Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: History based pruning question

Author: Ed Schröder

Date: 02:06:04 08/27/05

Go up one level in this thread


On August 26, 2005 at 23:11:18, Robert Hyatt wrote:

>On August 26, 2005 at 21:03:07, Uri Blass wrote:
>
>>On August 26, 2005 at 17:50:32, Robert Hyatt wrote:
>>
>>>On August 26, 2005 at 17:21:57, Uri Blass wrote:
>>>
>>>>On August 26, 2005 at 17:08:52, Robert Hyatt wrote:
>>>>
>>>>>On August 26, 2005 at 16:58:21, Uri Blass wrote:
>>>>>
>>>>>>On August 26, 2005 at 14:54:32, Robert Hyatt wrote:
>>>>>>
>>>>>>>On August 26, 2005 at 14:21:34, Alvaro Jose Povoa Cardoso wrote:
>>>>>>>
>>>>>>>>Hi,
>>>>>>>>some of you compare the number of times a move failed high to the number o times
>>>>>>>>the same move failed low in order to decide if a move can be reduced one ply.
>>>>>>>>I've tested this and also tested using the actual values of the history table
>>>>>>>>(using of course another history table for fail lows).
>>>>>>>>I couldn't reach a conclusion though.
>>>>>>>>What is your experience on this?
>>>>>>>>
>>>>>>>>best regards,
>>>>>>>>Alvaro
>>>>>>>
>>>>>>>
>>>>>>>My first thought is that the number of "fail lows" is irrelevant.  What you
>>>>>>>really want to avoid is a reduction on a move that might fail high.  Any move
>>>>>>>will fail low in some situations, but you want to handle the "typical" case
>>>>>>>correctly and not reduce if there is a reasonable chance the reduction will hide
>>>>>>>something.
>>>>>>
>>>>>>I think that it is relevant.
>>>>>>
>>>>>>If a move was never tried and never had an option to fail low then you do not
>>>>>>want to reduce it.
>>>>>>
>>>>>
>>>>>Chances of that happening is about zero.  There are only a finite (and small)
>>>>>number of different possible moves in the game.  "All the right moves" (PhD
>>>>>thesis by Ebeling) illustrated this.
>>>>
>>>>I agree that there is a finite number of moves but
>>>>I am sure that there are moves that are never tried during the first seconds of
>>>>a search simply because you need many moves to make them legal.
>>>>
>>>>It does not mean that in the first time that they are legal they should be
>>>>pruned.
>>>>
>>>>For example
>>>>[D]r1b3k1/1pp5/8/8/8/8/6PP/4KB1R w - - 0 1
>>>>
>>>>I doubt if you will find a move like Kf6-g7 at small depths but it does not mean
>>>>that the move should be pruned and this move can be logical in supporting passed
>>>>pawns.
>>>>
>>>>Uri
>>>
>>>
>>>by the time I get to _that_ position I could guarantee you every move has been
>>>tried millions of times. :)
>>
>>Even if you use statistics about all the game and not only about specific search
>>I do not think that every move has been tried millions of times because white
>>king from f6 to g7 is not something that you try in the opening when the white
>>king is at e1 or g1 and stupid lines when the king go forward to the direction
>>of g7 usually pruned by null move pruning.
>>
>>Uri
>
>
>Most don't do history like that.  Usually it is just a 12 bit index <from><to>.
>So it doesn't differentiate between a king move from f6 go g7, and a
>bishop/queen move from f6 to g7...

History Reductions is a working idea Bob, you should definitely try it.

Ed



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.