Author: Ron Murawski
Date: 11:14:05 07/29/01
Go up one level in this thread
On July 29, 2001 at 07:58:38, Tony Werten wrote: >On July 29, 2001 at 07:29:05, Uri Blass wrote: > >>On July 29, 2001 at 05:44:27, Tony Werten wrote: >> >>>On July 28, 2001 at 18:47:21, Ron Murawski wrote: >>> >>>> >>>>I have implemented a pin bitboard for king-pinned >>>>pieces and it has helped the strength of my program. >>>>My question is: Is it worthwhile to identify ALL >>>>pinned pieces? >>> >>>Ignore all. It is not a starting engine issue. You'll be adjusting it every time >>>you make some changes. >>> >>>The state you're engine is at now you should leave it up to quiescence to >>>determine wether a piece can or cannot capture. >>> >>>Thre is lots of basic stuff you still have to solve. Don't go into details to >>>early. TSCP and Gerbil are examples showing that just doing the basics but doing >>>it good still gives a strong program. >>> >>>cheers, >>> >>>Tony >> >>I think that it is better to think on your own algorithms and not copying >>algorithms from other people. >> >>I do not see the point of doing another strong program like everyone. >> >>I think that a significant progress in computer chess may be possible by >>thinking about different algorithms and the fact that almost everyone try to >>copy algorithms from other people is not productive for getting the progress. >> >>The problem is that the known way to develop chess programs was investigated too >>much so if you start a different way you can expect that your program is going >>to be weaker not because the different way is worse but because of the fact that >>the different way was not investigated and you have nobody to copy from. >> >>I believe that evaluating pinned pieces may help (I read that people found that >>this knowledge does not help much in the evaluation but I think that the main >>advantage of it may be not in the evaluation but in the search because it may >>help to get better order of moves or to know better which lines to extend(for >>example it may be a good idea to search first a move that is attacking a pinned >>piece and not a move that is attacking another piece). >> >>I do not say it from my experience of developing a chess program because I still >>do not have a chess program and only a move generator. > >That's basicly what I meant. You can't experiment with new ideas if you don't >have a decent playing program. Otherwise your results don't say anything about >your ideas, just about the gaps in the program. > >cheers, > >Tony > >> >>Uri Okay. Thanks for all the advice. I hadn't realized I was walking this road alone. My program isn't quite as weak as I've led you to believe. It seems to hang in there vs GnuChess for 70-80 moves. I'm losing on endgame knowledge now. My suspicion is that other pins, due to their tenuous nature, are best dealt with by the search. Again, thanks for all the replies. Ron Murawski
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.