Chess Variant tournaments: Crazyhouse, etc.

enhorning · Post by **enhorning** » Sat Dec 22, 2012 4:11 am

Crazyhouse

I'm going to run a Crazyhouse tournament - expected start in about a week, maybe slightly sooner (depends on when stuff I'm currently running ends) - using Winboard 4.6.2.

1-cpu (on i-970), 512 MB Hash, 40 moves in 20 minutes repeating (30s / move average).

Format will be 10-fold Round Robin, with each program playing both sides after the following 5 opening moves:
1. d4
1. e4
1. e3
1. Nc3
1. Nf3

My preliminary list of programs is:
Imortal 1.0 (Deuterium)
KKFChess v2.6.6
Sjeng 11.2
Sunsetter 7e
TJChess 1.1

Originally, I had planned to include Pulsar as well, but as it kept losing games due to illegal moves in my blitz test tournament, I decided to drop it from the list of participants. It lost 9 games out of 30 in the test tournament either due to illegal moves or due to falsely claiming the opponent's move was illegal (ignoring the fact that its own king was in check, in the games I checked up on).

If anybody is curious, here are the results of the blitz:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Sunse TJche Imort Sjeng Pulsa KKFCh
 1. Sunsetter                 ##### 11111 01011 11111 =1111 11111
                              ##### 10100 01110 10111 11111 11101   81%  40.5 (1095.0, 819.8)

 2. TJchess 1.1-x64           00000 ##### 11100 10111 11111 11111
                              01011 ##### 11101 11000 11101 11111   70%  35.0 (1150.0, 652.0)

 3. Imortal v1.0              10100 00011 ##### 10101 11110 11111
                              10001 00010 ##### 11111 11111 11111   68%  34.0 (1160.0, 609.5)

 4. Sjeng 11.2                00000 01000 01010 ##### 10111 11111
                              01000 00111 00000 ##### 11011 11111   50%  25.0 (1250.0, 378.5)

 5. Pulsar2009-9b             =0000 00000 00001 01000 ##### 11111
                              00000 00010 00000 00100 ##### 00111   25%  12.5 (1375.0, 163.3)

 6. KKFChess v2.6.6           00000 00000 00000 00000 00000 #####
                              00010 00000 00000 00000 11000 #####    6%   3.0 (1470.0,  65.5)

So, seems Imortal won't have as easy a time as its sibling Tadyak did! And while KKFChess by far performed worst, it still managed to steal a game from the top-scoring engine.

If anybody knows of any other Crazyhouse playing engines, please let me know!

Etc

The variant tournaments do not stop here. I preliminarily plan on Berolina Chess next, and in the future, after that some of these variants might happen:

Shatranj
Capablanca Chess (with various starting positions)
Spartan Chess
Grand Chess
Makruk
Nightrider Chess
Extinction Chess
Seirawan Chess
Three Checks Chess

There are other variants I would be interested in running as well (e.g. Atomic Chess, Cylinder Chess, Alice Chess, Janus Chess, Los Alamos Chess, Legan Chess, Racing Kings) but haven't found enough engines for to make it a worthwhile tournament.

If anybody has an opinion on what tournament to run next (or want to adapt their program for a particular variant), let me know! Feel free to suggest other variants as well, provided that there are engines that support them.

Ferdy · Post by **Ferdy** » Sat Dec 22, 2012 7:16 am

Those 2 are indeed very tough. You were given material but you will be mated by dropping its remaining on-hold material

I hope Imortal will give a good challenge at longer time control

yolin · Post by **yolin** » Sun Dec 23, 2012 12:14 am

Feuerstein can also play crazyhouse, but website is down. However source code is still available

http://wayback.archive.org/web/*/http:/ ... itzSchach/*

If anyone can compile and share this (the crazyhouse version), it would be very much appreciated!

enhorning · Post by **enhorning** » Sun Dec 23, 2012 12:50 am

yolin wrote:Feuerstein can also play crazyhouse, but website is down. However source code is still available

http://wayback.archive.org/web/*/http:/ ... itzSchach/*

If anyone can compile and share this (the crazyhouse version), it would be very much appreciated!

