English

On Matrix Multiplication and Polynomial Identity Testing

Computational Complexity 2024-04-18 v1

Abstract

We show that lower bounds on the border rank of matrix multiplication can be used to non-trivially derandomize polynomial identity testing for small algebraic circuits. Letting R(n)\underline{R}(n) denote the border rank of n×n×nn \times n \times n matrix multiplication, we construct a hitting set generator with seed length O(nR1(s))O(\sqrt{n} \cdot \underline{R}^{-1}(s)) that hits nn-variate circuits of multiplicative complexity ss. If the matrix multiplication exponent ω\omega is not 2, our generator has seed length O(n1ε)O(n^{1 - \varepsilon}) and hits circuits of size O(n1+δ)O(n^{1 + \delta}) for sufficiently small ε,δ>0\varepsilon, \delta > 0. Surprisingly, the fact that R(n)n2\underline{R}(n) \ge n^2 already yields new, non-trivial hitting set generators for circuits of sublinear multiplicative complexity.

Keywords

Cite

@article{arxiv.2208.01078,
  title  = {On Matrix Multiplication and Polynomial Identity Testing},
  author = {Robert Andrews},
  journal= {arXiv preprint arXiv:2208.01078},
  year   = {2024}
}
R2 v1 2026-06-25T01:23:39.190Z