Author: Robert Hyatt
Date: 11:02:02 01/26/00
Go up one level in this thread
On January 26, 2000 at 00:53:56, Christophe Theron wrote: >On January 25, 2000 at 23:27:40, Peter McKenzie wrote: > >>On January 25, 2000 at 20:52:28, Christophe Theron wrote: >> >>>On January 25, 2000 at 17:29:15, blass uri wrote: >>> >>>>On January 25, 2000 at 13:31:01, Christophe Theron wrote: >>>> >>>>>On January 24, 2000 at 15:50:15, blass uri wrote: >>>>> >>>>>>On January 24, 2000 at 15:34:39, Christophe Theron wrote: >>>>>> >>>>>>>On January 24, 2000 at 09:10:12, Robert Hyatt wrote: >>>>>> >>>>>>>>I suspect that one commercial program is doing no capture search, but may be >>>>>>>>using a static exchange evaluator to prevent the last move in a PV from >>>>>>>>being an outright blunder. Seems to be working. No names to protect the >>>>>>>>guilty, however. :) >>>>>>> >>>>>>> >>>>>>> >>>>>>>I'm really curious about this. Using a SEE to "close" the search is in my >>>>>>>opinion a beginner's mistake (no offense intended). >>>>>>> >>>>>>>It's the first thing I have done myself when I started programming chess, and I >>>>>>>suspect many of us have followed the same path. >>>>>>> >>>>>>>But if somebody tells me it really works, then I have, once again, to reconsider >>>>>>>seriously all I know about chess programming. >>>>>>> >>>>>>>It's not new anyway. Every now and then I realize how little I know. That's part >>>>>>>of the fun. >>>>>> >>>>>>The program hyatt talk about is probably Junior. >>>>>> >>>>>>Uri >>>>> >>>>> >>>>>Really??? >>>>> >>>>>Why do you think it's Junior? >>>>> >>>>>Junior is said to have a huge NPS. Using a SEE at the end of the tree (in every >>>>>node actually) is supposed to drop your NPS considerably... >>>>> >>>>> >>>>> Christophe >>>> >>>>I remember that amir ban said in the past that he does not think that Qsearch is >>>>a good idea. >>>> >>>>I also remember that there was a position that proved it when Junior never did a >>>>mistake at small depths that other programs did because of Qsearch. >>>> >>>>Uri >>> >>> >>>OK. I hope Amir reads this and can say if it is correct. >> >>Maybe they do a simplified qsearch where you just do recaptures on the >>destination square of the last fullwidth move? I played with this once, it >>isn't too bad. I eventually discarded it, probably for the wrong reasons :-) > > >I had something similar in Tiger 2 years ago: QSearch would, after some plies >(could be set to any depth), look only at captures on the last square. I >eventually discarded this after extensive testing. > > >Maybe you and I discarded it for the right reasons! :) > > > Christophe > > I don't know. It always worries me when I discard something for the "right reasons" (I don't think I _ever_ discard something for the wrong reasons, knowingly. :) ) and then find that someone else is still using the idea and making it work. It means I have to think about why my reasons were wrong after all. :) not always easy to do either... I discarded SE a long time ago because it hurt performance. When I tried them again in 1993 or so they were much better because of the increased speed on my machine (the Cray). Maybe we all just failed to find "the trick" that makes them viable... > > >>>I am surprised, but I like to be surprised this way. >>> >>> >>> 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.