 |
|
 Chess variants (10x8)Sam has closed his piano and gone to bed ... now we can talk about the real stuff of life ... love, liberty and games such as Janus, Capablanca Random, Embassy Chess & the odd mention of other 10x8 variants is welcome too For posting: - invitations to games (you can also use the New Game menu or for particular games: Janus; Capablanca Random; or Embassy) - information about upcoming tournaments - disussion of games (please limit this to completed games or discussion on how a game has arrived at a certain position ... speculation on who has an advantage or the benefits of potential moves is not permitted while that particular game is in progress) - links to interesting related sites (non-promotional) この掲示板でメッセージを作成にはポーン会員以上の会員レベルが必要となりますので、あなたは作成権限が有りません。 モード: 誰でも投稿可能
21. ９月 2006, 21:23:42 件名: TableBases This post is just for informational purposes only, and doesn't apply to any one person, program, or game. I think most of us are aware that the 6-piece tablebases for 8x8 chess are completed. This was a massive undertaking, and a DISTRIBUTED project that covered many years, and many computers being used in parallel to complete the task. As for 10x8 chess variants, only the 5-piece tablebases are completed. Just so everyone knows, the branching factor for 10x8 board with the 2 new pieces is just gigantic when compared to 8x8 chess. For example, the 4-piece tablebase set for 10x8 with the Chancellor and Archbishop takes only 26 minutes to compute. The 5-piece set required 34 days! The ratio of time involved was more than a factor of 1800! Also remember the longest 5-piece 10x8 win is 268 moves (not plies) which is longer than the longest 6-piece tablebase win (262 moves) for 8x8 chess. The 6-piece tablebase set has not even been started, nor does it make sense to begin. The time scaling for its completion will be roughly 75 (number of squares on which to place piece #6) x 2 (each "distance to win" must reserve 2 bytes whereas the 5-piece set only needed 1) x 343 (this times as many more tablebase slices to solve) = over 50,000 times as long as the 5-piece set. These numbers are mind boggling. The 60 gigabyte 10x8 tablebases are all that exist. There is no "1 terbayte" set anywhere. The average tablebase file for the 6-piece set will be 75 x 2 = 150 times the size of the 5-piece files. So, they will be about 300 GB each. This mysterious "1 terabyte" that was mentioned before would only be the size of 3 files, if they were completed (and they are not.) Everyone is probably aware that you have to solve the non-pawn files first, such as King + 2 Queens vs. King + 2 Queens, King + 2 Queens vs. King + 1 Queen + 1 Chancellor, King + 2 Queens vs. King + 1 Queen + 1 Archbishop, etc. Even if these tablebase files were completed, totalling about 1 TB, they would be statistically of no help whatsoever during the coarse of a game between 2 programs. 21. ９月 2006, 21:38:24 件名: Re: TableBases ChessCarpenter: Indeed such table bases are of noticable theoretical interest, and usually a good marketing argument, too. ;-) My opinon on such tables is, that they should stay away from valued games. Moreover, most games, wherein ever table bases could be successfully used, were already decided before that stage has been reached. Nevertheless having such tables could be of remarkable productive usability for answering questions of endgame theory. Thus the place for using them should be in post game analyses only. 21. ９月 2006, 23:53:57 件名: Re: TableBases SMIRF Engine: Now an interesting thought,harking back to a previous post, is that many would consider the consultation of these tablebases (for either 8x8 or 10x8) to be acceptable if they were in a book (some book!) but not if they are in a computer ... this is a dilemma (and argument) which I wonder will ever be resolved. 22. ９月 2006, 00:03:13 件名: Re: TableBases ChessCarpenter: This mind numbing, brute force number crunching is the complete antithesis of why I play games. There's not an ounce of romanticism or sportsmanship to it. I suppose it has it's place in the realm of theory and understanding of games without chance in them. I'm glad to hear the numbers are so large as to make it more or less impossible to know every single outcome possible. There's other ways to do it besides knowing every move possible. A set of rules for each group of remaining pieces might use less memory, but I have the feeling that's not the case in this instance. Have you tried this approach? And then there's SMIRF Engine's point about just what good will such tables be if the game is already decided when it gets to that point? It's amazing that just having the extra 16 squares and the addition of four pieces on the board makes the numbers get so much larger. Regular Chess itself has very big numbers when one wants to count all the moves possible. What about the 10 × 10 board for Grand Chess? It uses the same pieces as Embassy Chess. Will the number be slightly more, or lots more? I'm thinking lots more. Though the restricted Pawn promotion might keep it down more than the unrestricted Pawn promotion of Embassy Chess. What's a Chancellor and Archbishop? There's no game on this site with those pieces. Do you mean the Marshall and Cardinal or Janus? They are not new pieces. They're more than 100 years old. 23. ９月 2006, 21:43:34 件名: Re: TableBases Walter Montego: Two of my favorite endgames on the 10x8 board are Archbishop vs. Bishop + Knight, which is a mate in 202 for the Archbishop, shown here: http://www.gothicchess.com/endings/kakbn/game.htm And Chancellor and Pawn vs. Chancellor, a mate in 140, shown here: http://www.gothicchess.com/endings/kcpkc/game.htm Incredibly complex endings that really let you appreciate whatever your favorite 10x8 game is that gets you to these points. Even more interestingly, programs cannot play these endings properly without these precomputed tablebases. Strong human players have a better chance of winning the endings more rapidly than programs that do not have the tablebases, another one of the ironies of the world of computer programs. 25. ９月 2006, 12:39:03 件名: Re: TableBases ChessCarpenter: As interesting as the games are, links to the the site referenced are consiered inappropriate on this site. In future please refrain from sending patrons from this site to there. Thanks you. |
日時
オンライン友達
気に入り掲示板
同好会
今日のアドバイス
|
|
著作権 © 2002 - 2024 Filip Rachunek. 上へ |