On irregularities in the graph of generalized divisor functions

From MaRDI portal





For the divisor function \(d(n)\), \textit{P. Erdős} [Mat. Lapok 11, 26--32 (1960; Zbl 0100.27201)] proved that for each \(k\in\mathbb{N}\) there are infinitely many \(n\in\mathbb{N}\) satisfying \[ d(n)> \prod^k_{i=1} d(n- i)d(n+ i).\tag{1} \] This author generalizes this result to the function \(d(A,n)= \sum_{a\in A,a| n} 1\), where \(A\in\mathbb{N}\). The proof depends on a result from sieve theory. In the other direction, a straightforward necessary and sufficient condition on \(A\) is established in order that \[ d(A, n)< \min(d(A, n- 1), d(A,n+1))\tag{2} \] for infinitely many \(n\in\mathbb{N}\); in particular the conditions hold for \(A= \mathbb{N}\). The example \(A = 3\mathbb{N}\) shows that there are sets \(A\) for which (2) never holds. The author ends with a series of corollaries. One of these answers a problem posed by A. Sárközy that, for every infinite set \(A\), the sequence \(| d(A,n +1) -d(A,n)|\) is unbounded. In other corollaries, the main theorems are applied with \(A\) the set of primes or the set of prime powers to obtain results analogous to (1), and of the type (2) in the special case \(A= \mathbb{N}\), for the functions \(v(n)\) (or \(\omega(n)\)) and \(\Omega(n)\).











This page was built for publication: On irregularities in the graph of generalized divisor functions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4429652)