Author: Steffen Jakob
Date: 00:00:17 06/19/00
Go up one level in this thread
On June 19, 2000 at 02:17:48, Oliver Roese wrote: >On June 18, 2000 at 21:43:30, Robert Hyatt wrote: > >>On June 18, 2000 at 16:45:26, J. Wesley Cleveland wrote: >> >>>I had an idea about this. If you kept track of how many extensions you did in >>>the search, if you had an unusually high number of extensions the iteration >>>before, you could search the next iteration to a lesser depth, e.g. >>>next_depth = last_depth + k*(nodes/(nodes+extensions)) >>>where k is equal to or somewhat greater than 1. >> >> >>Now your task is to test that. Sounds at least worth some testing. >> >>:) > >Could you shortly give an idea, why this approach could be beneficial?! >Lots of extensions indicate lots of stuff happening. >Searching is designed to overcome tactical barriers. Why then >_reduce_ the search depth? > Hi Oliver! It might be a good idea in the last depth of the iterative deepening when you can estimate that you prob. wont be able to finish this depth if you increment a full ply. Before that I dont think that it is useful. In the contrary: I made experiments with incrementing > 1 ply in the first plies of the iterative deepening and in some positions this was much faster than the increment of one ply. BTW: on my todo list is to give Hossa extra time if the number of extensions > nodes/N which is related to your thoughts above. Best wishes, Steffen.
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.