Uniform Titchmarsh divisor problems (Q2054251)

The original Titchmarsh divisor problem asks for the evaluation of \[ \sum_{a<p\le X} \tau(p-a), \] where \(\tau(n)\) is the number of divisors of \(n\) function and \(a\) is a positive integer. The main term \[ X\frac{\phi(a)}{a}\prod_{p\nmid a} \left(1+\frac{1}{p(p-1)}\right) \] was found by Titchmarsh already, and the best error term known is \(O_{a,A}(x(\log x)^{-A})\) valid for all \(A>0\) and is an application of the Bombieri-Vinogradov theorem. In the paper under review the authors address the uniformity in \(a\). That is their aim is to replace the above main term by a more complicated sum of terms such as the error depends only on \(A\) but not on \(a\). Their main result is Theorem 1.1. In particular, if \(X\ge 2a\), then \[ \sum_{a\le n\le X} \Lambda(n)\tau(n-a) = c_0(a)((X-a)\log(X-a)+(2\gamma-1)(X-a)) \]\[ + c_0'(a)(X-a)+O_A(X(\log X)^{-A}), \] for all \(A>0\), where \[ c_s(a):=\prod_{p\mid a} \left(1-\frac{1}{p^{s+1}}\right)\prod_{p\nmid a} \left(1+\frac{1}{(p-1)p^{s+1}}\right). \] Their method works for other divisor like functions such as \(r_2(n)\) which is the number of representation of \(n\) as a sum of two squares. Theorem 1.2 in particular gives that the number of solutions of the equation \(n=p+x^2+y^2\) in primes \(p\) and non-zero integers \(x,y\) is \[ \pi Li(n)\prod_{p}\left(1+\frac{\chi_{-4}(p)}{p(p-1)}\right)\prod_{p\mid n} \left(\frac{(p-1)(p-\chi_{-4}(p)}{p^2-p+\chi_{-4}(p)}\right)+O_A(n(\log n)^{-A}) \] for all \(A>0\), where \(\chi_{-4}\) is the only nonprincipal Dirichlet character modulo \(4\). The proofs use bounds for exponential sums based on the Riemann hypothesis for varieties over finite fields, spectral theory of automorphic forms and representation theoretic methods for local considerations of Fourier coefficients.