11  00:00:43  + 1.21  H4-H5 D8-F7 B1-C2 B6-B5 C2-D3 B5-B4 D3-D4 B4-B3 A2xB3
                      A7-A5 E8xF8 G8xF8
12  00:01:06  + 1.16  H4-H5 D8-F7 B1-C2 B6-B5 C2-D3 B5-B4 D3-D4 A7-A5 D4-D5
                      F7-D8 B2-B3 G7-G5 E8xF8 G8xF8

It still doesn't see the whole line, I think, but it would probably play
the right moves in a game with longer timecontrol.
If this happens more often (I think you are right about Kc2 (?), it is a
draw) the tactical version will become the default version ;)

Cheers
Martin




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.