Author: Robert Hyatt
Date: 07:41:47 01/18/98
Go up one level in this thread
On January 18, 1998 at 01:29:31, Stuart Cracraft wrote: >I read Thomas Anatharaman's (sp?) paper on extensions. After many >weeks or months of testing on Deep Whatever, he concluded that >the best extension by far was the threat extension. > >I certainly believe that people in the micro field and Hyatt have a lot >more experience with extensions and may take issue with that but let >us forge onward. > >So I read that paper very carefully but am no closer to a good threat >extension than before reading the paper. There's something about his >explanatory style that just seems very obscure and opaque to me. > >To this august group, I say "What is threat extension and how do you >think it is best implemented?" > >As you can observe, I'm asking for feedback, going off and implementing, >and then reporting the result. Typically, the procedure will be a >solicitation with subject "XYZ" and the report will be "Result of XYZ". >I hope you enjoy! > >Is this computer chess programming by delphic process or what? :-) > >--Stuart the idea is to extend when you detect a strong threat. Usually you try to detect threats by the opponent side and extend by one ply when he has one, to be sure the threat it searched carefully. Detecting threats is not easy. The "null-move search extension" you asked about is one way. The danger is that you will detect non-existant threats and extend too much which will reduce your overall search depth accordingly...
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.