Near to the Top

[Dann Corbit]

Geschrieben von:/Posted by: Dann Corbit at 17 February 2000 00:14:09:
Als Antwort auf:/As an answer to: Re: Near to the Top geschrieben von:/posted by: Colin Frayn at 17 February 2000 00:05:56:

Hmmm.. I fear I may be a little out classed there

I've erased most of my weaker engines, which makes it a little difficult for ColChess.

Well it'll be interesting. ColChess isn't at all bad for a brute force engine, but considering it doesn't use null move or search extensions, and I haven't even begun to add in futility pruning and razoring, it should get thrashed by all of these other engines who are at least 200-400 ELO higher. Oh, and I use 8*8 board notation because it's easier to read and understand, but unfortunately that gives me a slow move generation function :{
Still - thanks for trying it out - I'll be interested to see if it manages to get a draw....

The latest version is actually pretty strong, I think. I predict it will finish towards the middle of the pack. You are a very promising chess programmer. If you put in the effort, you could probably come up with a world-beater.
It would be nice if you could modify ColChess to do standard EPD output. Then I could use it to double check some non-checkmate chess problems. Since ColChess does not use NULL-move, it will find some answers to problems that NULL-move pruners miss.


My FTP site

[Colin Frayn]

Geschrieben von:/Posted by: Colin Frayn at 17 February 2000 12:01:47:
Als Antwort auf:/As an answer to: Re: Near to the Top geschrieben von:/posted by: Dann Corbit at 17 February 2000 00:14:09:

Hmmm.. I fear I may be a little out classed there

The latest version is actually pretty strong, I think. I predict it will
finish towards the middle of the pack
If you put in the effort, you could probably come up with a world-beater.
It would be nice if you could modify ColChess to do standard EPD output
Then I could use it to double check some non-checkmate chess problems. Since >ColChess does not use NULL-move, it will find some answers to problems that >NULL-move pruners miss.

I'm extremely grateful for your praise and optimism, but looking at the ELO ratings I don't think it will win any matches in this one. I know that it only gets about 1 point out of 30 against GNUChess v4 and I'm not sure what GNUChess 5 is like. The others all seem to be around the same level.
However, in the large low level tournament it should do rather well - perhaps in the top third. Won't know until I see the first few game results.
If you only knew how much of my time I already spend coding
I don't think that ColChess will ever be up to the level of, say, GNUChess and its contemporaries because of some of my methods and the fact that I like to keep it brute force. However I think that if I add in extensions and perhaps some frontier node futility pruning then maybe it'll not be far off. My goal is 220 in WAC before the end of the year. I think it should be manageable.
I'm not sure what you have in mind, but there's a command 'dfen'
which gived a fen code for the current position.
Yeah that's true, though at the moment it might not be that useful for it. Hopefully in a few weeks time when I've managed to add in some of these ideas it'll be performing a lot better.
Cheers,
Colin

ColChess Homepage

[Mogens Larsen]

Geschrieben von:/Posted by: Mogens Larsen at 17 February 2000 13:39:45:
Als Antwort auf:/As an answer to: Re: Near to the Top geschrieben von:/posted by: Colin Frayn at 17 February 2000 12:01:47:

I'm extremely grateful for your praise and optimism, but looking at the ELO ratings I don't think it will win any matches in this one. I know that it only gets about 1 point out of 30 against GNUChess v4 and I'm not sure what GNUChess 5 is like. The others all seem to be around the same level.
However, in the large low level tournament it should do rather well - perhaps in the top third. Won't know until I see the first few game results.

GNUChess 5 seems to be significantly weaker than v4.14, at least concerning faster games. There might be a point to be gained. Ant, EXChess and Inmichess are leading with 2.5/3, which isn't surprising. I believe there's a lot of difference in strength, just not that noticeable at 40/5min.
I can tell you that ColChess just got its first point with a win against SSEChess. SSEChess overlooked an attack from which it wasn't able to recover. If it can get half a point against GNUChess (if they meet) then the program is doing better than your worst fears ).
Best wishes...
Mogens

[Colin Frayn]

Geschrieben von:/Posted by: Colin Frayn at 17 February 2000 16:52:11:
Als Antwort auf:/As an answer to: Re: Near to the Top geschrieben von:/posted by: Mogens Larsen at 17 February 2000 13:39:45:

I can tell you that ColChess just got its first point with a win against SSEChess. SSEChess overlooked an attack from which it wasn't able to recover.

