Author: Tony Werten
Date: 02:17:28 02/04/05
Go up one level in this thread
On February 04, 2005 at 04:15:26, Uri Blass wrote: >On February 04, 2005 at 04:05:52, Tony Werten wrote: > >>You'll find lots more with white bishop on A8 and black pawn on B7, sntm in >>check etc. >> >>What's your point ? >> >>Tony > >You are right that I will find better examples. > >These examples are better because they are illegal also in FRC or alternatively >illegal with 3 pieces > >B2K4/1p6/8/8/3k4/8/8/8 w - - 0 1 > >B2K4/1k6/8/8/8/8/8/8 b - - 0 1 > >The point is that it may be possible to reduce the size of the nalimov >tablebases by having only legal positions(unfortunately I guess that only a >small reduction is possible). Ah, I see. No, I think you could get a big reduction, the problem is just how to identify these positions (fast). In the TBS files, there is a count of "broken positions" wich are the illegal positions. Quite a big percentage for some files as you can see. There are also a lot of positions that can only be reached by first giving away a won or draw position. You want to get rid of these as well. Even better, recognize the won positions and you don't need the tablebases at all :) Just kidding, you don't seem to get the idea of tablebases. They are just a stupid way of enumerating all possible positions and their value. Eugene did make his generator a bit smarter by taking out some of the realy stupid/illegal stuff, but it's still a stupid enumeration. What you're talking about is making them smarter, but that's more about recognizers rather than tablebases. Tony > >Uri > >Uri
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.