Maverick 0.60 is Released...

[Steve Maughan]

I've been working on Maverick and it's now time for a new release.

Here are the changes:

+ Support for Chess960
+ Basic King safety
+ Fixed a problem with using the opening book with Arena
+ Fixed an obscure bug which could crash the engine after a stop command
+ Transferred source code to Github (https://github.com/stevemaughan/maverick.git)
+ Made a bazillion tweaks which may, or may not, help!

Based on self play the improvement in strength is about 50 ELO. I'm hoping this translates into at least +30 ELO in the real world.

You can download here

The king safety code seems to have made the style of play more attractive. In this game against SOS 5.1 Maverick 0.60 gets a nice win as a result of the kingside attack:

[pgn]
[Event "?"]
[Site "?"]
[Date "2015.03.20"]
[Round "1"]
[White "Maverick 0.60"]
[Black "SOS 5.1"]
[Result "1-0"]

1. d4 {book} Nf6 {book} 2. Nf3 {book} e6 {book} 3. e3
{book} c5 {book} 4. Bd3 {book} Nc6 {book} 5. O-O {book} b6
{book} 6. Nc3 {+0.10/12 19s} Bb7 {-0.01/14 20s} 7. b3
{+0.08/12 50s} cxd4 {+0.14/14 18s} 8. exd4 {+0.12/13 51s}
Rc8 {+0.19/14 38s} 9. Bb2 {+0.03/12 22s} Be7 {+0.15/14 17s}
10. a3 {+0.12/12 30s} d5 {+0.10/15 18s} 11. Re1 {+0.29/11
17s} Qc7 {+0.15/13 16s} 12. b4 {+0.39/11 23s} Qf4 {+0.17/13
15s} 13. Bc1 {+0.44/11 33s} Qb8 {+0.04/15 17s} 14. Qd2
{+0.32/11 24s} O-O {+0.14/14 12s} 15. Bb2 {+0.17/11 21s} a6
{+0.05/13 18s} 16. b5 {+0.18/11 18s} axb5 {+0.19/15 18s}
17. Nxb5 {0.00/12 23s} Na5 {+0.23/15 23s} 18. Qg5 {+0.01/12
34s} Nc4 {+0.27/14 18s} 19. Bc1 {-0.05/11 12s} Qa8
{+0.41/14 20s} 20. Bf4 {-0.04/11 26s} Qa4 {+0.50/14 20s}
21. Reb1 {-0.04/11 34s} h6 {+0.45/14 11s} 22. Qh4 {-0.03/11
10s} Bc6 {+0.46/14 26s} 23. Bxh6 {0.00/11 18s} gxh6
{-0.01/15 37s} 24. Qxh6 {0.00/11 9.2s} Qa5 {-0.01/14 22s}
25. Ng5 {+3.20/11 9.6s} Qd2 {-3.86/13 58s} 26. Bh7+
{+5.50/11 9.8s} Kh8 {-9.95/14 59s} 27. Bf5+ {+6.96/12 11s}
Kg8 {-9.97/14 19s} 28. Rb3 {+7.60/11 10s} Ne3 {-9.97/14
21s} 29. Rxe3 {+7.91/11 10s} Qxe3 {-10.70/13 8.6s} 30. Bh7+
{+11.98/14 14s} Kh8 {-13.73/15 40s} 31. Bd3+ {+12.01/15
46s} Kg8 {-18.22/15 34s} 32. fxe3 {+12.14/14 28s} Bxb5
{-21.47/14 23s} 33. Bxb5 {+12.14/13 37s} Rc7 {-M16/15 21s}
34. Bd3 {+M15/12 7.7s} Rc3 {-M14/15 5.3s} 35. Rf1 {+M13/13
3.3s} Rxd3 {-M12/17 5.7s} 36. Rxf6 {+M11/11 0.005s} Rd1+
{-M10/26 3.0s} 37. Kf2 {+M9/9 0.002s} Rd2+ {-M8/28 5.0s}
38. Kf3 {+M7/7 0.003s} Rf2+ {-M6/30 5.1s} 39. Kxf2 {+M5/5
0.001s} Rd8 {-M6/29 2.8s} 40. Qh7+ {+M3/3 0.002s} Kf8
{-M2/28 0.84s} 41. Qxf7# {0.00/1 0.002s, White mates} 1-0[/pgn]

[D]2r2rk1/4bpp1/1pb1pn1p/1N1p4/q1nP1B1Q/P2B1N2/2P2PPP/RR4K1 w - -

23. Bxh6!

[Henk]

Just play in the online blitz tourney this Saturday and we will believe you.

