Complete sequences in \(\mathbb{N}^ 2\) (Q2563527): Difference between revisions

It is shown that the classical results of \textit{J. Folkman} [Can. J. Math. 18, 643-655 (1966; Zbl 0151.03703)] on representation of integers as sums of distinct terms from a fixed sequence in \(\mathbb{N}\) cannot be extended without additional conditions to \(\mathbb{N}^2\). The first of the two ``negative'' results says that there exists an \(\alpha>0\) and \({\mathcal A}\subseteq\mathbb{N}^2\) for which \({\mathcal A}(n)>\alpha n^2\) and \(\mathcal A\) does not contain an infinite two dimensional arithmetic progression of the form \(\{\mathbf{x}_0+k\mathbf{d}\mid k\in\mathbb{N}\}\). In the second result it is shown that also an immediate analogue of a sufficient condition for the completeness in \(\mathbb{N}\), namely the bounded gaps together with containment of sufficiently large intervals does not imply the completeness in \(\mathbb{N}^2\). Two additional results are proved in the paper. The first one gives estimations from below and above of the maximal size of a discrete rectangle in the sum set of the initial segment of a sequence with gaps only of the form \((1,0)\) and \((0,1)\), and in the second one a sufficient condition for the so called \(p\)--completeness.