NEW! NeroWB6.0 engine now available!

[J_Huikari]

You can find my new NeroWB-engine from my new homepage http://www.kanetti.fi/jari.huikari/

[Uri Blass]

There is some bug in the program

It detects draw by repetition too late

Here are 3 of my games against it at 3 minutes/40 moves

[Event "Computer chess game"]
[Site "URI-AMD"]
[Date "2005.11.27"]
[Round "-"]
[White "nerowb60"]
[Black "agur"]
[Result "1/2-1/2"]
[TimeControl "40/180"]

1. Nf3 Nf6 2. g3 e6 3. Bg2 Be7 4. d3 O-O 5. Bf4 d5 6. Nc3 Bd6 7. Bg5 h6 8.
Bh4 g5 9. Nxg5 hxg5 10. Bxg5 Be7 11. e4 dxe4 12. Bh6 Re8 13. Qd2 Kh7 14.
Nxe4 Nxe4 15. Bxe4+ f5 16. Bf3 Rg8 17. Bf4 Nc6 18. Bxc6 bxc6 19. Qa5 Qd5
20. Qxd5 cxd5 21. Bxc7 Bf6 22. O-O-O Bd7 23. Rhe1 Rac8 24. Be5 Kg6 25. f4
Rh8 26. Bxf6 Kxf6 27. h4 Rhg8 28. Re3 Rg4 29. d4 Rcg8 30. Ra3 Ra8 31. Rdd3
a5 32. Rdb3 a4 33. Rb7 Rd8 34. Ra7 Be8 35. Re3 Bd7 36. a3 Kg6 37. c3 Kf6
38. Kd2 Ke7 39. Ra5 Rdg8 40. Rxd5 Rxg3 41. Re1 R3g4 42. Rxf5 Rxh4 43. Rg5
Rxg5 44. fxg5 Rg4 45. Re5 Kf7 46. d5 exd5 47. Rxd5 Ke6 48. Rc5 Rg2+ 49. Kc1
Re2 50. Ra5 Re5 51. Ra6+ Kf5 52. g6 Re7 53. Kd2 Rg7 54. c4 Rxg6 55. Rxg6
Kxg6 56. c5 Kf5 57. Kd3 Ke5 58. Kc4 Ke4 59. b4 axb3 60. Kxb3 Kd5 61. Kb4
Kc6 62. a4 Bc8 63. Kc4 Ba6+ 64. Kb4 Bb7 65. Kc4 Ba6+ 66. Kb4 Bb7 67. Kc4
Ba6+ 68. Kb4 Bb7 69. Kc4 Ba6+ 70. Kb4
{draw 3 rep.} 1/2-1/2

[Event "Computer chess game"]
[Site "URI-AMD"]
[Date "2005.11.27"]
[Round "-"]
[White "nerowb60"]
[Black "agur"]
[Result "0-1"]
[TimeControl "40/180"]

1. Nf3 Nf6 2. g3 d5 3. Nc3 c5 4. Ng5 h6 5. Nf3 Nc6 6. Bg2 Bg4 7. h3 Bh5 8.
Nh4 e6 9. g4 Bg6 10. Nxg6 fxg6 11. d3 Be7 12. g5 Nh5 13. gxh6 gxh6 14. Bd2
Qd7 15. Qc1 g5 16. h4 Nf4 17. Bxf4 gxf4 18. Qxf4 O-O-O 19. h5 Rdf8 20. Qd2
Bg5 21. e3 d4 22. Nb5 a6 23. Na3 dxe3 24. fxe3 Ne5 25. Qc3 Qc7 26. e4 Ng4
27. Rh3 Ne3 28. Bf3 Rh7 29. Nc4 Nxc4 30. Qxc4 Re7 31. Bg4 Qd6 32. Ke2 Qf4
33. Qxc5+ Rc7 34. Bxe6+ Kb8 35. Qxf8+ Qxf8 36. c3 Qd6 37. Bg8 Qf8 38. Be6
Qf4 39. Rd1 Qf6 40. Bc4 Qf4 41. Be6 Re7 42. Bg8 Qg4+ 43. Rf3 Qxh5 44. a3
Rg7 45. Be6 Bf4 46. Rf1 Rg2+ 47. Rf2 Rxf2+ 48. Kxf2 Qh2+ 49. Kf1 Qxb2 50.
c4 Qxa3 51. Kg2 Qb2+ 52. Rf2 Qe5 53. Rf3 Qxe6 54. Rxf4 a5 55. Rf8+ Ka7 56.
Rf5 b6 57. Kf3 a4 58. Rh5 a3 59. Rh2 Qf6+ 60. Ke3 Qb2 61. Re2 Qxe2+ 62.
Kxe2 a2 63. Ke3 a1=Q 64. Kd2
{White resigns} 0-1

