Upcoming: Opening Test Suite in Chess for Android

[AartBik]

I finished an upcoming new feature in Chess for Android where a tournament can be played from both sides of any external opening book (in PGN format), such as the famous "Nunn Positions" (published by GM John Nunn) or the "Silver Opening Suite" (published by Albert Silver).

As a sneak preview, here are two games between Stockfish 2.1 and Komodo 3.0 (32MB hash, 1s./move) from both sides of an opening line from the Silver Opening Suite.

[Event "Chess for Android Tournament"]
[Site "United States"]
[Date "2011.08.26"]
[Round "3"]
[White "Stockfish 2.1"]
[Black "Komodo32 3 AB"]
[Result "0-1"]
[PlyCount "122"]
[TimeControl "1/1"]
[Device "Nexus One"]

1. Nf3 Nf6
2. g3 g6
3. Bg2 Bg7
4. O-O O-O
5. d3 d6
6. e4 {from PGN 0.01} c5 {[10/15]4/b1c3 1.282}
7. Nc3 {[8]-15/b8c6 1.125} Nc6 {[11/15]12/a2a3 1.315}
8. Bd2 {[8]-25/c8g4 1.124} Qb6 {[11/15]12/c3a4 1.351}
9. Rb1 {[9]-13/c8g4 1.115} Bd7 {[10/14]12/d2e3 1.559}
10. Re1 {[9]-15/a8e8 1.127} e5 {[9/15]-8/d2g5 1.653}
11. Bg5 {[10]17/d7e6 1.118} Be6 {[12/14]0/d1d2 1.47}
12. Bxf6 {[10]37/g7f6 1.12} Bxf6 {[11/18]-12/c3d5 1.265}
13. Nd5 {[11]33/e6d5 1.121} Bxd5 {[12/17]-4/e4d5 1.323}
14. exd5 {[12]32/c6e7 1.125} Ne7 {[12/16]-4/f3d2 1.199}
15. Nd2 {[10]28/f6g7 1.126} Rfd8 {[11/13]-8/a2a3 1.342}
16. c3 {[9]40/b6c7 1.114} Qa6 {[11/15]4/d1c2 1.378}
17. Qc2 {[9]52/a8c8 1.122} Rac8 {[11/13]0/b1a1 1.278}
18. b4 {[9]52/c5b4 1.118} cxb4 {[11/16]0/b1b4 1.346}
19. Rxb4 {[9]38/f6g7 1.126} b6 {[11/15]-12/d2e4 1.233}
20. Rbb1 {[8]30/a6a3 1.124} Qa5 {[10/14]4/d2c4 1.347}
21. c4 {[8]3/e7d5 1.119} Nxd5 {[10/15]101/a2a3 1.243}
22. a3 {[8]-13/d5e7 1.129} Ne7 {[10/15]50/b1b4 2.025}
23. Ne4 {[9]-26/f6g7 1.122} Bg7 {[11/17]101/e4c3 1.275}
24. Qc3 {[9]-28/d6d5 1.13} Qa6 {[13/19]96/c3b3 1.958}
25. Bh3 {[8]-82/c8c7 1.138} Rc7 {[10/15]165/c3b3 1.44}
26. Ng5 {[8]-97/d6d5 1.114} h6 {[10/13]226/g5e4 1.316}
27. Nf3 {[10]-141/e5e4 1.117} e4 {[11/17]406/f3d4 1.289}
28. Nd4 {[10]-157/e7c6 1.123} Nc6 {[11/18]460/e1e4 1.166}
29. Rxe4 {[10]-179/c6d4 1.113} f5 {[12/21]460/e4h4 1.685}
