Slow Chess 2.94

[Jonathan Kreuzer]

Geschrieben von:/Posted by: Jonathan Kreuzer at 13 September 2004 07:50:30:

I was just going to keep this one private until I had time to work on it more and hopefully get some of the things I wanted to do done, but it will be many months until then, so here's a link to Slow Chess 2.94 for anyone who likes testing engines.
www.3dkingdoms.com/chess/slow294.zip
Despite the fact this one isn't released on the page, I've hopefully I've fixed more bugs than I created. I've only played a few test games to make sure it seems to run okay, though I'd guess 2.94 is slightly stronger than 2.93.
The main thing I worked on was rewriting code for clarity in case I choose to do a source release next version. The assembly move generation was converted to simple C code with no speed loss. But there have been many other changes.
-Various Speed Tweaks (~8% faster on my computer, I'm not sure if this holds for other computers.)
-Eval Changes, for code clarity and maybe more pleasing play.
-A few time control changes, mainly for aesthetic reasons.
ie. sometimes recaps quicker, or at strange times (something to do with fail-lows) it will think for a longer time.
-Fixed a couple bugs

[Thomas Lagershausen]

Geschrieben von:/Posted by: Thomas Lagershausen at 13 September 2004 16:37:57:
Als Antwort auf:/In reply to: Slow Chess 2.94 geschrieben von:/posted by: Jonathan Kreuzer at 13 September 2004 07:50:30:

Thx Jonathan,
your program is tactical really brilliant.
If you would teach him more positional things like the following two it would play like a grandmaster.
r2q1r1k/pppbn2p/6pb/2pPpp1n/4P3/1PN2NP1/PB3PBP/2RQR1K1 w - - 0 1
after 1.Nxe5 ! Bxc1 2.Qxc1 followed by Qh6 and the white bishop on b2 rules on the long diagonal a1-h8 and the white centralpawns will dominate the game.
r2q1rk1/pb1nbp1p/1p4p1/3p2Pn/2pP4/2N1PN2/PPQBBP2/1K1R3R w - - 0 2
with 1.Rxh5! gxh5 2.Rh1 Re8 3.Rxh5 Nf8 4.Ne5 is white playing a winning attack on the kingside.
Greetings
Thomas

[Jonathan Kreuzer]

Geschrieben von:/Posted by: Jonathan Kreuzer at 14 September 2004 06:03:18:
Als Antwort auf:/In reply to: Re: Slow Chess 2.94 geschrieben von:/posted by: Thomas Lagershausen at 13 September 2004 16:37:57:

Thx Jonathan,
your program is tactical really brilliant.
If you would teach him more positional things like the following two it would play like a grandmaster.
r2q1r1k/pppbn2p/6pb/2pPpp1n/4P3/1PN2NP1/PB3PBP/2RQR1K1 w - - 0 1
after 1.Nxe5 ! Bxc1 2.Qxc1 followed by Qh6 and the white bishop on b2 rules on the long diagonal a1-h8 and the white centralpawns will dominate the game.
r2q1rk1/pb1nbp1p/1p4p1/3p2Pn/2pP4/2N1PN2/PPQBBP2/1K1R3R w - - 0 2
with 1.Rxh5! gxh5 2.Rh1 Re8 3.Rxh5 Nf8 4.Ne5 is white playing a winning attack on the kingside.
Greetings
Thomas

Hi, Thanks for your interest. Even after over a year since the original release, there are still some types of positions that Slow Chess evaluates very wrong that I want to fix before really trying to tweak the eval for best play.
When I'm working on Slow again, king safety is the main evaluation element I want to rewrite for the next version. With more accurate king safety I hope Slow might be a more consistently effective attacker (and avoid some of the weird evals caused by misidentifying whether or not a king is in danger.) I'm not sure if it will solve the 2nd position but I think white's score will be higher on the attacking line.

[Thomas Lagershausen]

Geschrieben von:/Posted by: Thomas Lagershausen at 14 September 2004 11:02:31:
Als Antwort auf:/In reply to: Re: Slow Chess 2.94 geschrieben von:/posted by: Jonathan Kreuzer at 14 September 2004 06:03:18:

Thx Jonathan,
your program is tactical really brilliant.
If you would teach him more positional things like the following two it would play like a grandmaster.
r2q1r1k/pppbn2p/6pb/2pPpp1n/4P3/1PN2NP1/PB3PBP/2RQR1K1 w - - 0 1
after 1.Nxe5 ! Bxc1 2.Qxc1 followed by Qh6 and the white bishop on b2 rules on the long diagonal a1-h8 and the white centralpawns will dominate the game.
r2q1rk1/pb1nbp1p/1p4p1/3p2Pn/2pP4/2N1PN2/PPQBBP2/1K1R3R w - - 0 2
with 1.Rxh5! gxh5 2.Rh1 Re8 3.Rxh5 Nf8 4.Ne5 is white playing a winning attack on the kingside.
Greetings
Thomas

Hi, Thanks for your interest. Even after over a year since the original release, there are still some types of positions that Slow Chess evaluates very wrong that I want to fix before really trying to tweak the eval for best play.
When I'm working on Slow again, king safety is the main evaluation element I want to rewrite for the next version. With more accurate king safety I hope Slow might be a more consistently effective attacker (and avoid some of the weird evals caused by misidentifying whether or not a king is in danger.) I'm not sure if it will solve the 2nd position but I think white's score will be higher on the attacking line.

A good inspiration for you could be the book "Storming the Barricades" by GM Larry Christiansen (Gambit Publications 2000).
The games and the notes are very proffesional. I like this game from the book very much:
[Event "Memorial M.Chigorin"]
[Site "St. Petersburg (Russia)"]
[Date "1993.??.??"]
[Round "3"]
[White "Serper"]
[Black "Nikolaidis"]
[Result "1-0"]
[Termination "normal"]
[PlyCount "95"]
[WhiteType "program"]
[BlackType "program"]
1.c4 g6 2.e4 Bg7 3.d4 d6 4.Nc3 Nf6 5.Nge2 Nbd7 6.Ng3 c6 7.Be2 a6 8.Be3 h5 9.f3
b5 10.c5 dxc5 11.dxc5 Qc7 12.O-O h4 13.Nh1 Nh5 14.Qd2 e5 15.Nf2 Nf8 16.a4 b4
17.Nd5 cxd5 18.exd5 f5 19.d6 Qc6 20.Bb5 axb5 21.axb5 Qxb5 22.Rxa8 Qc6 23.Rfa1 f4
24.R1a7 Nd7 25.Rxc8+ Qxc8 26.Qd5 fxe3 27.Qe6+ Kf8 28.Rxd7 exf2+ 29.Kf1 Qe8 30.Rf7+ Qxf7
31.Qc8+ Qe8 32.d7 Kf7 33.dxe8=Q+ Rxe8 34.Qb7+ Re7 35.c6 e4 36.c7 e3 37.Qd5+ Kf6 38.Qd6+
Kf7 39.Qd5+ Kf6 40.Qd6+ Kf7 41.Qxe7+ Kxe7 42.c8=Q Bh6 43.Qc5+ Ke8 44.Qb5+ Kd8
45.Qb6+ Kd7 46.Qxg6 e2+ 47.Kxf2 Be3+ 48.Ke1 {} 1-0
I hope Slowchess 2.95 will find 20.Bb5!, because it is a beautifull move and a winner. At the moment only Junior and Tao in the computerchessworld are finding 17.Nd5 !!
So the future will be very interesting.
All the best
Thomas