Author: Robert Hyatt
Date: 07:35:08 08/25/02
Go up one level in this thread
On August 25, 2002 at 09:53:10, scott farrell wrote: >Robert, > >Am I correct, in that you still need to be careful in doing too many extensions, >and stopping from getting to the next depth? Maybe or maybe not. It depends on how well you do the extensions. IE if you extend _all_ the right lines, but don't extend bad lines, then you can't help but be better. But, in general, this is impossible to do, so you end up extending what you shouldn't, also, and you end up not extending everything you should. Not extending something doesn't penalize you in terms of time, although you will not see things that you should. But extending that which you should not is a serious problem, because it simply lowers the overall search depth. And search depth is what helps you compensate for not extending the things you should... because you simply see it without extensions. > >Many other types of extendions (like quiesence search) you never want to reduce >your chance to get to the next ply. Obviously getting to the next ply is better >value than doing more quisence searching. Again, it is an issue of how accurate you are doing things. YOu could, in theory, search exactly one node and win every game. What you hope is that your extensions make your program tactically more accurate... testing is the only way to see if this happens however... > >Is this the same case for SEs? or doesn't it matter if you are a ply less, but >solving some tactical positions better? > That is a good question. If the extension makes you find most things quicker, then it is a winner. It will almost always slow you down in places, but you are interested in the overall improvement, not the improvement on _every_ position. >Scott >sfarrell@icconsulting.com.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.