Author: Heiner Marxen
Date: 08:19:10 03/06/02
Go up one level in this thread
On March 06, 2002 at 09:19:12, Bernhard Bauer wrote: >On March 06, 2002 at 08:58:40, David Dory wrote: > >>On March 06, 2002 at 05:29:30, Bernhard Bauer wrote: >> >>>On March 06, 2002 at 05:17:55, Gian-Carlo Pascutto wrote: >>> >>>>On March 06, 2002 at 05:14:57, Bernhard Bauer wrote: >>>> >>>>>Do you really think the Washington Post gave the position in EPD? >>>> >>>>I don't know what the Washington Post gave, but the position >>>>as posted here was not correct. >>>> >>>>If they only have the diagram, it was ambiguous. >>>> >>>>-- >>>>GCP >>> >>>You don't know what the Washington Post gave? :-))) >>>The poster who started this thread did't convert the position to correct epd as >>>he was not aware of castling. >>>If they only give the diagram you have to assume every posibility. So a diagram >>>is *not* ambiguous. >>>In any case, table bases will not help, whether the epd string is right or >>>wrong. >>>Admit it, even it may be dificult, you were plain wrong. >>>Kind regards >>>Bernhard >> >>I understand what you mean about the diagram, and taking no assumptions, even >>though the posted FEN did state EXPLICITLY that no castling rights existed. >>We should all have taken a step back and looked at that possibility as you >>suggest. >> >>I don't have a single clue what you mean when you say "table bases will not >>help, whether the epd string is right or wrong".* >> >>Not only will the table bases HELP, but they will help the program play the >>position AS GOOD AS GOD, (rumor has it that's pretty good, even for playing >>chess with atheists or agnostics)! :-) >> >>When it comes to "difficult admitting", I'd hold up a mirror and not look to >>Terry, et. al.. >> >>Dave >> >> >> >> >> >> >>*(actually, it's an FEN string, but no matter). > >Dear chess friend, >When I said "table bases will not help, whether the epd string is right or >wrong".* >I meant it only for the given position. Since current table bases (Nalimov table >bases) do not know about castling rights they may not give the shortest move >sequence to mate. This holds wether you have an EPD or FEN string. So it dosn't >matter. >If castling is involved a program using table bases may play considarably worse >than GOD. Huh? I don't follow. A program using EGTBs does not do that unconditionally. It uses them only if the EGTB data fits the current position. Now, not using any EGTB while castling rights are still present is quite an easy test, and obviously the right thing to do. When further down the search tree castling rights are removed (e.g. by moving the king) then suddenly EGTBs fit the position and should be used (for that advanced position). In that way, some search is followed by EGTB accesses, which _does_ help to shrink the search tree dramatically. And if the search is exhaustive, the result can quite well be as decisive as a direct EGTB access. I hope this is clearer now. >I think that is common knowlege. If you don't believe me, ask Eugene or any >programmer. Well, I asked myself :-)) >Kind regards >Bernhard 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.