Computer Chess Club Archives


Search

Terms

Messages

Subject: Question about Tablebases, Namilov's Paper

Author: Ed Trice

Date: 10:26:34 12/31/03


I remember reading a paper by Eugene Namilov a while back. He compared his
compression results with that of Ken Thompson, et alia, demonstrating his scheme
was more efficient.

I recall the creating of a 2D array for storing only the King-King legal
positions as a baseline, in order to eliminate as many illegal/impossible
positions as possible.

My specialty is checkers endgame database construction, but something I learned
from my experience at building some large databases (over 132 billion positions
with "win-loss-draw" information compressed for runtime lookup, and over 21
billion Distance-To-Mate {can announce a win in 253 plies}) could apply to the
chess world very nicely.

I am wondering if anyone knows if positions with illegal checks have been
removed from the endgame databases?

If not, I figured out a way to store only legal positions for chess endgame
databases, while still alowing for a general indexing function (not complicated
at all) and some of the other superfluous positon removal (all pieces on the
long diagonal, etc.)

I am wondering if someone has Eugene's email address?



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.