Author: Miguel A. Ballicora
Date: 09:02:06 02/27/06
Go up one level in this thread
On February 27, 2006 at 11:15:50, Vasik Rajlich wrote: >On February 27, 2006 at 09:41:37, Tord Romstad wrote: > >>On February 27, 2006 at 06:48:18, Tord Romstad wrote: >> >>>I'll try to write an explanation on my web page, and refer to that one >>>in the future. I hope to have it ready today or tomorrow. Stay tuned! >> >>My first draft is ready: >> >>http://www.glaurungchess.com/lmr.html >> >>Comments, corrections and suggested additions or improvements >>are welcome. >> >>The only two engines I mention in my "Sample Code" paragraph are >>Fruit and Glaurung. This is because of ignorance, and not because >>of disrespect to other authors. If you are the author of an open source >>engine using history pruning (or whatever you prefer to call it) and >>wants it to be mentioned on my page, please let me know. >> >>Tord > >Tord, > >nice summary. > >A few minor things: First of all, congratulations for CCT8! >1) For me, the terms "pruning" and "reducing" are interchangeable. I'd guess >most CC programmers agree with this - for example, we have null move pruning. This is semantics, but I do not agree. Reducing is decreasing (depth--), pruning is removing (depth=0). You may argue that if I reduce a branch by 1, you are pruning the leaf. Not necessarily, because in the subtree you still have chances to extend and get back that leaf into the tree. When your prune, is gone. Maybe somet definitions chanced in the last 3 years, but before, it was understood that pruning was "I do not search this subtree anymore" and reductions was "I search it with reduced depth". Regards, Miguel >2) Prob-cut is not a fail-high-node selectivity algorithm. You might add SE >there instead. Yes, I know that SE extends rather than reduces - but it's the >same idea as multi-cut. The idea of categorizing selectivity into fail-high-node >and fail-low-node is good. > >Vas
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.