Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Please, say in few words what can reduce the "branching factor".

Author: Robert Hyatt

Date: 20:07:56 09/20/99

Go up one level in this thread


On September 20, 1999 at 22:45:44, leonid wrote:

>On September 20, 1999 at 22:25:41, KarinsDad wrote:
>
>>On September 20, 1999 at 18:02:23, leonid wrote:
>>
>>>On September 20, 1999 at 15:22:16, KarinsDad wrote:
>>>
>>>>On September 20, 1999 at 14:47:49, leonid wrote:
>>>>
>>>>[snip]
>>>>>
>>>>>My engin: 6 plys 0.22 sec.    8 plys 18 sec.     10 plys 14min. and 54 sec.
>>>>>
>>>>>Rebel 10:        2.6 sec             40 sec              12min. and 8 sec.
>>>>>
>>>>>If my "branching factor" between 6 plys and 8 is 82, for Rebel it is only
>>>>>15.3. This is how Rebel take over on the 10 ply. Number of positions per
>>>>>second are almost indentical but my are higher. Average number of nodes
>>>>>per ply is 36.
>>>>>
>>>>
>>>>Well, it appears that the branching factor for your program (2 ply) from ply 8
>>>>to 10 is 49.7 whereas Rebel's is 18.2. So, your branching factor from ply 8 to
>>>>ply 10 is dropping relative to your branching factor from ply 6 to ply 8. Rebel
>>>>on the other hand has a slightly increasing branching factor from ply 8 to ply
>>>>10 relative to it's ply 6 to ply 8.
>>>>
>>>>So, whatever the problem is, it does appear to be based on the position (i.e.
>>>>some set of positions give you a much greater branching factor relative to Rebel
>>>>than another set 2 ply higher). Since pure Alpha Beta should have a ballpark of
>>>>a 2 ply branching factor of 36 for the position you are examining, I would guess
>>>>that you have a mistake somewhere in that portion of the algorithm. Maybe
>>>>assigning Beta to a bound when you should assign Alpha or something along those
>>>>lines.
>>>>
>>>>KarinsDad :)
>>>
>>>Will be nice if you could explain me the last sentence of your response.
>>>And also what is the "Ballpark"?
>>>Thanks,
>>>Leonid.
>>
>>Basically, your branching factors (82 and 50) are much higher than normal Alpha
>>Beta (which for 2 ply should result in a value of about 36). Therefore, it seems
>>that either you are not using Alpha Beta, or you have a mistake in your Alpha
>>Beta such as assigning a Beta value to a bound on a cutoff instead of assigning
>>an Alpha value to it or some such.
>>
>>"Ballpark" is a reference from American baseball where if a ball is hit inside
>>the ballpark, it is still in play. If it is hit outside the ballpark, it is a
>>home run (and out of play). Somehow, this became an expression that means that
>>something is within a reasonable value (i.e. in the ballpark). It is often used
>>with guesses as per "My guess is x which I know is off, but it is in the
>>ballpark.". So in this case, a value of 36 is a reasonable value (in the
>>ballpark) for a ply 2 branching factor for the position that you examined.
>>
>>KarinsDad :)
>
>Thanks for your response!
>Your saying that around 6 nodes must be seen inside of 36 in the ply
>correspond 100% of what I found in the best games. My logic all the time did
>worst that this, around 8 from 36. Will try to figure out where those
>bounds are wrong.
>
>With my respect,
>Leonid.


It also might not be a bug, but might be caused by poor move ordering,
as that can blow up the search similar to what you are seeing..



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.