Re: advice sought on key/data histogram analysis of rijndael/128 and serpent

"lkcl" <lkcl@lkcl.net> wrote in message
news:1129995498.823045.73550@g49g2000cwa.googlegroups.com...

> what is the probability that, if the average count is supposed
> to be 5250 in a set of 512000x128x128 tests, that the
> input i=0 and j=0 produces a count of only 5177????

Your math looks to be a little off there. You're missing an enormously
important number there, how big is each sample? This is important because
the numbers above don't make any sense, they are self-referential, but the
numbers don't add up. In 512000 tests you should get 5120 failures on
average, 5177 is actually over, not under as you indicate, 5250 would
require 525000 tests, regardless that is within the experimental error of
code that has already been verified to be buggy at least once. That the
statistical code written by me reflects that as the number of blocks
increases any such correlation disappears reflects that this is almost
certainly not a flaw in Rijndael (in non-quantum space extra information
cannot reduce knowledge). Your graphs reflect only that you have the exact
same statistical anomalies that are required for random behavior. If you can
prove differently it would be of interest, but your tests so far do not show
the correlation you are trying to draw.
                Joe
Received on Mon Oct 24 02:08:05 2005