On a problem of J. H. Fang and Z. K. Fang (Q2226635)

Let \(\mathbb{N}\) be the set of all nonnegative integers. For any sequence of positive integers \(A=\{a_1<a_2<a_3< \cdots\}\), let \(P(A)= \{\sum_{i=1}^\infty \varepsilon_ia_i \in \mathbb{N}: \varepsilon_i\in \{0,1\}\}\). What can one say about the growths of the sequence \(B=\{b_1<b_2<b_3< \cdots\}\), if there is an infinite sequence \(A\) such that \(P(A)=\mathbb{N}\setminus B\)? It has been known by \textit{Y. G. Chen} and \textit{J. H. Fang}, Acta Math. Hung. 134, No. 4, 416--430 (2012; Zbl 1254.11014)] that for some values of \(b_1\) and \(b_2=3b_1+4\) there are no such \(B\) and \(A\) sequences. \textit{J. H. Fang} and \textit{Z. K. Fang} [Acta Math. Hung. 158, No. 1, 36--39 (2019; Zbl 1438.11020)] proved that if there are such \(A\) and \(B\) sequences, and \(b_2=3b_1+5\), then \(b_3\geq 4b_1+6\). The paper under review proves that if there are such \(A\) and \(B\) sequences, and \(b_2\geq 3b_1+5\), then \(b_3\geq b_2+b_1+1\), and shows that such sequences indeed exist.