Author: Robert Hyatt
Date: 10:24:19 02/28/01
Go up one level in this thread
On February 28, 2001 at 12:36:52, Leen Ammeraal wrote: >On February 28, 2001 at 07:28:47, Dieter Buerssner wrote: > >>On February 28, 2001 at 05:56:36, Leen Ammeraal wrote: >> >>>I am not sure about when to avoid nullmoves. >>>I omit it: >>>a. when in check >>>b. when there are less than 5 pieces (including pawns) on the board >> >>I think, it is very risky to do nullmoves in pawn endgames, especially when the >>position is blocked. >> >>>c. when the last move was a nullmove >> >>Vincent has described a method, to allways do exactly two nullmoves in a row. I >>do something similar in Yace.´I search the position at even more reduced depth, >>than depth-R-1 without doing a nullmove. Only when this gives a fail high, I >>allow a nullmove cutoff. Essentially, this is the same than the double nullmove >>of Vincent. With this method, Zugzwang situations will be detected eventually. >>The drawback is, that you will get less nullmove cutoffs. As seen in various >>postings here, i.e. Fritz won't use this more secure approach, and therefor may >>fail to see certain short mates. But I have no doubt, that its author is fully >>aware of this, and probably knows, that the more risky approach yields in better >>results. >> >>>d. at the root node >>>Should I also omit it in some other cases, >>>for example, when any hashmove (even with a low draft) was found, >> >>There are situations, where the hash information indicates, that the null move >>will not yield in a cutoff. I.e. when you have an upperbound value in the hash, >>that was searched with enoug depth, and the score is smaller beta. >> >>>or when beta = alpha + 1? >> >>I don't think, that nullmove should be avoided here. > >Neither did I, but in Heinz's book, p. 25, I read the >following, which made me wonder about null windows: > >"...null-move pruning generates selective cutoffs at >nominal full-width nodes..." > >Could you (or anyone else, preferably Heinz himself?) >please tell me what he means by 'nominal full-width nodes'? >By the way, thanks for your response, and my compliments >for your wonderful program YACE. >Leen I think that simply means that at a node that will _normally_ be searched full-width to depth D, null-move will cause us to terminate the search here after searching only to depth D-R, which is a form of selective searching since we search less deeply along this branch than we normally would.
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.