Bihasa v3.0 is released

[Ferdy]

Hi Ferdinand,
did you get my mail about the question for
Deuterium v11.02.29.107 ?

Just read today.

Any news about when the new release will be that has the bug fix?

For Bihasa could be 4 hrs from now

[George Tsavdaris]

Hi Ferdinand,
did you get my mail about the question for
Deuterium v11.02.29.107 ?

Just read today.

Any news about when the new release will be that has the bug fix?

For Bihasa could be 4 hrs from now

Super great!

[Ferdy]

New version is available.

Code: Select all

Bihasa v3.1 (27-Jan-2012)
1. Fixed analyze command
2. Added SEE-static exchange evaluator for move ordering and prunning.
3. Modify pawn piece square tables.
4. Increase evaluation scores for archbishop and chancellor that are close to opp king.
5. Modify king attack combination evaluation scores. 
6. Improved time control handling

Test results tc 40/5sec after 300 games, there could be some strength improvement.

Code: Select all

  
   1 Bihasa-v3.1   125 300.0 (214.0 :  86.0)
                       300.0 (214.0 :  86.0) Bihasa-v3.0   -49
   2 Bihasa-v3.0   -49 1300.0 (620.0 : 680.0)
                       300.0 ( 86.0 : 214.0) Bihasa-v3.1   125
                       1000.0 (534.0 : 466.0) Bihasa-v2.1   -76
   3 Bihasa-v2.1   -76 1000.0 (466.0 : 534.0)
                       1000.0 (466.0 : 534.0) Bihasa-v3.0   -49

Code: Select all

Rank Name          Elo    +    - games score oppo. draws 
   1 Bihasa-v3.1   125   26   25   300   71%   -49   11% 
   2 Bihasa-v3.0   -49   12   12  1300   48%   -30   12% 
   3 Bihasa-v2.1   -76   13   13  1000   47%   -49   13% 

[George Tsavdaris]

[quote="Ferdy"]New version is available.

Code: Select all

Bihasa v3.1 (27-Jan-2012)

Test results tc 40/5sec after 300 games, there could be some strength improvement.

[code]
Rank Name          Elo    +    - games score oppo. draws 
   1 Bihasa-v3.1   125   26   25   300   71%   -49   11% 
   2 Bihasa-v3.0   -49   12   12  1300   48%   -30   12% 
   3 Bihasa-v2.1   -76   13   13  1000   47%   -49   13% 

[/quote]

170 ELO improvement????
That is just amazing!!
Let's see if it can give me some challenge now. 170 ELO are way too much.

Many thanks.

[George Tsavdaris]

OK Bihasa 3.1 seems to be a worthy opponent.
I'm 1-1 in the first 2 games against it on 40/10 time control.

I crushed her(is it her? it sounds like female ) in the first game where she must had the biggest fail low of her life!! From +0.56 to -10.36 in one move!!!!! It's my classic style against computers. They don't see the attack until it's too late.
Gothic Vortex experienced much worse fail lows in the past from +14.0 to -mate in X

In the second game everything went as scheduled for me until i castled too prematurely and Bihasa wasted no time and attacked ferociously with very good and precise play where i stood no chance to do anything at all. Very humiliating loss.

[Event "Computer Chess Game"]
[Site "??"]
[Date "2012.01.27"]
[Round "1"]
[White "George"]
[Black "Bihasa v3.1"]
[Result "1-0"]
[TimeControl "40/600"]
[Variant "gothic"]
[Annotator "1... +0.04"]

1. f4 Nc6 {+0.04/14 0.9} 2. d4 d5 {+0.00/15 1.8} 3. c3 Nh6 {+0.00/15 4} 4.
g3 g6 {+0.04/14 2.5} 5. Nj3 Bg7 {+0.16/13 2.0} 6. h3 f5 {+0.32/14 7} 7. Bf3
Af6 {+0.24/14 4} 8. Ah2 O-O {+0.40/13 8} 9. Nh4 e6 {+0.28/13 12} 10. Cg2
Ae7 {+0.36/13 6} 11. i4 i6 {+0.64/13 7} 12. i5 Ng8 {+0.40/14 9} 13. j4 Nf6
{+0.32/12 36} 14. Ci1 Af8 {+0.28/11 54} 15. e3 Nh5 {+0.36/11 20} 16. Ai4 b6
{+0.56/11 32} 17. Qe2 a5 {+0.60/11 10} 18. Qj2 Bf6 {+0.80/9 26} 19. j5 Qe7
{-0.88/9 53} 20. jxi6 Ba6+ {+0.48/9 6} 21. Kf2 hxi6 {+0.56/8 46} 22. Bxh5
Cg7 {-10.36/12 6} 23. Aj6+ Kj8 {-4.44/3 5} 24. Axh8 Bxh4 {-14.04/12 17} 25.
gxh4 gxh5 {-15.08/12 11} 26. Cj3 j5 {-15.72/9 6} 27. Axi6+ Ki8
{-299.88/16 12} 28. Cxj5 Qxh4+ {-299.90/12 7}
{Black Resigns} 1-0


[Event "Computer Chess Game"]
[Site "??"]
[Date "2012.01.27"]
[Round "2"]
[White "Bihasa v3.1"]
[Black "George"]
[Result "1-0"]
[TimeControl "40/600"]
[Variant "gothic"]
[Annotator "1. +0.08"]

