Author: Christophe Theron
Date: 20:05:14 12/27/01
Go up one level in this thread
On December 27, 2001 at 21:05:01, Martin Giepmans wrote: >On December 27, 2001 at 18:53:04, Christophe Theron wrote: > >>On December 27, 2001 at 13:35:00, Martin Giepmans wrote: >> >>>On December 27, 2001 at 12:46:39, Christophe Theron wrote: >>> >>>>On December 27, 2001 at 12:07:35, Martin Giepmans wrote: >>>> >>>>>On December 27, 2001 at 08:43:35, Uri Blass wrote: >>>>> >>>>>>On December 27, 2001 at 08:15:08, K. Burcham wrote: >>>>>> >>>>>>> >>>>>>>I guess that you missed the following post about spidderchess that needed less >>>>>>>than 1 second >>>>>>> >>>>>>>http://www.icdchess.com/forums/1/message.shtml?203638 >>>>>>> >>>>>>>Uri >>>>>>> >>>>>>> >>>>>>>Uri, did you believe this? >>>>>> >>>>>>Yes >>>>>> >>>>>>>it has to be either for the wrong reason, or the program was run after >>>>>>>position learning. we know for the necessary calculations, this position >>>>>>>cannot be determined in one second. >>>>>> >>>>>>Maybe you know it >>>>>>I do not know it and I believe the programmer. >>>>>> >>>>>>I believe that it is not impossible to do it and >>>>>>the fact that other programs are not close to do it >>>>>>means nothing. >>>>>> >>>>>>It is possible that spidderchess use different search rules >>>>>>than other programs. >>>>>> >>>>>>I also plan to use different search rules >>>>>>that are going to help my program to solve it faster but >>>>>>I need to compare them with the known search rules >>>>>>also in other positions before deciding if >>>>>>it is a good idea to use them. >>>>>> >>>>>> >>>>>>Uri >>>>> >>>>> >>>>>It's indeed true that SpiderChess uses different searchrules. >>>>>I try to simulate the human chessplayer. I try ... >>>>>And in this position it works! Now all those other positions :) >>>>> >>>>>Martin >>>> >>>> >>>> >>>>Yes from looking at the PV and the nominal search depth it looks like >>>>SpiderChess does not stop at the nominal depth. For example IIRC at ply depth 5, >>>>the PV was 13 plies long (not ending with captures or checks). >>>> >>>>Or maybe ply depth 5 means that the first 5 plies are looked at in "brute force" >>>>mode, and the next plies with a selective search? That's how Genius displays its >>>>depth, so SpiderChess depth cannot be compared with other programs depths? >>>> >>>> >>>> >>>> Christophe >>> >>> >>>Hi Christophe, >>> >>>What you see beyond the nominal depth are extensions. There is some >>>selectivity before that but not much. The program doesn't use nullmove. >>>Extensions may go very deep along narrow, more or less human paths, so if the >>>program says "depth = 6" you cannot really trust that. >>>But it's probably not the same as Genius. >>> >>>Martin >> >> >> >>When you say "extension", does it mean the kind of extension everybody does on >>recapture or check (not counting the move in the ply count), or do you mean a >>different kind of extensions, something like "I have reached my horizon but this >>position is not quiet at all, so I do not stop here"? >> >> >> >> Christophe > > >Both, but there is still a lot of work to do on extensions of the >second type (don't stop unless the position seems quiet). >The detection of "quietness" that I use now is rather unsophisticated. >It's also slow, very very slow. >Because of this the program mainly relies on extensions of the first type >(not counting the move in the ply count). >This may change in the future ... > >Martin Yes, "extensions" of the second type are really tricky, and I don't know of any program getting a significant increase in playing strength from them. Christophe
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.