Author: KarinsDad
Date: 10:25:18 05/19/99
Go up one level in this thread
On May 19, 1999 at 02:40:34, Dave Gomboc wrote: [snip] > >What a waste of bits! :) The entire game should have been storable in far >fewer. To know exactly how many, take the number of legal moves from each >position, multiple them all together to get a big number, P. Find n such that P >/ 2^n < 0. n = the number of bits you need to store the raw positions (no >1/2-1/2, player names, etc.) Dave, This is a cool schema. For a position that has 40 legal moves, you just store which of the 40 is the current move. This gives you a new position where you then pick one move out of how many is legal in that position. So, you average a little under 5.333 bits per move (about 400 bits for this game) as opposed to the standard 12 bits per move (source and destination squares) * 76 moves = 912 bits to store this game. The division on 400 bit numbers (or even 1500 bit numbers on super long games) to calculate it out is probably a little rough. May want to make it 1 to 7 bits per move and allow the current position determine how many possible moves there are and hence how many bits in the structure are used to determine the next move (this makes the structure a little larger, but it makes the calculation much faster and easier). I like it! Thanks for showing it to me. KarinsDad :) > >I didn't do it for this game, but n will likely be in the low hundreds. You are >using several thousand bits to represent the game. > >Dave
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.