Chess Programmer's Tournament 2009

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

bob
Posts: 20943
Joined: Mon Feb 27, 2006 7:30 pm
Location: Birmingham, AL

Re: Chess Programmer's Tournament 2009

Post by bob »

Here's mine:

Code: Select all

               21->   1:39   0.45   1. fxe7 Qxe7 2. Rd3 Bc6 3. Rh3 Bxe4
                                    4. Qe3 g5 5. Rhf3 Rf8 6. Rf6 Qe8 7.
                                    Qd2 Bg6 8. Re2 Qd8 9. Qd4 Re8 10. Rxe8
                                    Qxe8 11. g4
               22     4:06   0.15   1. fxe7 Qxe7 2. Rd3 Bc6 3. Rh3 Bxe4
                                    4. Qe3 g5 5. Rhf3 Rf8 6. Rf6 Kg7 7.
                                    b4 Qe5 8. c5 dxc5 9. bxc5 bxc5 10.
                                    Rxa6 Qa1+ <HT>
               22->   4&#58;30   0.15   1. fxe7 Qxe7 2. Rd3 Bc6 3. Rh3 Bxe4
                                    4. Qe3 g5 5. Rhf3 Rf8 6. Rf6 Kg7 7.
                                    b4 Qe5 8. c5 dxc5 9. bxc5 bxc5 10.
                                    Rxa6 Qa1+ <HT>
               23     5&#58;08   0.28   1. fxe7 Qxe7 2. Rd3 Bc6 3. Rh3 Bxe4
                                    4. Qe3 g5 5. Rhf3 Rf8 6. Rf6 Kg7 7.
                                    a3 Qe5 8. Qh3 Qa1+ 9. Rf1 Qd4+ 10.
                                    Kh1 Bg6 11. Qf3 Qb2 12. b4 Qd4 13.
                                    Re6 Qxc4
              time=5&#58;26  mat=-3  n=5982697338  fh=91%  nps=18.3M
              ext-> check=161.0M qcheck=179.0M reduce=3121.1M/661.7M
              predicted=0  evals=309.9M  50move=0  EGTBprobes=0  hits=0
              SMP->  splits=421853  aborts=52460  data=59/512  elap=5&#58;26
hardware is a "bit" faster of course. :)
Top
User avatar
Eelco de Groot
Posts: 4567
Joined: Sun Mar 12, 2006 2:40 am
Full name:   

Re: Chess Programmer's Tournament 2009

Post by Eelco de Groot »

Gerd Isenberg wrote: [d] 1r6/2pbrp1k/pp1p1Ppp/3Rq3/2P1P3/1P6/P4RPP/4Q1K1 w - - 0 32
What do other programs prefer here?
It is probably an artefact of the search here but Blueberry has a slight preference for first playing Qf1, there is no safe place for the queen it seems, so on the second move White can still take it?

No fxe7 but these three moves are close together, Rybka 2.2n2 plays Rxe5, Naum 3.1 Qf1.

I'm not sure the game is always, or mosttimes, lost with a Queen against Rook and Bishop and a pawn structure like this, would like to see a statistical analysis, in some cases it might be a forceable win though, from a couple of shoot-outs played by Naum 3.1 there were two draws and one loss for White.

[D]1r6/2pbrp1k/pp1p1Ppp/3Rq3/2P1P3/1P6/P4RPP/4Q1K1 w - -

Engine: Blueberry Beta 4 DM70 Build 322 (256 MB)
by F. Letouzey, T. Gaksch, E. de Groot

1/05 0:00 +2.10 1.Rxe5 Rxe5 2.Re2 (198)

2/10 0:00 +2.12 1.Rxe5 Rxe5 2.Re2 Rbe8 (1.599)

3/23 0:00 +2.12 1.Rxe5 Rxe5 2.Re2 Rbe8 (70.647)

4/31 0:00 +2.08 1.Rxe5 Rxe5 2.h4 Bg4 3.Rf4 Bd7 (132.180)

5/32 0:00 +2.00 1.Rxe5 Rxe5 2.h4 Bg4 3.Rf4 Bd7 4.Rf3 Rbe8 (260.126)

6/42 0:02 +0.39 1.Rxe5 Rxe5 2.Re2 Bc6 3.h4 Rbe8 4.h5 Bxe4
5.Re3 gxh5 6.Qf2 Bf5 7.Qg3 (1.502.865) 590

6/42 0:16 +0.47 1.Qf1 Rf8 2.Rxe5 Rxe5 3.Qd1 Rfe8
4.Rf1 Bc6 5.h4 Bxe4 6.g4 Bc6 7.Qd2 (8.837.725) 540

