Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Branching factor

Author: Robert Hyatt

Date: 12:02:45 11/21/00

Go up one level in this thread


On November 21, 2000 at 13:04:08, J. Wesley Cleveland wrote:

>On November 21, 2000 at 09:31:33, Sven Reichard wrote:
>
>>On November 20, 2000 at 23:24:49, Daniel Kang wrote:
>>
>>>Also, alpha-beta pruning allows one to arrive at a correct minimax solution
>>>without having to search the entire tree, so your math is a bit off there as
>>>well.
>>>
>>>Dan.
>>
>>It still needs to search (b/2)^n nodes,
>
>I think you mean b^(n/2)
>
>> if b is the branching factor, and n is
>>the depth in plies, *even assuming* optimal move ordering. So the math wasn't
>>that much off after all.
>>Sven.


The right formula is this:

n=w^floor(d/2)+w^ceil(d/2)

floor -> round down, ceil -> round up.

For d even, it simplifies to

n=2*w^(d/2)

but that assumes that W and D are constants.  For today's programs, they
most definitely are not constant.



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.