Integers represented as a sum of primes and powers of two. (Q1811406)

In 1953 Yu. V. Linnik proved that there exists an absolute constant \(K\) such that the equation \(p_1 + p_2 + 2^{\nu_1} + \dots + 2^{\nu_K} = N\) has solutions in primes \(p_1, p_2\) and integers \(\nu_1, \dots, \nu_K\) for all sufficiently large even \(N\). While the value of \(K\) resulting from Linnik's method is computable, the corresponding calculation is quite technical, and thus no explicit numerical bound on \(K\) was known until the late 1990s. The first explicit bounds were given by J. Y. Liu, M. C. Liu, and T. Z. Wang in a series of papers, in which they showed that \(K \leq 54 000\) and that this bound can be reduced to \(K \leq 200\) under the assumption of the Generalized Riemann Hypothesis (GRH). Subsequently, the unconditional bound on \(K\) was reduced several times to \(K \leq 1906\) [see \textit{H. Z. Li}, Acta Arith. 96, 369--379 (2001; Zbl 0973.11088)]. In the paper under review, the authors obtain a significant improvement on earlier results. They show that \(K \leq 13\) and that this bound can be reduced to \(K \leq 7\) under GRH. As in earlier work on the subject, the proofs employ the Hardy--Littlewood circle method and rely on estimates for the exponential sums \[ S(\alpha) = \sum_{\delta N < p \leq N} \exp( 2\pi i\alpha p), \qquad T(\alpha) = \sum_{2^{\nu} \leq N} \exp( 2\pi i\alpha 2^{\nu}). \] In particular, estimates for the measure of the set where \(T(\alpha)\) is large play a very important role. New estimates for the measure of that set represent the first main novelty in the paper under review. Another important innovation appears in the application of the circle method. Using a very delicate analysis of the contribution of the major arcs, the authors demonstrate that hypothetical Siegel zeros of Dirichlet \(L\)-functions do not influence the major arcs in this problem. Thus, no recourse is needed to explicit numerical zero-free regions. \{Reviewer's remark: The same conditional result was obtained independently by \textit{J. Pintz} and \textit{I. Ruzsa} [Acta Arith. 109, 169--194 (2003; Zbl 1031.11060)]. Those authors also announced that they had an unconditional proof of the bound \(K \leq 8\), but to the reviewer's best knowledge, the details of that proof have not yet appeared.\}