An IPv6 address consists of a pair of doubles. A IEEE double-precision complex number, if you will.
Show this thread
-
want to see the map of IPv6 addresses treated as pairs of IEEE doubles
The vast majority of IPv6 addresses in use start with 2001 or 2400 or the like, so you'd have just a few small clusters in the vast exponential space, since those bits set the exponent. And then the imaginary part would be uncorrelated junk (MACs and the like) or denormals.
