The divisor function over integers with a missing digit

Jiseong Kim
2024-11-14
Abstract:In this paper, we study the sum of the divisor function over sets with digit restrictions. As a corollary, we prove a non-trivial bound for the version over short intervals of length $X^{3/4+\varepsilon}.$
Number Theory
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the sum of the divisor function on the set of integers with specific numerical limitations. Specifically, the paper studies the behavior of the sum of the squares of the divisor function on the set of natural numbers \(X^*\) when certain numbers are excluded. These natural numbers \(n\) do not contain a specific digit \(b\) in their representation in base \(g\). ### Main contributions of the paper: 1. **Research background**: - Consider the set \(X^*\), where \(X^*\) is the set of natural numbers in the range \(X < n\leqslant2X\) and whose \(g\)-ary representation does not contain the digit \(b\). - Since the elements in \(X^*\) lack a multiplicative structure, the traditional methods for summing arithmetic functions are not applicable. 2. **Main results**: - By introducing the Fourier transform and other analytical tools, the author proves a non - trivial upper bound for the sum of the squares of the divisor function in a short interval of length \(X^{3/4+\epsilon}\). - Specifically, the main theorem (Theorem 1.1) of the paper gives an accurate asymptotic formula that describes the main term and the error term of the sum of the squares of the divisor function on \(X^*\). 3. **Technical means**: - Use the circle method to deal with the problem of summing the divisor function. - Utilize the sieve method of Maynard et al. and the results of Fourier analysis to obtain strong upper bounds for \(\sum_{n\in X^*}e(\alpha n)\) and its moments. - Through these upper bounds, further study the problem of summing the divisor function in short intervals. ### Specific problems of the paper: - **Problem definition**: Study the behavior of the sum of the squares of the divisor function \(\sum_{n\in X^*}d_2(n)\) on \(X^*\). - **Objective**: Prove the asymptotic formula for this sum as \(X\to\infty\) and give an upper bound for the error term. ### Main conclusions: - **Theorem 1.1**: Under certain conditions, gives the asymptotic formula for \(\sum_{n\in X^*}d_2(n)\) and proves that the upper bound of the error term is \(O_\epsilon(X^{3/4 + 2\eta+\epsilon})\). - **Corollary 1.3**: Under more general conditions, gives the upper bound for \(\sum_{n\in X^*}d_2(n)\). - **Corollary 1.4**: Considers the problem of summing the divisor function in short intervals and proves a similar upper bound. ### Mathematical formulas: - The size of \(X^*\) is approximately \(X\frac{\log(g - 1)}{\log g}\). - The formula in the main theorem: \[ \sum_{n\in X^*}d_2(n)=\sum_{q\leqslant P}\int_{|\beta|<\frac{1}{qQ}}\int_X^{2X}p_2^*(x)e(\beta x)\sum_{n\in X^*}c_q(n)e(-n\beta)\,dx\,d\beta+O_\epsilon(X^{3/4 + 2\eta+\epsilon}) \] where, \[ p_2^*(x)=\frac{1}{q}(\log x + 2\gamma- 1-2\log q) \] \[ c_q(n)=\sum_{1\leqslant a\leqslant q,(a,q)=1}e\left(\frac{an}{q}\right) \] Through these results, the paper is for the study of the set of integers with numerical limitations.