Author: Peter Fendrich
Date: 07:13:34 08/17/04
Go up one level in this thread
On August 17, 2004 at 08:32:55, Uri Blass wrote: >On August 17, 2004 at 08:14:28, Gabor Szots wrote: > >>On August 17, 2004 at 05:20:46, Uri Blass wrote: >> >>>On August 17, 2004 at 02:29:03, Gabor Szots wrote: >>> >>>>On August 16, 2004 at 20:30:06, Tony Nichols wrote: >>>> >>>>> I'm looking for the best engine/settings for tactical analysis of my games. >>>>>Does anyone have any suggestions? >>>> >>>>Among freeware engines Gothmog is by far the best tactical program. I'd hazard >>>>the opinion it equals the best commercials. >>> >>>Is it based on analysis of games? >>>I am interested more in detecting blunders in comp-comp games and not in test >>>suites when the target is to find sacrifices. >>> >>>tactics is not only sacrifices. >>> >>>I think that it may be interesting to give engine the games of premier division >>>WBEC(40/40 time control) and to run engines on the games when the target is to >>>find blunders. >>> >>>You can use Fritz with every engine by tools->analysis->blunder check >>>You can decide about constant time and constant margin. >>> >>>Later you can have list of positions when at least one of the top programs >>>detected blunders and these positions are candidate positions and if all top >>>program still consider the move as blunder after more time and the evaluation >>>is not clearly winning for the same side before the blunder and after the >>>blunder you can decide that these blunders were correctly detected as blunders(I >>>do not detect a move as a blunder in case that the evaluation was changed from >>>+10 pawns to +20 pawns for the same side). >>> >>>Uri >> >>I agree that tactics is not only sacrifices. Therefor I base my claim on test >>suites of two types: >> >>(1) Tactical test suites such as ECP-GCP and IQ2. >>(2) The king attack part of the WM-test. >> >>These are totally different in nature (first for sacrifices, second for >>intuition) and Gothmog excels in both. >> >>Gábor > >king attack is often about sacrifices. > >If you are interested in positions that are not sacrifices then maybe arasan >test suite is better. > >I also think that tactics is about calculation and not intuition so the test >should not be about finding a move but about finding some minimal difference in >evaluation. In general I don't agree to that statement. Michail Tal, the former World Champion, said himself that he often made moves by pure intuition and those were som of the most beutiful tactical shots in history. Chess computers don't of course have any intuition, what I know about... I suppose that Gábor refered to human intution in this statement. /Peter >Another problem is that 0.6 pawns of program A is not equivalent to 0.6 pawn of >program B so we should have some divtionary for the meaning of advantage by >different programs. > >Basically it is better to translate evaluation to probabilities based on games >of the programs when blunders are defined as moves that has some minimal defined >influence on the expected result of the game. > >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.