Author: José Carlos
Date: 07:32:33 12/31/03
Go up one level in this thread
On December 31, 2003 at 08:50:07, Frank Phillips wrote: >On December 31, 2003 at 07:22:41, José Carlos wrote: > > >> >> >> As suggested in other post, you can simply return alpha then. I do it >>differently, I always try the first move (according to my move ordering), and >>allow pruning all the others. >> >> José C. > >Yes I have tried that too. Do you get a net benefit from this Heinz-style >pruning? I keep coming back to it (in various adaptations), but it never seems >more than 'a wash' as Bob might say. Yes, but only in my latest versions. I couldn't quantify it because I've been changing many things in Averno lately, but it reduces the tree a little. Nothing impressive, however. As I see it, futility pruning creates kind of a bridge between regular search and qsearch, like a selective part, but really too thin to make a difference. I don't like the idea of extended futility as described by Heinz because it doesn't make sense to me. Let's think of it: We are at depth 2 (to me, depth 0 calls qsearch directly, depth 1 tries all move with futility pruning) and we are about to make a move. Our material score is below alpha and this move won't raise the score above alpha-margin, being margin a constant which, for now, is not important. We know we will move and then force the opponent to make a move and well drop into qsearch. Then, it's me who will have the option to stand pat or try a capture, so being at depth 2, and assuming any move by the opponent won't worsen his position (this assumption we always make in null move), the only thing we have better than in depth 1 is that we'll be given the chance to stand pat. Since we assume the opponent won't worsen his position, we'll be given the option to stand pat in a bad position, so no advantage. Under this point of view, I would try futility with the same margin at depths 1 and 2, and then a bigger margin at 3 and 4, and so on. But I haven't had time to test all of this. For now, I stay on the safe side and use a bigger margin at depth 2 and, so far, both futilities seem to save some nodes to me. José C.
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.