On polynomial approximations over Z/2kZ

Abstract

We study approximation of Boolean functions by low-degree polynomials over the ring Z/2kZ. More precisely, given a Boolean function F:\0,1\n → \0,1\, define its k-lift to be Fk:\0,1\n → \0,2k-1\ by Fk(x) = 2k-F(x) 2k. We consider the fractional agreement (which we refer to as γd,k(F)) of Fk with degree d polynomials from Z/2kZ[x1,…,xn]. Our results are the following: - Increasing k can help: We observe that as k increases, γd,k(F) cannot decrease. We give two kinds of examples where γd,k(F) actually increases. The first is an infinite family of functions F such that γ2d,2(F) - γ3d-1,1(F) ≥ (1). The second is an infinite family of functions F such that γd,1(F)≤12+o(1) -- as small as possible -- but γd,3(F) ≥ 12+(1). - Increasing k doesn't always help: Adapting a proof of Green [Comput. Complexity, 9(1):16-38, 2000], we show that irrespective of the value of k, the Majority function Majn satisfies γd,k(Majn) ≤ 12+O(d)n. In other words, polynomials over Z/2kZ for large k do not approximate the majority function any better than polynomials over Z/2Z. We observe that the model we study subsumes the model of non-classical polynomials in the sense that proving bounds in our model implies bounds on the agreement of non-classical polynomials with Boolean functions. In particular, our results answer questions raised by Bhowmick and Lovett [In Proc. 30th Computational Complexity Conf., pages 72-87, 2015] that ask whether non-classical polynomials approximate Boolean functions better than classical polynomials of the same degree.

0

Turn this paper into a lesson

ArcXiv compiles a structured reading guide from this paper's metadata: plain-English importance, contributions, prerequisite concepts, which sections to read first, flashcards, and a quiz. Grounded in the abstract, never invented.

Discussion (0)

Sign in to join the discussion.

Loading comments…