Using Thompson`s endgame databases for chess engine in Chess Academy Playing Program

Thompson`s endgame databases contain all important endgames with five or fewer pieces on the board. An endgame also contains any 3- or 4-stone endgames that may result from exchanges. There is a separate file for each endgame. All possible positions are stored in the database files. When the program reaches such an endgame, the engine does not calculate any more moves but uses only the stored knowledge from the endgame databases.

Register Thompson’s Endgame Databases: Check the “Use Thompson endgame databases” option; click the button image131 to select the required CD or even a directory to use (if none(s) was already selected). The program will show a file selection window “Search for a directory”, where you have to specify the path to the Thompson`s Endgame Databases CD (or to the directory on your hard disk). Select a directory/drive and click the Open button. Alternatively, double-click the mouse pointer on the appropriate directory / CD drive or, after highlighting a directory, simply press the Enter key.

If you want to prevent the engine from continuing to play moves from the endgame library, simply uncheck “Thompson`s endgame databases” again. The program remembers the last set path.

If you want to use Thompson`s endgame databases again, you just need to check the option again.

Using Nalimov tablebases for chess engine in Chess Academy playing program

Chess programs usually simply do not “know” how to handle certain endgame positions or endgame types in order to be able to win. For this reason almost all commercial chess programs are delivered with the most necessary endgame tablebases by default or at least have an interface between program interface and endgame database to enable the chess engine, which supports such tablebases, to access them.

The endgame databases can be thought of as a huge reference book, in which specific and rated chess positions (endgames) are stored. The first endgame databases by Eugene Nalimov were published around the beginning of 1998 and have been continuously developed since then. The current “Nalimov tablebases” contain all possible positional constellations for all important three-, four- and five-pin games (currently there are also several six-pin tablebases available, but most of them contain several gigabytes, which makes the distribution of such files much more difficult), which are systematized according to a given scheme. Such tablebases are available for free access. Mr. Eugen Nalimov originally wrote the algorithms that enable generation of such tablebases. He then published the source files and corresponding instructions for generating and using tablebases, as well as for creating executables, for free access on the Internet. Mr. Andrew Kadash has published appropriate compression tools and algorithms that allow the creation / use of tablebases in compressed form. This information is also included for free access on the Internet.

Nalimov tablebases are conceptually superior to Thompson databases. For each possible position within a given piece constellation, information about “win”, “draw” or “loss” is stored. If the position is concretely won or lost, the maximum number of moves until mate is also given. A chess program with an interface to the “Nalimov-Tablebases” can therefore give an absolutely precise indication about all three-, four- and five-pointers (some programs currently also serve six-pointers) and, in contrast to the Thompson databases, will already consider the values from the endgame database in the search (variant finding, position evaluation).

In contrast to the Thompson databases, the “Nalimov tablebases” should be installed on the hard disk so that the chess program has quick access to the information during the search. The use of tablebases in compressed form is also possible, whereby instead of about 30 GB only about 7.5 GB of free space (as hard disk capacity or on the DVD, or on individual CDs) is required. In principle, the databases can also be used from CD-Rom or DVD, but due to the significantly slower access from CD or DVD, this results in considerable performance losses and slows down the chess program drastically. Nalimov tablebases know among other things also the “En passant” rule. There are also some limitations when using tablebases. For example, the “50 moves” rule is not taken into account. Also castling (which should not be very important in the endgame) is not known by the tablebases.

Chess Academy 7 usually supports three-, four-, five- and six-stone tablebases. Access to endgame databases (Nalimov tablebases) Nalimov tablebases are now available in different distribution packages, sometimes only 3-, 4- and some 5-men tablebases, sometimes 5-men complete. Some distribution packages even offer 6-steiner tablebases (selected files only). The CD edition of the Chess Academy 7 program does not include tablebases. To register Nalimov tablebases under Chess Academy 7 and thus be able to use them, please choose a distribution package of your choice in which Nalimov tablebases are available.

Under the program interface you can set the path to the Nalimov tablebases via a dialog window (Engine / Engine settings / Books / Nalimov tablebases).

Utilization of cache memory to certify access to data The cache memory is used to speed up access to data (tablebases) during the game. After installation, the memory is limited to 8 MB. However, we recommend setting the memory to 16 or 24 MB or more if you have 256 MB or more of memory.

 

The structure of endgame databases (Nalimov tablebases)

The filename of each database indicates the pieces that are on the board. The names correspond to the respective English-language names of the pieces: K = King (king) Q = Queen (queen) R = Rook (rook) B = Bishop (bishop) N = Knight (pronounced “Neit”; knight) P = Pawn (pawn) An endgame database is simply a table with values for any position. Thus, the position of the pieces on the board determine the value that is read from the table. Only this value identifies a position as won or lost. If the position is lost, the number of moves needed until the loss is also stored.