Author: Vincent Diepeveen
Date: 03:55:26 04/04/01
Go up one level in this thread
On April 03, 2001 at 04:40:28, David Blackman wrote: >Run 2 different killers: capture ones and other ones. Or let captures into the >killer list but not the history list. Or let captures into the killer list but >only if they are losing according to SEE. Or something like that. You can't let >captures wipe everything else out of the killer and history lists. That loses. >But it is probably a win to do captures early if they are bad according to SEE >but good according to killer or history heuristic. Actually i tried this years ago, there is one big problem however. Suppose a losing capture wins in a position X and suppose i put it in the capture killermove tree. I can assure you that it'll find a stupid capture which works great in position X. Now suddenly in many normal positions it goes try this stupid capture which only worked in position X very well. You can't do a stupid capture without storing why it's good. However as soon as you can store why it's good, then it's no longer a 'stupid' capture as you could also statically figure it out then...
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.