1. c4 {+0.08/15 4} g6 2. Nc3 {+0.08/16 2.6} e5 3. Nh3 {+0.08/16 2.3} c6 4.
d3 {+0.20/15 6} d6 5. j4 {+0.16/14 9} Nj6 6. e4 {+0.52/13 14} h6 7. j5
{+0.48/13 20} Ah7 8. g3 {+0.60/13 43} Be6 9. Be3 {+0.80/12 12} Na6 10. f4
{+0.80/12 26} Bg7 11. Bf3 {+1.32/12 25} Ce7 12. Bxj7 {+1.96/13 9} b6 13.
Bi6 {+2.12/13 25} Af6 14. fxe5 {+2.40/11 11} dxe5 15. Af2 {+1.96/11 9} Qd7
16. g4 {+2.40/13 21} Ni4 17. O-O {+2.34/13 1:04} O-O-O 18. Rf1
{+2.32/12 48} Kb7 19. Qb3 {+2.58/11 18} Cc8 20. a4 {+2.74/12 1:06} c5 21.
a5 {+4.20/12 10} Nb4 22. Nd5 {+4.30/12 16} Bxd5 23. exd5 {+4.12/10 1.0} Kb8
24. g5 {+4.46/12 1.4} hxg5 25. Ah1 {+4.60/12 10} Ae8 26. Nxg5
{+4.64/12 1.0} Nxg5 27. Bxg5 {+4.48/12 8} f6 28. Cg2 {+4.60/13 1.1} Cd6 29.
Bxi7 {+5.08/15 2.6} Rf8 30. axb6 {+5.18/12 6} axb6 31. d4 {+5.80/12 0.9} f5
32. Rfe1 {+6.32/12 0.8} cxd4 33. Qxb4 {+6.36/12 7} Rc8 34. b3 {+6.64/12 11}
Ac7 35. c5 {+9.22/14 6} Cxd5 36. Qa3 {+11.74/14 5} Rce8 37. Cxg6
{+12.94/13 5} Bf6 38. c6 {+22.76/15 4} Qxc6 39. Qa7+ {+24.46/16 13}
{Black resigns} 1-0

[Ferdy]

Thanks for the game, that is a model game for me to solve .
I have an initial idea on how to improve from this, let's see how Bihasa will progress.

[George Tsavdaris]

For the love of God!!!! What monster is this?

There is no point in testing it against Gothic Vortex since after 8 games it leads 8-0. And Vortex is using a good opening book that i was puzzled if i should use it or not because it would not be fair for Bihasa.
Yes 8 games are too few, but try to play more games manually with Vortex in its own GUI.

So since i am the only opponent for Bihasa that there is a sense to play against, i played another 4 games with Bihasa and i won the 2 of them, i drew one and lost the other. So all in all i lead 3.5-2.5 until now.
40/10 time control as always.

[George Tsavdaris]

Paste this position for Bihasa to analyze in analysis mode. do not analyze with it, this is not my point.
10/3k5p/R9/3N2p1N1/2p7/1a2P1P3/p3PP1PPP/6K3 b - - 0 1

Promote the a2 Pawn to a Queen. 1...a1Q+
See what happens.

Black promotes to a Queen and gives a check. Yet Bihasa ignores this and wants to play a move for white that doesn't remove the check!


It seems that promoting to a Chancellor has the same effect.

[hgm]

Another request: could you also implement the memory command for setting the hash size? Then I can no longer forget to update the ini file when I install a new version...

[George Tsavdaris]

Paste this position for Bihasa to analyze in analysis mode. do not analyze with it, this is not my point.
10/3k5p/R9/3N2p1N1/2p7/1a2P1P3/p3PP1PPP/6K3 b - - 0 1

Promote the a2 Pawn to a Queen. 1...a1Q+
See what happens.

Black promotes to a Queen and gives a check. Yet Bihasa ignores this and wants to play a move for white that doesn't remove the check!


It seems that promoting to a Chancellor has the same effect.

Strange but now after exiting and opening again Bihasa promoting to a Queen does not make this bug to appear and Bihasa wants to capture normally with the Rook in analysis mode. Yet promoting to a Chancellor makes the bug to appear again.

Well now by pasting the game with the above position and playing 10...a1Q+ makes Bihasa again to have this buggy behavior in analysis mode and it does not want to remove the check!


[Event "?"]
[Site "?"]
[Date "?"]
[Round "-"]
[White "?"]
[Black "?"]
[Result "*"]
[Variant "gothic"]
[FEN "9N/3R1kb2p/5p1n2/p1p3p1N1/1a8/1P2P1P3/4PP1PPP/5K4 b - - 0 1"]
[SetUp "1"]

{--------------
. . . . . . . . . N
. . . R . k b . . p
. . . . . p . n . .
p . p . . . p . N .
. a . . . . . . . .
. P . . P . P . . .
. . . . P P . P P P
. . . . . K . . . .
black to play
--------------}
1... Ke6 2. Rxg7 Ad2+ 3. Kg1 Axb3 4. Njh7 c4 5. Rb7 a4 6. Rb6+ Ke7 7. Nxf6
a3 8. Nd5+ Kd7 9. Rxh6 a2 10. Ra6