Interpolating classical partitions of the set of positive integers (Q2024959)

If \(\alpha\), \(\beta\) are irrational numbers such that \(\alpha^{-1}+\beta^{-1}=1\), then the Beatty theorem [Am. Math. Monthly 33, 159 (1926; JFM 52.0189.01)] states that the sets \(A=\{\lfloor k\alpha\rfloor\}_{k\geq1}\) and \(B=\{\lfloor k\beta\rfloor\}_{k\geq1}\) partition the positive integers. A relevant Beatty partition is given by \(\alpha=\phi\) and \(\beta=\phi^2\), where \(\phi\) is the golden ratio. The Beatty theorem is the main motivation of the present work. The authors define a class of partitions \(D_1,D_2,\ldots,D_n\) of positive integers into disjoint sets as follows. Let \(n\geq1\), \[ G=\left\{2^{n-1},2^{n-1}+2^{n-2},\ldots,\sum_{i=1}^n2^{n-i}\right\}, \] and let \(l(k)\) be a non-decreasing integer-valued sequence such that \(l(1)=2^{n-1}\) and \(l(k+1)-l(k)\in G\) for any \(k\geq1\). Then the authors prove that for \(D_1=\{l(k)\}_{k\geq1}\), the sets \begin{align*} &D_1,\quad D_2=\{k-2^{n-2}: k\in D_1\}\cup\{k+2^{n-2}: k\in D_1\},\\ &D_3=\{k-2^{n-3}: k\in D_2\}\cup\{k+2^{n-3}: k\in D_2\},\\ &\ldots,\quad D_n=\{k-1: k\in D_{n-1}\}\cup\{k+1: k\in D_{n-1}\} \end{align*} are disjoint and partition the set of positive integers. As a corollary, if \(l(k)=(2^{n-1}-1)h(k)+k\), where \(h(k)\) is an integer-valued sequence such that \(h(1)=1\) and \(h(k)-h(k-1)\leq2\), then the aforementioned condition is satisfied, and we obtain a partition of positive integers. Moreover, the authors also consider the interactions of Beatty partitions and the 3-set partitions. Let \(A=\{\lfloor k\phi\rfloor\}_{k\geq1}\) and \(B=\{\lfloor k\phi^2\rfloor\}_{k\geq1}\). Let \(D_1=\{d_{k1}\}_{k\geq1}=\{l(k)\}_{k\geq1}\), \(D_2=\{d_{k2}\}_{k\geq1}=\{D_1-2\}\cup\{D_1+2\}\), \(D_3=\{d_{k 3}\}_{k\geq1}=\{D_2-1\}\cup\{D_2+1\}\) be the 3-set partition created with \(l(k)={\left(2^{3-1}-1\right)}\lfloor k\phi\rfloor+k\). We classify the row \((d_{k1},d_{k2},d_{k3})\) as \((A,A,A)\) if \(d_{k1}\), \(d_{k2}\), \(d_{k3}\in A\); as \((A,B,A)\) if \(d_{k2}\in B\), \(d_{k1}\), \(d_{k3}\in A\) and so fourth. There are 8 such classifications: \((A,A,A)\), \((B,A,A)\), \((A,B,A)\), \((B,B,A)\), \((A,A,B)\), \((B,A,B)\), \((A,B,B)\), \((B,B,B)\). However, the authors prove that indeed just \(6\) out of \(8\) are possible. The classifications \((A,B,B)\) and \((B,B,B)\) never occur.