Chiron 1.1 bug?

[uaf]

Yes this is a bug related to en passant: unfortunately I forgot to handle ep captures in the blockage code. Thanks for reporting.

By the way, Merry Christmas to everyone!

[MM]

Yes this is a bug related to en passant: unfortunately I forgot to handle ep captures in the blockage code. Thanks for reporting.

By the way, Merry Christmas to everyone!

Thank you Mr Farina,

Happy Christmas to you too

[lucasart]

Hi,

This is from a Critter beta tester and I am not able to verify because I do not own Chiron. He submitted some logs when he tested Critter's pawn blockage detection (and compared with Chiron).

Starting from this position:
[d]8/6pp/1k2p3/p1p1P1P1/P1R2r1P/3P4/1P2P1K1/8 b - - 0 1

... both engines saw the blockage, and gave 0.00 evaluation. However, Chiron was not able to hold the draw...

[d]8/7p/1k2p1p1/p1p1P1P1/P1P4P/5K2/1P2P3/8 b - - 1 3
Here Chiron played 3. ...h5?? and lost the game. It's most probably caused by ignoring en passant in the pawn blockage detection code. Hope this helps...

pgn:

Code: Select all

[Event "Computer chess game"]
[Site "far_away_galaxy"]
[Date "2011.12.21"]
[Round "?"]
[White "Critter dev 64bit SSE"]
[Black "Chiron-1.1-64bit"]
[Result "*"]
[WhiteElo "2200"]
[TimeControl "60+1"]
[SetUp "1"]
[FEN "8/6pp/1k2p3/p1p1P1P1/P1R2r1P/3P4/1P2P1K1/8 b - - 0 1"]
[PlyCount "15"]

1. ... Rxc4 {(Rxc4 dxc4 g6 h5) 0.00/32 2} 2. dxc4 {(dxc4 g6 Kf3) 0.00/33 3}
g6 {(g6 h5 gxh5 e4) 0.00/34 2} 3. Kf3 {(Kf3) 0.00/64 0} h5 {(h5 gxh6)
0.00/42 2} 4. gxh6 {(gxh6 Kc6 h7 Kd7 h8Q Ke7 Qf6+ Kd7 Ke4 Kc7 Qxe6 Kd8 Qd6+
Kc8 e6 Kb7 e7 g5 e8Q gxh4 Qeb8+) +M11/22 2} Kc6 {(Kc6 h7 Kc7 h8Q Kb6 Qb8+
Kc6 Qe8+ Kc7 Qxg6 Kc6 h5 Kd7 h6 Kc7 h7 Kb7 h8Q Ka6 Qhg7 Kb6 Qxe6+) -M12/14
0} 5. h7 {(h7 Kd7 h8Q Ke7 Qf6+ Kd7 Ke4 Kc7 Qxe6 Kd8 Qd6+ Kc8 e6 Kb7 e7 g5
e8Q gxh4 Qeb8+) +M10/21 0} Kb6 {(Kb6 h8Q Kc7 Qg7+ Kc8 Qg8+ Kd7 Qxg6 Kc7 h5
Kd7 h6 Kc7 h7 Kb7 h8Q Ka6 Qhg7 Kb6 Qxe6+) -M11/9 0} 6. h8=Q {(h8Q Kb7 Qe8
Kb6 Qxe6+ Kc7 Qd6+ Kb7 e6 g5 Ke4 gxh4 e7 h3 e8Q Ka7 Qeb8+) +M9/21 2} Kb7
{(Kb7 Qg7+ Kc8 Qg8+ Kd7 Qxg6 Kc7 h5 Kd7 h6 Kc7 h7 Kb7 h8Q Ka6 Qhg7 Kb6
Qxe6+) -M10/6 0} 7. Qd8 {(Qd8 Kc6 b4 cxb4 Qc8+ Kb6 c5+ Ka7 Qd7+ Kb8 c6 Ka8
Qb7+) +M7/22 2} Ka7 {(Ka7 Qc8 Kb6 Qb8+ Kc6 Qe8+ Kc7 Qxg6 Kc6 h5 Kd7 h6 Kc7
h7 Kb7 h8Q Ka6 Qhg7 Kb6 Qxe6+) -M11/9 0} 8. b4 {(b4 cxb4 Qc7+ Ka8 c5 g5 c6
g4+ Kxg4 b3 Qb7+) +M6/23 2} g5 {(g5 hxg5 cxb4 Qc7+ Ka8 g6 b3 g7 b2 g8Q+)
-M5/10 0 Black resigns} 1-0

Hello Richard,

Seems that is an improvement since Critter 1.2 (v1.2 doesn't understand the pawn blockade here).

I'm just curious to know how Critter manages to understand it. Is it by search means (seeing somehow that no progress can be made after so many plies) or does the eval directly return zero from this position ?
[d]8/7p/1k2p1p1/p1p1P1P1/P1P4P/5K2/1P2P3/8 b - - 1 3

[rvida]

I'm just curious to know how Critter manages to understand it. Is it by search means (seeing somehow that no progress can be made after so many plies) or does the eval directly return zero from this position ?

Draw score is returned directly from eval. Implementation is based on this paper by Omid David Tabibi.

[lucasart]

I'm just curious to know how Critter manages to understand it. Is it by search means (seeing somehow that no progress can be made after so many plies) or does the eval directly return zero from this position ?

Draw score is returned directly from eval. Implementation is based on this paper by Omid David Tabibi.

Thank you! It's a *very* good paper. I'm reading it now

[Sylwy]

I'm just curious to know how Critter manages to understand it. Is it by search means (seeing somehow that no progress can be made after so many plies) or does the eval directly return zero from this position ?

Draw score is returned directly from eval. Implementation is based on this paper by Omid David Tabibi.

Thank you! It's a *very* good paper. I'm reading it now

Hi Lucas !

Better for you to study Falcon chess engine !
Much better !


SilvianR