Author: blass uri
Date: 19:54:53 03/28/99
Go up one level in this thread
On March 28, 1999 at 19:37:23, James T. Walker wrote: >On March 28, 1999 at 14:04:00, blass uri wrote: > >>The database cannot tell For some endgames like KRP vs KR if it is a win for the >>weaker side or a draw but can tell if it is a win for the stronger side or a >>draw. >> >>Usually wins for the weaker side are short wins so the right winning move for >>the weaker side is usually obvious. >> >>The usual problem is to find the right way for the stronger side to win and here >>the tablebases can help >> >>best Regards >>Uri > >Hello again Uri, >I'm puzzled by your remarks concerning the Fritz CD ROM. You seem to be calling >it a "database" and differentiating it from a "tabelbase". I'm not sure what >the difference is but I can't follow your reasoning that the "database" cannot >tell if it is a win for the weaker side but can tell if it is a win/draw for the >stronger side. It seems to me that all of this information is in the >"database". The Fritz engine is certainly not making this decision with "zero" >search time. If you look at the move list in Fritz when using the CD ROM it >shows (0) 50. Rd6 or something like that. Which means it's a draw when playing >50.Rd6. If you see (14) 50. Rd6 it means it's a win and conversion in 14 moves. > It does not matter which side you are playing it is a win in 14 for one side >and a loss in 14 for the other(Conversion in 14). This stuff is comming right >off the CD ROM not from Fritz or Junior. Usually you will see something like >this: >(13) 50. Ra8 >(12) 50. Rc7 >(12) 50. Rd7 >(12) 50. Kb7 >(11) 50. Rb7 >This is what it looks like for the losing side. The best move is at the top and >gives the maximum number of moves to conversion which therefore is the best move >if your intention is to last as long as possible. > >It's just the opposite for the winning side. The numbers in the () increase >when going down the list. For example: >(11) 50...Rf5 >(13) 50...Re5 >(14) 50...Kh6 >This means it 11 moves to conversion startin with 50.... Rf5 with best play. > >So why would you say the "database" cannot tell if it is a win for the weaker >side? Do you mean it cannot tell when there are errors on the CD ROM? In that >case it can't tell for either side since the data in the "database" is wrong. The point is that I did not find example when the "database" is wrong if it is win for the stronger side in KRP vs KR I found that the database may be wrong for KRP vs KR if underpromotion is the only way to win but I did not find an example for this case. The database may be wrong if it tell a draw for the stronger side and here is an example: 1r5K/3Pk3/3R4/8/8/8/8/8 w - - 0 1 tablebases say the losing d8B+ >Please explain to me what you mean. >Jim Walker "Fritz" probably uses tablebases based on Ken Thonpson's endgame databases for KRP vs KR These databases tell only the distance to conversion for the stronger side and do not discover another information. Stupid Fritz probably translate the no information in the database to draw. I think that the tablebases in the CD are based also on small wrong search. I posted an example when fritz say mate in 13 against itself instead of mate in 1 about underpromotion for a rook in 25/03 1:03 Fritz has no information about KQvs KRR and the only explanation is that the CD is based on small search of KxR after the underpromotion. Uri
This page took 0.01 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.