You have no idea how overjoyed I am to hear that
That's inspired me to get version v6 out as soon as possible. Lots of ideas - not enough time to code them.
Thanks a lot,
Colin

ColChess Homepage

[Frank Quisinsky]

Geschrieben von:/Posted by: Frank Quisinsky at 17 February 2000 17:06:36:
Als Antwort auf:/As an answer to: Re: Near to the Top geschrieben von:/posted by: pete at 16 February 2000 17:09:21:
Hallo,

Frank , hast Du schon Zeit gefunden , die neue Bringer-Version zu testen ? Sie ist denke ich mittlerweile stärker , als Du vermutest ; hier als kleiner Appetithappen ein nettes Spielchen gegen Diepx ( Dual Pentium 500 irgendwas ) + ein netter Blitzsieg gegen Hossa ( zum Vergnügen ) ; nach den Resultaten auf ICC scheint Bringer zu Postmodernist aufgeschlossen zu haben ( immer so ca. 50 %) und wenn man mal die Überflieger-Winboardprogramme Crafty , LGG , Comet , Insomniac + AnMon und evtl. Gromit ( in längeren Partien ) abzieht ( vielleicht habe ich eins vergessen :=) ) zur erweiterten Spitzengruppe zu gehören :
Während frühere Bringer-Versionen besondere Stärken im Blitzspiel hatten , habe ich den Eindruck dass die aktuelle Version etwas längere Bedenkzeit braucht , um richtig auf Touren zu kommen . Der Spielstil ist meistens recht ruhig und die meisten Punkte werden in Stellugen mit geklärtem Zentrum erzielt .

Nein, derzeit lasse ich AnMon und Comet ein wenig über Nacht blitzen. Wenn ich meine beabsichtigten Berichte abgeschlossen habe, werde ich wieder ein WinBoard Turnier mit Bringer spielen.
Es ist gut möglich das Gerrit etwas gefunden hat. Die Bringer Version 1.5 war sehr stark, nach den Ergebnissen zu urteilen (auf meiner Webseite) war die Version 1.6 Beta eher etwas schwächer.
Ja genau, im Blitz sind die älteren Bringer Versionen Super stark. Bei längeren Bedenkzeiten auch aber da fällt Bringer etwas ab. Das mit dem Zentrum ist interessant, ich habe mir nur wenige Bringer Partien intensiv angesehen. Hier stellte ich fest, daß Bringer sehr ausgeglichen spielt und im Endspiel etwas nachläßt. Auch taktisch gibt es gute Kombinationen.
Im Grunde bin ich derzeit aber nicht unbedingt auf dem neuesten Stand. Das wird sich ändern )
Gruß
Frank

[Dann Corbit]

Geschrieben von:/Posted by: Dann Corbit at 17 February 2000 20:05:43:
Als Antwort auf:/As an answer to: Re: Near to the Top geschrieben von:/posted by: Colin Frayn at 17 February 2000 12:01:47:
[snip]

It would be nice if you could modify ColChess to do standard EPD output

I'm not sure what you have in mind, but there's a command 'dfen'
which gived a fen code for the current position.

Some chess programs can process EPD (which is a subset of the PGN Standard by Steven J. Edwards). I use this data in the Chess Analysis Project.

