Author: Fabien Letouzey
Date: 07:33:12 03/04/04
Go up one level in this thread
Hello Tord, On March 04, 2004 at 09:12:38, Tord Romstad wrote: >On March 03, 2004 at 09:31:22, Matthias Gemuh wrote: >>xx depth without extensions >>yy includes extensions in main search, not quiescence >I don't think the standard specifies what the "seldepth" is supposed to mean, >and I don't think xx and yy have the above meaning in all UCI engines. My post was indeed about holes in the UCI document and other non-standard features, hence the title "in practice". I think UCI engine authors can agree about "common non-standard" features :) >In Gothmog, I use them in approximately the way you describe. xx is the >iteration number (which in my engine is not quite the same as "depth >without extensions", because my search is not even close to fixed-depth), >while yy is the longest line searched in the main search. Do you use a probability-based "depth decrement" if I may ask? One that estimates a probability distribution of moves at every node, and decreases the "remaining depth" by the log or something similar? This is anyway only a generalisation of SEX/fractional-ply algorithms. It also includes singular extensions as a special case. Sounds really slow, but I think worth a try. One could switch to normal ply-counting near the leaves anyway. You can answer by just yes or no :) About "yy", can you confirm that you never include quiescence plies in it? I don't really understand why stopping there, quiescence search is also a selective search ... By the way, I believe Gothmog has a *very* bright future; good luck :) Fabien.
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.