Author: blass uri
Date: 13:52:29 03/22/00
Go up one level in this thread
On March 22, 2000 at 16:25:39, Gian-Carlo Pascutto wrote: >On March 22, 2000 at 06:05:53, blass uri wrote: > >>I think that it is a mistake in correspondence games and that they should >>calculate evaluation for the second and third best move in the first iterations >>in order to know better which lines to extend(this is what humans with the help >>of computer do in correspondence games) but I do not know about one program >that does it. >> >>I have not nimzo and Fritz6 but I do not think that they are different and I >>guess they do the same typical mistakes of chess programs. > >You are wrong. > >Any program that uses the Fritz 5.32 or Fritz 6 interface (including >Hiarcs, Nimzo, Junior, ...) has a correspondence analysis mode that >works exactly as you describe. > >-- >GCP The correspondence analysis is not used in the correpondence games of steve otherwise Nimzo and Fritz could not get big depthes. The correspondence analysis does not work exactly as I described. My idea was not to avoid analyzing the root position for a long time but to analyze some options in the root in the first iterations(in the last iterations you can use the knowledge to decide how much to extend them). I have chessbase engines 1)The correspondence analysis gives a tree and does not give a move. It is possible to find a move based on the tree but it is better to give the program to calculate for a long time because the program may miss a surprising move close to the root of the tree. 2)The correspondence analysis does not use a good method about which lines to extend(it extends the same if the difference between the best and second best is 0.01 pawns or 0.5 pawns). I think that it is a joke and I do not use it for analyzing games. I use many other methods: 1) giving the program a long time to calculate without calculating the second best move 2)giving the program to calculate more than one option at the root to see which lines to analyze. 3)engine-engines games at slow time control from the root position or from other interesting positions close to the root(I prefer this method because I can see evaluations after every move when I cannot see evaluations after every move in correspondence analysis mode) Uri
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.