Author: Steve Maughan
Date: 03:21:25 04/26/99
Go up one level in this thread
Brett Hash tables are a programs search memory and generally the larger the hash table the better. Basically, when a program comes across a position it first looks into its hash table to see if it already knows something about the position. The type of information it may find is the best move it found the last time it encountered this position, the score of the position based upon a search of 'x' extra moves or a score bound eg 'this position looses a queen'. The program use this information to either ignore this position (eg a line that looses a queen probably will not need searching any deeper) or search the best move first. When it has completed searching from a position, it stores what it already knows in the hash table for future use. Generally hash tables make a big difference in the endgame where transpositions occur frequently in the search i.e. the stored knowledge in the hash tables is extremely useful. That's the basic idea - it can be one of the most complex aspects of chess programming. For a deeper understanding search Deja News for "Hash Zobrist" in the alt.games.computer.chess newsgroup and you sould find quite a bit (Zobrist is the algorithm used by most programs to hash a position). Hope that helps! Steve Maughan
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.