Here are some examples of what I mean:
Input for Crafty:
r1bq1rk1/ppn1ppbp/2p3p1/4P3/3P4/1QN2N2/PP3PPP/R1B2RK1 b - -
r2q1rk1/ppn1ppbp/2p3p1/4P3/3P2b1/1QN2N2/PP3PPP/R1B2RK1 w - -
r1bq1rk1/p1n1ppbp/1pp3p1/4P3/3P4/1QN2N2/PP3PPP/R1B2RK1 w - -
r1b2rk1/ppnqppbp/2p3p1/4P3/3P4/1QN2N2/PP3PPP/R1B2RK1 w - -
r1bq1rk1/pp2ppbp/2p3p1/3nP3/3P4/1QN2N2/PP3PPP/R1B2RK1 w - -
rb1q1rk1/pp2ppbp/2p3p1/3nP3/3P4/1QN2N2/PP3PPP/R1B2RK1 w - -
Output from Crafty:
r1bq1rk1/ppn1ppbp/2p3p1/4P3/3P4/1QN2N2/PP3PPP/R1B2RK1 b - - acd 13; acn 121730605; acs 1000; ce 35; pv Qd7 h3 Rd8 Bg5 h6 Bh4 Re8 Qc4 Qe6 Qd3 Bd7 Rac1;
r2q1rk1/ppn1ppbp/2p3p1/4P3/3P2b1/1QN2N2/PP3PPP/R1B2RK1 w - - acd 14; acn 133018260; acs 1001; ce 5; pv Qxb7 Bxf3 gxf3 Ne6 d5 Rb8 Qxa7 cxd5 f4 d4 Rd1 Ra8 Qb7 d3;
r1bq1rk1/p1n1ppbp/1pp3p1/4P3/3P4/1QN2N2/PP3PPP/R1B2RK1 w - - acd 12; acn 119806627; acs 1000; ce -34; pv Be3 Be6 Qc2 Nd5 Qd2 Nxc3 Qxc3 Bd5 Rac1 Rc8 Rfe1 e6 Bg5;
r1b2rk1/ppnqppbp/2p3p1/4P3/3P4/1QN2N2/PP3PPP/R1B2RK1 w - - acd 12; acn 116385354; acs 1001; ce -42; pv Bg5 h6 Be3 Rd8 Qd1 Qe8 Qd2 Qf8 Ne4 Bf5 Nc5 Rab8;
r1bq1rk1/pp2ppbp/2p3p1/3nP3/3P4/1QN2N2/PP3PPP/R1B2RK1 w - - acd 14; acn 135875436; acs 1001; ce -30; pv Nxd5 Qxd5 Qxd5 cxd5 Bd2 Bd7 Bb4 Rfe8 Rfc1 Bc6 Rc2 e6 Rac1 Rac8 h3 h6;
rb1q1rk1/pp2ppbp/2p3p1/3nP3/3P4/1QN2N2/PP3PPP/R1B2RK1 w - - acd 14; acn 110410305; acs 1001; ce 101; pv Qxb7 Qc8 Qb3 Qd7 Nxd5 cxd5 Bf4 Bc7 Qe3 Rab8 Rab1 Rfc8 Bh6 Bh8;

Input for Chest:
1B1Q1R2/8/qNrn3p/2p1rp2/Rn3k1K/8/5P2/bbN4B w - -
1B2K3/p2p1ppQ/N3k3/P1R1p2N/3p1b2/3P2p1/6P1/8 w - -
1B4k1/4pp2/p5pp/1r5P/2n3q1/2P2Q2/P1P3P1/2KR3R b - -
1B6/3K3p/5p2/pp2pr2/N2k3q/1Q5p/2R2b2/5N2 w - -
1B6/4QK2/r3P3/5kNn/2N1pr2/3P2pP/4Rq2/3n1R2 w - -
Output from Chest:
1B1Q1R2/8/qNrn3p/2p1rp2/Rn3k1K/8/5P2/bbN4B w - - acn 189; acs 0; bm Qf6; ce 32764; dm 2; pv Qf6 h5 Qg5#;
1B2K3/p2p1ppQ/N3k3/P1R1p2N/3p1b2/3P2p1/6P1/8 w - - acn 222; acs 0; bm Rb5; ce 32764; dm 2; pv Rb5 g6 Qxf7#;
1B4k1/4pp2/p5pp/1r5P/2n3q1/2P2Q2/P1P3P1/2KR3R b - - acn 905; acs 0; bm Qg5+; ce 32760; dm 4; pv Qg5+ Qf4 Rb1+ Kxb1 Qb5+ Kc1 Qb2#;
1B6/3K3p/5p2/pp2pr2/N2k3q/1Q5p/2R2b2/5N2 w - - acn 475; acs 0; bm Nc3; ce 32764; dm 2; pv Nc3 h2 Qd5#;
1B6/4QK2/r3P3/5kNn/2N1pr2/3P2pP/4Rq2/3n1R2 w - - acn 390; acs 0; bm Nf3; ce 32764; dm 2; pv Nf3 g2 Qg5#;
[snip]

My FTP site

[Mogens Larsen]

Geschrieben von:/Posted by: Mogens Larsen at 18 February 2000 00:05:04:
Als Antwort auf:/As an answer to: Re: Near to the Top geschrieben von:/posted by: Mogens Larsen at 16 February 2000 16:48:47:
Here is the standing after the first 6 rounds of my medium-tournament:
1. Ant4.16: 5 (not that surprising, but undefeated)
2. Dragon3.11: 4.5 (surprising?)
3. KnightX1.51: 4
4. EXchess3.11, Inmichess3.01, LDblanche2.0c: 3.5
5. Fortress1.61, Arasan5.3, LambChop7.10: 3
6. SSEchess2.02II: 2.5
7. GNUchess5, Cilian4.13: 2
8. Colchess5.4: 1.5
9. Sjeng5c: 1
This is how it looks so far. The last couple of rounds will be finished tomorrow.
Best wishes...
Mogens