Author: Colin Frayn
Date: 17:33:28 04/07/00
Go up one level in this thread
>Yes, I strongly disagree because normal futility pruning >is _theoretically sound_ for searchers with a capture-only >quiescence and a futility margin that is at least as large >as the maximal positional score produced by the static >evaluation function. Standard futility pruning simply "lifts" >stand-pat cutoffs from horizon nodes to frontier nodes. Yes, I think my program fails on the 'capture only' quiescence bit. I suppose I could always implement futility pruning deep in the quiescence search as soon as the "capture only" bit begins. Is the type of pruning you just described actually going to give you any improvements though, as that's a pretty large futility margin? >In your description of extended futility pruning it should >read "mat_gain(move)" and not "mat_gain(node)", Yes, sorry. I should have read over that one before I submitted it :) >If you do not restrict the pruning to quiet moves, you end up >doing real razoring which is clearly bad and produces exactly >the results Tom and you mention. I actually did restrict it to quiet moves, but I still got a drop in tactical accuracy. Admittedly it wasn't as strong as the one Tom mentioned, maybe 2-3% in WAC tests perhaps, but that was enough to put me off it. The speed bonus I got was maybe 30% however, which would have been nice...... >BTW, several strong commercial chess programs use normal and >extended futility pruning meanwhile with good success. So, >somebody must do something wrong ... I just don't want to implement anything which isn't totally theoretically sound. Admittedly I could implement a type of sound futility pruning, as you say, but I'm certainly not going to add anything which could get even 1% of tests wrong as that goes against the idea behind my program. Hence no null moving etc... Cheers, Colin
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.