Author: David Rasmussen
Date: 23:10:14 12/04/02
Go up one level in this thread
On December 04, 2002 at 20:05:14, Russell Reagan wrote: >Let me see if I understand what forward pruning accomplishes. I understand how >it works. Forward pruning only reduces the time to depth (and nodes to depth), >but it doesn't produce a more accurate score when searching to a fixed depth. In >other words, if you had a choice of the two following options: > >1. 10-ply search using no forward pruning >2. 10-ply search using null-move > >and all other things being equal, then choice 1 would potentially be the more >accurate 10 ply result. Is this correct? In practice, yes. But ideally, no. If the forward pruning was "perfect" it would only prune things that weren't important. That is the goal. A more realistic and practical goal would be to have PV's (specifically the first move of the PV) that are "as good as" a PV from a non-pruned search or at least "good enough". All of this is of course when searching to a fixed depth. Searching to this fixed depth should take (significantly) less time, or else there is no reason to prune. Specifically, we hope to be able to search at least on ply more in the same time, so as to enhance tactical and positional ability. /David
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.