Score:0

How to calculate inverse of an element using MPC and secret sharing

dm flag

Suppose we have multiple parties ${P_i}$ each holding a secret share $(y_i,k_i)$ of the secret $(y,k)$. How can we calculate the inverse value $(y+k)^{-1}$ using multiparty computation such that these secret shares are never revealed?

Score:3
sb flag

First, to simplify notation, note that computing $(y+k)^{-1}$ is equivalent to computing $z^{-1}$, where $z=y+k$, which can be locally computed by the parties. Furthermore, I assume that $z \in \mathbb{F}_p$ is a field element (e.g., integers mod $p$).

Now, to compute $z^{-1} \in \mathbb{F}_p$, each party can (locally) sample a random share of a value $r \in \mathbb{F}_p \setminus \{0\}$. Then, the parties reveal $zr$ to compute $(zr)^{-1} \in \mathbb{F}_p$, in the clear. Finally, the parties use their shares of $r$ to engage in a MPC protocol for multiplication to compute $(zr)^{-1}\cdot r = z^{-1} \in \mathbb{F}_p$. A similar protocol was described in Algorithm for computing modular inverse in MPC under a different context.

integrator avatar
cn flag
You seem to have made quite a few assumptions about the secret sharing scheme
Sacha Servan-Schreiber avatar
sb flag
My answer would probably have worked better as a comment. The main assumption is that @KanchanBisht is using an additive secret sharing scheme over a field, which appears reasonable based on how the question was formulated.
integrator avatar
cn flag
To clarify, I think your answer is good. I was just (clumsily) pointing out it would be better to explicitly mention you were assuming additivity (especially since you swept under the rug where that property is used: "which can be locally computed by the parties").
I sit in a Tesla and translated this thread with Ai:

mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.