Author: Thorsten Czub
Date: 19:01:30 12/26/97
Go up one level in this thread
>If evrybody uses capture search, this is no longer relevant. > >The words "used to" and "was" above imply that the recent Genius's no >longer act this way, true ? > >Chris Whittington The PURE asymmetrie in search is pretty good to see in OLD genius versions. The latet ones have to compete with Fritz in Blitz. Therefore Lang/Weiner were forced to give Genius some special extensions. These extensions often make the lack of the asymmetrie a little foggy. But there is still enough asymmetrie in Genius5 that I can show it in positions. The only problem is: I stopped any WORK on genius5 a few years, or was it months ago. WHen was genius5 released ? Oh man, thats really a long time ago. The similarity in the symptoms between genius and junior can be seen the best with older genius/Mephisto versions. When I take e.g. my Mephisto 68000 Roma i see 2 components: The M - search and the S-search. The M (= minimum??) search is very very slow: Maybe in 3 minutes 2 !! If this is plies ??! the S-search could be selective peak ! The interesting thing is now to study the LENGTH of the main lines while M and S increase ! It is obvious that these old mephisto machines have LONG mainlines although M is 0 or 1 or 2 (plies or whatever). Because of the hardware S cannot be brute force plies. Maybe M can be. Since I have these old Mephisto machines (Mephisto Roma build in 1988) I have studied these strange search-depths and main-lines etc. So I know the symptoms from this time. But in those times I had no contact with programmers and my knowledge about how chess programs work was very low. Later I increased it a little and was able to order and USE the things I have seen in a better context. Chris e.g. is very good in understanding these ideas and use the concepts for creating anti-concepts. It is 4 h clock in germany. I will go to bed now. Till tomorrow ! Maybe we could talk about Karpow-Topalov 1.Rxe6! and how genius versions have problems to find Rxe6 with white but not with black ! The same problem with CSTal vs. The crazy bishop Re1 (manouver R to e1-e3-g3) These moves are difficult to find with the 1,3,5,7 plies but easily to see for genius as thread with the brute-force-plies 2,4,6! Can somebody of you calculate how much the search tree is reduced when you do the odd/even asymmetrie and do maybe 4 or 5 branches in the 1,3,5,7 ... plies ?? and ALL or as much as alpha-beta needs in the 2,4,6,8. Is it possible to shift the whole thing AFTER you got some working main-line ? I mean, if you once have good moves in ply 2,4,6,8 is it possible to shift the problem and to get good branches in the 1,3,5,7,9 plies ? I think when I do mail-chess with my friends my chess-tree is also odd/even pruned. I select a few good moves in my own plies and let the computers refute my few "good" moves with their ideas. Due to the fact that I use 15 programs I get almost any THREAD of my opponents by the programs in the 2,4,6,8 plies and my few good moves can only survive if my plan or idea is ok. Step by step a GENIUS tree develops !! But with one main difference, chris mentioned it: genius uses static exchange evaluator-. This makes mistakes when exchanges happen. Genius5 is still full of these misevaluations in exchanges. Of course in mail-chess I do not use static-exchange evaluation. Why should I ? :-)
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.