Thermopylay Marathon 2011 (live!)

[hgm]

The Thermopylae Marathon 2011 has begun! Seven Spartan-Chess engines will battle each other the coming days for eternal glory!

Participants

Code: Select all

1. Catalyst          Richard Albert
2. ChessV            Gregory Strong
3. Fairy-Max 4.8R    H.G.Muller
4. Nebiyu 1.1        Daniel Shawul
5. Oberon            Pawel Koziol
6. Sjaak 92          Evert Glebbeek
7. Spartacus 0.23    H.G.Muller

Tournament format
First there will be a qualifier of 4 games per pairing, played in two cycles. The second cycle will probably start from a forced opening line to avoid duplicat games.
The first four qualify for the playoffs, and will play a full round-robin there (2 games per pairing). The score of the qualifier will keep counting (also results against non-qualified opponents).

Time Control
The games will be played at classical TC of 40 moves / 24 min. Nebiyu and Spartacus will face 6-fold time odds in the qualifier to make the tourney more interesting. That means they will get 4 min, when playing a non-handicapped opponent.

Web page
The games will be broadcast live on http://80.100.28.169/spartan/spartan.html , with the aid of the ChessLive! viewer, so they can be watched through any standard web browser.

[Richard Allbert]

Are updates allowed?

I can understand if not, as you've set time handicaps

[hgm]

In principle I wanted to allow updates only at the end of a complete cycle. But on the other hand, I don't want to discourage people improving their engines. But hasty updates within a tournament are a recipe for trouble. All versions playing now have been extensively tested for not being buggy, (whichis just as well, because some predecessors were), and I could not repeat such testing for replacements.

But if you are sure an update is significantly better than the current version, I would not have a major objection against replacing Catalyst now. In principle replacement by a stronger version would give unfair advantage to the engines having played the weaker version. But in this case it has only played against two handicapped opponents, so it could not really be considered unfair if they were not as much handicapped as intended.

[hgm]

Sjaak vs. Oberon was a very interesting game. At some point Oberon allows Sjaak to promote to a second Queen, so that for more than 50 moves it is facing two Queens (+ two Rooks + Knight). The heavier pieces of the Persians turn out to be quite powerless against the lighter Spartan pieces, which defend each other quite effectively. The majority of Hoplites vs. Pawns in the end decides the battle in favor of the Spartans.

Code: Select all

[Event "Thermopylae Marathon 2011"]
[Site "SCHAAKPC"]
[Date "2011.02.07"]
[Round "3.3"]
[White "Sjaak 92"]
[Black "Oberon"]
[Result "0-1"]
[TimeControl "40/1440"]
[Variant "spartan"]
[Number "18"]
[Annotator "1. +0.32   1... +0.39"]

