English

Univariate Ideal Membership Parameterized by Rank, Degree, and Number of Generators

Data Structures and Algorithms 2018-09-24 v2

Abstract

Let F[X]\mathbb{F}[X] be the polynomial ring over the variables X={x1,x2,,xn}X=\{x_1,x_2, \ldots, x_n\}. An ideal I=p1(x1),,pn(xn)I=\langle p_1(x_1), \ldots, p_n(x_n)\rangle generated by univariate polynomials {pi(xi)}i=1n\{p_i(x_i)\}_{i=1}^n is a \emph{univariate ideal}. We study the ideal membership problem for the univariate ideals and show the following results. \item Let f(X)F[1,,r]f(X)\in\mathbb{F}[\ell_1, \ldots, \ell_r] be a (low rank) polynomial given by an arithmetic circuit where i:1ir\ell_i : 1\leq i\leq r are linear forms, and I=p1(x1),,pn(xn)I=\langle p_1(x_1), \ldots, p_n(x_n)\rangle be a univariate ideal. Given αFn\vec{\alpha}\in {\mathbb{F}}^n, the (unique) remainder f(X)(modI)f(X) \pmod I can be evaluated at α\vec{\alpha} in deterministic time dO(r)poly(n)d^{O(r)}\cdot poly(n), where d=max{deg(f),deg(p1),deg(pn)}d=\max\{\deg(f),\deg(p_1)\ldots,\deg(p_n)\}. This yields an nO(r)n^{O(r)} algorithm for minimum vertex cover in graphs with rank-rr adjacency matrices. It also yields an nO(r)n^{O(r)} algorithm for evaluating the permanent of a n×nn\times n matrix of rank rr, over any field F\mathbb{F}. Over Q\mathbb{Q}, an algorithm of similar run time for low rank permanent is due to Barvinok[Bar96] via a different technique. \item Let f(X)F[X]f(X)\in\mathbb{F}[X] be given by an arithmetic circuit of degree kk (kk treated as fixed parameter) and I=p1(x1),,pn(xn)I=\langle p_1(x_1), \ldots, p_n(x_n)\rangle. We show in the special case when I=x1e1,,xnenI=\langle x_1^{e_1}, \ldots, x_n^{e_n}\rangle, we obtain a randomized O(4.08k)O^*(4.08^k) algorithm that uses poly(n,k)poly(n,k) space. \item Given f(X)F[X]f(X)\in\mathbb{F}[X] by an arithmetic circuit and I=p1(x1),,pk(xk)I=\langle p_1(x_1), \ldots, p_k(x_k) \rangle, membership testing is W[1]W[1]-hard, parameterized by kk. The problem is MINI[1]MINI[1]-hard in the special case when I=x1e1,,xkekI=\langle x_1^{e_1}, \ldots, x_k^{e_k}\rangle.

Keywords

Cite

@article{arxiv.1808.10787,
  title  = {Univariate Ideal Membership Parameterized by Rank, Degree, and Number of Generators},
  author = {V. Arvind and Abhranil Chatterjee and Rajit Datta and Partha Mukhopadhyay},
  journal= {arXiv preprint arXiv:1808.10787},
  year   = {2018}
}