Author: Johan Hutting
Date: 08:00:59 12/11/02
Go up one level in this thread
On December 11, 2002 at 10:08:40, Uri Blass wrote: >On December 11, 2002 at 10:00:38, Johan Hutting wrote: > >>On December 11, 2002 at 07:44:43, Sune Fischer wrote: >> >>>Chess is primarily tactics, >> >>Vincent does have a point, to get into a position where you can use those >>tactics you require positional knowledge. Any 1600+ rated player can tell you >>that :) > >search can help also for better positional moves. A deeper global search may help you find more tactical jokes, but if the position doesn't contain those due to sound positional play, where does that leave your engine? Don't get me wrong, the more unique engines/ideas, the better. However, Vincent does have a point that a player rated 2000+ would opt for the positional approach rather then a tactical one. I myself find it quite annoying to see Celes play bad positional moves on occasion. >> >>My guess is Uri uses a very simple patzer eval with big bonusses for good pieces >>along with a fast search. So far it makes movei play good, but if you wish to >>improve you'll run into a wall. > >Movei is still not optimized for speed. hmmm, why not? You're going for a speedy search with few eval, I would guess that a faster search is a big benefit? >I use no assembler in movei I don't use any Assembly in Celes, mainly because I think the speed benefit is rather small plus the code remains more maintainable this way. Modern compilers already produce fast (enough) assembly anyway. >and there are other things that can be improved. Good luck with those :) Celes main problems at the moment are a number of search instabilities and a rather solid evaluation. I'm attempting to get both fixed before CCT5. Which reminds me, Movei still isn't on the participants list... Even with a slow computer, why not participate? The experience is useful. >I believe that it is possible to improve the search algorithms. >I do not believe that I am going to run into a wall about improving the search. What I meant was that I believe that a better positional search will give you more benefit then a couple of 'small' search speed improvements. Just a difference between adding small and big improvements :) >>>if you have near _optimal_ extensions and near >>>_optimal_ pruning rules, then you could probably get those extra plies on your >>>opponent and could win WCCC using tactics. >> >>but, with a bigger eval and good pruning you will still search the promising >>lines deeper, right? > >yes but the point is that I believe that the search algorithm of the good >programs can be improved significantly. but, if you find such improvements with your small eval, perhaps they will not work with the bigger eval most/all strong programs have. That's the wall I was talking about. Regards, Johan
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.