Author: Mike Byrne
Date: 11:47:16 09/15/05
Go up one level in this thread
On September 15, 2005 at 11:08:35, Thorsten Czub wrote: >try out the following position: > >http://www.thorstenczub.de/Txe6.jpg > >genius has problems finding 1.Rxe6 (and later Rxg6) >this is because the key moves are in 1 and 3 ply. > >if genius would search them in ply 2 and 4 it would easily see it. > >you don't believe it ? > >try out the position one move before with black to move > >http://www.thorstenczub.de/einenvorTxe6.jpg > >and you see that 1. Rc8 Rxe6 Ra7 Rxg6+ fxg6 appears in an instant. > >genius search is pruning more in the own plies and following almost all moves in >the opponent plies. > >therefore it cannot find the solution here when it has white, but would see it >as >thread against itself if you take back one move and let it compute as black. > >this is also the reason genius plays very defensive, because it sees all threads >against itself but not all good moves for itself. > >this helped genius dominate the computerchess scene for a long time. >but when nullmove was used in computerchess, the asymetric search suddenly was >not the one with the fastest tree anymore. therefore there was no advantage >anymore and genius was not anymore number one. > >in all richard lang programs this method has never changed. > >its in the amsterdam module, in the roma, in the vancouver, in the genius >1,2,3,... 7.1 and in the mobile versions. > >he never changed this. >and so the programs strength never changed much. > > > > >all written on my website, sorry that it is in german... > >http://www.thorstenczub.de/antichrist.html it also makes it a tough opponent for humans - very nice program, for along time , it was king of the hill
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.