Author: Heiner Marxen
Date: 15:23:23 05/24/00
Go up one level in this thread
On May 24, 2000 at 17:23:27, blass uri wrote: >On May 24, 2000 at 17:05:38, Heiner Marxen wrote: > >>On May 22, 2000 at 19:01:38, Dann Corbit wrote: >> >>>On May 22, 2000 at 15:52:20, Daniel Chancey wrote: >>> >>>>I agree that Chessmaster is #1 in finding solutions. If someone could hold a >>>>checkmating contest (The organizers set up positons in which mate can be forced >>>>in a number of moves in the middlegame) Chessmaster might be on top. >>>> >>>>If anyone knows who's the best in 5 min, 15, 30, 60, and 40/120 times in >>>>Chessmaster, tell me. If not, I'm on it. >>> >>>Chest would slaughter them. >> >>Sorry to disappoint you, but: no, Chest will not slaughter them here, >>since finding *some* forced mate is a quite different job, >>than what Chest normally does. >> >>To find that there is no mate in 8, Chest on a PIII/550 with 500MB cache >>already burns over 20 minutes for the first of the three positions, >>and 48 minutes for the third. >> >>I have plans to introduce a sort of mate finder mode (what leonid >>calls "quick logic"), where the attacker is restricted to only >>perform checking moves, except for say one quiet move. > >I do not think that the attacker perform only checks move except one in the >examples that I posted. > >I think that you should let the attacker to do something more general then >checking moves execpt one case to solve the problems(I think that you should let >the attacker to play "quiet" moves that threat mate in 1 or mate in 2) Unfortunately, detection of a "threat" is not for free. It is not even defined with absolute clarity. Maybe some time I will consider it more closely, but currently there are many other things to change in Chest that are more important to me. Restricting the attacker to checking moves is not a bad idea: in real games that end with a mate there are often longer series of checking moves before the mate. A very old version of Chest already had this feature, and the results were quite good (IMO). >>Such a job will be much quicker, of course. >>In *that* mode, Chest probably slaughters them, but that remains >>to be seen. > >The challnage is to be faster than chessmaster(My experience is that other >programs are clearly slower than chessmaster in the positions that I posted and >they may be 100 times slower and not only 2 or 3 times slower). > >Uri Well, that is a really big challenge. When I have done that mate finder mode, I will try the 3 positions, and report my findings. Some time this year (I hope so). 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.