Computer Chess Club Archives




Subject: Re: Couple of chess programming questions

Author: Jon Dart

Date: 06:59:03 09/10/02

Go up one level in this thread

On September 10, 2002 at 09:26:14, Eli Liang wrote:

>A couple of chess programming questions:

Well, it's more than a couple, and many of these are more like research problems
than questions.

>(3) Reading Aske Plaat's search & re-search paper, it really seems like mtd(f)
>is something of a magic bullet.  But I note it seems that more programs don't
>use it than do (for example Crafty).  What is wrong with mtd(f) which Plaat
>doesn't say?

Aske Platt's paper suggests that considerable experimentation is required to
make it work efficiently, and that existing programs may require changes in
other areas than just the search in order to work well with this algorithm. But
there are people with more experience in this area who can answer this question
better than I.

>(4) What are the current thoughts concerning bitboards/rotated-bitboards versus
>conventional 0x88 or other algorithms?  Just looking at open source chess
>programs, it doesn't seem that the chess programming community has come to a
>consensus on relative performance... or have you?

It isn't readily possible to say which approach is better, because you can't
easily convert a program that uses bitboards to one that has a different board
representation, plus each choice makes some things more efficient and some less
efficient. The result is that an "other things being equal" comparison isn't
really possible, because you can't hold other things equal while changing
something as fundamental as board representation. It appears, however, that you
can build a high-performance chess engine with or without bitboards.

>(5) What is currently thought to be the best algorithm for autofitting the
>static evaluation function parameters? DT's least-squares approach seems
>simplest, but it seems as if a dozen other things have been tried too.  Is there
>any best-of-breed approach here?
>(6) Has anyone found any real "practical" benefits to fractional-ply extensions?

These are widely used and I believe do have benefits. The ability to use
fractional extensions allows you to extend in important cases without thereby
blowing the search tree up too much. I have done extensive testing with
different extension algorithms and extension values, and have never found fixed
1-ply depth extensions to be better overall. But your mileage may vary.

This page took 0.12 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.