Long Arithmetic Progressions in Sumsets and Subset Sums: Constructive Proofs and Efficient Witnesses

Abstract

Existence of long arithmetic progression in sumsets and subset sums has been studied extensively in the field of additive combinatorics. These additive combinatorics results play a central role in the recent progress of fundamental problems in theoretical computer science including Knapsack and Subset Sum. The non-constructiveness of relevant additive combinatorics results affects their application in algorithms. In particular, several additive combinatorics-based algorithms for Subset Sum work only for the decision version of the problem, but not for the search version. We provide constructive proofs for finite addition theorems [S\'ark\H{o}zy'89 '94], which are fundamental results in additive combinatorics concerning the existence of long arithmetic progression in sumsets and subset sums. Our constructive proofs yield a near-linear time algorithm that returns an arithmetic progression explicitly, and moreover, for each term in the arithmetic progression, it also returns its representation as the sum of elements in the base set. As an application, we obtain an $\tilde{O}(n)$-time algorithm for the search version of dense subset sum now. Another application of our result is Unbounded Subset Sum, where each input integer can be used an infinite number of times. A classic result on the Frobenius problem [Erd\H{o}s and Graham '72] implies that for all $t \geq 2a^2_{\max}/n$, the decision version can be solved trivially in linear time. It remains unknown whether the search version can be solved in the same time. Our result implies that for all $t \geq ca^2_{\max}/n$ for some constant $c$, a solution for Unbounded Subset Sum can be obtained in $O(n \log a_{\max})$ time.