Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: move ordering and node count

Author: Ralf Elvsén

Date: 06:47:58 03/30/04

Go up one level in this thread


On March 29, 2004 at 15:50:10, Dann Corbit wrote:

>I have not done the math, but I am guessing no matter what king of move ordering
>you have (purely randome or the pv move every time) you will get something like
>this:
>
>nodes = some_constant * sqrt(mini_max_nodes)
>
>If you have random move ordering, then the constant will be very large.
>If you have perfect move ordering, then the constant will be very small.
>

Your formula is

some_constant*mini_max_nodes^s

with s = 1/2 .

This is only true for perfect move ordering. I did some numerical investigation
of this for random move ordering. I got

some_other_constant*mini_max_nodes^s

where s = 0.71 +/- 0.01 (this uncertainty is an overestimation, can't remember
right now). Who knows maybe s = 1/sqrt(2)   :)  . According to Dave Gomboc,
Monty Newborn derived a result s = 2/3 but the result was flawed. I haven't seen
that paper.

Ralf





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.