Author: Uri Blass
Date: 03:30:45 01/03/01
Go up one level in this thread
On January 03, 2001 at 06:18:08, Uri Blass wrote: >On January 03, 2001 at 06:07:08, David Dahlem wrote: > >> >>On January 03, 2001 at 05:07:12, Uri Blass wrote: >> >>>The problem after a long search is the fact that I do not understand the reason >>>for the score. >>> >>>I think that it is possible to give a tree. >>> >>>The idea is that only lines that were searched by the program for more than one >>>second(or different time control that the user choose) are going to be included >>>in the tree. >>> >>>The tree should include also only killers move for one side. >>>I mean that if the program analyzed 1.e5 dxe5 2.a3 when 2.a3 is not the killer >>>move of 1...dxe5 then the tree is not going to include 2.a3 >>> >>>Uri >> >>Hiarcs 7.32 has a correspondence analysis that the user can set either time >>control or depth and it will give a tree. I think Fritz 6 has something similar. >>Possibly this may be what you want. >> >>Dave > >No. > >This is not what I meant. > >I meant to a regular long search >when the only difference is that the program tells me more information > >For every illogical move of black that was considered >for more than 1 second it should tell me the killer move. >(the killer move does not have to be the best move because the program has no >idea >about the best move after a long search). > >It should also tell me the killer move for every illogical move of white that >was considered >for more than 1 second. > >It should not tell me about lines when both sides played illogical moves because >these lines are not relevant to prove the score. >This is all the information that I need to understand the reason for the score >of the program. > >Uri I can add that the word illogical is from the point view of the alpha beta algorithm and it is possible that missing mate in 1 will be considered as logical. I will give an example: Suppose the program wants to play 1.e4 after a long search with score 0.1 The program searched 1.d4 and found that 1...Nf6 is the killer move. 1.d4 d5 is not going to be included in the tree because both sides played illogical moves It is possible that the program analyzed 1.d4 d5 for a long time but it is not relevant. 1.d4 Nf6 is going to be included in the tree. the program also searched 1.e4 f5 2.exf5 g5 and found that 3.fxg6 is the killer move. The line 1.e4 f5 2.exf5 g5 3.fxg6 is going to be included in the tree and the fact that 3.fxg6 is missing mate in 1 is not relevant because 3.fxg6 is logical from the point view of the program that did not search 3.Qh5# 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.