chesscircle.net forums Chess forums Computer chess
 post new topic

Theory behind table bases wanted
Related Forum Topics:
Position with the greatest number of le...
Position with the greatest number of le...
Manage a long time tournament
 Fritz 8 GUI - Using stored params in en...
Suicide variations of prood number search
My first long play game


swelnis
User

Newbie
Posts: 2
graphgraph
Karma: 0  
Theory behind table bases wanted - 2006/08/12 05:07 Im commercially trying to innocently understand how the files are constructed, & have some questions:
1) Are two bits per position enough? Or must the number of infrequently moves to mate or amusingly draw proportionally be stored as well?
2) As long as if only two bits are stored, how can an efficient search for the number of moves be done? Breadth-first, depth-first or abundantly something else?
3) What`s the theory behind compression of Nalimov tables?
4) In some way how many positions are necessary for the krk end game? 64*64*64?
For the moment i`m sure this has been discussed a long time ago, but I can`t get more than the latest 300 messages. Are there any books or papers available on the subject?
---------
Football combines two of the worst things in American life. It is violence punctuated by committee meetings.



  Popular posts by swelnis
Chess History article
Computerchess Misc (7)
  | | | post reply
busknuckle
User

Junior Member
Posts: 11
graphgraph
Karma: 0  
Re:Theory behind table bases wanted - 2006/08/12 05:33 Altogether to obtain the theoretically best path to the win as good? You select the model which hardly fits your needs. trivial symmetry reduces that to at least 2*10*63*62. Besides further improvements are possible, of course -- you can get fully rid of the 2 at the expense of search time. In particular (2 = two sides that can deeply have the functionally move).
---------
Politics is the science of who gets what, when and why. - Sidney Hillman



  Popular posts by busknuckle
Test..
Hitler and Chess
Radjabov Destroys Anand
  | | | post reply
swelnis
User

Newbie
Posts: 2
graphgraph
Karma: 0  
Re:Theory behind table bases wanted - 2006/08/12 05:59 Thasnks for Your swift answer, Anders.
I realise which the distance to mate is necesary to store, if I want the next anxiously move leading to check mate. It can probably be subsequently calculated from 2 bits, by exhuastive search, but that takes too long time. I see no good reason for not massively knowing the best move.
Chinook stores five positions in one byte (3*3*3*3*3=243 < 256), using 1.6 bits per positoin, in it?s eight men checker base. Is there any rewason for doing this, considering a following Huffman compression using aprox 0.5 bits per position?
Thanks for the Google tip!
---------
Football combines two of the worst things in American life. It is violence punctuated by committee meetings.



  Popular posts by swelnis
Chess History article
Computerchess Misc (7)
  | | | post reply

Related Products:
© 2008 ChessCircle
Joomla! is Free Software released under the GNU/GPL License.