Author: Heiner Marxen
Date: 08:53:19 02/17/04
Go up one level in this thread
On February 17, 2004 at 10:17:01, Janosch Zwerensky wrote: > >>I like your approach of compressibility. It is possible that any such rule may >>not yield substantial savings; i.e., the rule may be almost as large as the >>tablebases themselves. > >The program that generated the tablebase is sort of a compressed form of the >tablebase, and it's way shorter, so significant compression is certainly >possible (though there is no proof that you can do it without working a lot when >"uncompressing" the data...). So we talk about a kind of Kolmogorov complexity, here, right? That would be the shortest algorithm computing the wanted function. Unfortunately, we also (normally) demand that the algorithm be efficient. There have been several attempts to find rule sets (a restricted kind of algorithm) coding complete EGTBs, but not with much success. OTOH, personally I (mildly) believe, there are some short and efficient algorithms out there, waiting to be discovered/invented. Well... until some such beast has been demonstrated, that is just speculation. Cheers, Heiner
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.