1. e3 {+0.32/13 1:00} Lc6 {+0.39/12 28} 2. Nf3 {+0.24/14 57} Lf6
{+0.48/12 27} 3. Be2 {+0.24/13 55} Cd6 {+0.56/12 34} 4. O-O {+0.24/13 53}
Cee6 {+0.56/12 26} 5. Qe1 {+0.24/13 43} Ke8 {+0.85/11 33} 6. d3
{+0.16/13 49} Hge5 {+0.87/11 20} 7. Nc3 {+0.16/12 47} Hbd5 {+0.85/11 24} 8.
Rb1 {+0.16/12 43} Hhf5 {+0.86/11 21} 9. g4 {+0.08/11 38} Wh7 {+1.11/11 32}
10. h4 {+0.20/12 40} Hfe4 {+1.06/10 26} 11. Bd2 {+0.24/11 40} Hg6
{+1.37/11 36} 12. Kh1 {+0.16/12 38} Hg5 {+1.58/11 21} 13. h5 {+0.24/11 35}
Hxe3 {+1.55/10 22} 14. fxe3 {+0.16/11 35} Hxg4 {+1.56/11 21} 15. Rg1
{+0.12/12 34} Hgf5 {+1.47/11 38} 16. Rxg4 {+0.16/13 26} Hfe4 {+1.49/10 24}
17. Qg3 {+0.12/12 31} Hxe3 {+1.45/10 26} 18. Bxe3 {+0.12/12 30} Hf5
{+1.24/10 24} 19. Ng5 {+0.16/12 29} Lxg5 {+1.29/11 29} 20. Rxg5
{+0.16/12 27} Cdf6 {+1.31/9 33} 21. Qg1 {+0.12/11 26} Hd4 {+1.38/10 33} 22.
Bf3 {+0.72/12 22} Ce5 {+1.59/10 45} 23. Rg7 {+0.80/12 24} Wf8 {+1.08/10 38}
24. h6 {+0.72/12 22} We6 {+0.89/9 38} 25. h7 {+1.08/11 22} Cxe3
{+0.83/10 25} 26. h8=Q {+2.00/11 21} Cxf3 {+0.84/9 49} 27. Qg2
{+1.88/10 20} Cf8 {+0.78/10 49} 28. Qh7 {+1.80/11 20} Cf4 {+0.59/9 31} 29.
Qe2 {+1.68/10 19} Wd8 {+0.64/9 26} 30. Qhh2 {+1.80/11 17} Cf7 {+0.66/9 53}
31. Rg8 {+1.84/11 17} Cf8 {+0.77/9 45} 32. Rg2 {+1.80/13 17} Gb7
{+0.65/8 44} 33. Qe1 {+1.80/12 15} Hcb6 {+0.63/9 46} 34. Ne2 {+1.80/11 15}
Cf3 {+0.52/9 38} 35. Ng1 {+1.80/12 14} Cf4 {+0.46/10 1:01} 36. Rd1
{+1.80/12 14} C8f6 {+0.37/9 42} 37. Re2 {+1.76/11 13} Gd7 {+0.46/9 1:07}
38. Re5 {+1.76/11 13} Ce6 {+0.30/9 41} 39. Re3 {+1.68/14 12} Ld6
{+0.11/9 1:33} 40. Re2 {+1.72/11 10} Cff6 {+0.02/9 1:12} 41. Qhh4
{+1.60/12 1:11} Lf4 {+0.14/11 43} 42. Rg2 {+1.52/13 1:06} He3 {+0.15/10 49}
43. Nf3 {+1.52/13 1:06} Ced6 {+0.24/10 58} 44. Qc3 {+1.56/11 1:01} Cf8
{+0.17/9 37} 45. Qh5+ {+1.52/11 1:00} Cf7 {+0.16/9 31} 46. Qh3
{+1.52/12 58} Gc7 {+0.17/10 37} 47. Qe1 {+1.52/13 55} Ge7 {+0.19/10 41} 48.
Qa5 {+1.52/12 53} Wc6 {+0.17/10 46} 49. Qc5 {+1.52/13 51} Gc7 {+0.18/10 39}
50. Qh8 {+1.52/11 46} Cf8 {+0.09/10 31} 51. Qh5 {+1.52/12 47} Cf7
{+0.13/10 48} 52. Re1 {+1.56/10 45} We7 {+0.06/11 28} 53. Qa5 {+1.28/11 43}
Wf6 {+0.46/11 34} 54. Qh3 {+0.48/12 41} Wxb2 {+0.45/10 23} 55. Ng5
{+0.56/11 40} We5 {+0.33/10 23} 56. c4 {+0.48/11 33} Wc6 {+0.18/10 30} 57.
Qa3 {+0.48/12 32} Wd4 {+0.14/10 38} 58. Qa4 {+0.48/11 31} Cfd7
{+0.33/10 37} 59. Nf3 {+0.52/11 34} Wc6 {+0.21/10 35} 60. Qb3 {+0.56/12 32}
We7 {+0.20/10 25} 61. cxd5 {+0.52/11 31} C6xd5 {+0.20/9 32} 62. Qh8
{+0.52/10 30} Wd6 {+0.30/9 22} 63. Reg1 {+0.48/10 29} Wf7 {+0.27/10 29} 64.
Qa1 {+0.60/10 27} Gd6 {+0.73/10 17} 65. Rd1 {+0.00/10 26} Wh5 {+0.82/9 19}