6/42 0:18 +0.47 1.Qf1 Rf8 2.Rxe5 Rxe5 3.a4 (10.099.230) 551

6/48 0:22 +0.57 1.Qf1 Rf8 2.Re2 Rfe8 3.h4 Qc3 4.fxe7 Rxe7
5.e5 Bc6 6.Rd1 Qb4 7.exd6 Qc5+ 8.Kh1 Rxe2
9.Qxe2 cxd6 10.Qe7 (13.136.481) 573

7/48 0:26 +0.54 1.Qf1 Ree8 2.Rxe5 Rxe5 3.h4 Rbe8
4.Qd3 Bc6 5.Qh3 h5 6.Rd2 Rxe4 7.a4 Re2 (15.161.436) 576

8/48 1:05 +0.26 1.Qf1 Ree8 2.Rxe5 Rxe5 3.h4 Rbe8
4.Qd3 Bc6 5.Rd2 Bxe4 6.Qd4 Bf5 7.g3 Re2
8.Qd5 Re1+ 9.Kf2 (38.367.021) 582

9/52 10:56 +0.26 1.Qf1 Ree8 2.Rxe5 Rxe5 3.Re2 Rbe8
4.Re3 Rxe4 5.Rxe4 Rxe4 6.h3 a5 7.Qf2 a4
8.Kh2 b5 9.Qf3 axb3 10.axb3 (372.769.085) 567

10/52 18:25 +0.28 1.Qf1 Ree8 2.Rxe5 Rxe5 3.Re2 Rbe8
4.Qf2 Bc6 5.g4 Rxe4 6.Rxe4 Rxe4 7.h3 Re5 (614.549.486) 556

11/52 19:36 +0.28 1.Qf1 Ree8 2.Rxe5 Rxe5 3.Re2 Rbe8
4.Qf2 Bc6 5.h4 Rxe4 6.Rxe4 Rxe4 7.g3 Re5 (653.948.038) 555

best move: Qe1-f1 time: 66:56.860 min n/s: 508.425 CPU 100.0% n/s(1CPU): 508.425 nodes: 2.042.080.000

Later builds are a bit screwed up by experiments with the output, I hope I can reconstruct Build 322. I was trying to get the engine to output some sort of draw evaluation if there is no progress in the score, but I'm not so very good with strings, it slowed the engine down too much and I haven't yet found a good way to recognize string contents. All I would like is a = symbol at the end or beginning of a PV but then the GUI would start complaining about illegal moves :?


I would like to do something like this, the output is from Build 322, for the study by Timman but the = signs are post-added by me:

8/8/8/1p1k4/3P4/4b3/KPP5/8 w - -

Engine: Blueberry Beta 4 DM70 Build 322 (256 MB)
by F. Letouzey, T. Gaksch, E. de Groot

7/17 0:00 -0.90 1.c3 Bg5 2.Kb3 Be7 3.Kc2 Ke4 4.b3 Bd6
5.c4 (81.726)

8/17 0:00 -0.90 1.c3 Bg5 2.Kb3 Be7 3.Kc2 Ke4 4.b3 Bd6
5.c4 (148.344)

9/21 0:00 -1.03 1.c3 Kc4 2.Ka3 Bf4 3.Ka2 Bd6 4.Ka1 Kd5
5.Ka2 Bc7 6.Kb3 (349.216)

10/22 0:00 -1.03 1.c3 Kc4 2.Ka3 Bf4 3.Ka2 Bd6 4.Ka1 Kd5
5.Ka2 Bc7 6.Kb3 (484.988)

11/22 0:00 -1.12 1.c3 Kc4 2.Ka3 Bf4 3.Ka2 Bg5 4.Ka3 Be7+
5.Ka2 Kd5 6.b3 Ke4 7.c4 (695.234)

12/33 0:02 -2.09 1.c3 Kc4 2.Ka3 Bf4 3.Ka2 Bd6 4.Kb1 Kb3
5.Ka1 Bc7 6.Kb1 Bf4 7.d5 Bd6 8.Kc1 Kc4
9.Kc2 Kc5 10.Kd3 Kxd5 (1.952.281) 785

12/39 0:09 -2.08 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Bg3 7.Kd3 Kc5 8.b3 Bf4
9.Ke4 Bd6 10.Kd3 (8.593.195) 900

12/39 0:09 -2.08 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Bg3 7.Kd3 Kc5 8.b3 Bf4
9.Ke4 Bd6 10.Kd3 (8.696.520) 900

12/39 0:12 -2.08 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Bg3 7.b3 Bf4 8.Kd3 Kc5
9.Ke4 Bd6 10.Ke3 (11.009.838) 907

