Author: David Rasmussen
Date: 04:25:06 12/04/01
Go up one level in this thread
On December 04, 2001 at 03:51:28, Tony Werten wrote: > >Having good 64 bit keys doesn't mean you can chop them in half and have 2 good >32 bit keys. > >easy example: >1111 and 0011 are different in 50% of all bits, wich is quite good. Take half of >the bits and you end up with 11 and 11 wich is "not quite as good". ( although >the other half (00 and 11) scores 100%) > >Tony Good 64-bit _minimum_ hamming distance doesn't imply good 32-bit _minimum_ hamming distance, but good 64-bit _average_ hamming distance does statistically imply good 32-bit _average_ hamming distance. Especially when you have "many" keys. I have verified that my 32-bit values are quite good. If you use a decent PRNG, you _will_ get keys with good 32-bit properties as well as good 64-bit properties. Otherwise, the numbers aren't very random, but contain structure and redundancy. /David
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.