Discussion of anything and everything relating to chess playing software and machines.
Moderator: Ras
Henk
Posts: 7261Joined: Mon May 27, 2013 10:31 am
Post by Henk Wed May 20, 2015 4:55 pm
Evert wrote: Henk wrote: Evert wrote: Henk wrote: In a virtual world anything is possible. What's wrong with -1000. -1000 means -100 cps.
[d]r1bqkb1r/ppp1pppp/2np3n/4P3/3P4/5N1P/PPP2PP1/RNBQKB1R w KQkq - 1 2
You mean what's wrong with evaluating the above position as a pawn down for black? Other than black not being a pawn down?
Sorry mistake. I'm not used to thinking in centi pawns. It means 0.1 pawns = 10 cps
You have the value of a pawn set at 10000? Really?
No to be more precise:
Code: Select all
public static int PAWN_VALUE = 9 * 1024;
But it's value is updated dependent on the game phase. It's difficult to create a stupid engine.
Main problem is to get deeper searches without losing elo.
bob
Posts: 20943Joined: Mon Feb 27, 2006 7:30 pmLocation: Birmingham, AL
Post by bob Thu May 21, 2015 8:38 pm
Henk wrote: In this position Skipper plays Nf2.
[d] r1bqkb1r/ppp1pppp/2np4/4P3/3P2n1/5N1P/PPP2PP1/RNBQKB1R b KQkq - 0 1
[pgn]
1. e4 Nc6
2. Nf3 Nf6
3. e5 Ng4
4. d4 d6
5. h3 Nf2
[/pgn]
Code: Select all
Depth Value Time
1 -3500.00 0.002 86 g4h6
2 -4048.00 0.004 136 g4f2
3 -2148.00 0.009 421 g4f2
4 -2648.00 0.022 1685 g4f2
5 -1198.00 0.050 4042 g4f2
6 -2348.00 0.115 10215 g4f2 e1f2 d6e5 d4d5 c8e6 b1c3
7 -1646.00 0.328 27122 g4f2 e1f2 d6e5 d4d5 e5e4 f2e1 e7e5
8 -1648.00 0.890 78992 g4f2 e1f2 d6e5 d4d5 e7e6 b1c3 e6d5 c3d5
9 -2746.00 2.564 224170 g4f2 e1f2 d6e5 f1b5 e5e4 f3e5 c8d7 e5d7 d8d7
Strange when I load the position and try to reproduce it I get the normal:
Code: Select all
Depth Value Time
1 -1000.00 0.002 106 g4h6
2 -2000.00 0.005 282 g4h6 b1c3
3 -2252.00 0.013 1201 g4h6 c1h6 g7h6
4 -3252.00 0.029 2952 g4h6 c1h6 g7h6 b1c3
5 -2452.00 0.057 6140 g4h6 c1h6 g7h6 b1c3 d6e5
6 -3052.00 0.114 12947 g4h6 c1h6 g7h6 b1c3 d6e5 d4e5
7 -2552.00 0.293 31367 g4h6 c1h6 g7h6 b1c3 d6e5 d4e5 c8d7
8 -3052.00 0.761 84134 g4h6 c1h6 g7h6 b1c3 d6e5 d4e5 c8d7 f1e2
9 -2352.00 2.418 261383 g4h6 c1h6 g7h6 b1c3 c8d7 e5d6 c7d6 d4d5 c6b4
10 -2852.00 5.719 667175 g4h6 c1h6 g7h6 e5d6 c7d6 d4d5 c6e5 b1c3 f8g7
But that is not what it plays in a normal game. For I encounter this position very often. Difference in evaluation is not much so maybe I should reduce the penalty for losing castling rights a bit.
You think?
Henk
Posts: 7261Joined: Mon May 27, 2013 10:31 am
Post by Henk Thu May 21, 2015 9:19 pm
I got this position again today. I almost would believe Nf2 is the best move in this position. Maybe penalty for h3 is too big. I don't know. Already lowered penalty for castling rights and king not at border.
Sven
Posts: 4052Joined: Thu May 15, 2008 9:57 pmLocation: Berlin, GermanyFull name: Sven Schüle
Post by Sven Sun May 24, 2015 8:16 pm
Henk wrote: I got this position again today. I almost would believe Nf2 is the best move in this position. Maybe penalty for h3 is too big. I don't know. Already lowered penalty for castling rights and king not at border.
1...Nf2 is certainly *not* the best move in this position, it simply loses a minor piece vs. two pawns after 2.Kxf2 dxe5 3.Bb5 exd4 4.Bxc6+ bxc6 5.Qxd4 (or similar) while 1...Nh6 "only" leaves Black with a major positional disadvantage, for instance after 2.Nc3 g6 3.Bf4 dxe5 4.dxe5 Bg7 5.Bb5 0-0 6.0-0 a6 7.Bxc6 Qxd1 8.Raxd1 bxc6 9.Na4.
The real problem in this opening is 3...Ng4? which is already a losing move. Even 2...Ng6 is quite unusual although after 3.e5 Nd5 Black is not hopeless at least. White has a slight advantage, as always.
Maybe you should check your eval scoring for pawns in the center?
zd3nik
Posts: 193Joined: Wed Mar 11, 2015 3:34 amLocation: United States
Post by zd3nik Sun May 24, 2015 8:57 pm
Henk wrote: In this position Skipper plays Bd2 and thinks his position is worse -0.32. What do normal engines play here ? How smart is your engine ?
[d] r1bq1rk1/pppn1ppp/2n1p3/2PpP3/Pb1P4/2NB1N2/1P3PPP/R1BQK2R w KQ - 1 2
Code: Select all
Depth Value Time
1 -4100.00 0.018 394 c1d2
2 -4500.00 0.028 1917 c1d2 a7a5
3 -4100.00 0.042 3624 c1d2 h7h5 h2h4
4 -4300.00 0.108 15478 h2h4 h7h6 c1d2 a7a5
5 -4100.00 0.257 39956 h2h4 h7h6 c1d2 a7a5 b2b3
6 -4100.00 0.563 101565 h2h4 h7h6 c1d2 a7a5 h4h5 d8e7
7 -4100.00 2.068 404468 h2h4 h7h6 c1d2 a7a6 b2b3 h6h5 d1e2
8 -3900.00 9.292 1816195 h2h4 h7h6 h4h5 d8e7 c1d2 a7a5 b2b3 g8h8
9 -3200.00 40.067 7705559 c1d2 h7h5 c3a2 a7a5 d2c3 d8e7 d3b5 b4c3 b2c3
I feel your pain. Having written about a dozen chess engines over the years (all original code, but very few original ideas) I have experienced the frustration of seeing my engines miss tactics that other engines seem to find without any effort at all. I found the culprit was more often bugs in the code than bad design.
Is skipper open-source? I'd be interested in seeing how quiescence search was implemented since check and capture extensions in qsearch are essential in positions like this.
STC
Sven
Posts: 4052Joined: Thu May 15, 2008 9:57 pmLocation: Berlin, GermanyFull name: Sven Schüle
Post by Sven Sun May 24, 2015 11:54 pm
Sven Schüle wrote: Even 2...Ng6
Nf6 of course
rabbits23
Posts: 144Joined: Fri Sep 12, 2014 4:57 amLocation: Randwick Australia
Post by rabbits23 Mon May 25, 2015 3:58 am
Analysis by Rybka 4 960 x64:
