Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Branching factor, etc

Author: Andreas Herrmann

Date: 20:00:30 08/30/02

Go up one level in this thread


On August 30, 2002 at 21:03:25, Uri Blass wrote:

>On August 30, 2002 at 17:24:34, Andreas Herrmann wrote:
>
>>On August 30, 2002 at 14:33:08, Omid David wrote:
>>
>>>On August 30, 2002 at 14:27:53, Andreas Herrmann wrote:
>>>
>>>>On August 30, 2002 at 09:56:21, Omid David wrote:
>>>>
>>>>>On August 29, 2002 at 23:03:43, Dann Corbit wrote:
>>>>>
>>>>>>On August 29, 2002 at 22:50:53, Brian Richardson wrote:
>>>>>>[snip]
>>>>>>>Is move sorting turned off in Yace, GLC and Gnu for the depth = 6 searchs?
>>>>>>
>>>>>>Pretty irrelevant since all of them show a branching factor between 2 and 3 for
>>>>>>the opening position.
>>>>>>
>>>>>>The program described by the OP had a branching factor of 6-8, IIRC.
>>>>>>
>>>>>>IOW, something is clearly amiss.
>>>>>>
>>>>>>I wonder how often the OP finds the requested position in the hash table.
>>>>>>Usually, hash table alone would be enough to prevent a branch factor that
>>>>>>terrible.
>>>>>
>>>>>How do you calculate branching factor here?
>>>>
>>>>A branching factor of 3 means that each node has in the average 3 child nodes
>>>>(This is the description i have found on an internet page). So the formula must
>>>>be:
>>>>
>>>>bf = ( Nodes [ply n] - nodes [ply n-1] ) / nodes [ply n-1]
>>>>
>>>>Excample:
>>>>Whole nodes until ply 5 = 4000 and whole nodes until ply 6 = 20000.
>>>>Then you got a branching factor of
>>>>bf [ply 6] = ( 20000 - 4000 ) / 4000 = 4.0
>>>>
>>>>have a nice day
>>>>Andreas
>>>
>>>I know this :-)
>>>
>>>But there is the odd/even issue, so the b-factor can change drastically while
>>>moving from an odd ply to an even ply, and vice versa.
>>
>>I think the best is to calculate an average branching factor from all plys.
>>
>>bf[avg] = ( bf[2] + bf[3] + bf[4] ... + bf[n] ) / (n - 1)
>>
>>Andreas
>
>It is better to use
>( bf[2] * bf[3] * bf[4] ... * bf[n] )^(1/(n-1))
>
>Uri

Hi Uri,

have you found an official description of the calculation of the branching
factor in search trees? Because i don't know how to calculate the branching
factor correct. My formulas i have created only based on the description
"A branching factor of 3 means that each node has in the average 3 child nodes"
that i have found on an internet page.
Can you tell me where your formula is from (perhaps an internet link), because i
want to implement a calculation in Holmes.
I got every time smaller values with your formula:
Excample:
bf[2] = 2;  bf[3] = 4;  bf[4] = 3;  bf[5] = 7
bf[avg a] = 4.0
bf[avg b] = 3.6

I got only the same average bf if there is in all plys the same bf because then
is
  bf[avg] = bf[n]^(n-1)^(1/(n-1)) = bf[n] = maximum

Andreas








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.