Primes in arithmetic progressions to large moduli II: well-factorable estimates (Q7014530)

Let \(\phi\) denotes the Euler function, \(\pi(x)\) be the number of primes less than \(x\), and \(\pi(x;q,a)\) be the number of primes less than \(x\) congruent to \(a\) modulo \(q\). The Bombieri-Vinogradov Theorem states that for every \(A>0\) and \(B=B(A)\) sufficiently large in terms of \(A\) we have\N\[\N\sum_{q\le x^{1/2}/(\log{x})^B}\,\sup_{(a,q)=1}\Bigl|\pi(x;q,a)-\frac{\pi(x)}{\phi(q)}\Bigr|\ll_A\frac{x}{(\log{x})^A}.\N\]\NAn essential problem in analytic number theory is to widen the range for \(q\) in the above theorem, which is far from reach at this time. But, it is often sufficient to measure the difference between \(\pi(x;q,a)\) and \(\pi(x)/\phi(q)\) with well-factorable weights, which naturally appear in sieve problems, rather than absolute values. Based on this technical weakening, in their remarkable investigation in the theory of ``primes in arithmetic progressions to large moduli'', \textit{E. Bombieri} et al. [Acta Math. 156, 203--251 (1986; Zbl 0588.10042); Math. Ann. 277, 361--393 (1987; Zbl 0625.10036); J. Am. Math. Soc. 2, No. 2, 215--223 (1989; Zbl 0674.10036)] produced estimates analogous to Bombieri-Vinogradov Theorem, involving moduli up to \(x^{4/7-\varepsilon}\) with \(\varepsilon>0\).\N\NIn the present paper, which is the second part of his three impressive works, carrying same title, and remembering the above mentioned three works of Bombieri-Friedlander-Iwaniec, the author improves on this direction up to \(x^{3/5-\varepsilon}\). The initial idea is to generalize the notion of well-factorable weights. Letting \(Q\in\mathbb{R}\), we say a sequence \(\lambda_q\) is triply well factorable of level \(Q\) if for any choice of factorization \(Q=Q_1Q_2Q_3\) with \(Q_1,Q_2,Q_3\ge 1\), there exist three sequences \(\gamma^{(1)}_{q_1},\gamma^{(2)}_{q_2},\gamma^{(3)}_{q_3}\) the conditions\N\begin{itemize}\N\item \(|\gamma^{(1)}_{q_1}|,|\gamma^{(2)}_{q_2}|,|\gamma^{(3)}_{q_3}|\le 1\) for all \(q_1,q_2,q_3\),\N\N\item \(\gamma^{(i)}_{q_i}\) is supported on \(1\le q_i\le Q_i\) for \(i\in\{1,2,3\}\)\N\N\item We have\N\[\N\lambda_q =\sum_{q=q_1q_2q_3}\gamma^{(1)}_{q_1}\gamma^{(2)}_{q_2}\gamma^{(3)}_{q_3}.\N\]\N\N\end{itemize}\NAccordingly, the main result of the present paper states that letting \(a\in\mathbb{Z}\), \(A,\varepsilon>0\) and \(\lambda_q\) to be triply well factorable of level \(Q\le x^{3/5-\varepsilon}\), the following approximation holds\N\[\N\sum_{\substack{q\le Q\\\N(a,q)=1}}\lambda_q\Bigl(\pi(x;q,a)-\frac{\pi(x)}{\phi(q)}\Bigr)\ll_{a,A,\varepsilon} \frac{x}{(\log{x})^A}.\N\]\NThe author also proves that if \(\lambda^+_d\) is well-factorable upper bound sieve weights for the linear sieve of level \(D\le x^{7/12-\varepsilon}\), then\N\[\N\sum_{\substack{q\le x^{7/12-\varepsilon}\\\N(q,a)=1} }\lambda_q^+\Bigl(\pi(x;q,a)-\frac{\pi(x)}{\phi(q)}\Bigr)\ll_{a,A,\varepsilon} \frac{x}{(\log{x})^A}.\N\]\NThis estimate has capacity to improve several results based on sieves, more precisely when it is combined with techniques such as Chen's switching principle or Harman's sieve.\N\NThe proof of the main result is based on a generalization of the above mentioned works of Bombieri-Friedlander-Iwaniec, and several related results including approximations for exponential sums based on ideas from automorphic forms due to \textit{J.-M. Deshouillers} and \textit{H. Iwaniec} [Invent. Math. 70, 219--288 (1982; Zbl 0502.10021)].\N\NFor Part I see [the author, Primes in arithmetic progressions to large moduli I: Fixed residue classes. Providence, RI: American Mathematical Society (AMS) (2025; Zbl 08009107)].