Author: Reinhard Scharnagl
Date: 07:10:47 06/14/05
Go up one level in this thread
On June 14, 2005 at 08:43:22, Christopher Conkie wrote: >Hi Reinhard, > >I could just about feel OK about your reasoning behind the search itself >but.....I would say that I would expect a little more accuracy from very strong >engines. > >What do you think is wrong with the mate announcement in Comet and Exchess? > >#2 when showing mate in 3, in other words. > >Do they both count all mates as being one move less than they actually are? > >And if so, why? Is it a bug? > >Christopher It is always the same. People think a PV line has to consist of a sequence of optimal moves. If this would be possible at the end of this line, I would take that method and select my first move by that. But such a method could not (yet) exist. Thus a PV line is only an information line. E.g. in Smirf the PV line is created from the cached transposition table. Thus the calculated evaluation and the PV line are built at different times, and easyly the information of both could be slightly different. Reinhard.
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.