66. Qc2 {+0.40/11 25} Hbc5 {+1.49/10 34} 67. Nh2 {+0.04/11 24} Gh6
{+1.83/10 19} 68. Qf2 {-1.04/12 23} Hd4 {+1.85/10 23} 69. Kg1 {-1.00/9 22}
Hxd3 {+1.92/9 13} 70. Rc1 {-1.00/10 21} Cc7 {+2.11/9 22} 71. Nf1
{-1.52/9 20} Ce5 {+1.88/9 11} 72. Rg8 {-1.56/9 19} He2 {+1.84/8 12} 73. Qg2
{-1.52/9 19} Kcd7 {+1.93/7 21} 74. Rxe8 {-1.56/8 18} Wxe8 {+2.98/8 18} 75.
Qd4+ {-1.68/8 17} Ke7 {+3.14/7 19} 76. Qg4 {-2.64/9 16} Hc5 {+4.17/8 12}
77. Qgxf4 {-3.92/10 16} Hxf4 {+3.20/9 16} 78. Qxe5+ {-3.76/10 15} Ge6
{+3.26/8 14} 79. Qxf4 {-4.08/10 13} Wb5 {+3.47/8 8} 80. Qg5+ {-2.72/9 14}
Kf7 {+3.83/8 47} 81. Qf4+ {-3.84/11 1:13} Gf6 {+4.07/10 53} 82. Qxe3
{-4.24/14 1:10} Wc6 {+3.85/10 1:21} 83. Qxe2 {-4.24/14 1:02} Gg5+
{+4.51/10 1:17} 84. Kh2 {-4.28/14 1:05} Gh4+ {+4.55/11 0.1} 85. Kg1 Gh1+
{+4.50/11 0.1} 86. Kf2 Gg2+ {+4.55/11 1:06} 87. Ke1 {-4.32/14 57} Wb4+
{+4.65/12 1:10} 88. Kd1 {-4.36/16 55} Gxe2+ {+4.63/11 0.1} 89. Kxe2 Wxa2
{+4.72/11 1:06} 90. Rc2 {-4.32/15 50} We6 {+4.72/10 1:03} 91. Ke1
{-4.32/14 48} Wd5 {+4.70/10 36} 92. Rf2+ {-4.48/15 46} Kg6 {+4.81/10 51}
93. Ng3 {-4.52/15 44} Wc3+ {+4.94/10 44} 94. Kf1 {-4.52/16 42} Hb4
{+4.95/10 53} 95. Rg2 {-4.52/14 41} Kf7 {+5.06/10 27} 96. Rf2+
{-4.52/14 39} Ke6 {+5.15/10 49} 97. Re2+ {-4.52/14 37} Kd5 {+5.09/10 25}
98. Rc2 {-4.56/14 36} Wb5+ {+5.31/11 27} 99. Ne2 {-5.12/15 34} Cc5
{+5.38/11 38} 100. Rd2+ {-5.24/14 33} Ke5 {+5.48/10 52} 101. Ra2
{-6.40/13 31} Hc3 {+7.16/10 26} 102. Kg2 {-6.56/15 30} Wc4 {+7.26/11 36}
103. Nxc3 {-6.56/16 29} Cxc3 {+7.28/10 21} 104. Ra4 {-6.56/15 28} Wb6
{+7.30/10 24} 105. Rh4 {-6.56/15 26} We3+ {+7.34/9 18} 106. Kh2
{-6.56/16 25} Cd3 {+7.35/10 21} 107. Rh8 {-6.56/15 24} Wg4+ {+7.33/9 22}
108. Kg3 {-6.56/15 23} Wf6 {+7.33/9 31} 109. Rb8 {-6.56/14 22} Wd7
{+7.31/9 20} 110. Rg8 {-6.56/15 19} Ce3+ {+7.33/9 36} 111. Kf2
{-9.92/15 21} Ke4 {+7.36/10 16} 112. Kg2 {-10.48/15 20} Cf3 {+7.40/9 26}
113. Ra8 {-13.96/14 19} Wh3+ {+7.42/9 22} 114. Kh2 {-9.92/16 18} Wg4+
{+7.40/8 13} 115. Kg2 {-10.48/16 17} Cf2+ {+7.43/9 23} 116. Kg3
{-10.40/16 17} Wd7 {+7.42/9 16} 117. Kh4 {-10.00/15 16} Wf6+ {+7.40/9 18}
118. Kg3 {-10.48/16 15} Cf3+ {+7.36/8 24} 119. Kh2 {-10.48/15 15} We5+
{+7.39/8 13} 120. Kh1 {-14.04/13 14} Ke3 {+12.42/9 59} 121. Kg1
{-159.79/17 1:13} Cg3+ {+12.54/10 1:02} 122. Kh2 {-159.79/17 1:10} Cg5+
{+12.65/10 41} 123. Kg2 {-159.87/17 43} Wf3+ {+99.89/10 0.1} 124. Kf1
{-159.89/12 8} Wxa8 {+99.91/9 0.1} 125. Kg1 {-159.91/10 0.1} Cg4
{+99.93/6 0.1} 126. Kh2 {-159.95/6 0.1} Wf3+ {+99.95/3 0.1} 127. Kh3 Ch4+
{+99.97/2 0.1} 128. Kg3 We1# {+99.99/1}
{Xboard adjudication: Checkmate} 0-1

