Author: Marc Bourzutschky
Date: 06:24:56 03/09/04
Go up one level in this thread
On March 09, 2004 at 08:24:16, Tord Romstad wrote: >On March 08, 2004 at 16:07:47, Marc Bourzutschky wrote: > >>On March 08, 2004 at 13:46:44, Tord Romstad wrote: >> >>> >>> >>>In hexagonal chess, there is an interesting complication caused by the >>>fact that there are actually two kinds of wins: When a player is mated, >>>the result is 1-0, like in normal chess. But when the game ends in >>>stalemate, the result is 3/4-1/4. >>> >>>Apart from this, the only difference compared to normal chess is the >>>shape of the board and the movement of the pieces. >>> >> >>The different stalemate status is not hard to implement. You run the generator >>twice, where on the first run you resolve wins/losses, and on the 2nd run you >>resolve stalemates/draws. > >That's good news to me. :-) > >>With 91 squares on the hexagonal board, 5-piece endgames are comparable to >>6-piece endgames on a 8x8 board and will thus be the limit of what is reasonably >>doable today. > >But the hexagonal board also has more symmetries compared to a square >board. Wouldn't this help to reduce the size of the files, and the >amount of computations necessary to produce the tablebases? > If I remember my group theory correctly, you have at most a factor of 12 reduction due to symmetry. Looking at the piece movements it seems that for pawnless endings all 12 symmetries apply (compared to 8 on a square), while for endgames with pawns only a factor of two looks possible. For 5-man endgames, this leaves you with about 500 million position for pawnless, and over 3 billion positions with pawnful. This can still be done on today's PCs, but 6-man endings will be too large. -Marc
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.