Thanks, I downloaded this and compiled it (with the Free Pascal compiler) - needed to change one of the options near the top to enable crazyhouse, but that was all.

Not sure if it's the most efficient compile, as I have no experience with Free Pascal (... last time I used Pascal was some Turbo Pascal in the early 90's), but it works, and seems to be beating KKFChess in my test game.

Need to figure out how to change the hash size - the default is low - but browsing the source file, seems there is an option to change that.

Not sure how to share the compiled file - is it possible to do attachments here? I don't see any option for such... this forum software is so outdated

enhorning · Post by **enhorning** » Sun Dec 30, 2012 10:42 pm

Okay, I started the tournament yesterday!

I decided to include Pulsar after all... with so few programs available, I wanted to include all that I could.

Crazyhouse, 1-cpu (on i-970), 512 MB Hash, 40 moves in 20 minutes repeating (30s / move average).

Format: 10-fold Round Robin, each program playing both sides after these 5 opening moves: 1. d4, 1. e4, 1. e3, 1. Nc3, 1. Nf3.

Standings after 1 cycle out of 10:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              T I S S F K P
 1. TJchess 1.1-x64           # 1 1 0 1 1 1   83%   5.0 ( 16.0,  13.0)
 2. Imortal v1.0              0 # 1 1 1 1 1   83%   5.0 ( 16.0,  11.0)
 3. Sunsetter                 0 0 # 1 1 1 1   67%   4.0 ( 17.0,   7.0)
 4. Sjeng 11.2                1 0 0 # 1 0 1   50%   3.0 ( 18.0,   8.0)
 5. Feuerstein                0 0 0 0 # 1 1   33%   2.0 ( 19.0,   2.0)
 6. KKFChess v2.6.6           0 0 0 1 0 # 0   17%   1.0 ( 20.0,   3.0)
 7. Pulsar2009-9b             0 0 0 0 0 1 #   17%   1.0 ( 20.0,   1.0)

No bugs or weird games so far, as far as I can see.

enhorning · Post by **enhorning** » Mon Dec 31, 2012 8:43 am

Standings after 2 cycles out of 10: (1. e4 done)

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Su Im TJ Sj Fe KK Pu
 1. Sunsetter                 ## 01 01 11 11 11 11   83%  10.0 ( 64.0,  45.0)
 2. Imortal v1.0              10 ## 01 11 11 11 11   83%  10.0 ( 64.0,  45.0)
 3. TJchess 1.1-x64           10 10 ## 01 11 11 11   75%   9.0 ( 66.0,  40.0)
 4. Sjeng 11.2                00 00 10 ## 11 01 11   50%   6.0 ( 72.0,  21.0)
 5. Feuerstein                00 00 00 00 ## 11 10   25%   3.0 ( 78.0,   6.0)
 6. KKFChess v2.6.6           00 00 00 10 00 ## 01   17%   2.0 ( 80.0,   8.0)
 7. Pulsar2009-9b             00 00 00 00 01 10 ##   17%   2.0 ( 80.0,   5.0)

Everything still running smoothly - I wonder if the problems I had with Pulsar before was unlucky, or due to the short time control I was running tests at.

enhorning · Post by **enhorning** » Mon Dec 31, 2012 10:51 pm

Standings after 3 cycles out of 10:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Sun Imo TJc Sje Feu Pul KKF
 1. Sunsetter                 ### 011 011 111 111 111 111   89%  16.0 (141.0, 114.0)
 2. Imortal v1.0              100 ### 011 111 110 111 111   78%  14.0 (147.0,  96.0)
 3. TJchess 1.1-x64           100 100 ### 011 111 111 111   72%  13.0 (150.0,  82.0)
 4. Sjeng 11.2                000 000 100 ### 110 111 011   44%   8.0 (165.0,  41.0)
 5. Feuerstein                000 001 000 001 ### 100 111   33%   6.0 (171.0,  32.0)
 6. Pulsar2009-9b             000 000 000 000 011 ### 101   22%   4.0 (177.0,  16.0)
 7. KKFChess v2.6.6           000 000 000 100 000 010 ###   11%   2.0 (183.0,  12.0)

