A strong direct product theorem in terms of the smooth rectangle bound

TL;DR

This paper proves a strong direct product theorem in two-way public-coin communication complexity using the smooth rectangle bound, showing exponential decay in success probability when resources are insufficient, applicable to many key functions.

Contribution

It establishes a strong direct product theorem in terms of the smooth rectangle bound for all relations in two-way public-coin communication complexity.

Findings

Proves the theorem for relations with optimal lower bounds via the smooth rectangle bound.

Implicates near optimal results for functions separating classical and quantum communication complexity.

Unifies multiple lower bound methods under the smooth rectangle bound framework.

Abstract

A strong direct product theorem states that, in order to solve k instances of a problem, if we provide less than k times the resource required to compute one instance, then the probability of overall success is exponentially small in k. In this paper, we consider the model of two-way public-coin communication complexity and show a strong direct product theorem for all relations in terms of the smooth rectangle bound, introduced by Jain and Klauck as a generic lower bound method in this model. Our result therefore uniformly implies a strong direct product theorem for all relations for which an (asymptotically) optimal lower bound can be provided using the smooth rectangle bound, for example Inner Product, Greater-Than, Set-Disjointness, Gap-Hamming Distance etc. Our result also implies near optimal direct product results for several important functions and relations used to show exponential separations between classical and quantum communication complexity, for which near optimal lower bounds are provided using the rectangle bound, for example by Raz [1999], Gavinsky [2008] and Klartag and Regev [2011]. In fact we are not aware of any relation for which it is known that the smooth rectangle bound does not provide an optimal lower bound. This lower bound subsumes many of the other lower bound methods, for example the rectangle bound (a.k.a the corruption bound), the smooth discrepancy bound (a.k.a the \gamma_2 bound) which in turn subsumes the discrepancy bound, the subdistribution bound and the conditional min-entropy bound. We show our result using information theoretic arguments. A key tool we use is a sampling protocol due to Braverman [2012], in fact a modification of it used by Kerenidis, Laplante, Lerays, Roland and Xiao [2012].