Author: Tord Romstad
Date: 07:55:21 07/15/03
Like many others, I use static pruning in addition to null-move pruning in the last few plies of my search. Currently, the last 3 plies of my search are selective. I am not sure that this simple approach is the best. At slow time controls, it seems that a higher number of selective plies are better (although I still do not have enough data to make any certain conclusions about this). However, high selectivity causes too many tactical oversights when there is little time available. I have been tempted to try a Rebel-like selective search, and let the number of selective plies vary with the iteration number. But this seems to cause problems with the transposition table. To be able to check whether a transposition table entry has a sufficiently high search depth to be used for cutoff decisions, I will have to store both the brute force depth and the selective depth in the hash table entry. But what is the best way to compare two different depth/selective depth pairs? Is 1 ply full-width + 4 plies selective better than 2 plies full-width + 2 plies selective? I would be interested to learn how those among you who use variable selectivity based on iteration solves this problem. Tord
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.