Proving complexity lower bounds remains a challenging task: currently, we only know how to prove conditional uniform (algorithm) lower bounds and nonuniform (circuit) lower bounds in restricted circuit models. About a decade ago, Williams (STOC 2010) showed how to derive nonuniform lower bounds from uniform \emph{upper} bounds. Since then, a number of results of this kind have been proved. For example, Jahanjou et al. (ICALP 2015) and Carmosino et al. (ITCS 2016) proved that if NSETH fails, then $\text{E}^{\text{NP}}$ has circuit size $\omega(n)$. Just recently, it was shown how uniform \emph{lower} bounds can be used to derive nonuniform lower bounds: Belova et al. (SODA 2024) showed that if MAX-3-SAT cannot be solved in co-nondeterministic time $O(2^{(1 - \varepsilon)n})$, then there exists an explicit polynomial family of high arithmetic circuit size; Williams (FOCS 2024) showed that if NSETH is true, then Boolean Inner Product requires large $\text{ETHR} \circ \text{ETHR}$ circuits. In this paper, we continue developing this line of research and show that nondeterministic uniform lower bounds imply nonuniform lower bounds for various types of objects that are notoriously hard to analyze: circuits, matrix rigidity, and tensor rank. Specifically, we prove that if NSETH is true, then there exists a monotone Boolean function family in coNP of monotone circuit size $2^{\Omega(n / \log n)}$. Combining this with the result above, we get win-win circuit lower bounds: either $\text{E}^{\text{NP}}$ requires circuits of size $w(n)$ or coNP requires monotone circuits of size $2^{\Omega(n / \log n)}$. We also prove that MAX-3-SAT cannot be solved in co-nondeterministic time $O(2^{(1 - \varepsilon)n})$, then there exist small explicit families of high rigidity matrices and high rank tensors.
翻译:暂无翻译