Null Move heuristic
MANOHARARAJAH VALAVAN
According to an article in Artificial intelligence by Prof. Don Beal, the
improvements made to the program by incoporating the Null Move heuristics
was astounding....There was a huge saving in time mentioned.
Furthermore is it worthwhile incoporating it into a program with considerably
less chess knowledge (compared to other programs of course)?
--
-------------------------------------------------------------------------------
man...@ecf.utoronto.ca | 2nd year Comp Eng., University of Toronto
Valavan Manohararajah | OS/2 Warp "Operate at a higher level"
-----------------------------Team OS/2 Member----------------------------------
Peter Gillgasch
In article <DCEM6...@ecf.toronto.edu>, man...@ecf.toronto.edu (MANOHARARAJAH VALAVAN) writes:
|> How many of the chess programs these days use Null Move heuristics?
The question would be easier to answer if you had asked how many programs are
*not* using it... Nearly any microcomputer program uses it. The exceptions I
am aware of are all programs by Richard Lang and the Rebel program by Ed
Schroeder. On the "big iron" the only two programs I am aware of that do not
use it are Deep Blue Prototype and Zugzwang.
|> According to an article in Artificial intelligence by Prof. Don Beal, the
|> improvements made to the program by incoporating the Null Move heuristics
|> was astounding....There was a huge saving in time mentioned.
Correct. But to make sure we are talking about the same things... Don Beal's
article was about a generalized quiesence search. The term "null move search"
today is reserved for programs that do a *search* after the null move. This
was suggested in Don Beals article as an possibility.
The other articles on the subject I am aware of are a paper by Goetsch and
Campbell (reprinted in "Computers, Chess and Cognition") and Christian
Donninger's "Null move and deep search: Selective searching for obtuse
chess programs" in the ICCA Journal.
All of these papers are pretty valuable to the practioneer.
|> Furthermore is it worthwhile incoporating it into a program with considerably
|> less chess knowledge (compared to other programs of course)?
Yes it is. Searching deeper (faster) is especially important
when your program is a bit knowledge deficient. The null move
can even be used to make the program more knowledge deficient
(because the null move search can be used for threat detection)
to cut the execution time of the evaluation function down, making
the program faster. If you add null move searches to a knowledge
deficient program it is a self fullfiling prophecy: The program
was dumb, so I added null moves to increase the search depth -
then I removed evaluation terms that were no longer needed to
make it even more dumb. Strange thing is that it gets better
overall (^8
-- Peter
------------------------------------------------------------
Peter W. Gillgasch
Klosterweg 28/I-113 email gil...@ira.uka.de
76131 Karlsruhe/Germany Phone ++49/(0)721/6904255