Author: Sune Larsson
Date: 05:17:45 12/29/05
Go up one level in this thread
On December 29, 2005 at 07:42:33, Uri Blass wrote: >On December 29, 2005 at 07:08:31, Sune Larsson wrote: > >>On December 29, 2005 at 06:08:50, Joseph Ciarrochi wrote: >> >>>Sune Larsen caught the first position as being too simple and we have removed it >>>from future testing (8/4k3/1p2b1pp/p1p1p3/P1P1P3/1P3P1P/5KB1/8 b - - 0 37) >>> >>> >>>So it sounds liike we need some more complex endgames perhaps? I did expect >>>rybka to be worse than it did here. >>> >>>Sune, any responses to this? I'd be happy to test more compex endings, if you >>>generate them? >>> >>>best >>>Joseph >> >> >> I suggest we keep these 19 positions and just continue with the games. >> I find them extremly interesting. The point is of course not to specifically >> nail down Rybka's lack of endgame knowledge... Naturally we could then, by and >> by, expand the testsuite with new and more complex endings. That would be >> most interesting of course. They must be carefully chosen though. I also would >> like to know what theme a certain position will test. Maybe Uri has some >> contributions to make...;-) Only negative criticism is of very little value >> though... >> >> /S > >I can say that I do not see unbalanced positions when one side has knight for 3 >pawns. or queen for rook and pawns. > >I remember that fruit had problems in an endgame when it has queen and pawns >against rook and pawns but Fritz had pawn in the 7th and rybka could do nothing >better than perpetual check. > >When I say that the position are not balanced it means that I expect even weak >programs to score at least half points in some cases. The last 10 positions test the engines capability of *winning* won endgames with specific themes. If Fritz9 fails to win pos.10 with white vs Movei or Queen - it "only" shows that Fritz9 is lacking knowledge about "good knight vs bad bishop" - not that Movei or Queen played strongly as black. If you have balanced positions with minor piece vs 2 or three pawns - or Q+pawn vs R+B+2pawns etc - you are most welcome to share them with us. /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.