30. Rf4 {[8]-181/g6g5 1.133} g5 {[12/18]460/c3e1 1.149}
31. c5 {[8]-193/g5f4 1.122} Rf7 {[11/17]456/c3b3 1.478}
32. cxb6 {[7]-302/b6c5 1.125} gxf4 {[11/17]525/c3c6 1.414}
33. Qxc6 {[7]-316/ 1.122} Bxd4 {[11/16]553/c6b5 1.152}
34. Bf1 {[7]-340/f4g3 1.123} fxg3 {[10/17]565/h2g3 1.501}
35. hxg3 {[10]-336/a6a3 1.125} Qxa3 {[10/17]541/c6d5 1.189}
36. Qd5 {[9]-337/a3a4 1.123} Bxb6 {[10/19]585/d3d4 1.153}
37. d4 {[9]-329/a3c3 1.125} Kg7 {[12/18]622/f1e2 1.591}
38. Be2 {[8]-334/d8f8 1.118} Qa5 {[12/17]642/d5a5 1.644}
39. Qe6 {[9]-364/b6d4 1.125} Bxd4 {[11/18]723/e2h5 1.881}
40. Bh5 {[10]-371/f7c7 1.111} Rc7 {[11/19]731/b1e1 1.921}
41. Rd1 {[9]-403/a5e5 1.113} Qe5 {[10/17]832/e6g6 1.168}
42. Qg6+ {[9]-509/g7h8 1.122} Kh8 {[12/26]868/g6h6 1.468}
43. Kg2 {[10]-491/d8g8 1.124} Rg8 {[11/24]1026/g6h6 1.16}
44. Qxh6+ {[9]-638/c7h7 1.12} Rh7 {[12/26]1234/h6d2 1.432}
45. Qd2 {[11]-647/d4f2 1.116} Bxf2 {[13/30]1430/d2d6 1.341}
46. Qxf2 {[10]-640/h7h5 1.114} Rxh5 {[13/20]7973/d1d2 1.403}
47. Qd4 {[11]-656/g8g3 1.127} Rxg3+ {[12/27]1498/g2f2 1.155}
48. Kf2 {[11]-656/h5h2 1.133} Rh2+ {[15/35]8452/f2f1 1.355}
49. Kf1 {[12]-651/g3f3 1.12} Ra3 {[16/30]8533/f1g1 1.178}
50. Kg1 {[10]-660/h2c2 1.134} Raa2 {[17/25]8563/d4e5 1.401}
51. Qxe5+ {[11]-705/d6e5 1.116} dxe5 {[17/26]8383/d1d8 1.282}
52. Rc1 {[11]-747/h8g7 1.125} Rag2+ {[15/22]8678/g1f1 1.207}
53. Kf1 {[15]-903/g2c2 1.123} Rc2 {[17/24]9666/c1c2 1.349}
54. Rxc2 {[16]-1070/h2c2 1.112} Rxc2 {[16/24]mate:10/f1e1 1.37}
55. Ke1 {[17]-1128/h8g7 1.118} f4 {[16/18]mate:7/e1d1 1.163}
56. Kf1 {[15]-1277/h8g7 1.127} a5 {[18/12]mate:6/f1e1 1.29}
57. Ke1 {[13]mate:-12/f4f3 0.653} Rb2 {[24/10]mate:5/e1f1 1.211}
58. Kd1 {[9]mate:-4/a5a4 0.104} a4 {[40/8]mate:4/d1e1 1.303}
59. Ke1 {[5]mate:-3/a4a3 0.041} a3 {[100/6]mate:3/e1f1 0.986}
60. Kd1 {[2]mate:-2/a3a2 0.019} a2 {[100/4]mate:2/d1c1 0.689}
61. Ke1 {[1]mate:-1/a2a1q 0.018} a1=Q# 0-1

[Event "Chess for Android Tournament"]
[Site "United States"]
[Date "2011.08.26"]
[Round "4"]
[White "Komodo32 3 AB"]
[Black "Stockfish 2.1"]
[Result "1-0"]
[PlyCount "103"]
[TimeControl "1/1"]
[Device "Nexus One"]

