On suprema of convolutions on discrete cubes
José Gaitan, José Madrid
TL;DR
The paper determines the exact optimal constant for the infinity norm of the k-fold convolution of functions on the discrete cube {0,1}^d, revealing a precise formula for C_{k,1} and its d-dimensional generalization. The authors couple a diagonal-case analysis with a dimension-compression argument, employing a Poisson Binomial distribution viewpoint to show that the extremal configuration occurs when all inputs are equal. This yields sharp bounds on Sidon-like sets on hypercubes and sheds light on the continuous analogue problem, with explicit equality cases and strong structural insights. The work advances convolution inequalities on hypercubes and connects them to additive combinatorics through tight extremal bounds.
Abstract
We find the optimal constant $C$ such that \begin{equation*} \|f_1*f_2*\dots*f_{k}\|_{\infty}\geq C\prod_{i=1}^{k}\|f_i\|_1 \end{equation*} for functions $f_i:\{0,1\}^d\to\mathbb{R}$. As applications, we derive bounds for Sidon sets on hypercubes, and, we also obtain bounds for the continuous analogue problem.
