Author: Uri Blass
Date: 14:31:20 05/03/02
Go up one level in this thread
On May 03, 2002 at 13:24:24, Odd Gunnar Malin wrote:
>Hi.
>
>I have added some drawdetection and draw offer to my little engine and see what
>happend (remember PolarChess is only an 1800 engine):
>
>[D]4b3/p7/1p6/4N3/1k1K3P/8/8/8 w - - 0 63
>
>63.h5 Bxh5 64.Nc6+ Kb5 65.Nxa7+ Ka6 66.Nc8 b5 67.Kc5 Ka5 68.Na7
>{Game drawn by mutual agreement} 1/2-1/2
>
>And it was my engine who offered a draw in the end position!
>
>The hole game is here:
>http://www.chessclub.com/cgi-bin/view-game.pl?id=1040618103
>
>Odd Gunnar
>(proud father)
If you are interested in test positions for polarchess I can give you a lot of
positions from the game of my engine against it.
Polarchess did some mistakes that are not typical to computers
(for example wrong sacrifice or not seeing a simple mate in 4 in the endgame)
I post some notes about the game(the game is at the end of my post and also can
be downloaded at http://home.hccnet.nl/leo.dijksman/index.html after clicking on
the 5th division)
1)It was not a good game of my engine.
polarchess got a clear positional advantage when it suddenly sacrificed a piece
without enough compensation
[D]r2qk2r/1p2bppp/1n2b3/p2p4/P2Pn3/1QNB1N2/1P3PPP/R1B1R1K1 w kq a6 0 14
14.Qb5+ Bd7 15.Qb3 Be6 16.Nxd5?
2)My engine got a winning position that it spoiled by one blunder
[D]r2qk2r/1p2b2p/1nb1Q1p1/p4pN1/P2Pn3/3B4/1P3PPP/R1B1R1K1 b kq - 0 21
21...Nxg5 is a losing move.
My engine simply did not use null move pruning so it did not search deep enough
to see that 21...Nxg5 is losing(with null move pruning it can search 1 ply
deeper that makes a big difference).
It could also avoid it by evaluating pins but it does not evaluate pins directly
but only by the number of legal moves and it was not enough.
3)Polar had a simple win later but again it played the wrong line
[D]r3k3/2R1Br1p/6p1/p2bR3/P2P1p2/8/1P3PPP/6K1 w q - 0 32
The game continued 32.Bg5+ Kf8 33.Be7+ Ke8 34.Bf6+ Kf8 35.Rc3??
35.Be7+ could be the fastest way to win because movei does not check for
castling rights in the repetition detection and could lose the game after
35.Be7+ by 35...Ke8 with wrong 3 times repetition detection.
4)Polar also blundered later in the game when it missed a simple win by pawn
endgame.
[D]8/8/R4k2/3r2pp/PP3p2/8/5PKP/8 b - - 0 44
44...Kf5? 45.Ra5 Ke5 46.b5?
Note that movei's score got down when it played 45...Ke5 because it understood
that 46.Rxd5+ is good for white.
5)The last mistake of polarchess was 56.Kg1 when my program had no problem to
see the forced mate.
[D]1r6/RP6/8/8/P5pp/2k2p2/5P1P/5K2 w - - 0 56
Here is the full game:
Polar Chess 1.2 - Movei 0.06.02
WBEC3_5thDivision DUAL-P3-933 (8), 20.04.2002
1.e4 e5 2.Nf3 Nf6 3.Nxe5 d6 4.Nf3 Nxe4 5.d4 Bf5 6.Bd3 d5 7.0-0 Nd7 8.Re1 Be7
9.c4 c6 10.cxd5 cxd5 11.Qb3 Nb6 12.Nc3 Be6 13.a4 a5 14.Qb5+ Bd7 15.Qb3 Be6
16.Nxd5 Bxd5 17.Qb5+ Bc6 18.Qf5 g6 19.Qe5 f6 20.Qe6 f5 21.Ng5 Nxg5 22.Bxg5 Nc8
23.Rac1 Qd7 24.Qe5 Rg8 25.Bc4 Rf8 26.Be6 Qd6 27.Bxc8 Qxe5 28.Rxe5 Rf7 29.Bxb7
Bxb7 30.Rc7 Bd5 31.Bxe7 f4 32.Bg5+ Kf8 33.Be7+ Ke8 34.Bf6+ Kf8 35.Rc3 Bxg2
36.Rxa5 Rxf6 37.Rxa8+ Bxa8 38.Rc8+ Kg7 39.Rxa8 Rd6 40.d5 Kf6 41.Ra7 h5 42.Kg2
Rxd5 43.b4 g5 44.Ra6+ Kf5 45.Ra5 Ke5 46.b5 Kd6 47.Ra7 g4 48.Rh7 Kc5 49.Rh8 Rf5
50.Rh6 Kb4 51.Ra6 h4 52.Kf1 Rf8 53.b6 f3 54.b7 Rb8 55.Ra7 Kc3 56.Kg1 Rd8 57.h3
g3 58.fxg3 hxg3 59.h4 Rd1# Black mates 0-1
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.