1. Nf3 Nf6
2. g3 g6
3. Bg2 Bg7
4. O-O O-O
5. d3 d6
6. e4 {from PGN 0.008} c5 {[10]4/e4e5 1.126}
7. Nc3 {[11/13]0/b8c6 1.404} Nc6 {[10]18/c1g5 1.127}
8. Bd2 {[11/12]-12/d8b6 1.386} Bg4 {[9]14/h2h3 1.125}
9. h3 {[11/15]0/g4e6 1.611} Bxf3 {[9]15/g2f3 1.127}
10. Bxf3 {[11/15]20/d8b6 1.511} Nd4 {[9]4/f3g2 1.124}
11. Rb1 {[10/15]8/d8b6 1.463} Nd7 {[8]23/f3g2 1.11}
12. Bg2 {[11/14]16/e7e6 1.567} Rc8 {[9]-3/c3d5 1.104}
13. f4 {[10/14]20/d7b6 1.542} c4 {[8]21/d3c4 1.106}
14. Be3 {[11/14]32/d7b6 1.774} e6 {[7]39/d3c4 1.186}
15. Qd2 {[10/16]28/d8b6 1.779} Qb6 {[8]17/f1d1 1.111}
16. Kh2 {[10/12]44/c4d3 1.525} cxd3 {[6]19/d2d3 1.176}
17. Qxd3 {[11/20]60/b6b4 1.297} Qb4 {[8]-8/a2a3 1.113}
18. Rfd1 {[11/19]80/d7c5 1.299} Nc5 {[9]-81/d3d2 1.104}
19. a3 {[12/18]125/c5d3 1.263} Qa5 {[9]-53/d3d2 1.113}
20. Qd2 {[12/21]96/d4b5 1.268} Nb5 {[10]-57/c3b5 1.116}
21. Nxb5 {[11/16]109/a5b5 1.195} Qxb5 {[10]-61/b2b3 1.126}
22. b3 {[11/17]105/f8d8 1.163} Rfd8 {[10]-60/e4e5 1.107}
23. c4 {[11/17]109/b5a6 1.258} Qe8 {[9]-44/d2a5 1.124}
24. Qa5 {[10/15]72/e8e7 1.482} e5 {[9]-41/f4f5 1.13}
25. f5 {[9/14]56/e8c6 1.371} Qc6 {[8]-50/d1d5 1.128}
26. Qb5 {[10/16]48/c6c7 1.519} Qc7 {[9]-34/h2g1 1.122}
27. Bg5 {[11/16]48/f7f6 1.443} Re8 {[9]-24/h2g1 1.125}
28. fxg6 {[10/15]36/h7g6 1.448} hxg6 {[10]-14/g5e3 1.123}
29. Be3 {[12/15]40/e8d8 1.243} a6 {[9]-19/b5b4 1.124}
30. Qb4 {[11/15]56/c7c6 1.455} Qe7 {[8]-24/d1d5 1.14}
31. Qd2 {[10/14]60/e8d8 1.392} Rcd8 {[9]-23/d2d5 1.138}
32. h4 {[10/15]72/c5e6 1.274} Qc7 {[9]-23/d2f2 1.127}
33. Rf1 {[11/14]76/c5e6 1.492} Ne6 {[10]-8/d2d5 1.124}
34. h5 {[10/18]117/g6h5 1.259} gxh5 {[10]-23/g2f3 1.116}
35. Qe2 {[11/17]96/h5h4 1.213} a5 {[9]-12/e2h5 1.126}
36. Qxh5 {[10/19]165/e6d4 1.447} a4 {[10]-25/b3a4 1.109}
37. bxa4 {[11/16]202/e6d4 1.511} Nd4 {[9]-20/h2g1 1.124}
38. Rf2 {[10/15]189/e8f8 1.21} Ra8 {[9]-5/b1b4 1.118}
39. Rbf1 {[10/16]157/f7f6 1.386} Rf8 {[9]-5/e3d4 1.124}
40. Bxd4 {[10/19]387/e5d4 1.67} exd4 {[8]-151/e4e5 1.106}
41. e5 {[11/16]549/d6d5 2.1} d5 {[7]-162/g2d5 1.119}
42. Bxd5 {[11/20]1143/c7e5 1.629} Qxe5 {[8]-219/d5f7 1.136}
43. Rf5 {[12/22]1212/e5d5 1.685} Ra6 {[8]-657/d5f7 1.12}
44. Bxf7+ {[12/29]2227/f8f7 1.561} Rxf7 {[12]-1163/h5f7 1.131}
45. Qxf7+ {[13/28]10062/g8h8 1.387} Kh8 {[13]-1270/f5e5 1.116}
46. Rxe5 {[16/15]mate:7/g7e5 1.205} Bxe5 {[11]-1359/f7h5 1.125}
47. Qf8+ {[16/12]mate:6/h8h7 1.297} Kh7 {[10]mate:-5/f1f7 1.053}
48. Rf7+ {[18/10]mate:5/h7g6 1.303} Kg6 {[3]mate:-4/f8g8 0.034}
49. Qg8+ {[26/8]mate:4/g6h5 1.311} Kh5 {[2]mate:-3/g8h7 0.022}
50. Qh7+ {[51/6]mate:3/h5g5 1.557} Kg5 {[1]mate:-2/f7f5 0.015}
51. Rf5+ {[100/4]mate:2/g5g4 0.697} Kg4 {[1]mate:-1/h7h3 0.021}
52. Qh5# 1-0

