Author: Vincent Diepeveen
Date: 06:38:18 08/06/00
Go up one level in this thread
On August 05, 2000 at 11:37:01, Larry Griffiths wrote: >Which Algorithm is considered the best now-adays. Depends upon what kind of program you make. If you have an evaluation function that has patterns which all deliver very small penalties and bonusses, from which the summation also adds up to a near to material only evaluation, then MTD is an interesting alternative. If the evaluation function is either big, using a pawn as being worth 1000 points instead of 1 point, the eval is huge, or having high scores for for example king safety and or passers, then you have only 1 option that outperforms *anything*, and that's nullwindow search also called principal variation search which is pretty easy to implement. Usually at the start of your program MTD looks interesting, if your program gets better (more knowledge in eval, less bugs in search and better move ordering), then PVS usually outperforms anything. My advice is to start with PVS and not look to the rest. >NegaScout? MTD? PVS? Others? I am looking to implement one of the best search >type algorithms in my program. I would like to get it into the 2000 rated range >as this has been my lifetime goal. Then, maybe install winboard or something so >it can compete against other programs to get a rating. >I dont like MTD as it seems to be complex. > >Larry.
This page took 0.02 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.