Author: Martin Giepmans
Date: 10:30:39 04/19/04
Go up one level in this thread
On April 19, 2004 at 13:11:59, Andrew Wagner wrote: >Some have suggested this could possibly be a problem with how I store and >retrieve the bounds in my hash table, so let me explain what I do and see if the >logic is right. Much of it comes from Bruce Moreland's web site. > >My Search function calls a function called ProbeHash(). Probehash looks in the >Transposition table to see if the position has an entry stored there already. If >so, and the leaf distance of the hash entry is >= the current leaf distance, it >looks at the score and at what type of entry it is. If it's a beta entry (stored >when the move fails high), and the score is >=beta, Probehash returns beta. if >it's an alpha entry (stored after a fail high), and the score is <=alpha, alpha >is returned. If it's an exact score (the first time it was searched, the score >was between alpha and beta), the score itself is returned. If none of this is >true, it passes back a best move as a parameter. Once we arrive back at >Search(), if any of the above three cases with scores (alpha, beta, or the hash >entry score) were returned, Search immediately returns that score. Otherwise, it >continues on. > >I hope this makes sense. Does this sound correct to the rest of you? Dan, is >this similar to the logic of how you do it? Thanks! Andrew Looks okay to me. Just to be sure, what happens if you don't use the alpha-score and treat the exact score as if it were a beta score (or don't use it at all)? It might not find Kb1 fast in that case, so let it run for a while ...
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.