Author: Andrew Shapira
Date: 19:23:50 06/06/05
Go up one level in this thread
Hi Volker. What's the problem with hash tables and null move? Andrew On June 06, 2005 at 01:37:53, Volker Böhm wrote: >Hi, > >a chess program does not allways compute an exact result. The only thing to >bother is if the calculation error is small enought. >Hash entries are not allways exact. There are more problems with hash you didn´t >mention, expecially when using pruning (Nullmove, futility, other foreward >pruing) but also with the current search window. > >I suggest to ignore all of them except the distance to mate, but to not use hash >entries when searching the pv. Most search algorithm, example pvs, does >distinguish between "primary" thus pv move and non primary non pv move. You will >have some error conditions left because the non pv move could fail low because >of wrong hash entry and you will select the wrong pv. But the value of the >selected move is correctly the value of the position after playing the pv >including 50 moves rule. > >Greetings Volker
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.