On the order of magnitude of the divisor function (Q2508551)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On the order of magnitude of the divisor function |
scientific article |
Statements
On the order of magnitude of the divisor function (English)
0 references
13 October 2006
0 references
For \(\mathbb{D}\) an increasing sequence of positive integers, let \[ d(n,\mathbb{D})= \sum_{\substack{ d|n\\ d\in\mathbb{D},d\leq\sqrt{n}}} 1,\quad d_2(n,\mathbb{D})= \sum_{\substack{ [d,\delta]|n\\ d,\delta\in\mathbb{D},[d, \delta]\leq \sqrt{n}}} 1, \] where \([d,\delta]\) denotes the least common multiple of \(d\) and \(\delta\). Let \({\mathbf a}\) be a sequence \(a_1, a_2\dots\) of positive reals such that \(\sup_{n\geq 1}\,\sup_{m\geq cn}\,{a_m\over a_n}= \gamma_{{\mathbf a}}(c)<\infty\) for any real \(c> 0\). In Theorem 1.1 it is shown that \(\sum^\infty_{n=1} a_nd(n,\mathbb{D})\) converges if \(\sum_{d\in\mathbb{D}}{1\over d} \sum_{n\geq d^2} a_n\) converges. Theorem 1.2 establishes the analogous result that \(\sum^\infty_{n=1} a_n d_2(n,\mathbb{D})\) converges if \[ \sum_{d, \delta\in\mathbb{D}} {1\over[d,\delta]} \sum_{n\geq [d,\delta]^2} a_n \] converges. The proofs use probalistic arguments. These results are illustrated by considering some well known examples as well as an example for which standard arguments do not apply. In particular the author deduces that for \(\beta> 2\) both the infinite series \[ \sum d(n)n^{-1}(\log n)^{-\beta},\tag{i} \] \[ \sum \omega(n) n^{-1}(\log n)^{-1}(\log_2 n)^{-\beta}\tag{ii} \] converge, where \(d(n)= \sum_{d|n} 1\), \(\omega(n)= \sum_{p|n} 1\). [Note: the author starts the sums (i) and (ii) and other similar sums at \(n= 1\) where the denominator is zero, so the first term should be modified.] As a non-standard illustration he lets \(\mathbb{P}_1\) be such that \(\sum_{p_1\in\mathbb{P}_1} (\log p_1)^{1-\beta}\) converges for some \(\beta> 1\), and puts \(\mathbb{D}(\mathbb{P}_1,\mathbb{P})= \{d= p_1+ p: p> p_1, p_1\in\mathbb{P}_1, p\in\mathbb{P}\}\) where \(\mathbb{P}\) denotes the set of primes; then by Theorem 1.1 \(\sum_{n>1} {d(n,\mathbb{D}(\mathbb{P}_1, \mathbb{P}))\over n(\log n)^\beta}\) converges. From Theorem 1.2, by taking \(\mathbb{D}= \mathbb{P}\), \(a_n= n^{-1}(\log n)^{-1}(\log_2 n)^{-\alpha}\) and noting that \(d_2(n,\mathbb{P})\leq\omega^2(n)\leq d_2(n,\mathbb{P})+ 5\omega(n)\), the author deduces that \(\sum_{n>1} {\omega^2(n)\over n\log n(\log_2 n)^\alpha}\) converges for \(\alpha> 2\).
0 references
divisor function
0 references
prime divisors
0 references
Bernoulli random walk
0 references