Computer Chess Club Archives




Subject: Re: MTD(f) and storing the PV

Author: Andrew Williams

Date: 12:50:18 07/29/03

Go up one level in this thread

On July 29, 2003 at 14:58:48, Tord Romstad wrote:

>On July 29, 2003 at 05:42:11, Andrew Williams wrote:
>>On July 29, 2003 at 05:38:20, Tord Romstad wrote:
>>>I am currently experimenting with MTD(f).  There are still a few bugs left
>>>to fix, but it mostly seems to be working OK now.  However, there is one thing
>>>I haven't been able to figure out:
>>>Is there an easy way to store the PV in MTD(f)?  Currently I just construct a PV
>>>from the hash table after each iteration.  This works, but is somewhat messy and
>>>inconvenient.  Is there a better way to do it?
>>Hi Tord,
>>I do as you do, and extract the PV from the hash table. Some time ago, a guy
>>called Fabien Letouzey appeared here with an alternative suggestion:
>>I've never tried it.
>Thanks.  If building the PV from the hash table is good enough for you, I
>think I'll keep doing it like that myself (with Tony Werten's suggested
>improvements), and take a closer look at Fabien Letouzey's suggestion
>when my mtd(f) is good enough to be usable.
>There is still a long way left before mtd(f) works as well as traditional
>aspiration search in my program, but I will continue trying for some time.
>mtd(f) is fun and addictive.  :-)

Yes. There's just so much scope for experimentation. For example, in Aske
Plaat's online description, he raises and lowers the guess by 1 each time around
the MTD loop. I've never been able to make this work effectively. Far better to
accelerate this. If the score is going up, I add 1 the first time then 2, then 3
etc etc. I think I stop at 7 or something. I can't remember now. The same
applies to when the score is sinking.

Have you read Aske Plaat's PhD thesis? It's available online somewhere and is
very interesting and readable.


This page took 0.03 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.