Author: Uri Blass
Date: 01:39:24 01/16/03
Go up one level in this thread
On January 16, 2003 at 04:07:35, Uri Blass wrote: >On January 16, 2003 at 03:32:30, Uri Blass wrote: > >>On January 16, 2003 at 01:00:00, Scott Gasch wrote: >> >>>On January 16, 2003 at 00:15:24, Nathan Thom wrote: >>> >>>>Im an amateur chess player (around 1300), but love to program interesting >>>>problems. It seems that most of the programmers here are all very highly rated >>>>chess players. Most chess programs beat me easily, so I thought it would be >>>>interesting to see if I could write a program that could beat me aswell. >>>> >>>>In peoples opinion, will it be hard for me to write a program that can play very >>>>well (say 1800+) even if it only uses my basic knowledge of chess? >>> >>>In my opinion it's way more important to be a good programmer than it is to be a >>>good chess player in order to write a strong program. Writing a program to play >>>at an 1800 level is not hard at all... you can have a ton of bugs and it will >>>still do ok on a fast machine. I'm a terrible chess player and have an engine >>>that plays an fairly good game of chess... >>> >>>Scott >> >>You have a very complex evaluation for a terrible chess player. >>The last time that I read about your evaluation I could not understand your >>explanation when you evaluated position when no square near the king was >>attacked as more than +1 for black. >> >>I think to add some king safety evaluation to my program but the material that I >>read was too complex for me to understand. >> >>I could understand as a human that white had significant problems with king >>safety but I had no idea how to explain it to a computer and your explanations >>did not explain how do you do it(for example how to evaluate pawn storms). >> >>Uri > >I can add that I plan to add some code about kingsafety but only when squares >near the king are attacked by the opponent. > >I believe that in most cases that there is a king attack it is also possible to >see by search squares near the king that are attacked by the opponent. > >Movei is already a slow searcher and if the code does not do movei more than 10% >slower and first tests will suggest that the code is productive I will use it. > >Maybe I am taking risks here and I will produce a new bug that I will not >discover in the hours before the tournament but I will probably do it. > >Not taking risks and not changing the program except changing weights in the >last minute is good in most of the cases for the commercial programs that has no >big holes in evaluation. > >For Movei things may be different and I plan to do changes that I will use only >if first tests suggest that the changes are productive. > >Maybe it is a mistake to try it and Movei will suffer like a lot of other >programs in a previous tournament but I am not sure about it. > >It is also possible that I will decide in the last minute to use only the latest >version that I have now. > >I am interested to know what is the experience of other programmers with changes >in the last second. > >How many changes were good and how many changes were bad(I am not talking about >the top programs)?. > >Uri I can add that for movei the main problem is pawns missing near the king but code that says that the king is always in trouble when pawns are missing near the king is not good and I have also games when Movei won inspite of missing pawns. I will probably try to add the smallest scores that I can to avoid the errors that were done in games but first I need to write the code that I already started. The relevant 2 questions to evaluate are: 1)how many squares near the king are attacked by the opponent. and if it is more than 0 2)how many pawns near the king are missing(pawn at h3 instead of h2 g3 instead of g2 or f3 instead of f2 are going to be defined as partially missing). 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.