Re: Question.

[Martin Mačok]

On Tue, Apr 15, 2003 at 01:43:50PM -0400, Sternbach, William [IT] wrote:

> Is there any possible improvements to the alpha-beta pruning
> algorithms which might increase its ability to look ahead further
> and do more while using the same amount of CPU-time.  I have read
> some books on alpha-beta pruning algorithms (a long time ago when
> I was at Syracuse University).

AFAIK most of the best chess programs now use PVS/Negascout similar to
current GNU Chess (and crafty for example). There are some papers
which promises that MTDf() is a bit faster but it is not widely used.
(You can find it in Potato Chess for example...) Bob Hyatt tried it
once in crafty but wasn't happy with the result and switched back to
PVS/N. Unfortunately, commercial vendors don't publish their cookbooks
so we don't know what search algorithms are used in Fritz, Hiarcs,
Shredder etc. (I would be interested in Shredder especially) but
I think they probably use PVS/N too...

You can try to implement some special prunning algorithms (you can
find some in crafty source) but proving that they actually doesn't
make the engine weaker is not an easy task.

I think that the most easier way to improve GNU Chess is to do low
level optimizations and/or implement better positional evaluation.
It is believed that the search depth is somewhat close to adequate
using nowadays computers but they still play too much "greedy".

> is there any plans to improve the gnuchess algorithms to potentially
> bring it to a "master" level?

It is pretty strong engine, if it is not already it's very close to
"master" level IMHO... (but crafty is still stronger ;-)

-- 
         Martin Mačok                 http://underground.cz/
   martin.macok@underground.cz        http://Xtrmntr.org/ORBman/