Finite and infinite arithmetic progressions in sumsets (Q2498169)

The main result of the paper says that there exists a positive constant \(c\) such that for any positive integer \(n\) if \(A\) is a subset of \(\{1,2,\dots,n\}\) having at least \(cn^{1/2}\) elements, then the collection of its finite subset sums \(S_A=\left\{\sum_{x\in B}x : B\subset A, | A| <\infty\right\}\) contains an arithmetic progression of length \(n\). The authors introduce a new method to prove this result relying on inverse and geometrical arguments. They believe that their new method can be used to handle also other problems. As an application it is shown that there is a positive constant \(c\) such that any increasing sequence of density at least \(cn^{1/2}\) is subcomplete (i.e. \(S_A\) contains an infinite arithmetic progression). This result immediately implies via Folkman's partition argument a 40-year old Erdős-Folkman conjecture on complete sequences [\textit{P. Erdős}, Acta Arith. 7, 345--354 (1962; Zbl 0106.03805)].