Author: Sune Fischer
Date: 16:01:36 04/20/03
Go up one level in this thread
On April 20, 2003 at 18:42:05, Dieter Buerssner wrote: >On April 20, 2003 at 18:15:03, Sune Fischer wrote: > >>The question is if Crafty can find it _without_ TBs.... > >It would be very easy to check - just try it. I was explaining the point to Uri; Just because Crafty found it didn't prove Crafty had specific knowledge, any engine using TBs should find it. So you can say the question was rhetorical, the second time round :) >>If you program every bit of 4 vs 1 knowledge you are practicly building your own >>table bases. So what's the difference? > >It can be much faster than TBs, and need much less resources (for example space >on HD). Because (in this specific case at least) the knowledge is very fast, it >can be used deep in the search, where TBs cannot be used without big >disadvantages. It may only need very little code. Practically: 4-1 is always >won. True, but it may also slow down you code to check for this all the time, and you have to spend time implementing it. In short, it doesn't seem real valuable if you already have TBs. ...unless it can be made more general to extend further than to just the TB domain, which is basicly covered. >You "only" have to detect the exceptions. Perhaps KBPPK is the practically >most important exception. (I would think double h-pawn and wrong B happens more >often than triple h-pawn, the only other practically relevant exception I can >think of right now). Of course, there are many more exception - mainly beeing >stalemated. One ply search more is enough, to fix it. Or exceptions, where one >man hangs - search and "easier" knowledge will help there, too. With the KBPPK >-search cannot help easily. With many plies of search you will see - nice: I >have 2 Ps and a B. An important exception, perhaps. But even here it should be made as general as possible, with 1,2,..5 pawns on the rook file also being detected. -S. >Regards, >Dieter
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.