Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: What means lazy/plain alpha bounding?

Author: Tony Werten

Date: 19:45:11 02/06/01

Go up one level in this thread


On February 06, 2001 at 22:09:25, Ricardo Gibert wrote:

>On February 06, 2001 at 21:20:32, Robert Hyatt wrote:
>
>>On February 06, 2001 at 14:04:03, Severi Salminen wrote:
>>
>>>Hi!
>>>
>>>What means lazy and plain alpha bounding as used in DarkThought? These names are
>>>used in Heinz's book but not described. Thanks for info.
>>>
>>>Severi
>>
>>
>>This dates back to the late 1970's.  The idea is that when you use a PVS-like
>>search, and fail high at the root, you do not re-search the move immediately
>>but just accept it as best.  If a second move fails high, you have to resolve
>>one of them to a true score then search the second with a null-window search
>>centered on the new best score.
>>
>>On a single fail high, you don't do the research at all, you just proceed to
>>the next iteration.  Which is a bit of a problem in that you have no PV to
>>help order the next iteration.
>
>A nice clear explanation for what turns out to be a very interesting idea. I'll
>add it to my notes. Thanks!

Be carefull. PVS combined with hashtables and nullmove can give you false
failhighs. They can destroy your search.

Tony




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.