Author: Robert Hyatt
Date: 12:14:10 12/07/00
Go up one level in this thread
On December 07, 2000 at 14:38:05, Ernst A. Heinz wrote: >Hi Bert, > >>In a previous question about Futility pruning Vincent Diepeveen mentioned: >> >>>Futility pruning is very dubious. It sure speeds you up, but it's >>>dubious and tends to give you slight positional >>>differences that make a program play positional >>>a lot weaker. >> >>Can sombody explain me why Futility pruning tends to give slight positional >>differences? > >Vincent's statement is just wrong for futility pruning at >frontier nodes which is theoretically sound for suitable >"max_posn_score" margins, given that the search allows for >"stand-pat" beta cutoffs at horizon nodes (which typical >quiescence searches usually do). > >I have already explained this several times here and in my >publications but, unfortunately, Vincent does not seem to >understand. For more detailed information, please read my >article on extended futility pruning in the ICCA Journal >(a preprint thereof is available from my WWW pages at >http://supertech.lcs.mit.edu/~heinz/). > >However, if you use aggressive futility margins, the >forward pruning becomes selective and might erroneously >cut moves that actually affect the value of the frontier >node in question. As Bob already pointed out, selective >pruning always tends to lead to differences of *any* kind >(i.e., not only positional but also tactical ones). > >=Ernst= If you use _any_ kind of pruning, you can perturb the root score. Thanks to the transposition table that grafts parts of the tree into odd places. There is no way to anticipate which branches are useless _here_ but the hash results might improve the results over _there_. And as a result, you can get different root move scores no matter how "sound" the pruning.
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.