[AartBik]

Ooops. Never publish results before the feature is fully tested. In the games above, Stockfish is the winner for both games (due to a bug, the players were reversed in the header, although all analysis output is correctly reported). Sorry about that. In the final version this will work correctly

[thorstenczub]

cool, so the testing methods get more and more perfect.

can you make a tournament table that shows the results of the match opponents after or during the tournament ? so if e.g. 40 from 60 games have been played one can see which of the participants is leading so far ??

[AartBik]

Here are results of a tournament between the most recent Android versions of Stockfish, Komodo, and Critter (32MB cache, 1s./move) from both sides of the "Silver Opening Suite" (300 games total). Stockfish is the clear winner. Komodo now comes in third (probably because v3 honors fixed time per move better than previous versions, which may have skewed estimates of their strength a bit).

Code: Select all

1 Stockfish 2.1  **         60.0-40.0  76.5-23.5  136.5/200
2 Critter 1.2    40.0-60.0  **         72.5-27.5  112.5/200
3 Komodo32 3 AB  23.5-76.5  27.5-72.5  **          51.0/200

Thorsten, proper layout is a recurring issue when adding features. I simply do not have the space to print a cross-table during the tournament. Chess for Android shows a simple cross-table once the tournament is over though (see this screenshot), and you can upload the PGN of the tournament to your favorite chess application for further processing....

[thorstenczub]

yes aart. but i would like to see this "result" feature OPTIONAL by pressing a button or with a gesture or whatever during the match.

so the user who comes to his device can get an overview about the results of the match/tournament so far. in the moment he can only see the result of the paired opponents running the game in front.

I am not looking forward to a perfect crosstable. in big size.
a pure text line would IMO be enough, just showing the participants and the ranking
so far as you show it in the end.

[thorstenczub]

Stockfish is the clear winner. Komodo now comes in third (probably because v3 honors fixed time per move better than previous versions, which may have skewed estimates of their strength a bit).

in my opinion this fixed time per move is influencing the outcome in a negative way.

the gui should try to support Blitz game in X (optional + increment per move), and tournament games x moves in y minutes (optional + increment per move)
i doubt the engines handle this fixed time per move optimal enough.

[AartBik]

yes aart. but i would like to see this "result" feature OPTIONAL by pressing a button or with a gesture or whatever during the match

Fair enough. I added a feature where pressing ">>" during a tournament in progress dumps a simple cross-table of completed matches played so far. Expect this in version 3.0 as well (together with the opening suite feature, improved parsing of incomplete FEN strings, and some layout and formatting improvements in the UCI engine output).

[thorstenczub]

ah this is nice aart. reason is: if you have a LONG tournament that really takes days,
or hours, by pressing the "result" button you can get a picture about the progress of the engines.