Author: Robert Hyatt
Date: 06:40:59 08/01/01
Go up one level in this thread
On August 01, 2001 at 02:15:26, Pham Minh Tri wrote: >On August 01, 2001 at 00:57:01, Bruce Moreland wrote: > >>On July 31, 2001 at 23:24:13, Pham Minh Tri wrote: >> >>>I know that we could use endgame tables or endgame functions for end game >>>period. IMO, codes of EG function are much easier to develop, smaller and >>>quicker than tables. However, I see that almost all strong programs use EGTBs, >>>not those functions. Could someone explain to me why or advantages/disadvantages >>>of EGTB vs EG functions? >>>Many thanks in advance. >>>Pham >> >>Tables work all the time, and functions need to be tested on a huge number of >>cases. >> >>If you can write a function that will do it, you'd want to do this, because you >>save the table access. >> >>The problem is that it would be ridiculously hard to write a function that will >>do the work of one of the larger tables, while the smaller tables could reside >>in memory. > >But someone must write those functions to create the tables (correct me if I am >wrong). So instead of releasing tables, he could introduce functions. Other >people could learn and modify them, not use only in case of tables. Is it >correct? > No. The "builder function" does a huge tree search to mark every possible configuration of pieces as won, lost or drawn. There are no "heuristic scores" at all, just the perfect results obtained by searching very deeply (sometimes far beyond 200 plies) >> >>bruce
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.