Author: Dan Andersson
Date: 11:20:08 10/18/03
Go up one level in this thread
It might be more productive to have a dynamic criterion of term in the evaluation slightly biased in favour of open positions. Or if it is reasonably successful as a predictor to use it for enabling different features in the Q-search. Testing that would be reasonably easy. Do a test version of your search that compares the different Q-search results at each node you visit. And record the cases where they differ hugely. And make note of features in your eval that seem to have an affinity to upsets. It might even make a reasonable predictor to do this test Q-search at a node one or a few halfmoves before the leaf node. To mortgage the costs somewhat. Just off the top of my head. MvH Dan Andersson
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.