Author: Robert Hyatt
Date: 11:14:00 01/21/04
Go up one level in this thread
On January 21, 2004 at 13:34:27, Tord Romstad wrote: >On January 21, 2004 at 12:42:15, Sune Fischer wrote: > >>On January 21, 2004 at 06:54:49, Tord Romstad wrote: >> >>>On January 20, 2004 at 19:48:45, Sune Fischer wrote: >>> >>>>I'm surprised if a few PV moves can cut down the tree my any significant amount, >>>>but then again I never understood why IID works.. :) >>> >>>IID is one of the few concepts even I, who doesn't even adhere to the >>>"fundamental principle of chess programming" mentioned elsewhere in the >>>thread, think I am capable of understanding. :-) >>> >>>When the remaining depth is big, there is no best move in the hash table, >>>and it is likely that the search will fail high (typically because the >>>static eval is above or at least not too far below beta), it makes sense >>>to make some extra effort to obtain good move ordering for this node, >>>because it will probably cut down the size of the big subtree considerably. >>>A good way to do this is to first do a search with reduced depth. Most >>>people seem to reduce the depth by 2, but I always had better results with >>>a reduction of only 1 ply. >> >>I know how it is supposed to work, I just don't understand _why_ it works :) >> >>What if the window isn't fullwidth (alpha>-mate) and you fail low, you have just >>done an expensive D-2 search to get a "random" move to search. > >Yes. This is why it is a good idea not to do IID at all nodes, but only >at nodes where you are reasonably sure that there will be a fail high. >You should always check the value of the static eval before doing the >internal search. I don't even think you want to do it at fail-high nodes. Just at nodes where you expect a "true score". At least that is how _I_ have implemented it. > >There are some possible improvements to IID which I haven't found the time >to experiment with yet. You could for instance abort the internal search >if none of the n first moves fail high, or if one of the "sub-IID" searches >with a even more reduced search depth does not fail high. > >>Apparently that doesn't happen, but it should happen from time to time, >>shouldn't it? >> >>If it happens now and then, doesn't that eat away most of the savings? > >I am sure it does, if you are not careful about where you use IID. > >Tord
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.