Author: Gerd Isenberg
Date: 16:07:02 01/29/02
Go up one level in this thread
On January 29, 2002 at 18:13:47, Miguel A. Ballicora wrote: >On January 29, 2002 at 16:54:20, Dann Corbit wrote: > >>On January 29, 2002 at 16:43:28, Roy Eassa wrote: >> >>>On January 29, 2002 at 16:36:23, Thomas Mayer wrote: >>> >>>>Hi Roy, >>>> >>>>>>Wow, Fritz 7b is blind to this one. It does not see in advance that 37...Nxe5 >>>>>>is good for Black (it thinks Black is down more than 3 pawns). Must be >>>>>>null-move / zugzwang, right? >>>>> >>>>> >>>>>To be more specific, in the following position Black can play ...Re6 and be in >>>>>very good shape (certainly not behind). However, Fritz 7b evaluates this >>>>>position as being better for White by about 5 pawns: >>>>> >>>>>[d] 8/8/1p1r1k2/p1pRN1p1/P3K1P1/1P6/8/8 b - - 0 2 >>>> >>>>strange... Quark fully disagrees and thinks that black is little bit better... >>> >>> >>>Quark is right. Fritz is wrong. >>> >>>I guess this is a very good example of a blindness caused by using null move. >>> >>>My question is, do all zugzwang positions confound null move, or only some? >> >>Since Thomas uses NULL MOVE in Quark, it makes you wonder what could be causing >>the blind spot for Fritz? Using it past the time when it is appropriate >>perhaps? > >It depends on the implementation, I guess. Gaviota uses nullmove but it >has no problem to see that Re6 is not winning for white and soon it sees >that it is good for black by a small margin. In other words, it sees that the >knight falls. Gaviota does not prune after nullmove, it reduces depth. >So, zugswang problems are delayed only 2 plies. > I think that is, what Rainer Feldmann introduced as "Fail-High Reductions" in Advances in Computer Chess 8. I also tried it in the Dos-IsiChess for a while. Today a use a verification search, to confirm a NullMove Fail-High or to detect Zugzwang. I think similar to Vincent's double NullMove. Gerd
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.