[Modern Times]

Nice to see another Chess960 engine !

[Steve Maughan]

Hi Ray,

Adding chess960 support was more difficult than I anticipated. I may do a blog post about some of gotchas I encountered. I've tried to test the chess960 as much as possible. Please let me know if you see anything strange when playing chess960.

- Steve

[Evert]

Adding chess960 support was more difficult than I anticipated. I may do a blog post about some of gotchas I encountered. I've tried to test the chess960 as much as possible. Please let me know if you see anything strange when playing chess960.

Castling is a bitch when doing Chess960. SjaakII plays Chess960 too, if you want to check things against another (weaker) engine.

I've found it useful to include the following set of test positions for my movegen validation set (fields are FEN, depth, perft). They have perft at 3 different depths to make debugging a bit easier:

Code: Select all

   // Some FRC postions by Reinhard Scharnagl
   // (http://www.talkchess.com/forum/viewtopic.php?t=55274)
   { "r1k1r2q/p1ppp1pp/8/8/8/8/P1PPP1PP/R1K1R2Q w KQkq - 0 1", 1, 23 },
   { "r1k2r1q/p1ppp1pp/8/8/8/8/P1PPP1PP/R1K2R1Q w KQkq - 0 1", 1, 28 },
   { "8/8/8/4B2b/6nN/8/5P2/2R1K2k w Q - 0 1", 1, 34 },
   { "2r5/8/8/8/8/8/6PP/k2KR3 w K - 0 1", 1, 17 },
   { "4r3/3k4/8/8/8/8/6PP/qR1K1R2 w KQ - 0 1", 1, 19 },

   { "r1k1r2q/p1ppp1pp/8/8/8/8/P1PPP1PP/R1K1R2Q w KQkq - 0 1", 2, 522 },
   { "r1k2r1q/p1ppp1pp/8/8/8/8/P1PPP1PP/R1K2R1Q w KQkq - 0 1", 2, 738 },
   { "8/8/8/4B2b/6nN/8/5P2/2R1K2k w Q - 0 1", 2, 318 },
   { "2r5/8/8/8/8/8/6PP/k2KR3 w K - 0 1", 2, 242 },
   { "4r3/3k4/8/8/8/8/6PP/qR1K1R2 w KQ - 0 1", 2, 628 },

   { "r1k1r2q/p1ppp1pp/8/8/8/8/P1PPP1PP/R1K1R2Q w KQkq - 0 1", 5, 7096972 }, 
   { "r1k2r1q/p1ppp1pp/8/8/8/8/P1PPP1PP/R1K2R1Q w KQkq - 0 1", 5, 15194841 }, 
   { "8/8/8/4B2b/6nN/8/5P2/2R1K2k w Q - 0 1", 5, 3223406 }, 
   { "2r5/8/8/8/8/8/6PP/k2KR3 w K - 0 1", 5, 985298 }, 
   { "4r3/3k4/8/8/8/8/6PP/qR1K1R2 w KQ - 0 1", 5, 8992652 },

[Steve Maughan]

Thanks Evert - I think I did use these positions when implements chess960. I was amazed at the code which needed to changed.

- Steve

[Charly]

Hi,

I'm very happy that you work again on your engine. I'll test the updated version in the League "Euwe" in my 4th edition.

Regards

Arnaud Lohéac

[Steve Maughan]

Thanks Arnaud - let me know how it does!

- Steve

[Graham Banks]

I've been working on Maverick and it's now time for a new release.

Here are the changes:

+ Support for Chess960
+ Basic King safety
+ Fixed a problem with using the opening book with Arena
+ Fixed an obscure bug which could crash the engine after a stop command
+ Transferred source code to Github (https://github.com/stevemaughan/maverick.git)
+ Made a bazillion tweaks which may, or may not, help!

Based on self play the improvement in strength is about 50 ELO. I'm hoping this translates into at least +30 ELO in the real world.

You can download here

Thanks Steve. Sounds exciting!

[Bloodbane]

Good to see that you have kept on working on your chess engine. When I started working on my chess engine my first objective was to reach 2500 before you (I hope that doesn't offend you) so I have a sort of special relationship with Maverick, and I'd hate to see it be abandoned. I'm sure you'll reach 2500+ soon.