Author: Vincent Diepeveen
Date: 02:32:15 07/14/02
Go up one level in this thread
On July 13, 2002 at 21:31:14, Omid David wrote: there are so many things that work better in just diep than the 'official' implementation as is in publications. Now i didn't hit the subject 'forward pruning' yet. Most of the commercials have put insane much time in forward pruning. anyone wanting to post something about it: Ed? Christophe? Stefan? Johan? Frans? To just name a few guys who 100% sure read this posting without majority ever commenting on this, and for sure having a form of forward pruning which works for them :) >On July 13, 2002 at 11:57:00, Vincent Diepeveen wrote: > >>On July 13, 2002 at 11:17:12, Omid David wrote: >> >>For diep the adaptive nullmove, which i btw used in diep >>around 1995-1998 already, it was pretty interesting thought >>back then, but in the end R=3 always worked better. > >Although Heinz was the first to publish adaptive null-move pruning formally, >many people used it even before that. Donninger told me that in 1993 he used the >very same idea in NIMZO. In his article Heinz points out that Donninger hinted >at using R=2 in upper parts of the tree and R=3 in lower parts; but in fact >Heinz misunderstood Donninger's 1993 article since Donninger suggested exactly >what Heinz did. > > >> >>I have just to get a good compare run at Jan Louwman's auto232 >>players with diep with 2 versions, the only difference being >>R=3 versus adaptive nullmove (last 4 plies R=2), and that >>DIEP then nearly gets a ply less deeply, it definitely was >>reflected in score. A big problem is to get above that >>10 simply for DIEP. >> >>>On July 13, 2002 at 10:50:52, Robert Hyatt wrote: >>> >>>>On July 13, 2002 at 02:07:17, Vincent Diepeveen wrote: >>>> >>>>>I still do not understand which positions you talk about which R=2 >>>>>is finding and R=3 isn't. >>>> >>>>Note that he used fixed-depth. This is therefore not surprising since >>>>some lines will be searched one ply shallower.. >>> >>> >>>Of course. I merely did something like Heinz in "adaptive null-move pruning" (he >>>searched to fixed depths of 8, 10 and 12). But to get a clearer distinction, I >>>turned off checks in qsearch (that means in practice R=3 would perform better >>>than it did in those tests)
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.