A generalization of a theorem of Nagell

TL;DR

This paper generalizes Nagell's theorem by proving that certain reciprocal power sums involving arbitrary positive integers and linear sequences are never integers for n ≥ 2, using analytic and p-adic methods.

Contribution

It extends Nagell's theorem to more general reciprocal power sums with arbitrary exponents and linear sequences, providing a broader non-integer result.

Findings

Reciprocal power sums with arbitrary exponents are never integers for n ≥ 2.

The proof employs analytic and p-adic techniques.

Generalizes previous theorems by Nagell and Erdős-Niven.

Abstract

Let $n$ be a positive integer. In 1915, Theisinger proved that if $n\ge 2$, then the $n$-th harmonic sum $\sum_{k=1}^n\frac{1}{k}$ is not an integer. Let $a$ and $b$ be positive integers. In 1923, Nagell extended Theisinger's theorem by showing that the reciprocal sum $\sum_{k=1}^{n}\frac{1}{a+(k-1)b}$ is not an integer if $n\ge 2$. In 1946, Erd\H{o}s and Niven proved a theorem of a similar nature that states that there is only a finite number of integers $n$ for which one or more of the elementary symmetric functions of $1,1/2, ..., 1/n$ is an integer. In this paper, we present a generalization of Nagell's theorem. In fact, we show that for arbitrary $n$ positive integers $s_1, ..., s_n$ (not necessarily distinct and not necessarily monotonic), the following reciprocal power sum $$\sum\limits_{k=1}^{n}\frac{1}{(a+(k-1)b)^{s_{k}}}$$ is never an integer if $n\ge 2$. The proof of our result is analytic and $p$-adic in character.