Computer Chess Club Archives




Subject: Re: calculating the PV using MTD

Author: martin fierz

Date: 17:22:42 09/11/02

Go up one level in this thread

On September 11, 2002 at 04:08:52, Fabien Letouzey wrote:

salut fabien,

>Sorry for the awkward style, it is my first posting here.  I tried to be
>precise to avoid misunderstanding.

merci beaucoup! is all i can say. there was nothing awkward about that post!

my whole interest in getting a pv out of the search rather than from the
hashtable was that i wanted a guarantee that i got to the end of my PV, for
debugging purposes, like if i want to find out why my program makes a certain
bad move with a high score - what the hell was it thinking there?
retrieving from the HT is ok most of the time, but obviously for this purpose,
it is not so good because sometimes it will not give me the full PV because
something was overwritten in the HT. in my implementation, i will also miss the
entire qsearch because i don't store that in the HT. so your way of doing this
really helps :-)
and if i understand you correctly, then i can just get your "semiPV" for my last
MTD test, and that actually IS what my program was thinking when it made the
move, right?


This page took 0.04 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.