Author: Robert Hyatt
Date: 05:56:17 04/05/00
Go up one level in this thread
On April 05, 2000 at 07:55:09, Dezhi Zhao wrote: >On April 04, 2000 at 23:09:21, Robert Hyatt wrote: > >>On April 04, 2000 at 07:50:18, Jan Pernicka wrote: >> >>>On April 03, 2000 at 15:35:57, Robert Hyatt wrote: >>> >>>>>> >>>>>> Excuse me, could someone explain to me in several sentences what is >>>>>> SEE function good for? Thanks >>>>>> Jan >>>>> >>> >>>>What it does is ask the following question: >>>> >>>>If I capture the piece on square X, what will I gain (or lose) in terms of >>>>material. You just enumerate the white and black pieces that attack square >>>>X, and then do the exchanges to see if it is good or bad to do. Just like a >>>>human asks "Can I move my knight to e5?" by noticing the pieces for both sides >>>>that are currently attacking e5... >>> >>> So, what's the difference between SEE and quiescence search (when you >>> take into account only capturing moves "on 1 square")? >>> I think, they could be implemented in very similar way... >>> >>> Thanks, >>> >>> Jan >> >> >>if you limited the q-search to the single square, you would (a) supposedly get >>the same score as the SEE code, but (b) much slower, as the SEE is not doing >>any sort of move generation, recursive search stuff, etc. Way faster... >>And (IMHO of course) it would be prone to serious errors. I once did this >>in a 1970-72 version of 'blitz' but when I went from selective to full-width, >>I started doing a traditional capture search which is more comprehensive in >>detecting overloaded pieces, multiple hung pieces on different squares, etc.. > >If I read your posts correctly, the SEE function in Crafty only plays >with a score list but does not check the legality of the capture moves. >What Crafty do if the 1st or 2nd move in the list is in fact illegal >(the piece is pinned)? Does Crafty get a wrong SEE score in these cases? Yes. You can try making SEE "legal" for testing, but you'll find that the cost in speed isn't offset by the increased accuracy. Remember that in the normal search, SEE is not throwing moves away, just helping to order them. If it is wrong, it doesn't affect the chess, only the speed. > >By the way, CCC is now a lot faster for me than before:) > >dzhao
This page took 0.01 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.