Author: Robert Hyatt
Date: 05:05:24 01/22/98
Go up one level in this thread
On January 22, 1998 at 00:10:44, Andrew Walker wrote: >On January 21, 1998 at 23:54:49, James Long wrote: > >>On January 21, 1998 at 22:27:42, Andrew Walker wrote: >> >>> Here's an idea I came up with recently, it may be old news >>>but I've never heard of it being used so here goes: >>> >>> One thing I've noticed with chess programs (mainly at a fixed time per >>>move) is that often the search will stop when they have spent ages >>>considering the main move, or when several replys to a none best >>>move have been examined and it looks like it may become the new best >>>move. Generally the best thing that can happen in a search is when a new >>>move becomes the main move and with a reasonably higher evaluation. >>>So we would like this to happen sooner if at all possible. >>> The way searches normally work is that when the depth is increased, the >>>best move will be searched first. My idea is to search >>>all of the other moves first. >> >>When searching, it is important that the correct move be very close >>to the top of the list. This way, all other moves will be cut off >>with minimal work. If you searched in the opposite direction, you >>would get absolutely no benefit from a/b cutoffs. >> > Did you read the next paragraph of my post? As I have said, the >alternate moves may be all cut off! The direction is not reverse, the >alternates >are searched in the same order as they would be normally. > How can they be cut off since the real PV move has not yet been searched, so we have no idea what the true score should be? >>> Lets say a depth 6 search has been done for white and the best move is >>>e2-e4 with a score of +0.1. What we could do is start the depth 7 search >>>looking at all moves except e2-e4, and make them try to beat a score of >>>0.1+c, where c is a small positive constant such as 0.2. > >>> Andrew Walker >>> ajw01@uow.edu.au
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.