An improved bound on $\ell_q$ norms of noisy functions

TL;DR

This paper improves bounds on the $\, ext{ell}_q$ norms of noisy boolean functions for integer $q \,\geq 2$, leading to tighter error correction thresholds for Reed-Muller codes over binary symmetric channels.

Contribution

It provides a sharper inequality for the $\, ext{ell}_q$ norms of the noise operator on boolean functions for integer $q \,\geq 2$, with applications to coding theory.

Findings

Tighter bounds on $\, ext{ell}_q$ norms for integer $q \,\geq 2$.

Improved error correction threshold for Reed-Muller codes.

The inequality is tight for characteristic functions of subcubes.

Abstract

Let $T_{\epsilon}$, $0 \le \epsilon \le 1/2$, be the noise operator acting on functions on the boolean cube $\{0,1\}^n$. Let $f$ be a nonnegative function on $\{0,1\}^n$ and let $q \ge 1$. In arXiv:1809.09696 the $\ell_q$ norm of $T_{\epsilon} f$ was upperbounded by the average $\ell_q$ norm of conditional expectations of $f$, given sets whose elements are chosen at random with probability $\lambda$, depending on $q$ and on $\epsilon$. In this note we prove this inequality for integer $q \ge 2$ with a better (smaller) parameter $\lambda$. The new inequality is tight for characteristic functions of subcubes. As an application, following arXiv:2008.07236, we show that a Reed-Muller code $C$ of rate $R$ decodes errors on $\mathrm{BSC}(p)$ with high probability if \[ R ~<~ 1 - \log_2\left(1 + \sqrt{4p(1-p)}\right). \] This is a (minor) improvement on the estimate in arXiv:2008.07236.