Author: Reinhard Scharnagl
Date: 11:46:06 03/05/05
Go up one level in this thread
On March 05, 2005 at 12:00:13, Chris Welty wrote: >r3k3/8/8/8/8/8/8/4K3 w q - 0 1 > >and > >r2k4/8/8/8/8/8/8/4K3 w - - 0 1 > >get encoded to the same number in your scheme, so while it is compact it is also >a lossy compression scheme. No. First parts A are different e.g. because of different castling rights. Second the rook of first example does not occur in the filling matrix, whereas it does in the second example. >Also, you're using your "estimated average" length of 152.5 instead of your >upper bound of 172 bits so it's not really an upper bound calculation. > >>Average saving of color Bits = 2; >>Average saving of piece code Bits: = 6; > >>filling matrix redundances: - 1.1 Bits >>e.p. bit redundances: - 1.0 Bits > >what are these? a) filling matrix redundances: the filling matrix selects at maximum 30 of 62 possibilities. This is equivalent to about 60.8 Bits. b) e.p. bit redundances: because the number of positions where specifying e.p. to be possible will make sense is very marginal compared to those where the content of that byte has to be zero, so that Bit has nearly no relevance. Reinhard.
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.