Aart's Blog

I ran tournaments between different versions of the same engine to determine which one would be their representative. Below I show the results of a fast round-robin tournaments using Chess for Android on a Nexus One, Nexus S, and Galaxy Tab with, where applicable, 8MB hash table and 1MB tablebases cache (Nalimov, Gaviota, Scorpio, and Robbobases reside on SD card). Each version played each other version from both sides of the Nunn opening suite . The number one engine will play in the longer time control tournament ( which is not always the latest  version; engine authors let me know if you prefer otherwise since some results were pretty close ). Also, Pablo Vazquez and Jim Ablett kindly worked together to fix the issue in the sungorus . It now runs correctly and will participate in the tournament. 1   BikJump v2.1P       10.5/20 2   BikJump v1.8         9.5/20 1   Diablo 0.5.1b JA    17.5/20 2   Diablo 0.5.1         2.5/20                   1   DoubleCheck 1.3     13.5/20 2

The list of UCI and XBoard engines for Android is getting longer, and I am pondering about a more serious tournament at longer time controls in Chess for Android . As a first step, I tested all available Android binaries. The list is shown below. An "X" indicates that the engine has trouble running in Chess for Android (if you are the engine author and are interested in fixing this, please contact me for details). Next, I probably do a few quick tournaments to divide the engine in groups by strength, followed by tournaments at longer time control. Please feel free to contact me if you are interested in participating, if you don't know how to compile your engine for Android (ARM), or if you have good ideas about the format of this tournament. I am not really an experienced tournament manager, but merely do this for fun to get an idea of the relative strength of these engines. apilchess-106-ja           X    U. Lorenz, C. Donninger  bikjump1.8                      Aar

Version 3.0.1 of Chess for Android is available at the Android market and as direct download . It introduces the first, albeit simple support for the Chess Engine Communication Protocol (XBoard/WinBoard). Many features that are already supported for UCI (e.g. options setup, infinite analysis) as well as some XBoard specific features (e.g. resign) are still missing for this first version. I hope to add these features soon. Nevertheless, it is already possible to run tournaments between engines.

Chess for Android now contains sufficient functionality to perform the first inter-protocol tournament between the built-in Java engine, the UCI engine BikJump, and the WinBoard / XBoard engine Fairy-Max. Ten random opening book games between these engines at one second per move on a Nexus S ran without problems to completion with the following results.                     1          2          3           1 BikJump v2.1P     ********** 111½111111 1111111101 18.5/20 2 Fairymax 4.8Q     000½000000 ********** 1½10½11½01  7.0/20 3 Chess for Android 0000000010 0½01½00½10 **********  4.5/20

I am exploring extending Chess for Android with some rudimentary support for the Chess Engine Communication Protocol (often simply called the XBoard or WinBoard protocol), which will enable importing not just UCI but also XBoard/WinBoard engines. A very first prototype is functioning. I made an Android binary of the engine FairyMax (written by H.G. Muller , who was also very helpful providing more background), and imported this in Chess for Android. Some screenshots are shown below. I still have to deal with a lot of details, but it will be fun to support both protocols, and even run tournaments between UCI and XBoard/WinBoard engines.

Miguel A. Ballicora released Gaviota v0.84 . I tested the Android version against an earlier release with a quick one-second-per-move tournament on a Nexus S, setting the hash size to 32MB and using the Silver Opening Suite and some Gaviota tablebases on SD (the latest release fixes an issue with tablebase setup on Android). Below are the results. The new version seems much stronger! 1   gaviota v0.84  +66/-17/=17 74.50%   74.5/100 2   gaviota v0.83  +17/-66/=17 25.50%   25.5/100

Michel van den Bergh posted a new release of GNU Chess (5 branch) . To test the Android binary, I ran a one-second-per-move, 32MB hash tournament on a Nexus S, using the Silver Opening Suite , with Gaviota tablebases on SD. Below are the results. Although the latest release was mainly for FRC support, it also seems slightly stronger. GNU Chess               1         2         3  1 5.07.173b-libgtb-32       *     52.5-47.5 57.0-43.0  109.5/200 2 5.07.170.7b-libgtb-32 47.5-52.5     *     55.5-44.5  103.0/200 3 5.07.153.3b-32        43.0-57.0 44.5-55.5     *       87.5/200

We just came back from a visit to New York, where we visited ground zero. So tragic to remember what happened there ten years ago, but encouraging to see that America is rebuilding.

Running tournaments on diverse test suites gave me more confidence in the correctness of my FEN and PGN parser. This time Chess for Android found bugs in the " Rebel ELO Test ". Positions ELO 30, 121, 170  do not set the castling rights for either black or white properly, even though castling is the first given move (that is, the move the engine is supposed to find when used as a tactical test suite; when used as an opening suite, this move is obviously played right away). The results of (mis)using the corrected test suite as an opening suite in a fast match between the built-in Java engine and Stockfish 2.1 on a Nexus One is shown below.                    1 Stockfish 2.1    +421/  -4/=27 96.13%  434.5/452 2 Chess for Android  +4/-421/=27  3.87%   17.5/452 Here is one of the sparse victories of Chess for Android playing white from a very advantageous position. Nevertheless, I am happy that it found the right continuation as 1. Qc8+ (given in the suite) Kg7 2. Qf8+ Kg6

I ran various tournaments using the " Max Euwe suite ", from the hand of the famous grandmaster who inspired many Dutch aspirant chess players, including myself. During the first tournament, Chess for Android detected bugs in positions 146 and 148 , where the castling rights are not set properly for white, even though white actually castles in the first given move (that is, the move the engine is supposed to find when used as a tactical test suite). For example, in position 146, the embedded FEN string in the PGN file: r1b1kb1r/p3q1pp/2pp1p2/4n3/3N4/8/PPP1BPPP/R1BQK2R w kq - 0 1 should really be: r1b1kb1r/p3q1pp/2pp1p2/4n3/3N4/8/PPP1BPPP/R1BQK2R w KQkq - 0 1