Broken links

I am aware that all the old links toward file are broken: Blogger only store the pictures, so I should find a way to store the files. I think google drive seems a solution that deserve to be tried. So I will transfer all the material on google drive. I thanks all the persons that reported broken links in the comments and I apologize for this. 

Milady 4.05b

Finally, I'm back in computer chess with a new version of Milady, 4.05b . In this version, evaluation is simple and rewritten from scratch. It takes into account material, material imbalance, king attack, passed pawn, control of cases, control of centre, development, double pawns,.. Most of all, the evaluation is smoothly normalized such that there are no jump in the evaluation. I completely removed all the cut-off based on Alpha or on Beta that were in the evaluation before. In addition, the analysis feature is improved and it can follows move input during move analysis. Framework 4 required to work.

This version reached 2270 against my Novag Star Diamond (rated 2150 international ELO). All the games can be find here. These are examples of position reached during this match :

4Nb1k/5p2/2Q4p/1p2P1p1/2bP3P/4BP2/1q4PK/8 w - - 1 50  bm hxg5; During game 3, it is white to move after a blunder from black (Qb2), there are several good moves for white but the best is hxg5, leading for instance to  1.hxg5 Bf1 2.f4 Qe2 3.Nd6 Bxd6 4.Qe8+ Kg7 5.gxh6+ Kg6 6.Qg8+ Kf5 7.Qh7+ Kg4 8.Qg7+ Kf5 9.exd6 Qxe3 10.Qg5+ Ke4 11.Qe5+ Kd3 12.Qxb5+ Ke4.  

7k/pb2q3/8/2pppBb1/6Q1/P3PN2/1PK3P1/8 b - - 1 29 bm Bh6 ; game 5 Star Diamond played here Bxe3 (maybe not a bad move, Bh6 or Bf6 may be better although). It is hard to believe for a human, but here black are lost. Mate menace, checks, will lead to win material. 

Note that in engine tournament, the strength will be lower since there are some program crash due to GUI command missing: I should investigate on that...

The program is available in the download section, where you can find the three last versions. 

Milady_3.25 - Fidelity Phantom (19-1)

Game over! Milady won by a large score. I wonder why the score is so high, that is not normal. The same phenomena arose against the Mephisto Expert. The score leads to a rating of 2346. All the games are available here.So that maybe there is a phenomena, when a engine is tactically much stronger than an other, the stronger engine will win anyway? I have not the answer but I have seen this hypothesis on SchachComputer.info. To be clarified...

End of the match Phantom - Mephisto MMII : 13 - 7

I will update the page both of Phantom and MM II (double work :P ) . The Phantom won 13-7, giving a performance of 1877 +- 72, that is consistent with its ranking of 1835. Performance of MM II is of 1727+- 72.

 

Total games	20	Mean rating	1802,5	Error Barre (%)	8,9
	Score	Rating	Perf (%)	Rating opponents	Perf (elo)	Error Bar (Elo)	Elo
Mephisto MMII	7,0	1770,0	35,0	1835,0	1715,0	71,6	1727,5
Fidelity Phantom	13,0	1835,0	65,0	1770,0	1890,0	71,6	1877,5

 

  

During the match, I had the feeling that the Phantom was stronger than MM II. Some of the games were very pleasant, as can be seen below (game 17). Some of the games were just a succession of blunders from both sides. An other interesting position is this one :

 

6k1/p2b2p1/p1pb1p1p/Pr1p4/R2P3P/3NP3/5PPN/6K1 b - - 7 32 ; game 18 MMII-Phantom

 

After 31. Ra4 ?? there are several winning moves for blacks. The Phantom found one of the winning move. Could you find it?

4 positions from game 17 Fidelity Phantom- MM II

During the 17th game, I have taken 4 positions which can be used as tests for your engine (or you).

The solutions are given at the end.

 

1k2r2r/pp2b1pp/2p3n1/Q3P3/PP1N2q1/2P5/3B2PP/R3R1K1 b - a3 0 25 ; game 17 Phantom - MMII black wins (hard)

 

1k6/p5pp/1b2r3/3QP3/PP4qn/2P5/R2B1rPP/4R2K w - - 7 32 ; game 17 Phantom - MMII white is lost even with their best move (hard)

1k6/p5pp/1b4r1/3QP3/PP4qn/2P5/R4rPP/2B1R2K w - - 9 33 ; game 17 Phantom - MMII white wins (hard)

 

4k3/4P1p1/p2Br3/P1P4p/1P5n/8/5b1P/5R1K w - - 0 44 ; game 17 Phantom - MMII white wins (easy)

 

 Solutions :

 

1/ unseen winning move 25...Nh4

2/ even the best move 32.... Bc1 (or Be3) is loosing, for instance look at the Fruit 2.1 analyses lines:

 

14/47 00:37   39 267 399 1 058 211 -2,48 1.Bc1 Rxa2 2.Qxa2 Rxe5 3.Qg8+ Kb7 4.Qf7+ Bc7 5.Rg1 Rh5 6.a5 a6 7.b5 Rxb5 8.Be3 Rb2 9.Bf2

14/48 00:52   56 288 295 1 072 223 -2,26 1.Be3 Rxa2 2.Qxa2 Bxe3 3.h3 Qxg2+ 4.Qxg2 Nxg2 5.Kxg2 Bd2 6.Rf1 Bxc3 7.Rf8+ Kb7 8.Rf7+ Kb6 9.Rxg7 Bxb4 10.Rxh7 Rxe5 11.Rd7

15/51 02:11  139 916 580 1 062 940 -2,25 1.Be3 Rxa2 2.Qxa2 Bxe3 3.h3 Qxg2+ 4.Qxg2 Nxg2 5.Re2 Nf4 6.Rxe3 Nd5 7.Rf3 Rxe5 8.Kg2 Re7 9.Kg3 Kb7 10.b5

3/ after 32...Rg6?? winning move 33. Rxf2
4/ after 43...a6?? winning move 44. b5

Picture of Phantom - MM II match

It was the 13th game actually in progress (Phantom drives white pieces):

Challenger Chess Champion 7.5 - 2.5 Chess Titan

The game number 10 lost by Chess Titan which gives two figures for nothing (as it is the case in some games). The performance of Chess Challenger comes near from it's true level (around 1500):

[Date "2013.01.01"]
[Round "10"]
[White "Chess Titan"]
[Black "Champion Chess Challenger"]
[Result "0-1"]
[BlackElo "1535"]
[ECO "A41"]
[Opening "Neo-Old Indian"]
[Time "16:56:16"]
[Variation "Wade Defence, 3.c4 Nd7 4.Nc3"]
[WhiteElo "1300"]
[TimeControl "1800"]
[Termination "adjudication"]
[PlyCount "52"]
[WhiteType "program"]
[BlackType "program"]
1. c4 c5 2. Nc3 Nc6 3. e3 d6 4. d4 cxd4 5. exd4 Qb6 6. Nd5 Qd8 7. Nf3 Bg4
8. h3 Bxf3 9. gxf3 e6 10. Bg5 f6 11. Bd2 exd5 12. cxd5 Nb8 13. Bc4 Qe7+ 14.
Be3 Nd7 15. f4 Qe4 16. O-O Nb6 17. Qh5+ Kd8 18. b3 Ne7 19. Rfd1 Qg6+ 20.
Qxg6 hxg6 21. Be2 Rxh3 22. Bg4 Rh4 23. f3 f5 24. Bf2 Rh8 25. a4 Nbxd5 26.
Bg3 fxg4 {User Adjudication} 0-1