English

Polynomial Threshold Functions for Decision Lists

Computational Complexity 2022-12-22 v2

Abstract

For S{0,1}nS \subseteq \{0,1\}^n a Boolean function f ⁣:S{1,1}f \colon S \to \{-1,1\} is a polynomial threshold function (PTF) of degree dd and weight WW if there is a polynomial pp with integer coefficients of degree dd and with sum of absolute coefficients WW such that f(x)=sign(p(x))f(x) = \text{sign}(p(x)) for all xSx \in S. We study a representation of decision lists as PTFs over Boolean cubes {0,1}n\{0,1\}^n and over Hamming balls {0,1}kn\{0,1\}^{n}_{\leq k}. As our first result, we show that for all d=O((nlogn)1/3)d = O\left( \left( \frac{n}{\log n}\right)^{1/3}\right) any decision list over {0,1}n\{0,1\}^n can be represented by a PTF of degree dd and weight 2O(n/d2)2^{O(n/d^2)}. This improves the result by Klivans and Servedio [Klivans, Servedio, 2006] by a log2d\log^2 d factor in the exponent of the weight. Our bound is tight for all d=O((nlogn)1/3)d = O\left( \left( \frac{n}{\log n}\right)^{1/3}\right) due to the matching lower bound by Beigel [Beigel, 1994]. For decision lists over a Hamming ball {0,1}kn\{0,1\}^n_{\leq k} we show that the upper bound on weight above can be drastically improved to nO(k)n^{O(\sqrt{k})} for d=Θ(k)d = \Theta(\sqrt{k}). We also show that similar improvement is not possible for smaller degrees by proving the lower bound W=2Ω(n/d2)W = 2^{\Omega(n/d^2)} for all d=O(k)d = O(\sqrt{k}). \end{abstract}

Keywords

Cite

@article{arxiv.2207.09371,
  title  = {Polynomial Threshold Functions for Decision Lists},
  author = {Vladimir Podolskii and Nikolay V. Proskurin},
  journal= {arXiv preprint arXiv:2207.09371},
  year   = {2022}
}

Comments

14 pages in total (11 for article + 3 for references and appendix)

R2 v1 2026-06-25T01:03:20.488Z