I wrote:

In GF(2^n), inversion behaves very much like a change between conjugate observables

and made a vague comparison to a Fourier transform; after a little more thought, I guess it's more like convolution, since it's the sum of products going in opposite directions. So (x+h)^-1 = x * h = sum over even k ( x^{N-k} h^k ) where N=2^n-2. I guess I need to figure out whether this definition of convolution gives anything useful... -- Mike Stay staym@clear.net.nz http://www.xaim.com/staym