Degree Reduction
- Goal: compute a multiplication gate on shares
- Setup:
- $a$ is t-shared via $A(x)$ , $b$ is t-shared via $B(x)$
- Want each party to hold a t-share of $ab$
- Naïve idea: compute $A(\alpha_i)B(\alpha_i)$
- Problem 1: product is degree $2t$
- Problem 2: polynomial not random (factorable into $A,B$ ; a fresh random sharing shouldn’t reveal such structure)
- Fix: degree reduction
- The degree- $2t$ product polynomial $C(x)$ is uniquely determined if $n>2t$
Interpolate:
\[C(x) = \sum_{i=1}^n C(\alpha_i)\,\ell_i(x),\quad \ell_i(x)=\prod_{j\ne i}\frac{x-\alpha_j}{\alpha_i-\alpha_j}\]
But! We only need the secret:
\[C(0) = \sum_{i=1}^n \lambda_i\,C(\alpha_i),\quad \lambda_i=\ell_i(0)\]where $\lambda_i$ is a constant.
This becomes a linear relation, and we know how to deal with linear relations.- Trick (reshare + recombine):
- Each party $i$ (who knows $C(\alpha_i)$ ) t-shares $C(\alpha_i)$ with fresh randomness
- Everyone linearly combines these t-shares using the public weights $\lambda_i$
- Result: a fresh t-sharing of (C(0)=ab)
- Intuition recap:
- Multiplication → degree blow-up
- Interpolation focuses only on the constant term
- Identify that the problem becomes a linear relation
- Resharing restores randomness
- Weighted recombination yields a valid t-share of the product
This post is licensed under CC BY 4.0 by the author.