Author: Vincent Lejeune
Date: 16:04:52 07/21/99
Go up one level in this thread
On July 21, 1999 at 17:52:40, Christophe Theron wrote: >On July 21, 1999 at 01:33:46, Paulo Soares wrote: > >>On July 21, 1999 at 00:56:56, Christophe Theron wrote: >> >>(snip) >>> >>>Some programs don't see the mate in 2 (with both pawns on the 7th rank) because >>>after 1. Ra6! black is in zugzwang. >>> >>>Looks like nobody has read my previous post and I'm just losing my time trying >>>to explain it... >>> >>>Both Hiarcs and Fritz would fail on some zugzwang positions. Not on this one, >>>which is solved by Hiarcs, but there are positions where Hiarcs fails completely >>>because of its null move algorithm. >>> >>>...OK, just losing my time... :) >>> >>> >>> Christophe >> >>Christophe, this question was made by Eric Guttemberg, and was without >>reply. Could you answer, I have the same doubts of Eric. >> >>"With regard to the zugzwang problem, I am a little confused. I know that >>many basic endgames techniques rely repeatedly on the zugzwang theme. In >>many of them the weaker side would never lose if he could waive his move >>when it suited him(mating with K and R vs K is a whole different ballgame >>if the weaker side can not be forced to move) >>Yet Fritz does not seem to have any trouble with any of the basic endgame >>positions. What kinds of zugzwang positions does Fritz and other null-movers >>have trouble with?" >> >>Paulo Soares, from Brazil > >I don't know exactly which kind of positions Fritz will have trouble with. But >every programmer that uses null move knows about the "overlooking zugzwang" >disaster. > >So they try to avoid it by disabling the null move algorithm in positions where >a zugzwang is likely. Obviously, in endgame positions. > >They set some kind of material thresold, and if the current material on the >board is under this thresold, the null move selectivity is simply disabled. > >This can even be smarter. For example, it is unlikely that a side will be in >zugzwang if he has a queen. It is also unlikely that it will happen if a side >has more than 2 pieces (excluding pawns and king), though it is more risky to >assume this. Needless to say, you can find positions where one side, even with a >queen and another piece, is in zugzwang. But it is very very rare. > >Such a position has been posted recently, and guess what... Hiarcs fails to >solve it for example (I'm almost sure Fritz would fail too). [cut] I've found the original post I've verified the position but Fritz find the right move (h4 !) without problem, so I doubt that's a nullmover killer position, or Fritz have special algorithm for this kind of position ... original post: On July 12, 1999 at 12:48:08, Vincent Lejeune wrote: >When I watched a game a reach this position > >2k1q3/2p3n1/p1Np4/3P3p/P2P1p2/1Q5P/1KP5/8 w - - > >and I notice that Hiarcs 7.32 >don't find h4 after 4 min, depth (10/30). (the best it find is Qc4 +1.00) >but after h4 is played on the board, it instantly see an high score (+2.20) > > >(time on celeron 464 Mhz) Thanks Vincent! This is a nice zugzwang position. I think that after 1.h4 black is in zugzwang! It is very interesting since each side has still one queen and one knight, and the position comes from a real game. Try with Fritz and you should see that he has the same problem. Now we know that Hiarcs uses null move!!! Christophe
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.