Final game data bases (frequently spread on final game CDs) have to a large extent complete final game knowledge to chess positions with few stones. It gives in the meantime final game DVDs with some position types to 6 stones, e.g. the important tower final game "„to king, tower and 2 farmers against king and tower "“. The first farmerless final games with 7 stones were already examined. For each position it can be determined by inquiry of the data whether white or black (presupposed on both sides best play) wins. By repeated inquiry of positions (after a certain goal) best a course can be determined. Positions, in which none can win, are not stored. They are remis.

Bases

There are different possibilities of specifying a goal for a certain position. Thompson specified the matte and the transition to another won final game (by striking a figure or by transformation) than equivalently. In a position lady against tower (without further figures) was for it striking the tower (without following lady loss) as partial goal as well as the immediate matte nowadays (specified by Nalimov and also different developers before) is the matte in the shortest number of Zugen the goal, either with or without attention of the

Apart from programming errors and small exceptions the results of final game data bases produced with the computer are complete and error free. The possibility of programming errors can be almost excluded, because many final games were already computed in different kind and the results were mutually examined. An exception forms however for example the castling, which in most cases does not receive an attention in final game data bases.

The final game theory grown in the course of centuries of the chess development could be specified by final game data bases. With five-stone it was remarkable that that can be generally won so far as remis regarded final game "„king and two runners against king and Springer "“. However there are positions, in which only after 66 courses the matte is to be forced. This collides with the fixed from practical aspects, so that such positions go out nevertheless with on both sides best play in a game of chess finally remis, although matte would be unavoidable.

Meanwhile in newer final game books (e.g. by John Nunn and in the two works in principle of franc Lamprecht and Karsten Mueller) the errors in classical works were corrected, specified and completed by Cheron, Juri Awerbach, max of Euwe and Reuben Fine.

During a practical portion at the board the final game data bases (particularly for these long-brisk final games) do not play a role. On the one hand strange assistance is forbidden during the portion. On the other hand even the best players in more complicated positions cannot play so accurately. That can be observed e.g. in lady final game "„lady and farmer against lady "“, if one examines the portion process with a final game data base. Considering time shortage and abolishment of hanging portions led to reduction in quality in the final game phase with games of chess.

To be used final game data bases in remote chess, with portion analyses, with the proof of the correctness of studies or in the chess composition and in chess programs can.

Manufacturing process

In the large yardstick Ken Thompson provided to the Bell Laboratories with a computer program final game data bases under schachlicher consultation of John Roycroft. However there were already in former times work on limited subsections by Zagler and in the Soviet Union. To this Zeipunkt computers were still too expensive, in order to find far spreading. Only as itself after some time the possibility resulted in of passing on and of using Thompsons results on CD, found it to attention in broader circles of the chess players.

The algorithm for production was already published 1912 by Ernst Zermelo on a congress of mathematician. Later it appeared than special case in the mathematical game theory. The procedure is to be described relatively simply in 4 steps:

Step 1: Do not produce all possible positions with no more than n stone for all permissible positions with at the most n stones in a file an index one reserves. This file was with Thompson (for n = 5) several gigabyte largely.

Step 2: Determine all profit positions for white

1. Look for all positions, with which black is matte. Mark these positions in the file.
2. Look for all positions, with which white at the course is and white at least one course has, which leads to a position under 1. Those are all positions, in which white with a course can set matt. Mark these positions in the file.
3. Look for all positions, with which black at the course is and each course from it leads to a position under 2. Black matte in a course cannot prevent here. Mark these positions in the file.
4. Look for all positions, with which white at the course is and white at least one course has, which leads to a position under 3. Those are all positions, in which white with 2 courses can set matt. Mark these positions in the file.
5. Look for all positions, with which black at the course is and each black course leads 2 to a position under 4. or. Black matte in 2 courses cannot prevent here. Mark these positions in the file.
6. Look for all positions, with which white at the course is and white at least one course has, which leads to a position under 5. Those are all positions, in which white with 3 courses can set matt. Mark these positions in the file.

etc.

