The Mean Square of Divisor Function

TL;DR

This paper improves the error term estimate for the sum of the squares of the divisor function, achieving a tighter bound without assuming the Riemann Hypothesis, advancing understanding of divisor sums.

Contribution

The paper proves a sharper bound for the error term in the divisor function sum without relying on unproven hypotheses, refining previous results.

Findings

Established that E(x)=O(x^{1/2}( ext{log} x)^5) for the divisor function sum.

Improved the unconditional error estimate compared to earlier works.

Enhanced the understanding of divisor sum behavior without assuming RH.

Abstract

Let $d(n)$ be the divisor function. In 1916, S. Ramanujan stated but without proof that $$\sum_{n\leq x}d^2(n)=xP(\log x)+E(x), $$ where $P(y)$ is a cubic polynomial in $y$ and $$ E(x)=O(x^{{3\over 5}+\epsilon}), $$ where $\epsilon$ is a sufficiently small positive constant. He also stated that, assuming the Riemann Hypothesis(RH), $$ E(x)=O(x^{{1\over 2}+\epsilon}). $$ In 1922, B. M. Wilson proved the above result unconditionally. The direct application of the RH would produce $$ E(x)=O(x^{1\over 2}(\log x)^5\log\log x). $$ In 2003, K. Ramachandra and A. Sankaranarayanan proved the above result without any assumption. In this paper, we shall prove $$ E(x)=O(x^{1\over 2}(\log x)^5). $$