Author: Uri Blass
Date: 08:31:47 12/10/02
Go up one level in this thread
On December 10, 2002 at 10:45:04, Vincent Diepeveen wrote: >On December 10, 2002 at 09:59:56, Uri Blass wrote: > >>On December 10, 2002 at 09:20:02, Nicolas GUIBERT wrote: >> >>>>hello, the size of the evaluation of napoleon is only limited by the >>>>time i spent to it. I have 40 pages of evaluation patterns and ideas >>>>from Marcel to the left of this keyboard. For this tournament i could >>>>only implemet page 1 and 2. So i added 40 new patterns/bugfixes to >>>>the evaluation for last tournament the 2 evenings before the blitz >>>>tournament. >>> >>>Good luck then for these 40 pages. I also have a long, long, long, never-ending, >>>to-do list... :o))) >>> >>>>In C code expressed it is in total let me check: >>>> >>>>4000 source code lines. >>>> >>>>That ain't much IMHO, knowing i have a multiple of that in DIEP. >>> >>>How much in Diep ? >>>4000 is indeed way insufficient. >>> >>>>But it is probably more than most draughts programs (and way more >>>>than most chess programs). >>> >>>Really ? How much do you think it is for other top level chess programs ? >> >>He talked only about most program and not about most top program. > >i don't know about buggy (Nicolas plays draughts at the same >level that i play chess) but when compared to the rest for sure yes. > >>The latest version of movei(that is not freeware) is not a top program but I >>believe that it is better than most chess programs(there are a lot of free chess >>programs and it is hard to know if it is better than most of the chess programs >>because the question is what is the level of the private programs) >> >>It has only about 200 lines of C codes that are used for evaluation. >>I am interested more in search improvements then in evaluation improvements. >>Uri > >Then you will of course never make it to the top. > >Best regards, >Vincent I am not saying that I will not improve my evaluation. It is possible that in the future I am going to change my mind and decide that evaluation is more important but today it is not the case. I also plan to improve the evaluation but most of the development time is going to be for improving the search. I believe that good pruning ideas that are not null move pruning and good extensions and better order of moves can help significantly. I know your opinion against forward pruning. I have a different opinion. I believe you that your tries in forward pruning was productive in test suites but not productive in games but I believe that I have different ideas of forward pruning that are productive in games and not only in test suites. I believe that size of evaluation is not the only thing that is important. I know amateurs do stupid mistakes because of wrong weights(one example is that I saw a lot of games of amateurs when they lost 2 pieces for rook and pawn because of the stupid weights) I believe that 200 lines are not enough to have a top program but 2000 lines with the right knowledge and good search algorithm may be enough. It may be interesting to know how many lines do programs like Junior or Ruffian have. Uri
This page took 0.02 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.