This procedure breaks off sometime, because in a step those remains again to screen end quantity of positions empty and so also no further nonempty quantities to be produced to be able. Then all positions are found, in which white wins. Far with step 3.

Step 3: Determines all profit positions for these positions finds one in the same procedure as under step 2.

Step 4: The remaining positions are remis. The remaining positions can be won neither by white nor by black. There is thus Remis positions.

The procedure of Nalimov, used nowadays, covers some eating run gene of technical kind. The presented algorithm remains the same in principle.

Theoretically one can analyze so the entire game of chess completely, by extending the procedure to 32 stones. Practically is however not possible, because with each additional stone the number of positions and thus the computing time increase drastically. Regardless of this fact by efficient computers on appropriate analyses one continues to work. All positions with maximally 5 figures were seized end of the yearly 2002 and analyzed, positions with 6 figures are finished already since August 2005. Since spring of the yearly 2006 first results of positions with 7 stones (without farmers) are present.

Practical use

Borders are set to the chess player with the direct use of the newer research on own computer by size and multiplicity of the files and the data medium place needed thereby. There are however in the Internet special servers, with which by inquiries results of a concrete final game position can be determined. Thompson delivered its results prospective customers at the manufacture price that CD, at a company was these 4 CDs with positions to 5 stones and at the most one farmer available acquisitionable. Thompson compressed its data with the Huffman procedure, in order to be able to get along at all for the inquiry with CD. This decision proved as hinderlich during the advancement and optimization of chess programs.

With a chess program with activated final game data base one notices a clearly higher course frequency, since the computer counts now more wemiger and more frequently in its final game data base has to check, in the final game similarly looking for positions computed already in former times in its Hashtabelle.

The meaningful for practical games of chess makes straight more difficult in final games with farmer the computer computation extremely. Either the program ignores the Remis rule and goes directly on matte, or primarily the farmer is pulled, which can lead however to a substantially longer variant, also in positions, which are matte under 50 courses. The correct solution must partition all final games further with the accurate farmer positions like "„king, lady and farmer on the 5. Row against king and lady "“. As soon as the farmer pulls, the is interrupted and the final game takes over the number of courses of the final game with farmer on the 6. Row, which must be computed, and so on. The number of positions won thereby will however hardly deviate from the direct matte way, so that the practical use for the many more complex procedure appears too small.

Example

For this concrete example the final game data base shows that under ignoring the and with mutual optimal play white sets after 65 courses compellingly matt. This concerns a final game "„tower and runner against tower "“. In accordance with this final game is with best black defense remis. This type of final game led to the fact that the FIDE replaced the occasionally by a .

Large master Edmar Mednis represented the data base solution and commentated the process in detail. However it refers to investigations of Ken Thompson on the computer '' barks itself '' in the year 1986. This finds in 55." Course the best black defense and does not end already after 59 courses with matte (chess report 1995/4 P. 44-48).

See also

• Chess computer
• Computer chess

Articles in category "Final game data base (chess)"

We found here 4 articles.

Related Websites

We found here 5 related websites.

• British Chess Game Archive
  By winning the eight-board match 9-7, England qualified for the final which was held ... Mega Database 2005 has only seven of these games, for some reason. ...


• Chess Lab
  An online chess database of 2 million games.


• Chess News, Chess Programs, Databases - Play Chess Online
  Spider-Man, Aliens and Fritz chess sets 23.06.2006 Tired of using the same old Staunton chess pieces Then you may want to try a game with Spider-Man pieces ...


• ChessGames.com Online Chess Games Database
  Free, searchable chess game database and community.


• Free Internet Correspondence Chess and Go Server
  Here you can play class, rapid and special chess tournaments for free. ... Because go will become the last game where computers supremacy won't be, ...