Had the first couple of buggy games early in cycle 4. One false illegal move claim from Pulsar, and an illegal move by Sjeng:

[Event "Computer Chess Game"]
[Site "ENHORNING"]
[Date "2012.12.31"]
[Round "23"]
[White "Imortal v1.0"]
[Black "Sjeng 11.2"]
[Result "1-0"]
[TimeControl "40/1200"]
[Variant "crazyhouse"]
[Annotator "2. +0.35 1... +0.02"]

1. d4 Nf6 {+0.02/11 29} 2. Nf3 {+0.35/14 1:00} Ne4 {+0.16/10 28} 3. Nbd2
{+0.67/13 56} Nxd2 {-1.04/11 29} 4. Qxd2 {+1.38/13 52} e6 {-1.14/9 29} 5.
Ne5 {+1.64/10 52} Qe7 {-0.44/9 29} 6. N@b5 {+1.10/10 34} d6 {+0.80/9 29} 7.
Qg5 {+0.37/11 22} N@f6 {+0.27/10 29} 8. Ng4 {+0.54/11 42} Na6 {+0.97/9 22}
9. Nxf6+ {+0.21/11 41} gxf6 {+0.23/9 34} 10. Qf4 {+0.58/11 25} e5
{+0.97/9 30} 11. dxe5 {+0.95/10 23} dxe5 {+0.64/9 22} 12. Qf3 {+0.66/9 33}
P@g4 {+0.86/8 34} 13. Qd3 {+0.85/9 22} N@b4 {+0.27/8 23} 14. Qd2
{+0.47/10 37} Bd7 {-0.22/9 35} 15. N@h5 {+0.83/9 35} Bxb5 {+0.53/9 31} 16.
P@g7 {-3.50/9 33} N@e4 {+2.48/8 29} 17. gxh8=Q {-3.91/9 32} Nxd2
{+3.16/8 30} 18. Qxf8+ {-3.03/7 25} Qxf8 {+6.38/7 21} 19. Nxf6+
{-3.61/8 25} Kd8 {+6.69/7 34} 20. B@f5 {-4.26/7 25} Nxf1 {+6.68/7 31} 21.
R@g8 {-0.49/7 25} Nxc2+ {+6.85/7 59} 22. Bxc2 {+0.61/8 25} B@b4+
{+4.55/7 25} 23. P@c3 {+0.92/8 25} P@d2+ {+4.68/8 32} 24. Bxd2
{+1.14/10 24} Nxd2 {+2.75/8 59} 25. cxb4 {+1.14/8 21} Qxg8 {+1.99/6 48} 26.
Nxg8 {+1.14/7 25} B@h4 {+2.47/6 27} 27. Q@d5+ {+2.99/6 25} R@d7
{+5.49/7 37} 28. Qxb5 {+3.95/7 25} Q@f1+ {+9999.94/7 15} 29. Rxf1
{+18.38/2} Nf3+ {+9999.96/5 0.1} 30. exf3 {+33.55/6 27}
{Xboard: Forfeit due to invalid move: q@d2 (_1_1) res=24} 1-0

Interestingly enough, if the variant had been Loop Chess (which I do prefer to Crazyhouse, personally), that would have been a legal move... so I wonder if Sjeng is actually trying to play Loop Chess and not Crazyhouse! Might at some point run a few games between Sjeng and TJchess (which as far as I know is the only engine that supports Loop Chess) later on to test that theory out.

enhorning · Post by **enhorning** » Tue Jan 01, 2013 5:20 am

Standings after 4 cycles out of 10 (1. e4 and 1.d4 done):

