[Cryptography] Keyed affine transform

[Pierre Abbat]

On Friday, August 7, 2026 1:56:39 AM EDT Pierre Abbat wrote:
> I've written a preliminary version in https://github.com/phma/Kaftor.jl . I
> haven't yet run statistics or cryptanalysis on it; I've been tuning the
> multithreading.

I've graphed statistics, and they don't look good. I expected five-byte 
messages to be badly mixed, because byte 3 is affected by neither jumble nor 
mix3, but the last byte or two of sixteen-byte messages are also badly mixed, 
even though jumble affects all bytes of a sixteen-byte message (because 16+3 is 
prime) and shufflePairs affects every byte in every round (because 16 is a power 
of 2). I'm going to replace mix3, and probably also jumble, with something 
that mixes bits more thoroughly.

Pierre

-- 
Don't buy a French car in Holland. It may be a citroen.