Author: Vincent Diepeveen
Date: 09:00:12 04/10/01
Go up one level in this thread
On April 10, 2001 at 08:58:14, Ulrich Tuerke wrote: >On April 10, 2001 at 07:05:59, Tony Werten wrote: > >>On April 09, 2001 at 23:40:21, Jon Dart wrote: >> >>>>On April 09, 2001 at 17:04:09, Vincent Diepeveen wrote: >>> >>>> An interesting thing is lazy evaluation, as the problems of it are >>>> very similar to futility pruning. >>> >>>As another poster has said, it is not really risky to exit the eval early if >>>you're certain you are going to be have a score outside the search bounds. I >>>have a test mode where it goes through the full eval always but also checks to >>>see if it would have done a lazy exit, and if so, if it would have done so in >>>error. I run it once in a while and expect no errors. >>> >>>> A possible compromise i found in tests was to increase the margin. >>> >>>Yes. Crafty apparently uses a 1 pawn margin for futility pruning in the qsearch. >>>I tried this and didn't like the results. Mine is almost 2 pawns. >>> >>>I also account for trade bonuses and other adjustments to the material score >>>that would be made as a result of the capture. So I'm pretty conservative, at >>>least in the qsearch (I do other somewhat riskier pruning in the main search). >>>I've been trying some alternative strategies lately but haven't found reason to >>>make major changes so far. >>> >>>> My big question was: what score to return for example if evaluation in this >>>> position is e and e+ 3.5 pawns <= alfa ? >>> >>>> Must one return alpha, estimated evaluation or evaluation+3.5 pawns, >>>> when talking about e+margin <= alfa (idem story for e-margin >= beta) ? >>> >>>I return the estimated evaluation. But I fail to see that it makes a lot of >>>difference. If it's below alpha, you're not going to propagate this evaluation >>>up the tree, anyway. >> >>With failsoft alfabeta it easily could. >> >>Tony >> >>> >>>--Jon > >I think that it is a bad idea to return the estimated eval here because it can >result in some root-value far outside the aspiration window. It would be >interpreted as a corresponding bound which may cause serious trouble (fail low >in fail high verification search for instance). >IMO, it's much safer to return the corresponding beta or alpha. In case of >traversion of this value to the root this would cause a continous re-search >(i.e. over a neighbour region of the original window). > >Uli Exactly, to say polite and use a big understatement it is very dubious to return the estimated value instead of alpha or beta. However if i return alpha or beta then my tree gets blown up bigtime which removes all lazy evaluation advantages... ...apart from that i still need plies more to solve many king safety related positions, which i now solve by high scores (like -20 pawns for king safety) in king safety. Best Regards, Vincent
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.