[Event "Computer chess game"]
[Site "URI-AMD"]
[Date "2005.11.27"]
[Round "-"]
[White "agur"]
[Black "nerowb60"]
[Result "1-0"]
[TimeControl "40/180"]

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Be2 e5 7. Nf3 Qa5 8.
O-O Bg4 9. h3 Bxf3 10. Bxf3 h6 11. Nd5 Nbd7 12. Bd2 Qd8 13. Nxf6+ Nxf6 14.
Bc3 Qb6 15. b3 Be7 16. Qd2 Qc6 17. Rfe1 Kf8 18. Bb2 b5 19. Rac1 Qb7 20. c4
bxc4 21. Rxc4 Kg8 22. Rec1 Re8 23. Qa5 Bd8 24. Qb4 Qxb4 25. Rxb4 Be7 26.
Rb7 Kh7 27. Rcc7 Bf8 28. Rxf7 Kg6 29. Rfc7 a5 30. Ra7 a4 31. Rxa4 Re6 32.
Raa7 h5 33. a3 h4 34. b4 Re8 35. b5 Kh6 36. b6 g5 37. a4 Kg6 38. a5 Bh6 39.
a6 Re6 40. b7 Ree8 41. Ba3 d5 42. exd5 e4 43. Bg4 e3 44. fxe3 Rb8 45. d6
Rhe8 46. Rc8 Nxg4 47. hxg4 Rexc8 48. bxc8=Q Rxc8 49. d7 Rd8 50. Be7 Rb8 51.
d8=Q Rxd8 52. Bxd8 Bf8 53. Rb7 Bc5 54. Bb6 Bd6 55. a7 h3 56. gxh3 Kf6 57.
a8=Q Bf8 58. Qxf8+ Ke5 59. Re7+ Kd5 60. Qd8+ Kc4 61. Rc7+ Kb3 62. Qb8 Kb2
63. Bd4+ Ka3 64. Ra7#
{White mates} 1-0

[J_Huikari]

Yes. I have noticed that too. It announces draw by 3 repetitions always after 4 repetitions. I'll fix this for next release. (And it only looks for same position to occur after its own move, not after opponents move which may cause draw one half move earlier.)

Same problem exists in draw by insufficient material. It announces that one move later than it should.

But at least the game won't continue much further anymore.

Thanks for posting those games!

[J_Huikari]

Now it shows draw by 3 repetitions, not by 4 rep.

Also it recognizes draw by insufficien material right away.

Showing the thinking line works better now too, and I think the
playing strenght has improved a little too after some minor changes
in it's thinking.

http://www.kanetti.fi/jari.huikari/

[Uri Blass]

Now it shows draw by 3 repetitions, not by 4 rep.

Also it recognizes draw by insufficien material right away.

Showing the thinking line works better now too, and I think the
playing strenght has improved a little too after some minor changes
in it's thinking.

http://www.kanetti.fi/jari.huikari/

I think that one improvement that you may make in playing strength is
simply before calling qsearch to extend by one ply in case that the king is under threat.

You can use nero for analysis by the following steps.
1)choosing long time control under winboard(you can use 1000 minutes/40 moves)
2)copy the following game to winboard

[Event "5'/40"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "New game"]
[Black "Uri Blass"]
[Result "*"]
[PlyCount "39"]
[TimeControl "40/300:0/0:0/0"]

{64MB, Fritz8.ctg, URI-AMD} 1. e4 e5 2. Nf3 Nc6 3. d4 exd4 4. Nxd4 Nf6 5. Nxc6
bxc6 6. Bd3 d5 7. e5 Ng4 8. O-O f6 9. exf6 Nxf6 10. g3 Bd6 11. f4 O-O 12. Kf2
Qd7 13. Ke2 Qh3 14. Rh1 Bc5 15. Be3 Bd6 16. Nd2 Bb4 17. Nf1 Bc5 18. Kd2 Re8 19.
Qf3 Bg4 20. Qf2 *

