Author: Dave Gomboc
Date: 22:42:08 10/21/00
Go up one level in this thread
On October 20, 2000 at 09:29:50, Robert Hyatt wrote: >On October 20, 2000 at 00:02:19, Dave Gomboc wrote: > >>On October 19, 2000 at 09:33:48, Robert Hyatt wrote: >> >>>On October 19, 2000 at 05:57:30, Pham Minh Tri wrote: >>> >>>>I see some chess programs retrieve information based on color (side to move), >>>>this means only half of hash table is useful. I wonder if we could retrieve >>>>useful data from opponent part of hash table (I think when we implement Alpha >>>>Beta function, it is easy to convert information between two sides). >>>> >>>>Pham >>> >>>The problem is that in many positions, side to move is critical. And it would >>>be wrong should you do what you suggest. >>> >>>In endings it would be fatal. >> >>I think he means everything, e.g. rotating the board 180 degees and flipping the >>colours of all the pieces. Essentially, always hash the position from the >>perspective of the side to move, and assume that they are player A. Sounds >>costly. >> >>Dave > > >How could you do this? The score in the hash entry was derived after a long >search that was based on the current side to move. Which means that it could >be a zugzwang problem that makes the current side lose since he _must_ move. > >This same identical position, with the other side to move would have a >totally different outcome. And I don't see any way to correct this position >other than searching so that the zugzwang problem can be found... > >In the middlegame, it would probably be semi-workable, maybe. But in endgames, >it would lose badly. No, no... the position is always evaluated for the side that is to move. Anyway, it doesn't make any sense, forget about it. Dave
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.