On the Density of Some Sequences of Numbers

From MaRDI portal





Let \(f(m)\) be a non-negative arithmetical function satisfying \[ f(m_1m_2) = f(m_1)+f(m_2) \text{ if } (m_1,m_2)=1, \tag{1} \] \[ f(p_1) \neq f(p_2) \tag{2} \] for two different primes \(p_1,p_2;\) and let \[ N(f;c,d) = \sum _{\substack{{m\leq n} \\ {c\leq f(m)\leq d}}} 1, \qquad N(f;c) = N(f;c,\infty). \] The main result of this paper, which constitutes a wide generalization of the author's work on abundant numbers (Zbl 0010.10303), is that \(\lim_{n\to \infty} N (f;c)/n\) exists and is a continuous function of \(c\). The case of the abundant numbers is obtained by taking \(f(m) = \log {\sigma(m) \over m}\), \(c = \log 2\). Suppose first \(f(m)\) satisfies the more stringent conditions: \((3)\;f(p^\alpha) = f(p)\), \((4)\;\sum_p {f(p)\over p}\) converges. Defining \(f_p(m) = \sum_{{p|m}, {p\leq P}} f(p)\) it is easily seen that \(\lim_{n\to \infty} N(f_p;c)/n = A_p\) exists, and as \(A_p\) in non-decreasing and \(\leq 1\), \(\lim_{p \to \infty} A_p = A\) exists. That \(\lim_{n \to \infty} N(f;c)/n = A\) follows easily from the two lemmas: (I) For any \(\varepsilon >0\) there exists a \(\delta\) such that \(N(f;c,c+\delta) < \varepsilon n\) for all sufficiently large \(n\); (II) For any \(\varepsilon,\delta > 0\) there exists a \(P(\varepsilon,\delta)\) such that for \(P > P(\varepsilon,\delta)\) and all \(n\), the number of integers \(m\leq n\) for which \(f(m)-f_p(m) > \delta\) is less than \(\varepsilon n\). The main difficulty lies in the proof of (I), which uses the same idea as the paper already cited. The author then sketches the proof when (3) is not assumed. As regards (4), he shows that it can be replaced by a weaker condition \((4')\) and that if \((4')\) does not hold, then \(\lim_{n \to \infty} N(f;c)/n = 1\).











This page was built for publication: On the Density of Some Sequences of Numbers

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4761179)