Author: Robert Hyatt
Date: 20:04:21 12/30/03
Go up one level in this thread
On December 30, 2003 at 22:27:24, Ed Trice wrote: >Hello again Dr. Hyatt, > >> >>That is the very reason that I make the source public. And yes, 80 bit words >>won't be a serious problem.. Remember that the _original_ 64 bit program, >>chess 4.0, ran on a 60 bit computer, so it had to use two words to get 64 >>anyway. You will just use 3 on 32 bit machines, or 2 on 64 bit machines. >> > >Thank you very much! I must say, I am *very* impressed with the concept of the >static exchange evaluator. It is like a tree-search without all of the overhead >calling of the move generator! Idea came from Dennis Cooper and Ed Kozdrowicki of the old COKO program. I knew them pretty well back in the early 1970's and Ed gave me the idea and the original SEE code I started using back then... The idea has stuck around for 30+ years now as Coko used this in 1970 when I met the two authors at the first ACM event. > > >> >>Sounds like a fun project... Keep us posted here... > >The first step will be using C++ operator overloading for the bitwise routines >for the SHIFT, AND, OR functions, etc. > >As for the evaluation function for the values of the new pieces, I have already >come up with a derivation based on Taylor's work (1876). > >You can see the detailed explanation of it here: > >http://www.GothicChess.org/piece_values.html > >Taylor derived equations for "safely checking a king" to roughly come up with >values for the pieces. On a 10x8 board, all of the pieces lose strength. The >Knight loses the most (-16.67%) and the Rook the least (-5.50%) I could see why. Knight is already stretched pretty thin on 8x8 in some endgames. > >Thanks again for allowing me to use Crafty as a baseline. > >When I have a test version ready, I will let you know. > >--Ed
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.