I'm curious about the homomorphic properties of Paillier. So, basically if I have $\textsf{Dec}(\textsf{sk}, \textsf{Enc}(\textsf{pk}, \alpha) \cdot \textsf{Enc}(\textsf{pk}, \alpha^{-1}))$, I will get as result $\alpha + \alpha^{-1}$. But, does it also mean that if I have $\textsf{Dec}(\textsf{sk}, \textsf{Enc}(\textsf{pk}, \alpha)^{\textsf{Enc}(\textsf{pk}, \alpha^{-1})})$, then the result will be $\alpha \cdot \alpha^{-1}$, which will basically cancel each other, and will be left with 1?
2 Answers
No, there is no reason that $\textsf{Dec}(\textsf{sk}, \textsf{Enc}(\textsf{pk},\alpha)^{\textsf{Enc}(\textsf{pk}, \alpha^{-1})})$ would be $\alpha\cdot\alpha^{-1}$, including when we spread $\bmod N$ or $\bmod N^2$ here and there.
What does apply is: for overwhelmingly most $\alpha$ and $k$ in $\Bbb Z$, it holds that $\textsf{Dec}(\textsf{sk},\textsf{Enc}(\textsf{pk}, \alpha)^k\bmod N^2)=k\cdot\alpha\bmod N$. We could take $k=\alpha^{-1}\bmod N$ and get $\textsf{Dec}(\textsf{sk},\textsf{Enc}(\textsf{pk},\alpha)^{(\alpha^{-1}\bmod N)}\bmod N^2)=1$, but that's not useful anyway, since $\alpha^{-1}\bmod N$ reveals $\alpha\bmod N$.
Criticism of the question:
- It is not defined in which group it is computed $\alpha^{-1}$, and that matters.
- $\textsf{Dec}(\textsf{sk}, \textsf{Enc}(\textsf{pk}, \alpha) \cdot \textsf{Enc}(\textsf{pk}, \alpha^{-1}))$ will be $\alpha+\alpha^{-1}\bmod N$, which may or may not be $\alpha+\alpha^{-1}$.
Given two plaintexts $\alpha$ and $\beta$, Pailler cryptosystem $\mathcal{E}$ homomotphic property is: $\mathcal{E}(\alpha)\times \mathcal{E}(\beta)=\mathcal{E}(\alpha+\beta)$. So, $\mathcal{E}(\alpha)^n=\mathcal{E}(n\alpha)$. In your example, $n=\mathcal{E}(\alpha^{-1})$ and thus after decryption you will have $\mathcal{E}(\alpha^{-1})\times \alpha$ ans not $\alpha^{-1} \times \alpha$. This is a high level answer, but you need to define the modulo $N$ of your system.