Author: Tony Werten
Date: 08:36:35 08/01/01
Go up one level in this thread
On August 01, 2001 at 11:27:29, Erik Evenson wrote: >Excellent points -- especially the emphasis on generating the tablebases via a >retrograde approach. I haven't built the code to generate moves in reverse, but >I believe that is a minor mod to my forward move generation code, which is done >using bitboards (rotated and otherwise). > >This should also speed the analysis since mates in 2 become mates in 1 after >flagging all the mates in 1. > >Very interesting...now I just need to determine how to store the resulting move >table efficiently... I am starting to understand why the Nalimov tables are set >up the way they are. They are very efficient. One main thing is the placing of the kings. 1 king is confined to a corner (a1,d1,d4) which are 10 positions. Now place the second king on the board but only in legal places. This reduces the number of possible positions from 10x64 to 462. (or so) If there are pawns you can put the first king on the left side (in a1,d1,d8,a8) With the 2nd king this gives 1806 positions (or so) cheers, Tony > >If only I didn't have a real job to take me away from this... :)
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.