Author: Uri Blass
Date: 02:20:38 05/11/05
Go up one level in this thread
On May 11, 2005 at 02:17:57, Sune Fischer wrote: >On May 10, 2005 at 18:11:52, F. Huber wrote: > >>On May 10, 2005 at 17:43:15, Sune Fischer wrote: >> >>>On May 10, 2005 at 14:00:25, F. Huber wrote: >>> >>>>IMO at least 2 conditions should always be true: >>>>1) exactly one king for each side >>>>2) a maximum of 16 pieces for each side >>> >>>Number 2 is not strictly necessary, I think many engines can >>>play perfectly sensible "chess" with more than 32 pieces. >> >>That depends on the engine´s implementation details - it would certainly make >>problems, if the engine uses some kind of piece tables. > >Yes it will depend on the implementation but I don't think there is >anything in the game rules as such that prohibits it, other than the >fact that the position is illegal because you can't reach it from the >opening position. > >>>However, to be well defined at least the following must be true >>> >>> *) no pawns on 1st and 8th rank >> >>Here we would have to differentiate: >>A white pawn on 1st rank (or a black one on 8th rank) shouldn´t really be a >>problem (at least not for the move generator) - it could simply move like any >>other pawn, and if reaching the 2nd rank even make a double-step. > >If you start to define new rules a lot of things can suddenly make sense, but >there is currently no rule that tells us how such a pawn would move. > >>For a white pawn on 8th rank (or black on 1st) I would have a good idea for a >>new chess variant: >>Whenever a pawn reaches the 1st/8th rank, it is not necessary to promote him >>immediately - he simply waits here until he´s needed, and must/can be promoted >>to any usual piece only when he is moved the first time! > >So again we would have to redefine the game. :) > >>Isn´t that a good idea? >>What must I do to patent it? ;-) >>Any hints, Ed Trice? :-) >> >>>>That are e.g. the necessary requirements for Chest, and seem quite logical >>>>to me - everything else should not really confuse a well programmed engine! >>> >>>Hmm.. :) >> >>How should I understand this? Could it be that Frenzee _gets_ confused? ;-) > >Frenzee can handle 15 pieces of each type for each side (only one king though). >It's just exploiting the observation that the number 9 fits in 4 bits :) > >There are other observations to be made and I think writing a strong engine >requires one to take advantage of many of those observations. >Therefore I'm not surprised that if the stronger the engine the less >flexible the code will be. I think that people care too much about speed differences. The important thing is the algorithm. I believe that it is possible to use the same algorithm and handling every position without being more than 2 times slower and an engine that can beat shredder9 with 2:1 time handicap is a very strong engine that can be commercial(I believe that most of the commercial programs cannot do it). Movei may also crash in position when the king is threated from 2 bishops direction but I could do it 1% slower without that problem. Unfortunately I am too lazy to change it now. I decided that speed is not important at this moment and I could probably do latest movei at least 10% faster without allowing the users to change parameters of Movei personality. If Movei will be able to beat the best commercial program with 2:1 time handicap I may start to think about speed optimizations. I do not think to spend time about it before doing it. 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.