Author: Robert Hyatt
Date: 13:25:02 04/15/02
Go up one level in this thread
On April 14, 2002 at 14:25:25, Pat King wrote: >On April 13, 2002 at 00:36:07, Will Singleton wrote: > >[99% snipped because I'm tired of seeing "Pat King said" seven replies >after my replies have been snipped out] >>btw, thanks for addressing this. I think that people toss around chess terms >>that are not well-defined, with resulting confusion. I still think that the >>collision term is best defined by Beal, as I have indicated. Two hashkeys >>competing for the same table slot, or index, colliding. As you define it, is >>really a bug, and should not occur; hence the term is not meaningful. >> >>Will > >Why argue about the meaning of a word, as long as that word is defined within a >given thread? Is not a rose as sweet, etc, etc? > >I don't think any of us disagree about what constitutes a bug in this thread, >just what to call it! > >Pat There are two words that are misused all the time. "collision" and "branching-factor". Collision technically means just what I said. Branching factor is _not_ something that a program can reduce. It is very precisely defined as the "average" number of successors from any node in the game tree. This can't be changed unless a program does true forward- pruning and at every node, throws out some moves without searching them in any way... IE the classic approach called "selective search". Null-move doesn't do this. And the better term is "effective branching factor". We really don't have a good term to mean "two different hash signatures mapping to the same hash table addresss" unfortunately... Other terms are often "bent" a bit like "ply depth" (ie what is a ply if your program doesn't count one level in the tree as one ply like Junior?) At times, "bending" doesn't hurt. At other times, it causes confusion.
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.