Author: Andrew Williams
Date: 07:27:28 06/27/02
Go up one level in this thread
On June 27, 2002 at 08:28:47, José Carlos wrote: >On June 27, 2002 at 08:26:06, Andrew Williams wrote: > >>On June 26, 2002 at 16:21:11, Dieter Buerssner wrote: >> >>>On June 26, 2002 at 13:53:11, Andrew Williams wrote: >>> >>>>This is a null-move problem, caused by a stupid bug in PM. My null-move >>>>verification search was not working in this position. Now that I've fixed it, >>>>this particular error doesn't occur. I'm still interested in what others see >>>>here. >>> >>>For Yace, I cannot see much influence of null move here. I tried without null >>>moves at all, with default null move usage (R=2, including a verification >>>search) and with R=2 without any verification search. Each time I see a big fail >>>low at depth 9 for Kg3. Of course without null move, many more nodes are used. >>>Without verification search, somewhat less nodes are used than in the analysis >>>shown here. >>> >> >>This is interesting. For PM, if I make my null-moving less agressive (ie don't >>null move as much with few pieces, or don't use R=3 if there are few pieces), it >>sees the problem pretty much immediately. But with PM's default null move >>settings, it takes a *very* long time to see the problem. No matter what else I >>do, PM doesn't react until ply 12. >> >>Hmmmm.. perhaps 9 days before WCCC isn't the best time to radically re-think my >>null move strategy!! >> >>Thanks for your response. >> >>Andrew > > Don't know if this can help: Averno needs 11 plies to fail low with default >null-move (R=3/2) and 10 without null move. I didn't test other configurations. > > José C. All information about this issue is helpful, Jose! Cheers. Andrew
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.