arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2003.05874cs.CC
收藏

有界个体度的四圈深度指数大小的新下界

New Exponential Size Lower Bounds against Depth Four Circuits of Bounded Individual Degree

Suryajith Chillara

摘要:

Kayal,Saha和Tavenas [计算理论,2018]表明,对于所有足够大的整数$n$和$d$,使得$d\geq ω(\log{n})$,任何计算迭代矩阵乘法多项式($IMM_{n,d}$)的有界个体次数$δ= o(d)$的语法深度为4的电路必须具有大小$n^{Ω\left(\sqrt{d/δ}\right)}$。不幸的是,随着$δ$值的增加,该界限恶化。此外,只有当$δ$是$o(d)$时,边界才是超多项式。人们自然会问,是否可以削弱绑定中对$δ$的依赖。为此,在早期的结果[STACS,2020]中,我们证明了对于所有足够大的整数$n$和$d$,使得$d = Θ(\log^2{n})$,任何计算$IMM_{n,d}$的有界个体度$δ\leq n^{0.2}$的语法深度为4的电路必须具有大小$n^{Ω(\log{n})}$。本文进一步证明了:对所有足够大的整数$n$和$d$,以及绝对常数$a$和$b$,使得$ω(\log^2n)\leq d\leq n^{a}$,任何计算$IMM_{n,d}$的句法深度为4的有界个体度$δ\leq n^{b}$的回路的大小必须为$n^{Ω(\sqrt{d})}$ .我们的界限是通过仔细调整Kumar和Saraf的证明[SIAM J. Computing,2017]来获得的,以适应我们早期工作中引入的复杂性度量[STACS,2020]。

英文摘要:

Kayal, Saha and Tavenas [Theory of Computing, 2018] showed that for all large enough integers $n$ and $d$ such that $d\geq ω(\log{n})$, any syntactic depth four circuit of bounded individual degree $δ= o(d)$ that computes the Iterated Matrix Multiplication polynomial ($IMM_{n,d}$) must have size $n^{Ω\left(\sqrt{d/δ}\right)}$. Unfortunately, this bound deteriorates as the value of $δ$ increases. Further, the bound is superpolynomial only when $δ$ is $o(d)$. It is natural to ask if the dependence on $δ$ in the bound could be weakened. Towards this, in an earlier result [STACS, 2020], we showed that for all large enough integers $n$ and $d$ such that $d = Θ(\log^2{n})$, any syntactic depth four circuit of bounded individual degree $δ\leq n^{0.2}$ that computes $IMM_{n,d}$ must have size $n^{Ω(\log{n})}$. In this paper, we make further progress by proving that for all large enough integers $n$ and $d$, and absolute constants $a$ and $b$ such that $ω(\log^2n)\leq d\leq n^{a}$, any syntactic depth four circuit of bounded individual degree $δ\leq n^{b}$ that computes $IMM_{n,d}$ must have size $n^{Ω(\sqrt{d})}$. Our bound is obtained by carefully adapting the proof of Kumar and Saraf [SIAM J. Computing, 2017] to the complexity measure introduced in our earlier work [STACS, 2020].

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