Author: Sune Fischer
Date: 10:47:15 06/22/02
Go up one level in this thread
On June 22, 2002 at 11:42:31, Robert Henry Durrett wrote: > >Upon reflection on the subject, it seems to me that all successful chess engines >would make extensive use of any and all techniques which could speed up the >overall process. You might say "Time (clock cycles) is of the essence" in chess >engines. I would expect the idea of using indicators would have widespread >application in chess engines. > >Just as a fun thing to do, consider the possibility of indicators which would be >solely for the purpose of deciding whether or not to initiate execution of more >complicated "indicator" blocks of code. > >Bob D. I don't think there is a program not using it. Extension, pruning, what and how much to evaluate, those are examples currently in use in chessengines. But as my argument before showed, you need billions of code sections to handle all the individual cases. What you want is a actually an EGTB for all positions, and then a probe into the evaluator. The probing can be done more or less elegantly, but who should write all that evaluation code? Only if we can extract common set of rules can we hope to evaluate a significant subsection of those positions. Maybe it is not impossible, humans seem to be able to recognize some things very fast. If you ever figure out how to implement that process, then please let me know :) Think of the fortress positions, easy to see for a human, not so for a computer. Are they important, well yes Smirin showed us that, but how do we evaluate it staticly? -S.
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.