BP, in an n-plyiterative-deepening process, does some forward pruning even at ply 1. There are good, interesting, and bad moves. Good ones appear to win material. Interesting ones exchange material, or are the best move in a previous iteration. Bad moves appear to lose material. There is also consideration given to tactical conditions, such as whether there is a piece under attack or if the machine is retreating from check. At ply 1 during early iterations all possible moves are examined. At ply 1 and 2 the good and interesting moves are fully explored, but the bad moves are pruned at n-2. Near the bottom of the tree some "serious pruning" is enacted. BP has a 65K position table and would like to make this larger for endgames. It runs on a 33 MHz 486 processor. The iterative deepening goes in steps: 2,4, ... n-2, n. When they announce a 6-ply analysis, it is a pseudo-6 ply of full width. They examine approximately 1600 nodes/sec. Part of the BP philosophy is expressed: "If we prune on a even ply, and omit a good move, it is a shame. If we prune on an odd ply and omit a good move, it is a disaster."
Table of Contents
BP, (Mulder BP)
a chess program by Robert Cullum. BP was a selective program written in C and x86 assembly language to run on a Compaq 386 IBM PC or compatible [1].
Descriptions
WCCC 1989
WCCC 1989, from the booklet Kings Move - Welcome to the 1989 AGT World Computer Chess Championship [2]ACM 1991
given by Garth Courtois Jr., who had the opportunity to talk with some programmers during the ACM 1991 [3]:Tournament Play
BP played five ACM North American Computer Chess Championships, the ACM 1987, ACM 1988, ACM 1989, ACM 1991 and ACM 1993, as well the WCCC 1989 in Edmonton, Alberta, Canada [4] operated by Kevin O’Connell.Selected Games
WCCC 1989, round 2, Mulder BP - Pandix [5]External Links
References
What links here?
Up one level