Author: Robert Hyatt
Date: 07:28:38 05/15/99
Go up one level in this thread
On May 15, 1999 at 08:34:20, William Bryant wrote: >OK, 1st big problem which is probably something I don't understand. > >The program fails low at the root resetting alpha to -INFINITY. > >You now make your first root move, and search with alpha now equal to -beta, >and beta equal to -alpha, which makes beta equal to INFINITY. >If you call a null move at this point, the null move window is >INFINITY, INFINITY +1 which obviously does nothing but return a score of >INFINITY which is greater or equal to beta ..... > > >What am I doing wrong. Do you screen for this before doing a null move? > you have a bug. (a) _no_ score should ever exist _below_ -infinity, in fact no score must ever equal -infinity. Which may mean your -infinity isn't small enough. (b) if (a) is satisfied, how can you fail low on -infinity? The most common bug with null move is a fail high, and on the research where beta has been relaxed, you get a fail-low. _that_ can be a serious problem if you don't ignore the false fail-high at the root. You will occasionally play a totally garbage move. >William >wbryant@ix.netcom.com
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.