Author: Tim
Date: 01:27:41 01/29/00
Go up one level in this thread
On January 29, 2000 at 00:11:31, David Blackman wrote: >On January 28, 2000 at 19:33:14, Tim wrote: > >> >>O.K. Value returned is just bound. >>But how do we know bound is correct bound? >>Window size is so small. > >I think this code fragment must be part of a "fail-soft" algorithm. The idea of >fail soft is that you make sure anything that fails low returns an upper bound >on the value, and you make sure anything that fails high returns a lower bound >on the value, and you make sure the bounds you return are the best ones you can >possible get without searching more nodes. > >The main thing is to initialise the value at the top of the routine to -infinity >instead of the lower bound. How can you make sure anything that fails low returns an upper bound on the value? How can you make sure anything that fails high returns a lower bound on the value? If window is alpha ~ beta, we can make sure it is correct bound. But window is alpha ~ alpha+1, we can not make sure it is correct bound.
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.