Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Two Mate Positions

Author: Peter Berger

Date: 14:25:34 08/21/01

Go up one level in this thread


On August 21, 2001 at 13:52:01, Dieter Buerssner wrote:

I tried your positions with the program "Der Bringer" (
http://www.reubold.onlinehome.de) by Gerrit Reubold on an Athlon 1333 computer.

>Two mates from an actual game postition with equal material on the board.
>The first one may not be difficult to find for playing programs. It is a
>mate in 9 or better.
>
> r2q1r1k/1pbbn2B/p1n1p1p1/3p2NQ/1P1p1P2/P1P1B3/6PP/R4RK1 w - - 0 18

This one is very easy as you expected.

0:00:05.2  ( 9/28)    1294302   8.69  h7xg6   (Mat=-235,50=0)
0:00:11.4  ( 9/36)    2801304   Matt in 9  h7xg6  h8-g7  h5-h7  g7-f6  g2-g4

>
>The second one is mate in 19 or better. This positions occured 2 moves
>earlier in the game.
>
>[D] r2q1rk1/1pbbn1pp/p1n1p3/3p2N1/1P1p1P2/P1PBB3/6PP/R2Q1RK1 w - - 0 16
>
>Yace cannot find this mate, without help by first analysing some lines.

This is called cheating Dieter, isn't it ;-)) ?

OK - without any tricks :) :

0:14:03.6  (13/42)   222073875   18.88  d3xh7  g8-h8  d1-h5  f8-f6
(Mat=-335,50=3)
0:17:04.9  (14/42)   268965501   19.28  d3xh7   (Mat=-335,50=0)
0:32:29.9  (14/46)   517708825   19.68  d3xh7   (Mat=-335,50=0)
1:17:51.5  (14/46)   1269029346   20.42  d3xh7  g8-h8  d1-h5  g7-g6
(Mat=-335,50=0)

Not solved - I somehow expect this should be solveable in less than 10 hours for
Bringer at ply 15 but this remains to be tested ( hardware advances are a thrill
btw - a nice potential bug : what will Bringer do after about 11 hours when the
internal node counter in the GUI can't display the positions searched anymore ?!
) .


>With French Winawer, I have seen quite a few short computer chess games,
>also often, losses by Yace. But I have also seen fast wins for Black.
>Chester expected 15. fxe5. I think, 15. Ng5 is already winning, but Yace could
>not see it yet in the game. Actually, the two moves were rather close in score,
>which also shows, that there can be quite some luck involved.
>
>Regards,
>Dieter

You won't see the real problem with the Winawer against computers.

The Winawer might or might not be a good opening against other silicon GMs -
this is probably mainly a book tuning issue or luck . I think if you avoid the
positions with the penetrating queen on g7,h7 this should be very playable
against computers.

Problem is : all of them don't understand the positions resulting from the main
variations well.

Against a strong human the Winawer is a very bad choice I am sure. Btw also
playing against it with white is a hard challenge for the same reasons.

But the question which openings are a good choice for a computer seems to remain
open: there doesn't seem to be any publication about that.

And even most of the professionals don't seem to do too well.

Look at Maastricht and our world master . Round 1 and 3 showed Shredder playing
lines that should be very well known for not being suitable for a game against a
computer IMHO - and this was prepared by one of the world's leading experts ..

The only ones with really good opening books seem to be Quest,Junior,Tiger,Rebel
and Diep ( this one looks very much improved to me) this time.

The Crafty book looks miserable again ( sorry, no offense intended although it
probably will sound like that anyway) . The Ferret book doesn't impress much
either.

It seems to be a trial and error issue though : at last most seem to have come
up with the Najdorf and the Semi-Slav ( although 1.d4 seems to be very uncommon
- why ? ) .

Some personal observations ( I'd really like to discuss this but I don't expect
many answers ) :

a.) Playing 1. ...e5 against 1. e4 will be good if White opts for the mainline
Ruy Lopez only - this is a bad line for a computer as the black play requires
only knowledge computers do have while doing well with White will either require
book tuning to the extreme or knowledge I haven't seen yet.

b.) Many openings that look very promising in computer play are waiting for
discovery IMHO . Sveshnikov should be well for black in computer matches and how
about 1. e4 c6 2.d4 d5 3. Nd2(c3) de 4. Nxe4 Bf5 ? Computers don't seem to be
able to play this well with white but do very well with black.

I have searched the archives for every interesting idea about opening books
posted but the results are lame.

Why is this such a secret issue I ask myself ?

Cheers.

pete








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.