Small gaps between almost primes, the parity problem, and some conjectures of Erdős on consecutive integers II. (Q2220453): Difference between revisions

The authors show that, for any integer \(n\geq 1\), there are infinitely many integers \(x\) such that \(x\) and \(x+n\) have the same fixed ``exponent pattern'', i.e., there exists a multiset \(\{k_1,\ldots, k_j\}\) depending on \(n\) such that infinitely often both \(x\) and \(x+n\) have a prime factorization of the form \(p_1^{k_1}\cdots p_j^{k_j}\) with \(p_i\) distinct primes. The authors deduce in particular that, for every \(n\geq 1\), there is \(A\geq 1\) such that \(d(x)=d(x+n)=A\) for infinitely many integers \(x\), where \(d(x)\) is the divisor function. Previously, four of the authors had shown in Part I [the fist author et al., Int. Math. Res. Not. 2011, No. 7, 1439--1450 (2011; Zbl 1241.11114)] that \(d(x)=d(x+1)=24\) infinitely often. The proofs in the current paper use the main theorem proved in [loc. cit.] on triples of linear forms attaining two semiprime values infinitely often (Theorem 1), and additionally an elementary argument for constructing admissible tuples such that the coefficients in the linear relations between them have certain desirable properties (stated precisely in Theorem 2).