Author: Tord Romstad
Date: 01:45:50 01/10/04
Go up one level in this thread
On January 09, 2004 at 18:53:00, Anthony Cozzie wrote: >On January 09, 2004 at 16:47:26, Tord Romstad wrote: > >>On January 09, 2004 at 16:39:28, Bob Durrett wrote: >> >>>This begs the question: "Is it feasible to build a playing program which can >>>play more than one game well, INCLUDING classical chess? >> >>Of course it is. Gothmog will soon support Glinski's hexagonal chess in >>addition to >>classical chess. You could of course argue that it doesn't even play classical >>chess >>well today, but that is only because of the incompetence of its author, not >>because >>it is impossible. >> >>Tord > >seems to me 99% of the code will be different: different move generators, >different eval functions, different make_moves, different pruning. No. The same move generator will be used for both games, but with different lookup tables. I also have to disable generation of castling moves in hexagonal mode, but that is of course trivial. make_move and unmake_move also do not need to be changed. All the pruning I do now should work equally well in hexagonal chess, I hope (except that I need a different and probably a bit more complicated static exchange evaluator). In addition, most of the boring IO and xboard interface code can be used in both games. The evaluation function will obviously have to be completely different. But in the first version, the hexagonaly evaluation function will probably be very rudimentary, including piece square tables, some mobility and king safety, and almost nothing else. >About the only things that would stay the same are the transposition table and >the search itself. But those are trivial in terms of code size. In Gothmog, the search is far from trivial in terms of code size. It's about the same size as the eval and the move generator added together. Tord
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.