Computer Chess Club Archives




Subject: Re: MTD(f) and Wikipedia

Author: Alessandro Scotti

Date: 15:04:31 01/19/06

Go up one level in this thread

Hi Joachim,
from a pragmatic point of view, all of the strongest engines are known to use
PVS (which IIRC is not completely identical to Negascout) although many authors
did experiment with MTD(f) when Plaat paper came out. I'm one that fell for it
BTW, and Kiwi still uses MTD(f).

As for practical issues:
1) the overhead of re-searches is not so small as it appears in the academic
papers, because there is a lot of extra stuff going on in a "real" engine;
2) many of the useful search "tricks" and heuristics depends on the existence of
the alpha and beta bounds, and it is not so easy to adapt them for MTD(f);
3) even with an "accelerated" driver there are situations with a big jump in the
score that cause *a lot* of researches, each getting just a little bit closer to
the score... this often proves to be overkill in practice.

I think any engine that uses MTD(f) and is "serious" about it had to find a way
around at least the above issues.

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.