Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Anmon question ! Is anmon using alpha-beta ?

Author: Tom King

Date: 12:32:02 08/17/99

Go up one level in this thread


On August 17, 1999 at 02:51:41, Frank Schneider wrote:

>On August 16, 1999 at 19:45:32, Bruce Moreland wrote:
>
>>
>>On August 16, 1999 at 16:32:45, Thorsten Czub wrote:
>>
>>>Question to the programmer (s) [of anmon]:
>>>The evaluation in anmon is very strange. it does not jump.
>>>it scrolls or stretches not in intervall-jumps but nearly fluently
>>>up and down.
>>>
>>>Uli Tuerke thought that this could be because anmon uses no alpha-beta
>>>but an algorithm used in cilk-chess.
>I think the algorithm is called MTD(f). It is a kind of alpha-beta, where
>always alpha == beta-1. To establish the real score of a move you do a
>kind of binary search for the score, using null-window-searches.
>
>But I don't know if AnMon uses MTD(f).
>
>>>I have no idea which algorithm this is, but would like to know about
>>>the stuff.
>>>Can we maybe discuss about it ? Or when this is a secret information
>>>maybe talk about this algorithm in general no matter if it used or not.
>>
>>One of the programs mine played against in Paderborn did this too, I think it
>>was Gromit but perhaps it was Francesca or maybe I am completely wrong.
>It wasn't Gromit and probably not Francesca.
>

No, it wasn't Francesca. I use standard aspiration alpha-beta. Here's question,
would PVS buy me a speedup, and if so how much? My guess: a couple of %
Rgds,

Tom

>Frank
>
>>
>>bruce



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.