13/39 0:16 -2.08 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Bg3 7.b3 Bf4 8.Kd3 Kc5
9.Ke4 Bc7 10.Ke3 Kd6 (13.706.978) 848

14/41 0:34 -2.08 1.Kb3 Bxd4 2.Kb4 Kc6 3.Kb3 Bc5 4.Kc3 Kd5
5.Kd3 Bd4 6.c3 Be5 7.b3 Kc6 8.c4 b4
9.Ke4 Bd6 (26.346.892) 765

15/47 1:14 -2.07 1.Kb3 Bxd4 2.Kb4 Kc6 3.Kb3 Bc5 4.Kc3 Kd5
5.Kd3 Bd4 6.c3 Be5 7.b3 Kc6 8.Kd2 Bd6 (52.459.949) 703

16/47 2:08 -2.08 1.Kb3 Bxd4 2.Kb4 Kc6 3.Kb3 Bc5 4.Kc3 Kd5
5.Kd3 Bd4 6.c3 Be5 7.Kc2 Ke4 8.Kd2 Bd6
9.Ke2 Bf4 (86.809.494) 676

17/47 2:21 -2.08 1.Kb3 Bxd4 2.Kb4 Kc6 3.Kb3 Bc5 4.Kc3 Kd5
5.Kd3 Bd4 6.c3 Be5 7.b3 Kc6 8.Kc2 Bd6 (96.147.057) 679

18/53 3:23 -2.08 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Bc5+ 7.Ke4 b3 8.Kd3 Bb4
9.Kd4 Bd2 (137.910.842) 677

19/53 5:15 -2.08 = 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Bf4 7.Ke4 Bg3 8.b3 Bb8
9.Kd3 Bc7 10.Ke3 Kc5 11.Kd3 Be5
12.Ke3 Bd4+ (223.161.952) 706

20/55 9:25 -2.08 = 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Be7 7.b3 Kd6 8.Ke4 Kc5
9.Ke5 Bg5 10.Ke4 Bh6 11.Kd3 Kd6 (399.614.582) 706

21/67 23:48 -2.08 = 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Be7 7.b3 Kd6 8.Ke4 Kc5
9.Ke5 Bg5 10.Ke4 Bh6 11.Kd3 Bf4
12.Ke4 Bd6 (989.951.247) 693

22/67 25:44 -2.08 = 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Be7 7.b3 Kd6 8.Ke4 Kc5
9.Ke5 Bg5 10.Ke4 Bh6 11.Kd3 Bf4
12.Ke4 Bd6 (1.122.752.816) 726

23/71 30:17 -2.08 = 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Be7 7.b3 Kd6 8.Ke4 Kc5
9.Ke5 Bg5 10.Ke4 Bh6 11.Kd3 Bf4
12.Ke4 Bg5 13.Ke5 Kc6 (1.439.109.655) 791

24/75 50:24 -2.08 = 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Be7 7.b3 Kd6 8.Ke4 Kc5
9.Ke5 Bg5 10.Ke4 Bh6 11.Kd3 Bf4
12.Ke4 Bg5 13.Ke5 Kc6 (3.038.336.678) 1004

25/85 78:55 -2.08 = 1.Kb3 Bxd4 2.Kb4 Kc6 3.c4 Bc5+ 4.Kc3 b4+
5.Kd3 Bd6 6.Kd4 Be7 7.b3 Kd6 8.Ke4 Kc5
9.Ke5 Bg5 10.Ke4 Bh6 11.Kd3 Bf4
12.Ke4 Bg5 13.Ke5 Be3 14.Ke4 (5.184.573.698) 1094

For the moment I haven given up on this, I know it can be done by using the UCI 'string' command, hopefully also before sending the PV info to the Shredder GUI, it should be doable even for a non-programmer like me but first I have to reconstruct my precious Build Numero 322 8-)

Eelco
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan
Top
Richard Allbert
Posts: 792
Joined: Wed Jul 19, 2006 9:58 am

Re: Chess Programmer's Tournament 2009

Post by Richard Allbert »

Hi Richard,

Thanks very much for organising the tournament - I had a great time.

Although Lime isn't the strongest engine ("mid tournament change" induced bugs don't help) it was a lot of fun.

The main reason for attending was to finally meet the people involved - I'm glad I did.

Everyone is was nice, helpful, and all together a real mix of character!! I recevied a lot of good advice, which will prove very helpful in trying to make Lime something other than a weak engine.

I would highly recommend taking part in your tournaments.

See you in Leiden (hopefully!!).

Thanks again, and thanks to everyone else there,

Richard
Top

Return to “Computer Chess Club: General Topics”