Nair-Tenenbaum bounds uniform with respect to the discriminant

Kevin Henriot
DOI: https://doi.org/10.1017/S0305004111000752
2011-02-09
Abstract:A common problem in analytic number theory is to bound the sum of an arithmetic function over a set of integers. Nair and Tenenbaum found a very general bound that applies to short sums of a multivariable arithmetic function over polynomial values, under certain standard conditions on the growth of that function. Their bound features an implicit dependency on the discriminant of the relevant polynomial. In our paper we obtain an analogous bound with an explicit dependency on the discriminant, which is optimal in the discriminant aspect.
Number Theory
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to obtain results that are uniform with respect to the discriminant \(D\) in Nair and Tenenbaum's upper bound estimates for short sums of polynomials. Specifically, the paper aims to improve the bounds proposed by Daniel [3] to make them more precise in terms of the dependence on the discriminant \(D\). ### Background and Motivation Nair and Tenenbaum [10] obtained an upper bound estimate in their work, but the implicit constant in this estimate depends on the discriminant \(D\) of the polynomials \(Q_1,\ldots,Q_k\). This dependence is undesirable in some number - theoretic problems. For example, when dealing with the Quantum Unique Ergodicity problem, it is necessary to maintain a uniform estimate with respect to the discriminant \(D\). Holowinsky [5] needed such a uniform estimate in his research to deal with shifted convolution sums, that is: \[ \sum_{n\leq x}\lambda_f(n)\lambda_f(n + \ell) \] where \(\lambda_f(n)\) is the normalized Hecke eigenvalue of the Hecke eigenform \(f\). To ensure the uniformity of the estimate with respect to \(\ell\), it is also necessary to maintain uniformity with respect to the discriminant \(D\). ### Main Contributions of the Paper The author of this paper, Kevin Henriot, improved the results of Nair and Tenenbaum by introducing a new factor \(\Delta_D\) and made the bound estimates uniform with respect to the discriminant \(D\). Specifically, the author proved the following theorem: **Theorem 3.** Under the assumptions of Theorem 1, further assume that \(F\) is multiplicative and \(\varepsilon\leq\frac{\alpha}{50g(g+\frac{1}{\delta})}\), then for \(x\geq c_0\|Q\|^\delta\) and \(x^\alpha < y\leq x\), we have: \[ \sum_{x < n\leq x + y}F(|Q_1(n)|,\ldots,|Q_k(n)|)\ll\Delta_D y\prod_{p\leq x}\left(1-\frac{\rho(p)}{p}\right)\sum_{n_1\cdots n_k\leq x}\frac{(n_1\cdots n_k,D) = 1}{n_1\cdots n_k}F(n_1,\ldots,n_k)\rho_{Q_1}(n_1)\cdots\rho_{Q_k}(n_k) \] where, \[ \Delta_D=\prod_{p|D}\left(1+\sum_{\nu_j\leq\deg(Q_h)}\frac{F(p^{\nu_1},\ldots,p^{\nu_k})\#\{n\bmod p^{\max_j(\nu_j)+1}:p^{\nu_j}||Q_j(n)\forall j\}}{p^{\max_j(\nu_j)+1}}\right) \] The implicit constant and \(c_0\) depend at most on \(g,\alpha,\delta,A,B\). ### Summary The main purpose of this paper is to improve the existing bound estimates by introducing a new factor \(\Delta_D\) to make them uniform with respect to the discriminant \(D\). This not only improves previous work but also provides more precise estimates in some important number - theoretic problems.