3 ms·
I think the parent's point is that it's not `f(x) = x + g(x)`, but `f(x) = x + g(c)` for some c. Addition is invertible if one side is constant, no matter how t
by tgv 7y ago
I think the parent's point is that it's not `f(x) = x + g(x)`, but `f(x) = x + g(c)` for some c. Addition is invertible if one side is constant, no matter how that constant is determined: g can be bijective or constant, it doesn't matter for the invertibility of the addition.
- thaumasiotes 7y agoYes, the inclusion of the Feistel function makes the raw xor (a substitution cipher) into a Vigenère cipher instead, where the encryption key differs between locations in the text instead of being constant throughout. But Vigenère ciphers aren't really an improvement on substitution ciphers. (And they're not more non-invertible than simple substitution ciphers.) Most of the work is being done by the transformation that swaps two halves of the text between rounds.