Author: GuyHaworth
Date: 11:19:22 12/31/03
Go up one level in this thread
Maybe Eugene's e-address is to be found via one of his contributions to CCC. The paper was I suspect a Nalimov/H/Heinz collaboration based on a presentation by Ernst Heinz at ACG-9 (2000), and the result of a subsequent reworking by Eugene and myself. Certainly, EN's index to an EGT is a set of sub-indexes, one for each legal Kk-position. There are 462 anticipating no Ps and 1806 anticipating Ps. Eugene then prevents stm pieces giving an unblockable check to the sntm-King, though other checks will subsequently render the position 'broken' by virtue of being 'clearly illegal'. Thus, the subranges-size is KK-position-specific. There is no general algorithm for avoiding all positions with sntm in check, and therefore there is major 'waste' in the index in some endgames: this seems to increase with number of men on the board, so any impovement on this would be welcome. A downside of EN's index-function is that an inverse-index function, as would be needed for the Wu/Beal fully-retro EGT-generation algorithm, involves integer-division rather than 'division' by 64. This has performance implications for a method combining the Wu/Beal algorithm and Nalimov indexing. g
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.