Author: Heiner Marxen
Date: 13:13:05 09/05/01
Go up one level in this thread
On September 04, 2001 at 22:14:27, leonid wrote: >On September 04, 2001 at 20:51:23, Peter Berger wrote: > >>On September 04, 2001 at 19:16:29, leonid wrote: >> >>>Solution time is excellent! >>> >>>Chessmaster 8000 for sure found mate by selective and its selective search is >>>very effective. But how about brute force search? I am curious to know how he do >>>on this sense. >>> >>>To spot its brute force search (just my suggestion) find what is the Chessmater >>>time for 8 moves for the first position and in 12 moves for the second. In both >>>he will not find the mate but probably will end by searching by brute force. >>> >>>If you could indicate its brute force search for two close moves (for 7 and 8 >>>moves deep, for instance) it will be very revealing. This will permit to see his >>>branching factor for this kind of search. >>> >>>Some programs are very efficent in brute force search but not that tuned for >>>selective. The best brute force searcher that I saw here until now was Heiner's >>>mate solver. His time is close to miracle. >>> >>>Cheers, >>>Leonid. >>> >> >>Hi Leonid, >> >>to the best of my knowledge Chessmaster 8000 can't be set up to do a "brute >>force search" by an ordinary user - don't forget : it isn't a mate solver but a >>normal chess playing program. > >I remember that I tried on my own to find, in one previous version, if Chess >Master was using "brute force" but never found direct indication about this in >its description. It is good that you put this clear for me. > >Some programs do search of mate by brute force (at least it is my impression) >even if they are not mate solvers. Between them I tried Genius 4, Rebel 10 and >Mate 2.0. Some even indicate in their choice "selective search". > > > >>So Chessmaster's results are always "selective search". >> >>Chessmaster 8000 seems to be the best when it is about "selective search" so far >>: if it is about your positions with zillions of queens and other pieces - or >>more regular ones ( like in the chess newspapers) . > >This last was "normal". Too many pawns. I did this expecting that everybody, >with every program will be able to solve them. Then I also will be able to see >if mine could compete with usual program in "almost natural environment" . > >>Chest is without competition itself when it is about "brute force search " - I >>definitely agree ! Hi Leonid, Hi Peter! Thanks for all the kind words! >>Although it's not that bruce force at all - Heiner Marxen will use some neat >>tricks to improve his "brute force" results : and my impression was that many of >>them never would be thought of as brute force anyway . For example Chest does >>pruning ( to my _very_ miserable understanding at least ) . > >This is big surprise for me! But this can be seen only by directly using his >mate solver for solving many mate containing positions when shortest mate is >known in advance. Well, in some sense Chest does some pruning, but if it does, Chest is absolutely 100% sure that the pruned tree does not contain a forced mate. Those prunings are "safe". E.g. when the white side (which has to deliver mate) is left with just one piece except for the king, Chest knows that white must not loose that piece, and will not try any move that leaves that piece at a square where it may be captured by black. (That can become even more complex :-) The result will always be the same as for a complete "brute force" search, although the method Chest uses to arrive there may be called "not brute force" by some. For me it is a tuned version. >I hope that Heiner will be not that long in producing, finally, his mate solver >with handy and friendly graphics. For me it will be best possible reference that >could better suggest what practical limite in speed I can expect. Before I did a >mistake by looking into few others, less speedy programs. Unfortunately, after a long illness now my disk completely broke down... and some of my recent changes are lost :-( and almost all computations done in the last year. Hrmpf. I still hope to release something in December, or may be January. >My mate solver brute force contain no short cuts, no composed in advance "look >up table". It goes completely after the strict logic and see all possibilities >inside of given position in order of saying its final result. > >Selective use almost identical logic, as brute force, but search only through >promissing moves. > > >>And - Heiner said he is or will be working on his "selective search" soon , too >>. So I expect Chessmaster to be up for some thrilling competition when he is >>ready :) ! > >I hope so! That will last a bit longer, most probably :-( Cheers, Heiner
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.