On the generalised Dirichlet divisor problem

Chiara Bellotti,Andrew Yang

2023-04-06

Abstract:We improve unconditional estimates on $\Delta_k(x)$, the remainder term of the generalised divisor function, for large $k$. In particular, we show that $\Delta_k(x) \ll x^{1 - 1.889k^{-2/3}}$ for all sufficiently large fixed $k$.

Number Theory

What problem does this paper attempt to address?

This paper attempts to solve the problem of estimating the remainder \(\Delta_k(x)\) in the Generalised Dirichlet Divisor Problem, especially for larger values of \(k\). Specifically, the goal of the paper is to improve the unconditional estimate of \(\Delta_k(x)\), that is, the bound of the remainder of the general divisor function.

### Problem Background

For an integer \(k\geq2\), define \(d_k(n)\) as the number of ways to write \(n\) as a product of exactly \(k\) factors. The partial sum \(\sum_{n\leq x}d_k(n)\) satisfies the asymptotic formula:

\[
\sum_{n\leq x}d_k(n)=xP_{k - 1}(\log x)+\Delta_k(x)
\]

where \(P_{k-1}(t)\) is a polynomial of degree \(k - 1\), and \(\Delta_k(x)\) is the remainder. The Generalised Dirichlet Divisor Problem focuses on the order of \(\Delta_k(x)\) as \(x\rightarrow\infty\). The best current conjecture is:

\[
\Delta_k(x)\ll_{\varepsilon}x^{1/2-1/(2k)+\varepsilon}
\]

This result implies the famous Lindelöf Hypothesis.

### Main Contributions of the Paper

1. **Improving the Karatsuba Constant**: By using Heath - Brown's new estimation method and combining it with an improved \(B\) - constant, the author obtains a new Karatsuba constant \(D\approx1.224\).

2. **Proposing a New Estimation Method**: The author develops a method that directly uses exponential sum estimates instead of relying on the bounds of the Riemann zeta function. Combining the improved exponential sum estimates, they obtain a Karatsuba constant \(D\approx1.889\) which is applicable for sufficiently large \(k\).

### Main Results

Theorem 1.1 gives an estimate for \(k\geq30\):

\[
\Delta_k(x)\ll x^{1-1.224(k - 8.37)^{-2/3}}
\]

In addition, for \(k\geq15\), an estimate of the \(L^2\) norm is also given:

\[
\|\Delta_k(x)\|_2\ll x^{1-1.421(k - 4.18)^{-2/3}}
\]

Overall, this paper significantly improves the estimation accuracy of \(\Delta_k(x)\) by improving existing methods and introducing new techniques, especially when \(k\) is large.