Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Definition of branching factor?

Author: Tony Werten

Date: 04:46:39 02/04/01

Go up one level in this thread


On February 03, 2001 at 15:24:51, Larry Griffiths wrote:

>On February 03, 2001 at 13:14:12, Tony Werten wrote:
>
>>On February 03, 2001 at 12:00:55, Larry Griffiths wrote:
>>
>>>On February 02, 2001 at 19:35:47, Tony Werten wrote:
>>>
>>>>
>>>>Quite normal when you evaluate with material only or eval=0
>>>>
>>>>>They tend to be much
>>>>>higher on even plys.
>>>>
>>>>Comes with (plain) alfa-beta. You try to get around it with nullmove, extensions
>>>>and an elaborated evaluation function.
>>>>
>>>>Some work left for you,
>>>>
>>>>Tony
>>>>
>>>Thanks Tony,
>>>
>>>I have not been able to get null move to work, but I will keep on trying.
>>
>>I've found that the easiest way is to actually have a make_nullmove function
>>where you store the move, chance color_to_play, get rid off the enpassant
>>square, store hashkey etc. just as you would normally do with a make_move.
>
>It probabably would be better to say that I do not understand NullMove.  I will
>get out Ernst's book and look at it again.  I also might search the web to look
>for articles on NullMove.  I sort of do hashing, but that is another thing I am
>not too whippy at.
>
>>
>>>I have done some extensions before, but dropped them when I converted from
>>>alpha/beta to negamax.
>>
>>Don't you mean from negamax to alfa-beta ?
>
>Its negamax WITH cutoffs, whatever it is called :)

alfa-beta :)

>
>>Easy extensions are "in check" and "passed pawn to 7th rank"
>>
>>>I have no idea what elaborated evaluation means.
>>
>>A "big" evaluation function.
>
>OK, Thanks Tony.
>
>If I can understand some of this stuff within the next 10 years, then I might be
>able to reach my lifetime goal of a program rated over 2000.  It seems as I am
>getting older, that the elevator does not go to the top floor anymore.
>

The nice part of chessprogramming is that you can take one step at a time and
see the improvement for quite a long while.
I would suggest to get the hashtables correct and then go for nullmove. They are
very big winners.

Tony

>Larry :)
>
>>
>>Tony
>>
>>>
>>>Larry.



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.