Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Building the Principal Variation in MTD(f) searches

Author: Dave Gomboc

Date: 13:42:08 07/19/99

Go up one level in this thread


On July 19, 1999 at 16:08:38, Vincent Diepeveen wrote:

>On July 19, 1999 at 13:28:27, Dave Gomboc wrote:
>
>>On July 19, 1999 at 06:14:45, Vincent Diepeveen wrote:
>>
>>[snip]
>>
>>>I'm quite positive my 8 probe hashtable works excellent.
>>>
>>>I'm quite sure i investigated MTD very well.
>>>
>>>So what are we talking about?
>>>
>>>i give 2 reasons why MTD sucks bigtime for my program namely:
>>>     - it searches space PVS doesn't search; called overhead
>>>     - it's worst case behaviour is simply losing games for me,
>>>       as in *some* positions where it's already tough to search with PVS
>>>       it is only busy researching and researching to get a bound.
>>>
>>>The only thing you say then is that i'm not using hashtables?
>>>Comon drop dead.
>>
>>This isn't what I said.  I think it's clear you didn't read what I wrote very
>>carefully.  Oh well, it's not my problem.
>
>I posted a position with analyzes here.
>I'll post it again if you need, please let your proggie search 20 ply
>just using NULLMOVE R=3 and your beloved MTD, then we can talk.
>
>>[snip]
>>Dave

My program searches _far_ deeper than 20 ply with mtd(f), and without nullmove
at all.  Of course, it's not playing chess, but you can't have everything.

Dave



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.