3)Telling nero machine black

If you do it you get the following analysis:


17015 <first : 1 -900 0 34 h3h4
17015 <first : 1 -300 0 67 h3h5
17015 <first : 1 -100 2 527 c5d6
17015 <first : 1 -10 2 627 c5e3
17015 <first : 1 0 2 662 c5b4
17031 <first : 2 68 3 1607 c5b4 d2c1
17031 <first : 2 80 3 1720 c5e3 f1e3
17031 <first : 2 131 3 1969 g4f3 f2f3
17047 <first : 3 -181 5 3379 g4f3 f2f3 h3g4
17062 <first : 3 -142 5 3917 c5a3 b2a3 f6e4
17078 <first : 3 91 7 4536 c5b6 e3b6 c7b6
17125 <first : 3 98 11 8314 f6e4 d3e4 c5e3
17187 <first : 4 40 19 13658 f6e4 d3e4 c5e3
18297 <first : 5 41 130 92004 f6e4 d3e4 c5e3
19218 <first : 5 60 222 154505 c5b6 d2c1 e8e7
29703 <first : 6 39 1271 930003 c5b6 d2c1 h7h6
45547 <first : 6 46 2855 2122090 c5b4 d2c1 g4d7
61093 <first : 6 58 4410 3278099 d5d4 e3d4 e8e2
67468 <first : 6 83 5047 3746770 e8e3 f1e3 a8e8
166812 <first : 7 124 14980 11026214 e8e3 f1e3 a8e8
316109 <first : 7 307 29911 22006009 d5d4 d3c4 g8h8
473890 <first : 8 355 45686 33214916 d5d4 a1e1 d4e3
GameEnds(0, (null), 2)
493640 >first : force
493640 >first : quit

I can see that Nero needs depth 7 to find d4
The problem is probaly that Nero calculates the following line at depth 6

d4 Bxd4 Bxd4 Bc4+ Kh8 Qxd4 Qg2+

Bc4+ is extended but Qg2+ is not extended.
I could prove that Qg2+ is not extended by copying the following game

[Event "5'/40"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "New Game"]
[Black "Uri Blass"]
[Result "*"]
[PlyCount "45"]
[TimeControl "40/300:0/0:0/0"]

{64MB, Fritz8.ctg, URI-AMD} 1. e4 e5 2. Nf3 Nc6 3. d4 exd4 4. Nxd4 Nf6 5. Nxc6
bxc6 6. Bd3 d5 7. e5 Ng4 8. O-O f6 9. exf6 Nxf6 10. g3 Bd6 11. f4 O-O 12. Kf2
Qd7 13. Ke2 Qh3 14. Rh1 Bc5 15. Be3 Bd6 16. Nd2 Bb4 17. Nf1 Bc5 18. Kd2 Re8 19.
Qf3 Bg4 20. Qf2 d4 21. Bxd4 Bxd4 22. Bc4+ Kh8 23. Qxd4 *

Here nero needs depth 2 to find Qg2+ with a winning score and not depth 1.

18969 <first : 1 -1001 0 43 h3h4
18969 <first : 1 -401 0 85 h3h5
18969 <first : 1 -100 0 182 h3g2
19000 <first : 2 405 3 1631 h3g2 d2c1

one of the advantage of the better programs is that they extend Qg2+ so at depth 1 they search Qg2+(check extension so still depth 1) king escape depth 0 call to qsearch QxR inside the qsearch so they can see that black captures the rook at depth 1.


Uri

[J_Huikari]

I'm running now testgames. Current version versus modified version,
which values bishops and knights a little higher and does more
extensions. And doesn't resign so easily. After 6 games the modified
version leads 4-2 but it's far too early to say anything about the result.

My decision has been not to resign in games of type all moves in N mins.
Because opponent may be short of time.

Nero resigns in N moves in M minutes mode if it has had bad enough
score for many enough consecutive moves. I modified this that it must
be at least 2 rooks down for ten moves. But it doesn't resign if there
hasn't been progress in the game for many moves. (Near draw by 50
move rule.) And finally it doesn't resign if it finds mate in N for the
opponent. Game won't last long after that anyway. (now it's 4.5-3.5)