Computer Chess Club Archives




Subject: Re: MTD, IID, fail-low, root-research

Author: Russell Reagan

Date: 16:07:25 08/14/03

Go up one level in this thread

On August 14, 2003 at 17:17:19, Juergen Wolf wrote:

>>I do not think that IID is important.
>i read threads were exactly the opposite is stated. as far as i seen these
>are from chessprograms not using MTD and stating MTD doesn't help

Yes, but you have to be careful who is doing the posting. For instance, I recall
that Vincent has posted about how MTD(f) isn't an improvement. While Vincent is
a very smart guy and knows lots of stuff, he also says some things that are
questionable from time to time.

I'd take the word of someone like Rudolf Huber, who has successfully used MTD(f)
to create a strong program. I am sure MTD(f) is like many things in computer
chess. You have to struggle with it for some time to get it to work well. I get
the impression from Vincent that the period of struggle isn't sufficient
sometimes, which is why he doesn't think bitboards or MTD(f) are worth anything.
It may be the case that neither are drastic improvements over other approaches,
but it is clear that both can be used to create strong programs.

My experience in general is that you can post questions and gather the
experiences of others, and discuss things for weeks and months and end up with
more questions than when you started. If you really want answers reading source
code and writing source code is the only sure way to get answers, and then post
questions when you run into difficulties.

This page took 0.02 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.