Author: Aivaras Juzvikas
Date: 07:40:25 05/27/04
Go up one level in this thread
On May 27, 2004 at 05:12:25, Peter Fendrich wrote: >On May 27, 2004 at 01:36:50, Aivaras Juzvikas wrote: > >>On May 27, 2004 at 01:09:10, Matthias Gemuh wrote: >> >>>On May 26, 2004 at 21:51:34, Andrei P wrote: >>> >>>>This is a hypothetical question. I would welcome any rough ideas from the >>>>experts writing the engines on importance of different things that contribute to >>>>a strong engine. So, here we go. >>>> >>>>Let's say there is an engine ~2200 elo. it is a well-rounded engine with no >>>>striking gaps in the book,tactics,strategy, etc. Everything is just average. >>>>Now, the plan is to take it to 2600. If we separate the engine's play into >>>>parts: book+middlegame+endgame, or book+positional >>>>knowledge+tactics(calculations). How would you describe different part >>>>contributions as the engine gains 400 points (e.g. book improvements can gain >>>>max 50 elo, etc.)? >>> >>> >>> >>>Of the 400 points, 300 would come from improved search. >>> >>>/Matthias. >> >>so eval function improvement influence is so low?? > >...but improved search requires improved evaluation... >When search alg's reach a certain level they need to be assisted by smarter >evaluation terms in order to improve further. >/Peter thats why its so hard to tune eval function, sometimes variables with very obvious bonuses or penalties must be reduced or increased dramatically, because otherwise search tree grows bigger and the overall performance is worse.
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.