1gk1k3/h1h3RP/2l1wc2/3hch2/3h4/2NPBB2/PPP5/1R4QK b - - 0 25

With N=Warlord, Q=General, B=Lieutenant, R=Captain, and P=Hoplite for the Spartans:
[d]1qk1k3/p1p3RP/2b1nr2/3prp2/3p4/2NPBB2/PPP5/1R4QK b - - 0 25

Oberon decides to allow the promotion of the h-Pawn, for annihilating Sjaak's Bishop pair, by playing 25... Cxe3 26. h8=Q Cxf3. The Persians keep the extra Queen to move 77, when they are forced to sacrifice it for Lieutenant + Captain + Hoplite, to prevent promotion of one of the advancing Hoplites (77 Qgxf4 Hxf4 78. Qxe5+ Ge6 78. Qxf4).

4w3/2c1k3/7g/2h1ch2/3Q1lQ1/4h3/P3h3/2R2NK1 w - - 4 77

[d]4n3/2r1k3/7q/2p1rp2/3Q1bQ1/4p3/P3p3/2R2NK1 w - - 4 77

[Evert]

Sjaak vs. Oberon was a very interesting game. At some point Oberon allows Sjaak to promote to a second Queen, so that for more than 50 moves it is facing two Queens (+ two Rooks + Knight). The heavier pieces of the Persians turn out to be quite powerless against the lighter Spartan pieces, which defend each other quite effectively. The majority of Hoplites vs. Pawns in the end decides the battle in favor of the Spartans.

I was able to follow that game on and off for a bit, indeed very interesting. My impression shortly after the promotion was that white's position was going to be very difficult because it wouldn't be able to break through black's defences. Especially with two Spartan kings on the board.
Sjaak gave it a positive score, which from what I saw didn't improve after that point and slowly dwindled until the game was lost, but I haven't looked at the replay in detail. It'll be interesting to come up with an evaluation term that helps to correctly evaluate that position.
I also want to take a look at the second game against Fairy-Max, where Sjaak (with black) traded all major pieces so it ended with two kings, a captain and a lieutenant against knight, bishop and rook. It seemed to think the trade was good, but I don't think two kings quite make up for a rook in the end game. The game ended as a draw, but I haven't reviewed it yet.

[Evert]

In principle I wanted to allow updates only at the end of a complete cycle. But on the other hand, I don't want to discourage people improving their engines. But hasty updates within a tournament are a recipe for trouble. All versions playing now have been extensively tested for not being buggy, (whichis just as well, because some predecessors were), and I could not repeat such testing for replacements.

I've just found a major bug in Sjaak's evaluation function, where it didn't correctly evaluate mobility for sliders (leapers are fine). A quick fix seems to make quite abig difference, even without properly tuning evaluation weights (I think mobility is valued too much now) it's a clear plus in testing.
I don't know whether I have time to send you an update though, and I have my doubts about whether that'd be fair anyway. When would the next cycle start?

[hgm]

The next cycle will start tomorrow somewhere around this time. We have nearly completed round 4 of 7 now.

[Evert]

Ok. Well, I'll see if I have time to test and send on an update. If I don't, that's just too bad.

[Richard Allbert]

You're right, better with testing, especially after what happened last time!

I'm not sure I'll have the hash done by tomorrow, including testing.

Hopefully end of the week.

Ciao

Richard

[Daniel Shawul]

Did you see the games of catalyst against fairy max ? In one of the games, it was winning but then lost due to bad passed pawn evaluation I think. Add that too