English

Generalized relations between arithmetic functions

Number Theory 2026-01-22 v1

Abstract

The aim of this article is to present in a self-contained way identities arising in elementary number theory, among which the following one: dnμ2(d)φ(d)ds=pn(1+1(p1)ps). \sum_{d\mid n}\frac{\mu^2(d)}{\varphi(d)\,d^s}=\prod_{p\mid n}\left(1+\frac{1}{(p-1)p^s}\right). This formula expresses a non-trivial divisor sum involving the M\"obius function μ\mu and Euler's totient function φ\varphi as a simple and explicit multiplicative expression. This is a generalization of the remarkable Dineva formula, which corresponds to s=0s=0 and gives n/φ(n)n/\varphi(n) on the right-hand side. We explain why only squarefree divisors are involved, show how multiplicativity naturally comes into play, and interpret the identity as a finite Euler product. Beyond this one-parameter family of generalizations, we describe a general method for constructing similar formulas and present several examples. Finally, we reformulate these identities in terms of partial zeta functions, thus emphasizing their close relationship with the classical theory of Euler products and the Riemann zeta function. The connection with the Selberg sieve is briefly outlined.

Keywords

Cite

@article{arxiv.2601.14521,
  title  = {Generalized relations between arithmetic functions},
  author = {Jean-Christophe Pain},
  journal= {arXiv preprint arXiv:2601.14521},
  year   = {2026}
}