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Suns TJch Imor Sjen Feue Puls KKFC
 1. Sunsetter                 #### 0110 0111 1110 1111 1111 1111   83%  20.0 (256.0, 189.0)
 2. TJchess 1.1-x64           1001 #### 1001 0111 1111 1111 1111   79%  19.0 (260.0, 173.0)
 3. Imortal v1.0              1000 0110 #### 1111 1101 1111 1111   75%  18.0 (264.0, 157.0)
 4. Sjeng 11.2                0001 1000 0000 #### 1100 1111 0111   46%  11.0 (292.0,  83.0)
 5. Feuerstein                0000 0000 0010 0011 #### 1001 1111   37%   9.0 (300.0,  58.0)
 6. Pulsar2009-9b             0000 0000 0000 0000 0110 #### 1011   21%   5.0 (316.0,  24.0)
 7. KKFChess v2.6.6           0000 0000 0000 1000 0000 0100 ####    8%   2.0 (328.0,  16.0)

enhorning · Post by **enhorning** » Tue Jan 01, 2013 11:13 pm

Standings after 5 cycles out of 10:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Sun TJc Imo Sje Feu Pul KKF
 1. Sunsetter                 ### 011 011 111 111 111 111
                              ### 01  11  01  11  11  11    87%  26.0 (395.0, 313.0)

 2. TJchess 1.1-x64           100 ### 100 011 111 111 111
                              10  ### 11  11  10  11  11    77%  23.0 (410.0, 263.0)

 3. Imortal v1.0              100 011 ### 111 110 111 111
                              00  00  ### 11  11  11  11    73%  22.0 (415.0, 231.0)

 4. Sjeng 11.2                000 100 000 ### 110 111 011
                              10  00  00  ### 01  11  11    47%  14.0 (455.0, 125.0)

 5. Feuerstein                000 000 001 001 ### 100 111
                              00  01  00  10  ### 10  11    37%  11.0 (470.0,  97.0)

 6. Pulsar2009-9b             000 000 000 000 011 ### 101
                              00  00  00  00  01  ### 11    23%   7.0 (490.0,  41.0)

 7. KKFChess v2.6.6           000 000 000 100 000 010 ###
                              00  00  00  00  00  00  ###    7%   2.0 (515.0,  21.0)

Had another game with an illegal move, Feuerstein ignored that it was in check. So that's one forfeit each from Feuerstein, Sjeng and Pulsar so far.

Sunsetter pulled ahead a little bit in this cycle. Still, quite a close race between Sunsetter, TJchess and Imortal!

enhorning · Post by **enhorning** » Wed Jan 02, 2013 3:49 am

Standings after 6 cycles out of 10:

Code: Select all

Cross table, sorted by score percentage, Buchholz, SB

                              Sun TJc Imo Sje Feu Pul KKF
 1. Sunsetter                 ### 011 011 111 111 111 111
                              ### 011 110 010 111 111 111   83%  30.0 (576.0, 436.0)

 2. TJchess 1.1-x64           100 ### 100 011 111 111 111
                              100 ### 111 111 101 110 111   75%  27.0 (594.0, 380.0)

 3. Imortal v1.0              100 011 ### 111 110 111 111
                              001 000 ### 111 111 111 111   75%  27.0 (594.0, 352.0)

 4. Sjeng 11.2                000 100 000 ### 110 111 011
                              101 000 000 ### 010 110 111   44%  16.0 (660.0, 189.0)

 5. Feuerstein                000 000 001 001 ### 100 111
                              000 010 000 101 ### 101 111   39%  14.0 (672.0, 144.0)

 6. Pulsar2009-9b             000 000 000 000 011 ### 101
                              000 001 000 001 010 ### 111   28%  10.0 (696.0,  95.0)

 7. KKFChess v2.6.6           000 000 000 100 000 010 ###
                              000 000 000 000 000 000 ###    6%   2.0 (744.0,  26.0)

Had another forfeit from Feuerstein: "{Error: second chess program (feuerstein0461.exe) exited unexpectedly}"

Chess Variant tournaments: Crazyhouse, etc.

Chess Variant tournaments: Crazyhouse, etc.

Re: Chess Variant tournaments: Crazyhouse, etc.

Re: Chess Variant tournaments: Crazyhouse, etc.

Re: Chess Variant tournaments: Crazyhouse, etc.

Re: Chess Variant tournaments: Crazyhouse, etc.

Re: Chess Variant tournaments: Crazyhouse, etc.

Re: Chess Variant tournaments: Crazyhouse, etc.

Re: Chess Variant tournaments: Crazyhouse, etc.

Re: Chess Variant tournaments: Crazyhouse, etc.

Re: Chess Variant tournaments: Crazyhouse, etc.